ultraviolet mutagenesis andinducible dna repair escherichia …ultraviolet mutagenesis andinducible...

BAcERIOLOGICAL REVIEWS, Dec. 1976, p. 869-907Copyright © 1976 American Society for Microbiology

Vol. 40, No. 4Printed in U.S.A.

Ultraviolet Mutagenesis and Inducible DNA Repair inEscherichia coliEVELYN M. WITKIN

Department ofBiological Sciences, Douglass College, Rutgers University, New Brunswick, New Jersey 08903

INTRODUCTION .............................................................. 869Enzymatic Repair of UV Damage in E. coli ......... .......................... 869Error-Proof and Error-Prone Pathways of DNA Repair ....... ................. 871The "SOS" Hypothesis: the Regulatory Role of DNA Damage ...... ............ 874

EVIDENCE FOR THE INDUCIBILITY OF ERROR-PRONE REPAIR ("SOSREPAIR") ACTIVITY ..................................................... 875

SOS Repair of Bacteriophage DNA ............ ............................... 875SOS Repair of Bacterial DNA ................ ................................ 876Evidence from studies of repair-deficient mutants ....... .................... 876Evidence from studies of postreplication repair ........ ...................... 879

MANIFESTATIONS OF SOS REPAIR AND THEIR SIGNIFICANCE ..... ...... 879Mutation Frequency Response to Increasing Fluence of UV Radiation ..... ..... 879The UV Lesion Responsible for Induction of SOS Functions.................... 880Kinetics of Induction and Decay of SOS Repair Activity ....... ................ 882Time of Action of SOS Repair in UV Mutagenesis ...... ........................ 882Cryptic Premutational Lesions Susceptible to SOS Repair ...................... 884Mutator Effect of SOS Repair on Undamaged DNA ....... ..................... 885

PLASMIDS AND SOS REPAIR ................................................ 886MECHANISM OF SOS REPAIR ............................................... 886Mechanism of SOS Repair in Bacteriophage ......... .......................... 886Mechanism of SOS Repair in Bacteria ........... ............................. 887

IMPLICATIONS OF SOS REPAIR FOR CARCINOGENESIS ................... 888REGULATION OF SOS REPAIR AND OTHER UV-INDUCIBLE FUNCTIONS . 889PROTEASES AND THE EXPRESSION OF SOS FUNCTIONS .................. 894CONCLUSIONS ............................................................... 895APPENDIX................................................................... 896LITERATURE CITED.......................................................... 898

INTRODUCTIONMutagens change the genetic script by pro-

moting errors in replication or in repair of de-oxyribonucleic acid (DNA). Ultraviolet (UV)light owes its powerful mutagenic effect inEscherichia coli primarily to misrepair of ra-diation damage to the bacterial chromosome(26, 125, 230). Recent developments indicatethat the error-prone repair activity responsiblefor UV mutagenesis in E. coli and some of itsphages is repressed in undamaged wild-typecells, but is expressed, as one of a metabolicallydiverse but coordinately regulated group of in-ducible functions, in response to UV radiationor other agents that damage DNA or interruptits replication (53, 77, 174, 240). This articlesummarizes current evidence for the inducibil-ity ofmutagenic DNA repair and explores someof its implications. Other aspects of UV muta-genesis in microorganisms have recently beenreviewed (3, 62, 66, 67, 130).

Enzymatic Repair of UV Damage in E. coliEnzymatic repair mechanisms operating in

bacteria which minimize the damaging effects

of UV radiation are described in detail else-where (88, 91, 93, 102). The brief summarygiven here is intended only to provide sufficientbackground for discussion of UV mutagenesis.UV radiation produces a variety of photo-

products in DNA (169, 197, 202), among whichthe intrastrand cyclobutane-type dimer of adja-cent pyrimidines (8) has been identified as amajor cause of lethal, mutagenic and tumoro-genic effects in a wide spectrum of organisms(92, 95, 216, 225). Wild-type strains of E. colieffectively neutralize the potentially lethal ef-fects of as many as a thousand pyrimidine di-mers by utilizing one or more of three types ofenzymatic DNA repair, which are representedin Fig. 1. Photoreactivation (120) is accom-plished by "photoreactivating enzyme" (PRE),which binds pyrimidine dimers in the dark butrequires "photoreactivating light" (PRL) in therange of wavelengths mainly between 310 and400 nm to "split" or monomerize them in situ(183, 184). Since PRE acts specifically on pyrim-idine dimers (193), photoreversibility by thisenzyme indicates that a UV-initiated biologicalphenomenon requires the more or less persist-ent presence of at least one pyrimidine dimer.

869

on October 8, 2020 by guest

http://mm

br.asm.org/

Dow

nloaded from

http://mmbr.asm.org/

870 WITKIN

A B C

PHOTOREACTIVATION

UV pyrimidineA climer

photoreativotingenzyme binds dimer

I

photoreactivating light

I_D _ dimer

_ split

enzymer released

rv

ntactDNA

replication(dimers split)

EXCISION REPAIR

("short patch" pathway)UV

A'A n

incision bycorrendonucleose II

5- -3:3 5

excision and repairreplication by DNApolymerose I

3' 5

sealing by polynu-cleotide ligase

I

repai redDNAlIA

replication(dimers removed)

POSTREPLICATION REPAIR

(recombinational)UV

A ~~~~~~~~~~3'.

replication

5' : 3'3 V(discontinuous daughter strands)

recombinational exchangesby recombination enzymes

5 3:3'- V 5

repair replicationand ligase sealing

5, A

(continuous daughter strands)

replication(dimers still present)

FIG. 1. Major pathways ofenzymatic repair ofdamage to DNA caused by UV radiation in E. coli. All threeof these repair mechanisms are error-proof in principle. Symbols: -A... or -v, pyrimidine dimer or othernoninstructive UV photoproduct; heavy lines, daughter strands produced in first postirradiation DNAreplication; light lines, parental (UV-irradiated) strands of DNA; wavy lines, DNA polymerized duringrepair replication. Adapted from Hanawalt (91). See text for additional references.

In the dark, pyrimidine dimers are not chem-ically reversed, but may be removed from theDNA by a multienzymatic mechanism, excisionrepair (20, 195). The distortion of the doublehelix caused by a pyrimidine dimer (or by someother kinds of DNA lesions) is recognized bycorrendonuclease II, the product of the uvrAand uvrB genes of E. coli (22), which makes asingle-strand nick, or incision, on the 5' side ofthe dimer. An exonucleolytic excision releasesthe oligonucleotide bearing the pyrimidine di-mer plus some bases on either side of it, and the"excision gap" is patched by repair replication(170), the intact region of the strand oppositethe gap serving as template for accurate re-placement of the missing nucleotides. In wild-type E. coli, the excision and resynthesis stepsare probably effected simultaneously by DNApolymerase I (50, 78, 119), although efficientexcision requires additional gene products, in-cluding those ofuvrC (103), uvrE (200, 201, 213)and mfd (73, 74) genes. The final step is sealing

of the sugar-phosphate linkage by polynucleo-tide ligase. Alternate pathways of excision re-pair, utilizing DNA polymerase III (244) orDNA polymerase II (144), have been demon-strated in mutants lacking activity of DNApolymerase I. In addition to the "short-patch"type of excision repair shown in Fig. 1B, a"long-patch" pathway has been identifiedwhich requires the products of the recA+ andlexA + gene products, among others, and occursonly in growth-supporting media (49, 245, 248).

Pyrimidine dimers, that are neither photoen-zymatically split nor removed from the DNA byexcision repair (as in excision-deficient Uvr-mutants kept in the dark) block the continuousprogress of the DNA replication complex, butdo not prevent subsequent reinitiation of DNAsynthesis at a point beyond the dimer (185).When pyrimidine dimers or other noninstruc-tive lesions are present in the template strand,daughter strands are detected initially as seg-ments of relatively low molecular weight, their

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from


UV MUTAGENESIS AND DNA REPAIR IN E. COLI 871

continuity interrupted by gaps about 1,000 nu-cleotides long (114) which correspond approxi-mately in number and position to the dimers inthe parental strand (7, 105). These "daughter-strand gaps" are secondary UV lesions, causedby the replication of DNA containing primaryUV photoproducts that prevent base pairing.The third major type of enzymatic DNA repair,postreplication repair (Fig. 1C), operates toconnect the daughter-strand segments, therebyendowing them with the continuity required forfurther replication (185). The formation andjoining of daughter-strand segments synthe-sized on UV-irradiated templates should not beconfused with the discontinuous synthesis andmuch more rapid joining of Okazaki fragmentsat normal DNA replication forks (128). Al-though first demonstrated in Uvr- strains,postreplication repair occurs also in Uvr+strains capable of excising pyrimidine dimersefficiently (192, 203), implying that ability toperform excision repair does not prevent somenoninstructive UV lesions from passingthrough a replication fork.

In excision-deficient (Uvr-) strains that areotherwise normal, the major mechanism ofpostreplication repair is recombinational. AfterUV irradiation followed by DNA replicationand postreplication repair, specifically labeledparental DNA is found covalently inserted intothe daughter strands, the number of insertionscorresponding approximately to the number ofdaughter-strand gaps initially produced (186).The precise mechanism of recombinational ex-change operating to rejoin daughter strands isnot known, but it must be such that some py-rimidine dimers originally present in the pa-rental strands are exchanged into daughterstrands during postreplication repair (71). Fi-gure 1C shows a plausible but hypotheticalscheme for recombinational postreplication re-pair.E. coli is capable of performing postreplica-

tion repair via a number of distinct pathways(72, 182, 189, 190, 192, 209, 247), all of which,however, require the recA + genotype. In recA -mutants, no postreplication joining ofdaughterstrands occurs (203), and the UV sensitivity ofuvrA - recA double mutants, which can per-form neither excision repair nor postreplicationrepair, is such that a single pyrimidine dimerper strand ofDNA kills (103). The requirementfor the recA + gene product, which is necessaryfor any type of genetic recombination in E. coli(42), does not necessarily imply that all path-ways of postreplication repair are recombina-tional. The phenotype of recA- mutants ishighly pleiotropic, and these mutants lackmany wild-type functions in addition to recom-bination ability (Table 1). The specific nature of

some pathways of postreplication repair will beconsidered below, as they may relate to UVmutagenesis.

Error-Proof and Error-Prone Pathways ofDNA Repair

The possibility that UV-induced mutationsoriginate as errors in the repair of radiationdamage (223) has been substantiated by thefinding that UV mutability is eliminated bymutations in either the recA (124, 151, 229) orlexA (159, 227) genes of E. coli, which alsocause extreme UV sensitivity and abolish somepathways of DNA repair. Although they areUV-nonmutable, recA- and lexA- strains arespontaneously mutable and are also mutable byagents believed to cause only replication errors(125, 241). Since recA- and lexA- mutants per-form some kinds ofDNA repair efficiently (e.g.,the "short-patch" pathway of excision repair,photoreactivation) after UV irradiation, yetproduce no UV-induced mutations, the enzy-matic repair occurring in these strains must beaccurate, whereas one or more of the pathwaysof DNA repair requiring both the recA + andlexA + gene products are probably error-proneand responsible for UV mutability (26, 230).Enzymatic photoreversal of pyrimidine di-

mers is accurate in principle and unlikely tocause changes in the base sequence of photore-paired regions of the DNA. In fact, exposure toPRL after UV irradiation prevents most of themutagenic effect, as well as most of the lethaleffect, caused by the same UV treatment in"dark" controls (121, 167). There is no reason tobelieve that monomerization of pyrimidine di-mers by PRE is mutagenic, per se, although thevisible light used to activate PRE may exertmutagenic effects of its own (218, Fig. 5).The "short-patch" pathway ofexcision repair,

at least as it occurs in recA - and lexA - strains,must be accurate, since it does not contribute toa detectable level ofUV-induced mutation. Thelow frequency of mutations which has been as-cribed to errors in the repair of excision gaps(30, 164, 165) may be associated with "long-patch" excision repair, which, like UV muta-genesis, requires the recA + lexA + genotype.When Uvr+ and Uvr- strains are exposed to thesame low fluence ofUV radiation, causing nei-ther detectable killing nor detectable mutagen-esis in the excision-proficient Uvr+ strain, theUvr- strain, unable to excise UV photoprod-ucts, not only shows drastically reduced sur-vival but also produces a high frequency ofUV-induced mutations among the survivors (32, 98,225). The UV hypermutability of Uvr- strainsdemonstrates that unexcised UV photoproductsare much more mutagenic than the same photo-products removed from the DNA by excision,

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


872 WITKIN

Utjce t-- 1o r-q r.4 eq - cn

Q t- tc~~~~~~ o

o4 '4 r~ - CDO "CY^ ^ o Fs

m C Le zscvq "I" o- o

r- LO T-4

cq'--4'--4

co COcq~-o

rI 00 = c,- N4O1:- 00 0)00Lo 00 C -

00 CDI" LOcq4 '-4

&o~Q ,I" I" I

CZ Cd W -W

0-

Q o4

Oa 4. o o C o

M 0 com co En -4At

+ + ++ + +l1+I+ +c*-.

+ + ++ + ++ + +

++++

++ +++ + +~~~~~~( ++

oa),aBi

>

C d

COCCO~~~~~~~'

-4-

i

> >.^

Cd

.k-

14-0

'

0 0OC

z

+

++ + + +

a)so

C

0

°) >

-.e CO,

> ,;>

.!2Co

O

'0

E.L.

0C.)

Co

Co

C.)

L.

.0

Co

C.)

,0,

*S

mOEH

m

^00C8 2

.

+-4

t

t 2

04

0

CO

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



and implies that postreplication repair (at leastas it occurs in Uvr- strains) is far more error-prone than excision repair. In Uvr- strains, inwhich postreplication repair is the only type ofdark repair ofUV damage known to occur, it isprobable that UV mutagenesis is due primarilyto errors introduced into the DNA in the courseof joining discontinuous daughter strands (26,230).The existence of two distinct pathways of

postreplication repair, one error-proof andjeA + independent, the other error-prone and1exA+ dependent, was postulated (227) to ac-count for the properties of certain derivatives ofUV-sensitive, UV-nonmutable lexA- strains.These derivatives had been selected for in-creased UV resistance, and although somewere nearly as UV resistant as jDA + strains,they were still nonmutable by UV radiation.The UV-resistant, UV-nonmutable strainswere assumed to have acquired improved abil-ity to perform an error-proof type ofpostreplica-tion repair, enabling them to tolerate nearly asmany unexcised pyrimidine dimers as lexA+strains without undergoing UV mutagenesis.The ability of similar UV-resistant, UV-non-mutable strains to perform postreplication re-pair much more efficiently than their lexA-parents has been confirmed (215). Biochemicalevidence for a pathway of postreplication repairrequiring the recA+ and lexA + gene productsand associated with UV mutagenesis has re-cently been obtained (189).Two hypotheses concerning the nature of er-

ror-proofand error-prone pathways ofpostrepli-cation repair have been considered. Witkin(227) initially proposed that recombinationalrepair is accurate, and that the lexA+-depend-ent mechanism of daughter-strand gap-fillingis an error-prone DNA polymerase activity ca-pable of inserting nucleotides opposite pyrimi-dine dimers or other noncoding UV photoprod-ucts without template instruction. A similarmechanism was suggested by Bridges et al.(27). The lexA + gene was envisaged (227) eitheras conferring template independence upon anormal DNA polymerase or as itself coding anerror-prone DNA polymerase such as terminaldeoxynucleotidyl transferase (TDT), an enzymecapable ofrandom "end-addition" ofnucleotidesto single-strand ends of DNA (15). In eithercase, gaps opposite pyrimidine dimers could befilled by repair replication with a high probabil-ity of error. Failure to find TDT activity in E.coli, and the later demonstration that DNApolymerases I, II, and III (the only known con-stitutive DNA polymerases) are strictly de-pendent upon template instruction (7, 128; M.Radman, personal communication), tended to

discourage this hypothesis, at least until therequirement for an inducible function was rec-ognized.A second hypothesis (27, 125, 228) proposed

that UV mutagenesis may be due to errors in alexA +-dependent error-prone pathway of re-combinational repair. Three experimental ob-servations made this interpretation seem rea-sonable. One was the dependence of both re-combination ability and UV mutagenesis onthe recA+ gene product (43, 124, 151, 229) andthe apparently commensurate reduction ofbothby recB- and recC- mutations (232). Anotherwas the reduction of the yield of recombinantsto as little as 30% of normal when lexA- mu-tants were used as recipients in conjugation(230), an observation later confirmed but as-cribed to effects other than reduced ability toperform genetic recombination (159). Most im-portant was the approximate equality observedbetween the number of daughter-strand gapsrepaired and the number of recombinationalexchanges detected after postreplication repair(186). This relation implied that either all ornearly all postreplication repair is associatedwith recombination, although it did not ruleout the possibility that a small fraction of thedaughter-strand gaps is repaired by a nonre-combinational mechanism.Evidence against recombinational postrepli-

cation repair as the error-prone gap-fillingmechanism, however, began to accumulate in1971, when Bridges and Mottershead (30)showed that the UV mutability of chemostat-grown E. coli is not enhanced by chromosomalmultiplicity. In recB and recC mutants, whichshow reduced ability to perform genetic recom-bination (42, 68) and postreplication repair(247), yields of UV-induced mutations per sur-vivor at a given UV fluence were either thesame as those in wild type (99; T. Kato, R.Rothman, and A. J. Clark, personal communi-cation) or were somewhat lower (232), depend-ing upon the strain used and on the mutationscored. Even when reduced yields of UV-in-duced mutations were obtained, neither Witkin(232) nor Hill and Nestmann (99) consideredthe reduction to be conclusive evidence that theprobability of error per mutagenic lesion re-paired was affected by absence of exonucleaseV, now known to be the product of the recB andrecC genes (211). The demonstration that E.coli can utilize a second pathway of geneticrecombination requiring the recF+ but not therecB+ recC+ genotype (101) left open the possi-bility that UV mutagenesis might be associatedwith the recF+-dependent pathway. Recent ex-periments (T. Kato, R. Rothman, and A. J.Clark, personal communication) have shown

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


BACTERIOL. REV.

that UV mutagenesis occurs in recA + recB -recF- strains, which are unable to performeither of the two pathways of genetic recombi-nation. This important result suggests that therecA + gene product is required for UV muta-genesis for some reason other than its role inpromoting genetic recombination and that nei-ther of these pathways of recombinational post-replication repair is very error-prone. However,these experiments do not rule out the possibil-ity that residual recombinational repair, eitherdue to leakiness of the rec mutations or inde-pendent of both recB and recF gene products,may be at least partly responsible for bacterialUV mutagenesis. Recent work has demon-strated that error-prone DNA repair activity,whatever its mechanism, is not a constitutiveproperty of undamaged wild-type E. coli, but isinduced in response to UV radiation and othermutagenic treatments. A full description of theexperimental evidence which has led to thisconclusion will follow an account of the "SOS"hypothesis, an hypothesis that has stimulatedmuch of the work to be discussed in this articleand that continues to provide a unifying theo-retical framework for an impressively diverseyet interconnected array of biological and bio-chemical investigations.

The "SOS" Hypothesis: the Regulatory Roleof DNA Damage

Defais et al. (53) proposed that UV mutagen-esis in E. coli might depend upon an induciblefunction which, like X prophage, is repressed inhealthy wild-type cells but is expressed in re-sponse to UV irradiation. One basis for thissuggestion was the dependence ofUV mutagen-esis upon recA + and lexA + gene products, whichare also required for prophage induction by UVradiation, and for the pleiotropic expression ofother UV-inducible functions now recognized tobe numerous (Table 1). The hypothesis that therecA + and lexA + gene products jointly control acoordinately regulated group of inducible func-tions, including an error-prone DNA repair ac-tivity, was further developed by Radman as the"SOS" hypothesis (174, 175). The designation"SOS" (the international distress signal) im-plies that damage to DNA (or stalled DNAreplication) initiates a regulatory signal thatcauses the simultaneous derepression of var-ious functions, all of which presumably pro-mote the survival of the cell or of its phages. Ishall refer to the group of inducible functionsbelieved to belong to this regulatory unit as"SOS functions." "SOS repair" is defined hereas recA + lexA +-dependent error-prone DNA re-pair activity induced in E. coli by UV radiation

or other agents having in common the ability todamage DNA or interrupt its synthesis.The SOS hypothesis incorporates and inte-

grates many observations and ideas that goback as far as a quarter of a century to thedemonstration by Lwoff et al. (134) that UVradiation initiates mass induction of prophagein lysogenic bacteria. Other treatments, shar-ing with UV radiation the ability to arrestDNA replication, were also found to cause lyso-genic induction, including exposure to X rays(129), starvation for thymine (147, 199a), incu-bation with mitomycin C (167c), and tempera-ture elevation in certain mutants unable tosynthesize DNA at high temperatures (154,166, 188).Another early observation contributing to

the SOS hypothesis was Weigle's finding (220)that exposure of the host population of E. colicells to a low fluence of UV radiation prior toinfection substantially increases the survival ofUV-irradiated X phage, and is required for ahigh level ofphage UV mutagenesis. This "UV-reactivation" and the mutagenesis that accom-panies it require the recA + (151) and lexA+ (53)genotype in the host. Defais et al. suggestedthat UV irradiation of the bacteria induces thefunctions responsible for UV reactivation andUV mutagenesis ofthe irradiated phage, via aninduction pathway at least partially common toother recA + lexA +-dependent phenomena, suchas prophage induction and bacterial UV muta-genesis. Radman (174, 175) proposed that a sin-gle inducible error-prone repair replication sys-tem (SOS repair) may be responsible for boththe mutagenic reactivation of UV-irradiatedphage and for error-prone repair of bacterialDNA. I shall use the term "SOS repair" todenote inducible error-prone repair activity act-ing either on phage DNA or on bacterial DNAwithout implying that only one such activity isinduced in E. coli, although there is at presentno reason to invoke more than one. Since UVradiation is only one of many host treatmentscapable of stimulating mutagenic reactivationof UV-irradiated phage, the term "UV-reacti-vation" is misleading. Following Radman's sug-gestion (174), the terms "W-reactivation" and"W-mutagenesis" are used in Table 1 and else-where to refer to the Weigle effect.Another facet of the SOS hypothesis was an-

ticipated in an interpretation of the parallelsobserved between the conditions that promoteor prevent prophage induction in E. coli lyso-gens and those that promote or prevent fila-mentous growth in lon- strains such as E. coliB (226). Witkin proposed that certain bacterialfunctions, including the synthesis of a septum-inhibiting protein, may be governed by repres-

874 WITKIN


http://mm

br.asm.org/

Dow

nloaded from



sors similar enough to A repressor to respond tothe same inducer, an inducer produced onlywhen DNA replication is interrupted. The sub-sequent report (122) thatE. coli K-12 strain T44not only induces prophage spontaneously at ele-vated temperatures but also grows in long fila-ments at 420C, without requiring any of theusual inducing treatments, further indicatedcoordinate regulation of these two functions.Table 1 lists some of the extremely pleio-

tropic effects of recA and lezA- mutations,only one of which (loss of ability to performgenetic recombination) is caused by recA butnot by lexA mutations, and for this and otherreasons is not considered to be an inducibleSOS function. Table 1 also shows that the tif-1mutation, which is responsible for the thermalinducibility of prophage and filamentousgrowth in strain T44 (39), causes constitutiveexpression of other (perhaps of all) SOS func-tions at elevated temperature, without requir-ing any other inducing treatment. When incu-bated at 420C, tif-1 mutants show no abnormal-ity of DNA structure or replication (39), buttheir ability to express SOS functions at thistemperature requires the recA+ kexA+ genotype(40). The tif-1 mutation maps in or very nearthe recA locus (39), and may affect the sameprotein that is altered by recA mutations. Twoother mutations that are tightly linked to therecA locus and may map within it are lexB- (11)and zab- (40). Both lexWB- and zab- mutationscause a phenotype like that oflexA - mutations:noninducibility ofSOS functions without loss ofrecombination ability. The recA locus maps at58 min on the recently recalibrated linkagemap ofE. coli K-12 (4), and the lexA locus mapsat 90 min.Evidence that SOS functions are actually de-

repressed in response to UV radiation andother inducing agents (or in tit mutants at4200) is indirect in all cases except X prophageinduction, in which inactivation (198) and pro-teolytic cleavage (180) of X repressor have beendemonstrated after such treatments. A require-ment for new protein synthesis after the induc-ing treatment has been shown for the expres-sion of most of the SOS functions, and is ob-vious in the case of protein X.

EVIDENCE FOR THE INDUCIBILITY OFERROR-PRONE REPAIR ("SOS

REPAIR") ACTIVITYSOS Repair of Bacteriophage DNA

The increased survival of UV-irradiatedphage that occurs when the host is also irradi-ated before infection (W-reactivation) is accom-panied by a high level of phage UV mutagene-

sis (220). A comparable degree of reactivationand mutagenesis of UV-irradiated X phage oc-curs in unirradiated tit hosts incubated beforeinfection at 420C for about 40 min (39), but notin recA or lexA- derivatives of the tit strain(40). When wild-type strains are UV-irradiatedand incubated in growth-supporting mediumfor 30 min before infection, W-reactivation andW-mutagenesis of UV-irradiated X are maxi-mal, but neither occurs if chloramphenicol ispresent during the preinfection incubation (52).Like other recA+ lexA+-dependent functions,W-reactivation of X requires protein synthesisafter the inducing treatment for its expression.

Further evidence for common regulation ofW-reactivation and prophage induction is thatboth can be induced "indirectly." "Indirect" in-duction of prophage occurs when an unirra-diated F- ("female") lysogen is mated with aUV-irradiated F+ or F' ("male") donor strain(18, 19, 56). The UV-damaged DNA of the Fepisome (an E. coli sex factor), transferred byconjugation into a recipient lysogen whose ownDNA is undamaged, triggers prophage induc-tion. Indirect induction of prophage is also pro-moted by transfer of other UV-irradiated repli-cons, such as colicin I factor (152, 153) or P1bacteriophage (181). Indirect induction of W-reactivation and W-mutagenesis has been dem-onstrated by George et al. (77), who mated anunirradiated F- host with UV-irradiated do-nors (F' or Hfr) before infection with UV-irrad-ited X phage. Mutagenic reactivation of thephage occurred as efficiently as ifthe host itselfhad been irradiated, whether the DNA trans-ferred during conjugation was that of a UV-irradiated sex factor or a fragment of UV-irra-diated Hfr chromosome, and even if the DNAtransferred was restricted in the recipient. Inspite of some puzzling differences (e.g., pro-phage is indirectly inducible only by transfer ofa complete UV-irradiated replicon and not by afragment of Hfr DNA), the indirect inducibilityofboth prophage and W-reactivation points to acommon induction pathway. Indirect inductionof another SOS function (filamentous growth)has also been observed (152; J. Donch, personalcommunication).

W-reactivation (SOS repair) of UV-irradi-ated X phage occurs in both Uvr+ and Uvr-strains, and is independent of ability to performexcision repair (176). However, the range ofUVfluences in which both W-reactivation and Xprophage induction can be expressed is about 10times lower in Uvr- strains (53, 155), as shownin Fig. 2. The photoreversibility of prophageinduction (115), W-reactivation (220), and otherSOS fimctions indicates that pyrimidine dimeris necessary for their induction. Dimers that

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


876 WITKIN

0~~~~~~~~~~~~~~~~~1

10WV<

°1: t&P A or poIA-1 -0

wvrA po/A-i \ l1.

i 0.0110 20 30 40UV FLUENCE (J/m2)

FIG. 2. Relation between survival and inductionof two UV-inducible functions in polA-i, uvrA, andwild-type strains of E. coli. Dashed lines, survivalpercentage. Solid lines, composite curves for induc-tion of X prophage (refer to "free phage") and for W-reactivation of UV-irradiated X (refer to 'W-reactiva-tion factor") (53, 70, 155, 240).

are repaired rapidly by the excision repair sys-tem, however, evidently fail to generate an

effective SOS induction signal.Blanco and Devoret (10) and Devoret et al.

(55) have ruled out recombinational repair asthe mechanism of W-reactivation, and haveconcluded that it expresses the activity of apreviously unrecognized type of dark repair,involving neither excision ofDNA damage nor

recombinational exchanges, which is either in-duced or activated in the host by W-reactivat-ing treatments. The occurrence of W-reactiva-tion in single-stranded phages (167a, 209a) im-plies a nonrecombinational mechanism, bar-ring multiple infection, since only one copy ofthe phage genome is present even after a repli-cation. Experiments with the single-strandedphage qX174 (M. Radman, P. Caillet-Fauquet,M. Defais, and G. Villani, personal communi-cation) have demonstrated that W-reactivationis accomplished by an inducible DNA replica-tion mechanism, capable of polymerizing DNApast pyrimidine dimers in the template strand.These crucial studies will be described in detailin connection with the mechanism of SOS re-

pair.

SOS Repair of Bacterial DNAThe proposal that bacterial UV mutagenesis

and that of UV-irradiated X phage undergoingW-reactivation may depend upon the same in-ducible error-prone repair activity (174, 175)was initially based only upon their commonrequirement for the recA + lexA + genotype. Evi-

dence for inducibility ofa function necessary forbacterial UV mutagenesis has come from twokinds of studies: those in which mutant strainsexhibiting anomalous induction of prophageand other SOS functions were examined forcorresponding anomalies ofUV mutability pre-dicted by the SOS hypothesis, and those inwhich an inducible pathway of postreplicationrepair has been demonstrated biochemicallyand 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 inthe DNA polymerizing activity ofDNA polym-erase I (54, 87). The repair replication usuallyperformed by this enzyme can be taken over byDNA polymerase III (192, 209, 244, 247) and tosome extent by DNA polymerase II (144). Al-though polA-1 mutants excise pyrimidine di-mers more or less normally (21), they are slowto complete the patching and sealing ofexcisiongaps (117, 168). The UV sensitivity of polA-1mutants is relatively slight (4 to 5 times wildtype) compared with that of Uvr- strains una-ble to excise pyrimidine dimers (10 to 20 timeswild type). Nevertheless, as Fig. 2 shows, exci-sion-proficient polA-1 strains express at leasttwo UV-inducible SOS functions (prophage in-duction and W-reactivation) in the same lowrange ofUV fluences as Uvr- strains. Filamen-tous growth, another SOS function, is also in-duced by UV in the same low range of UVfluences in excision-proficient polA-1 mutants(J. Donch, personal communication). SincepolA-1 strains are slow to complete the repairreplication step of excision repair, it appearsthat a persistently open excision gap is equiva-lent to an unexcised pyrimidine dimer as ameans of initiating an effective SOS inductionsignal.Witkin and George (240) suggested that both

the UV sensitivity and the UV mutability ofpolA-1 strains can be explained by the anoma-lously low range of UV fluences in which thesestrains express inducible SOS functions, as-suming that all SOS functions are expressed inthe same fluence range. The UV survival curveof a Uvr+polA-1 strain is superimposable uponthe falling portion of induction versus fluencecurves for prophage and W-reactivation, asshown in Fig. 2. If SOS repair or any otherinducible SOS function is necessary for survivalat higher UV fluences, the survival of UV-irradiated polA-1 populations may be limitedby their ability to induce the vital function(s).Survivors, at any UV fluence, would then con-sist only ofthe fraction ofthepolA-1 populationin which SOS functions are expressed. Since aninducible inhibitor of exonuclease V is one of

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



the SOS functions shown in Table 1, the imme-diate cause of death inpoLA-1 strains may wellbe double-strand breaks in DNA, produced atlower UV fluences than in wild-type strains(16, 17). However, the increased probability ofproducing a lethal break may be traced to amore primary cause: the inability of poLA-1cells to express SOS functions in the samerange of UV fluences as wild-type strains.The UV mutability of polA-1 strains (231)

and of other strains now known to be deficientin the polymerizing activity of DNA polymer-ase I (125, 127) is the same as that of the wildtype, when the frequency of induced mutationsper survivor is compared at the same UV flu-ence. This means that a polA-1 survivor and aPoll survivor, exposed to the same UV fluence,have the same probability of undergoing muta-genesis, although the surviving fraction ismuch smaller in the polA-1 population. Al-though other interpretations are possible (231),a polA-1 survivor in which SOS functions areexpressed may be no different from an SOS-induced Pole survivor exposed to the same UVfluence in those parameters that determine theprobability of undergoing UV mutagenesis.The relevant parameters are: (i) the amount ofpremutational UV damage per survivor, (ii)the probability that such damage will be re-paired by an error-prone mechanism, and (iii)the probability of error per lesion repaired bythe error-prone repair system. Unless the sur-viving fraction inpolA-1 populations is selectedby UV radiation as the least damaged fraction,polA-1 and Polk survivors exposed to the sameUV fluence should be alike in these respects.They would be alike, for instance, if successfulSOS induction in a polA-1 survivor dependsupon metabolic factors unrelated to the amountof DNA damage a particular cell may havesustained. Cells that start transcription late,for example, may have the best chance of tran-scribing information for SOS functions fromfully repaired templates and surviving.

Since the induction of SOS functions is moreeffective inpolA-1 strains that in the wild typeat UV fluences below about 5 J/m2 (Fig. 2),Witkin and George postulated that the DNApolymerase I-deficient mutant should be moreUV mutable at these very low UV fluences, ifUV mutagenesis depends upon inducible SOSrepair. Significantly elevated UV mutability atsublethal UV fluences was demonstrated inpolA-1 strains, as were other properties pre-dicted by the SOS hypothesis (235, 240).

(ii) tiftI mutants. Studies of tif-1 mutants, inwhich SOS functions are thermally induciblewithout insult to DNA, have provided thestrongest evidence that an inducible SOS func-

tion is required for bacterial UV mutagenesis(234, 238). Induction of the postulated SOS re-pair activity by incubation at 420C should en-hance the UV mutability ofa tif- mutant, whencompared with a largely uninduced populationcontaining the same amount of potentially mu-tagenic UV damage. UV damage is obviouslynecessary in an assay for enhanced UV muta-genesis, and therefore the tif- bacteria and tiffcontrols must be UV irradiated either before orafter the incubation at 42"C. Since UV radia-tion itself is an efficient inducer of SOS func-tions, thermally induced SOS repair activityshould be expected to enhance UV mutagenesisonly at extremely low fluences, well belowthose required for mass UV induction of pro-phage or W-reactivation (Fig. 2). In Uvr-strains, UV fluences below about 0.5 J/m2 in-duce prophage or W-reactivation in fewer than1% of the irradiated population, yet about fiveto six pyrimidine dimers per genome are pro-duced inE. coli per 0.1 JWm2. Assuming that allSOS functions are induced at a given UV flu-ence to the same extent, exposure of tif bacte-ria to UV fluences below about 0.5 J/m2 (Uvr-)or 5 JWm2 (Uvr+) should introduce some poten-tially mutagenic UV photoproducts into theDNA of almost every cell, while inducing SOSfunctions (including the postulated SOS repairactivity) in an extremely small fraction of thepopulation.A 10-fold enhancement of the yield of UV-

induced Trp+ mutations was obtained in a tryp-tophane-requiring uvrA- tif-1 derivative of E.coli B/r (strain WP44) when UV irradiationwith about 0.1 J/m2 was followed by a 45-minincubation at 420C, compared with the yieldobtained from similarly treated tiff controls orfrom tif- cells maintained after UV irradiationat 300C (234). The magnitude of the enhance-ment was later increased to a maximum ofabout 50-fold at the lowest UV fluence used(238). Typical results are shown in Fig. 3. Theamount of enhancement of UV mutagenesisprovided by thermal post-treatment in the tifstrain diminishes as the UV fluence increases,and no significant enhancement is observed atUV fluences above about 3 J/m2. UV inductionof X lysogens and of W-reactivation in Uvr-strains approaches 100% at about the same flu-ence (Fig. 2). Once the UV treatment itself issufficient to induce SOS functions in every cell,thermal post-treatment should no longer en-hance the yield of UV-induced mutations.These results are in good agreement with theexpectation that postirradiation incubation at420C (which should induce SOS functions inevery tif cell) should enhance UV mutagenesisat UV fluences too low to cause mass UV-induc-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


878 WITKIN

+S1037 A/-o10U(/0

M >~~~~~~~/1

loo,

.~~~~~~~~Ti-

.1 0.2 0.4 0.8 1.6 3.2 6.4 1.28UV FLUENCE WJ

FIG. 3. Effect of postirradiation temperature ele-vation on the UV mutability of strains WP2, (tit)and WP44j-NF (tif-1 Sfi-). Symbols; A, survival per-centage (average for both strains., whether or notthermal post-treatment was given); *, frequency ofUV-induaced Trp+ mutations (average ofdata for tif+strain, whether or not thermal post-treatment wasgiven, and for tif-1 strain not given thermal post-treatment); O. frequency of UV-induced Trp+ muta-tions (average of data for tif-1 strain given thermalpost-treatment). (a) - - -, theoretical 'one-hit" curve(slope I); (b) ---, theoretical 'two-hit' curve(slope =2). Thermal post-treatment consisted of in-cubation for 60 min at 42C immediately after UVirradiation and plating on 5% SEM agar. All plateswere incubated at 301C for2 days (survival) or 3 days(mutations) starting either immediately after ther-mal post-treatment (if given) or immediately afterplating, if thermal post-treatment was not given.Points are averages of data from four experiments.Trp+ mutation frequencies shown have been cor-rected for spontaneous mutations by subtracting fre-quencies obtained from similarly treated unirra-diated controls. Titers of log cultures used were 1 x10e to 3 x1i0 cells per ml. Forda(uivalescription ofstrains, media and methods used, see references 234and 238.

tion, by permitting error-prone repair ofgVlesions that could otherwise not cause muta-tions. In uninduced cells, such lesions presum-ably either remain unrepaired or are repairedby error-proof mechanisms. The significance ofthe slopes of the curves in Fig. 3 will be consid-ered later.Not only was UV mutability enhanced in the

tifa mutant by thermal post-treatment, but

other predictions generated by the SOS hypoth-esis were fulfilled as well. Thermal enhance-ment of UV mutability was abolished by thepresence of chloramphenicol during the heattreatment, indicating a requirement for newprotein synthesis after the inducing treatment.Additives that had been shown to promote orprevent thermal induction of prophage in tif-lysogens (80, 123) had parallel effects on ther-mal enhancement of UV mutability: the en-hancement was abolished by pantoyl lactone orby a mixture of cytidine and guanosine and waspromoted even at 300C by adenine. Althoughthe basis of these effects is not understood, theyprovide strong evidence for common regulationof prophage and an inducible SOS function re-quired for bacterial UV mutagenesis.

(iii) dnaB's mutants. Temperature-sensitivednaB mutants ofE. coli, which stop synthesiz-ing DNA instantly at 42CC (219), also induceprophage at elevated temperatures when lyso-genic (166, 188). The SOS hypothesis predictsthat all SOS functions should be thermally in-ducible in dnaBts strains, including SOS repair.Enhancement of UV mutagenesis at very lowfluences, virtually identical with that describedfor tiff mutants, has been demonstrated indnaB-43 derivatives of E. coli B/r (237). Al-though tif and dnaBIs strains are very similarin their expression of SOS functions at 420C,the reasons for their thermal inducibility arequite different. In dnaBIs mutants, induction istriggered by interrupted DNA replication,whereas heat treatment oftiF mutants inducesSOS functions without damaging the DNA orarresting its synthesis.

(iv) Other mutant strains. The hyperinduci-bility of polA-1 strains for SOS functions atvery low UV fluences suggests that persistentlyopen gaps in DNA may generate effective SOSsignals, whereas gaps that are rapidly closed donot. Significantly elevated UV mutability atextremely low fluences (possibly indicative ofhyperinducibility for SOS repair activity) hasbeen observed in mfd mutants, which close ex-cision gaps very slowly (73) and are also some-what hyperinducible at low UV fluences forprophage and W-reactivation (74). ElevatedUV mutability in the extremely low UV fluencerange has been found as well in recF mutants(E. Witkin, unpublished observation), whichare deficient in postreplication repair (72), andin UVrA poLA-1 mutants (235), in which post-replication repair may be slow. All of theseobservations support the hypothesis that an in-ducible function is required for bacterial UVmutability, its induction coordinated with thatof prophage and other SOS functions. In addi-

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



tion, the hyperinducibility of strains in whichgaps in DNA are unusually persistent impli-cates such gaps as possible sources of the SOSinduction signal.Evidence from studies of postreplication re-

pair. A correlation between UV mutagenesis inE. coli and an inducible lexA+-dependent path-way of postreplication repair has been demon-strated by Sedgwick (189), who showed thatchloramphenicol, added just before UV irradia-tion of strain WP2S (a uvrA- trp- derivative ofB/r), prevents both the irreversible "fixation" ofUV-induced Trp+ mutations against enzymaticphotoreversal and the occurrence of a smallfraction of postreplication repair. Inhibition ofprotein synthesis after UV irradiation thuscauses a small number of daughter-strand gapsto remain open, and also blocks the occurrenceof UV mutagenesis. Chloramphenicol does notprevent the completion of daughter-strand join-ing, however, in irradiated bacteria which havebeen allowed a 20-min period of growth beforeaddition ofthe antibiotic, or in uvrA- tit bacte-ria which have been incubated for 70 min at420C before UV irradiation. No chlorampheni-col-sensitive pathway of postreplication repairwas detected in UV-nonmutable uvrA- lexA-bacteria. These observations establish that UVmutagenesis is associated with the occurrence

ofa minor lexA +-dependent pathway ofpostrep-lication repair that is UV inducible, or requiresa UV-inducible component. Sedgwick's studyalso shows that the major pathway of postrepli-cation repair in Uvr- E. coli, which is known tobe recombinational (186), is constitutive, lexA+-independent and error-free, at least in the pres-

ence of chloramphenicol. The existence of a1exA + - dependent chloramphenicol - sensitivepathway of postreplication repair has also beendemonstrated by Youngs and Smith (247).

MANIFESTATIONS OF SOS REPAIR ANDTHEIR SIGNIFICANCE

Mutation Frequency Response to IncreasingFluence of UV Radiation

At low fluences ofUV radiation, UV-inducedmutations in bacteria usually increase as thesquare of the UV fluence, as in the lower curvein Fig. 3. This "fluence-squared" (F2) relationimplies that two independent photon absorp-tions (UV "hits") are required to induce a muta-tion. The nature of the two "hits" has been asource ofspeculation for nearly two decades (25,33, 61, 64, 148, 224, 240, 249).

Since most UV-induced mutations are photo-reversible, it is generally assumed that at leastone of the UV photoproducts required for muta-

genesis is a pyrimidine dimer and that at leastone must be located in the gene in which theparticular mutation scored occurs. Two hy-potheses have been argued most recently. Oneis the "two-lesion" hypothesis, according towhich a UV-induced mutation requires the in-teraction of two closely spaced UV photoprod-ucts near the site of the mutation (25, 64, 148).The other is the "one lesion + SOS induction"hypothesis (61, 240), which proposes that onlyone UV photoproduct is a premutational lesionin the gene scored, whereas the other is anSOS-inducing "hit," required for induction ofSOS repair activity. The tit mutant, in whichSOS functions are thermally inducible, seemedan ideal instrument for determining which ofthe two hypotheses is correct. The "one-lesion +SOS induction" hypothesis predicts that titbacteria, UV-irradiated and then given a ther-mal post-treatment, will not require an SOS-inducing "hit," but will show a linear increaseof mutation frequency as the UV fluence in-creases, indicative of the accumulation of pre-mutational UV photoproducts caused by theabsorption of single photons. The frequency ofthermally enhanced yields of UV-induced Trp+mutations rises with an unmistakably linearslope, whereas controls show the typical F2 re-sponse, in both tif (233, 234) and dnaBts (237)strains, as can be seen for a tit strain in Fig. 3.This result seemed to provide unambiguoussupport for the idea that UV mutagenesis re-quires an SOS-inducing "hit" plus a premuta-tional photoproduct, the latter remaining cryp-tic in uninduced cells. However, if all UV mu-tagenesis involves SOS repair of damagecaused by single UV photoproducts, mutationfrequencies in normal strains should begin torise linearly with increasing UV fluence oncethe level ofUV radiation required for mass SOSinduction is reached. The mass-induction flu-ence for both X prophage induction and W-reac-tivation is approached at about 3 J/m2 in Uvr-strains (Fig. 2), the same level at which ther-mal post-treatments no longer enhance UV mu-tagenesis in tit (Fig. 3) and dnaB's (237) popu-lations. However, as Fig. 3 shows, the yield ofUV-induced mutations continues to rise ap-proximately as the square ofthe UV fluence, inboth tip and tifr strains, thermally post-treated or not, up to UV fluences as high as 9 J/m2. Bridges et al. (33) have also reported muta-tion frequency response curves that continuetheir P slope well above the mass-inductionfluence level. The data shown in Fig. 3 seem tosupport the "one-lesion + SOS induction" hy-pothesis at low UV fluences and the "two-le-sion" hypothesis at higher UV fluences. Both

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


880 WITKIN

Witkin (238) and Doudney (63) have recentlyconcluded that some UV-induced mutations inE. coli are caused by single UV photoproducts,whereas others are caused by the interaction oftwo closely spaced UV lesions, and that neitherhypothesis is wholly correct nor wholly incor-rect.Witkin (238) has distinguished between sin-

gle premutational UV lesions that require er-

ror-prone SOS repair but are not, in them-selves, sufficient signals for the induction ofSOS repair activity ("SOS-mutable, SOS-non-inducing" lesions), and UV lesions generatedby the interaction of two UV photoproducts,which are not only targets for SOS repair activ-ity but are also effective inducers ofSOS repair("SOS-mutable, SOS-inducing" lesions). Thedistinction between these two types ofpremuta-tional lesions was based on the kinetics of ther-mal enhancement of UV mutagenesis at 42°Cand the kinetics of decay at 300C of susceptibil-ity to such enhancement in tif strains. It wasconcluded that in uninduced cells SOS-muta-ble, SOS-noninducing lesions are cryptic, butcontribute to overt UV mutagenesis if SOS ac-tivity is induced by some other means, such astemperature elevation in tif mutants or expo-sure to UV fluences high enough for mass UVinduction of SOS repair. Unless such crypticpremutational lesions are present in UV-irradi-ated but uninduced cells, thermal post-treat-ments could not enhance UV mutagenesis intif or dnaBIs strains. Unless the cryptic lesionsare due to single UV photoproducts, the ther-mally enhanced yield should not rise linearlywith increasing UV fluence. However, unlessone assumes that some UV-induced mutationsare caused by two UV photoproducts, the F2accumulation of induced mutations at fluencesfar above the mass-induction level (33, 63; Fig.3) would be hard to explain. Variation in theproportions of the two kinds of premutationallesions might be expected in different genes, oreven in the same gene under different condi-tions, and could account for the complexity andvariability characteristic ofmutation frequencyresponse curves (63).

The UV Lesion Responsible for Induction ofSOS Functions

The composite induction curves for prophageinduction and W-reactivation shown in Fig. 2show a steep rise in the low UV fluence rangesimilar to the increase in frequency of UV-induced mutations with the square of the UVfluence in the same range. When measuredcarefully, prophage induction can be seen toincrease strictly as the square ofthe UV fluence

BACTERIOL. REV.

in both Uvr+ (141) and Uvr- (E. Witkin, unpub-lished data) E. coli lysogens, in striking con-trast to the linear increase of induced lysogenswith increasing exposure to X rays (140). If allSOS functions are assumed to respond to acommon induction signal, UV radiation clearlyrequires two photon absorptions to generate aneffective induction signal.Sedgwick (189, 190a) has suggested that a

target for SOS repair (and, by implication, anSOS-inducing lesion) may be a daughter-strandgap that partially overlaps another daughter-strand gap in the sister DNA molecule. Such apair of overlapping daughter-strand gapsshould be refractory to any type of recombina-tional repair, in principle, and could not bepatched by the repair replication activity of theknown constitutive DNA polymerases, becauseof the pyrimidine dimer or other noncoding UVphotoproduct located opposite each of the gaps.This type of structure (Fig. 4A) would beformed when two pyrimidine dimers on oppo-site strands of the parent molecule are closeenough so that the daughter-strand gaps pro-duced by them overlap, either as they are origi-nally formed or after some DNA degradation.In Uvr- strains, such gap structures could ac-count entirely for the F2 relation betweenexpression of SOS functions and UV fluence inthe low fluence range. Mass SOS induction, onthis basis, should occur at the fluence at whichevery irradiated cell, on the average, will con-tain one pair of overlapping daughter-strandgaps after DNA replication.

In Uvr+ strains, unexcised UV photoproductscould also generate overlapping daughter-strand gaps that would be refractory to consti-tutive repair systems, as in Uvr- strains. How-ever, Uvr+ strains could, in addition, form thekind of gap structure shown in Fig. 4B, whichwould also require the production oftwo excisa-ble UV photoproducts close to each other onopposite strands of the irradiated DNA mole-cule. Excision of one of the lesions, perhapswith some DNA degradation enlarging the ex-cision gap, could place the second lesion oppo-site the excision gap, thus preventing patchingby DNA polymerase I. Since correndonucleaseII requires a double-stranded substrate for inci-sion near a pyrimidine dimer, the second lesionwould remain unexcised at least until the firstexcision gap is somehow filled and sealed. Suchstructures, if produced in a part of the DNAreplicated before UV irradiation, might be re-pairable by recombination, but should be re-fractory to any known type of constitutive re-pair if produced in an unreplicated part of thechromosome. Bresler (23) has suggested that


http://mm

br.asm.org/

Dow

nloaded from



A B

POSTREPLICATION REPAIR

(chloramphenicol-sensitive pathway)UV

3' ,5'; replication

5'- /.3

3' 55'daughter-strand gaps overlap

s recombinational repair impossiblegaps enlarged by degradation

3,5 ' 33' ---------V 5'SOS repair induced, poly-

; merizes DNA past dimer,causing mutation (x)

2nd gap repaired byrecombination

5' 3'

31' 3

one daughter moleculeproduces overlappinggaps in next replication

or 2nd gap also repairedby SOS polymerase

5i 333' x'

3' 5'no overlapping gapsproduced in next re-plication

EXCISION REPAIR

("long-patch' pathway)UV

3 5s excision,degradation

5- - 3:3' 5'repair replication stops

sat 2 nd dimer, degroda-tion continues

5 -V -53' , 5'..

SOS repair induced,; polymerizes DNA postdimer, causing mutation(x)

5' 3'31 5'

2nd dimer r,%pairedby1f short-patch excisionrepair

(both strands carrymutant information)

FIG. 4. Possible pathways oferror-prone repair (SOS repair) ofdamage to DNA caused by UW radiation inE. coli. Symbols: -- or-or 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" basesequence). SOS repair activity is assumed to permit replication pastpyrimidine dimers or other noninstructiveLW photoproducts with a high probability of error (see section of text under heading, Mechanism of SOSRepair).

this type of structure could cause UV mutagen-esis by promoting replication errors duringrecA+-dependent excision gap resynthesis, buthe does not believe that error-prone repair ac-tivity is inducible. Polymerization past a py-rimidine dimer, with or without errors, is notwithin the capacity of any of the three knownconstitutive E. coli DNA polymerases (7, 128;P. Caillet-Fauquet, M. Defais, and M. Rad-man, personal communication).

Excision gaps opposite a dimer (Fig. 4B) oroverlapping daughter-strand gaps (Fig. 4A)would seem to confront E. coli with types ofdamage not repairable by constitutive repairsystems, and therefore lethal in strains not ableto induce a type of repair activity capable of

closing such gaps. Unless and until such activ-ity was induced, these otherwise nonrepairablegaps would tend to persist. Several strains inwhich repair defects cause unusually slow clo-sure of gaps in DNA (polA-1, recF, mfd) havebeen found to be hyperinducible for at leastsome SOS functions (see above). It is a reasona-ble possibility that an SOS signal may be initi-ated by either ofthe types of gaps shown in Fig.4, although there is no direct evidence for thisidea. It should be emphasized that the probabil-ity that two UV photoproducts would generatesuch a gap structure would depend not onlyupon the distance between them, but also onthe extent of DNA degradation occurring afterUV irradiation.

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


882 WITKIN

Several otherwise puzzling features of UVmutagenesis can be explained if it is assumedthat the two kinds of gaps shown in Fig. 4 arethe primary targets for SOS repair activity inE. coli. Although all UV-induced mutationsexamined in Uvr- strains become irreversibly"fixed" against initially antimutagenic post-treatments such as PRL (164) or caffeine (239)coincidentally with DNA replication, some UV-induced mutations in Uvr+ strains lose theirphotoreversibility as dimer excision is com-pleted, well before DNA replication, indicatingthat some mutations in Uvr+ strains originateas rare errors in the repair of excision gapsbefore DNA replication (30, 164, 165). The kindof excision gap shown in Fig. 4B could provide asite requiring prereplication SOS repair inUvr+ strains. Since induction of SOS repairrequires considerable time (see below), gapsrequiring this repair should persist as targetsfor exonuclease V degradation and could be-come considerably enlarged by the time theinducible repair system is available. This couldaccount for the "long patch" character of therecA+ lexA+-dependent (i.e., inducible) path-way of excision repair, which may well be thesame SOS repair activity that also repairs over-lapping daughter-strand gaps. As Bresler (23)has pointed out, subsequent excision of the sec-ond UV photoproduct, once the mutagenicevent has closed the first excision gap, wouldlead to a "pure clone" of the mutant type, atleast in cells containing only one surviving copyof the genome.

Sedgwick's overlapping daughter-strand gaphypothesis (190a) could also explain why detect-able UV-induced mutations are produced pri-marily and probably exclusively in the firstpostirradiation cell division (60, 230, 238), al-though pyrimidine dimers persist and are ex-changed into daughter strands in Uvr- strains(71). Closely spaced UV photoproducts that pro-duce overlapping daughter-strand gaps in thefirst postirradiation cell division would eitherfail to do so at all in subsequent replications, orwould do so with a reduced probability, depend-ing upon the mechanism of SOS repair (e.g., ifboth gaps are repaired by error-prone polymeri-zation, the photoproducts would remain in theparental strands and no overlapping daughter-strand gaps would be produced after the firstreplication). Even if mutagenic gap structuresare generated after the first DNA replication,the more rapid proliferation of DNA moleculesnot containing such damage might prevent theexpression of potential second-round muta-tions, since most methods used to detect in-duced mutants limit postirradiation growth inone way or another (3).

Kinetics of Induction and Decay of SOSRepair Activity

The kinetics of induction and decay of induc-ible error-prone repair activity have been deter-mined in two very different systems with re-markably similar results.

Witkin (238) investigated thermally inducedSOS repair activity in tif- B/r derivatives thathad been rendered viable at 420C by selectionfor a secondary mutation (Sfi-) that suppressesthermally inducible filamentous growth with-out altering other thermally inducible tif-me-diated functions (75). Kinetics of induction ofSOS repair activity were assayed in both Uvr+and Uvr- tif- Sfi- derivatives by determiningthe degree of enhancement of UV mutabilityobtained, after exposure to a low fluence of UVradiation, with increasing periods of incubationat 420C (0 to 30 h) before completing incubationat 300C. No significant thermal enhancementwas obtained if the incubation at 42°C wasshorter than 30 min. Optimal enhancementwas obtained with incubations at 420C for 45min after UV irradiation or longer. The abilityto enhance UV mutagenesis induced by a 60-min incubation at 420C was found to decay at300C (the noninducing temperature) to abouthalf its optimal level in 30 min, and was nolonger detectable after 90 min at 300C. Kineticsof induction at 42°C and of decay at 300C ofthermally induced ability to enhance UV muta-bility (i.e., of thermally induced SOS repairability) were virtually identical in Uvr+ andUvr- tif Sfi- strains. The time required forinduction of SOS repair is consistent with theobservation (69) that no error-prone gap-fillingoccurs during the first 20 min after UV irradia-tion.

Defais et al. (52) determined the kinetics ofinduction and decay ofUV-inducible capacity toW-reactivate UV-irradiated X phage. Irradi-ated X were reactivated maximally when infec-tion was delayed for 30 min after UV irradia-tion of the host cells. The UV-induced capacityfor W-reactivation decayed with a half-life of 30min once the optimal level was achieved. Al-though there are some differences, not surpris-ing in view of the radically different inducingtreatments and assay systems used, the twosets of kinetics obtained by Witkin and by De-fais et al. are similar enough to encourage thehypothesis that the same inducible error-pronerepair activity effects bacterial UV mutagene-sis and W-reactivation of UV-irradiated X.

Time of Action of SOS Repair in UVMutagenesis

Two recent studies have attempted to deter-mine just when UV-induced mutations are es-

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



tablished by the action of error-prone repair.Doubleday et al. (60) have used the kinetics ofloss of photoreversibility (LOP) as a measure ofirreversible commitment to UV-induced muta-tion caused by pyrimidine dimers in E. coliWP2S (uvrA- trp-). Irradiated bacteria, incu-bated for two generations in each of three dif-ferent media before plating on medium selec-tive for Trp+ mutants, produced the same num-ber of induced mutations at the same UV flu-ence, but lost photoreversibility of the inducedmutations at very different rates. In rich me-dium, photoreversibility was lost before thepopulation had completed one cell division; inminimal medium plus tryptophane, some pho-toreversibility of the UV-induced Trp+ muta-tions was still present after an average of twocell divisions; and in rich medium plus pantoyllactone, LOP kinetics were extremely slow. UVfluence had no effect on LOP kinetics, but onlyon the final yield of induced mutations. Theauthors considered their results compatiblewith either of two interpretations: (i) that er-ror-prone repair may occur after one or morerounds of DNA replication in which daughter-strand gaps are repaired by an error-free mech-anism, or (ii) that error-prone repair occursonly in the first round of postreplication repairat daughter-strand gaps, which are refractoryto error-free postreplication repair, perhaps be-cause they form part of an overlapping gapstructure such as that proposed by Sedgwick asthe target for SOS repair activity. In that case,the premutational gaps would remain open un-til the error-prone system is induced and/orcompletes its action. According to the latterinterpretation, the different rates of LOP re-flect medium effects on the rate of inductionand/or on the rate of action of the error-pronerepair system. Overall population increase,then, need not reflect the specifically delayeddivision of potential mutants, and all detecta-ble UV mutagenesis may actually occur onlyduring the first round of postreplication repair.This study also provided clear evidence, basedon segregational analysis, that most or all UVmutagenesis is "terminal," occurring only onceat a given site within a clone.Witkin (238) used tif Mfi strains (both Uvr-

and Uvr+) to analyze the timing of thermallyinduced error-prone repair activity, confirmingthe conclusion of Doubleday et al. that UVmutagenesis is "terminal." These experimentsalso showed that thermally enhanced yields ofUV-induced Trp+ mutations are produced pri-marily (possibly exclusively) before the comple-tion of the first round of postirradiation celldivision, when SOS activity is present (at420C). In populations consisting almost entirely

of cells not containing SOS repair activity (be-cause they were exposed to an extremely lowfluence of UV radiation and were incubatedafterwards at 3000), the capacity to produce anoptimally enhanced yield of UV-induced Trp+mutations in response to a delayed thermalpost-treatment was fully maintained at 300C forat least 2 to 3 h.

Photoreversibility of the enhanced yield ofUV-induced mutations, rapidly lost when incu-bation after UV irradiation is at 420C, was alsofully maintained for at least 2 h at 300C. It wasconcluded that sites for SOS repair activity arestable in uninduced cells. If such sites aredaughter-strand gaps, they must be refractoryto constitutive error-proof recombinational re-pair, remaining open and unrepaired for atleast 2 to 3 h in the absence of SOS repairactivity. Since the thermally enhanced yields ofUV-induced mutations increase linearly withUV fluence, however, the premutational sitesmust be gaps produced by single UV photoprod-ucts, which, in spite of being refractory to con-stitutive postreplication repair, are unable togenerate an SOS repair-inducing signal. Theseare the "SOS-mutable, SOS-noninducing" le-sions discussed above in connection with muta-tion frequency response curves.

If the target ofSOS repair activity is a gap inDNA, the "SOS-mutable, SOS-noninducing" le-sion responsible for thermally enhanced yieldsof Trp+ mutations must be a single daughter-strand gap, not overlapping another in the sis-ter molecule of DNA. In principle, nonoverlap-ping gaps should be susceptible to constitutiverecombinational repair, which Sedgwick hasfound to occur efficiently at 300C in the sametif Sfi- strain (189). The refractoriness of thisparticular group of lesions to constitutive post-replication repair may be related to uniqueproperties of transfer ribonucleic acid-codinggenes, in which most of the Trp+ mutationsinduced by UV in these strains (primarilyochre suppressors) probably occur. UV-inducedsuppressor mutations are uniquely refractoryto excision repair when incubation after UVirradiation is in rich medium, a phenomenonrelated to "mutation frequency decline" (12,230, 242). The same specific features that makesome transfer ribonucleic acid-coding genes lesssusceptible to excision repair than other genesunder certain conditions may also interferewith recombinational postreplication repair.For instance, possible "cloverleafing" of thenontranscribed strand during transcriptioncould place the premutational UV photoproductin a configuration inaccessible to repair en-zymes. Another possibility is that UV photo-products causing this type of suppressor muta-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


884 WITKIN

tion have an unusual primary structure, asindicated by the action spectrum for their in-duction, which is quite different from the actionspectra for the induction of other mutations orfor the production of pyrimidine dimers (194).

Cryptic Premutational Lesions Susceptible toSOS Repair

The existence ofSOS-mutable lesions incapa-ble of inducing error-prone repair activity hassome important implications. Although suchlesions are cryptic in uninduced cells, they arecapable ofundergoing mutagenesis in cells thatalso receive an effective induction signal. Acase in point is the fluorescent light used in mylaboratory as a source of PRL. Visible light isknown to be mutagenic (218), but the exposuresrequired to demonstrate this are usually muchlonger than the times used in photoreversalexperiments. However, as Fig. 5 shows, thesame exposure to PRL is far more mutagenic intif cells given a 60-min incubation at 420Cimmediately after illumination than in controlsincubated at 300C. This implies that PRL mayproduce SOS-mutable, SOS-noninducing le-sions in DNA which are cryptic in uninducedcells, but contribute to overt mutagenesis inSOS-induced cells. Recognition that somesources of visible light may produce such le-sions can provide an explanation for some oth-erwise enigmatic effects in experiments such asthose (33, 148) in which UV irradiation is fol-lowed by exposure to PRL and then by a secondUV irradiation. In both studies, pretreatmentwith UV + PRL enhanced the mutagenic effectof the second UV exposure and this effect wasattributed to nonphotoreversible UV photo-products, still present in the DNA after thePRL treatment. Actually, SOS-mutable, SOS-noninducing lesions produced by the PRL treat-ment itself may have caused or contributed tothe enhanced mutagenesis. SOS repair activ-ity, induced by the second UV exposure, couldhave potentiated the otherwise cryptic muta-genic effect of the PRL.The production by PRL of SOS-mutable,

SOS-noninducing photoproducts in DNA re-quires caution in the interpretation of photo-reversal experiments, especially those deter-mining kinetics of LOP. When PRL is givenimmediately after UV irradiation, dimers aresplit before SOS repair activity can be induced,and potentially mutagenic DNA lesions pro-duced by the PRL treatment, which do not trig-ger SOS induction, remain cryptic. However,when the PRL treatment is delayed for aboutan hour after UV irradiation, UV-induced SOSrepair activity is fully expressed by that time,and PRL treatment could contribute signifi-

a.

1-J

0

&200

4

+kI0-10~~ ~ ~ ~ ~ ~~~-

z 0-

Cr --0

0 5 10 15 20 25 30PRL EXPOSURE (MIN)

FIG. 5. Mutagenic effect of visible light (PRL) instrain WP44,-NF (tif-1 Sfi-). Aliquots (1 x 107 to 3 x107 cells) from log culture were plated on 5% SEMagar, exposed to PRL 0 to 30 min, and then incu-bated at 42TC for 60 min (a) or not (0). All plateswere incubated for 3 days at 309C starting eitherimmediately after post-treatment (ifgiven) or imme-diately after illumination (if no thermal post-treat-ment was given). Source ofPRL was a pair of West-inghouse "cool white" fluorescent bulbs. Trp+ muta-tion frequencies shown have been corrected for spon-taneous mutations by subtracting frequencies ob-tained in similarly treated 'dark" controls. Survival(monitored by plate-wash assays) did not depart sig-nificantly from 100% under any of the conditionsshown. (For detailed description of strain, methods,and media used, see references 234, 238, and 240.)No significant effect of thermal post-treatment wasobtained in similar experiments using the tiff strainWP2, (not shown).

cantly to the mutations scored as nonphotore-versible UV-induced mutations. A suitable con-trol would be equivalent PRL treatment of aheated but unirradiated tiff population, whichwould permit enumeration of the PRL-inducedcontribution to the "nonphotoreversible" frac-tion of mutations. Various sources of PRL maydiffer in the efficiency with which they produceSOS-mutable, SOS-noninducing photoprod-ucts. The mutagenic effect in heated tif cells ofthe PRL source used in my laboratory was re-duced to about 50% by a filter that eliminatedwave lengths below 350 nm.Heated tif populations may be useful in

screening tests for environmental mutagens,some of which, like fluorescent light, may be

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



far more mutagenic in combination with anefficient SOS-inducing agent that could be sus-pected from their nonmutagenic or weakly mu-tagenic response in the usual test systems.

It should be emphasized that photoreversibil-ity of UV-induced mutations gives no informa-tion about the nature of the premutational pho-toproduct. Pyrimidine dimers may be requiredfor the induction of SOS repair activity, butonce induced, the error-prone repair systemmay cause mutations during repair of nondi-mer damage as well. In Uvr+ strains, the split-ting of pyrimidine dimers may also reduce UVmutagenesis indirectly by promoting subse-quent excision of nondimer UV photoproductsthat would otherwise compete poorly with di-mers for excision enzymes.

Mutator Effect of SOS Repair on UndamagedDNA

An unexpected byproduct of studies of tif-1derivatives of B/r was that incubation of un-treated control populations under SOS-induc-ing conditions (i.e., at 4200 for 45 min or longer)resulted in a significant increase in sponta-neous mutability, estimated as about 60 timesthe normal rate of mutation from Trp- to Trp+,indicating that the tif-1 mutation is a mutatorat 420C (233, 234, 238). Mutator activity hasbeen described also in K-12 tif-1 strains (75).The elevated spontaneous mutability in unirra-diated controls does not obscure thermal en-hancement of UV mutability, since thermalpost-treatments required to demonstrate suchenhancement are brief (45 to 60 min), and theheated populations are small enough (2 x 107 to3 x 107 cells per plate), so that the actual num-ber of mutant colonies on heated but unirra-diated control plates is small compared with thenumber on platings of UV-irradiated popula-tions, and is subtracted before plotting UV-induced yields as in Fig. 3. Prolonged incuba-tion at 4200 of unirradiated tif-1 cells, however,reveals the mutator effect much more dramati-cally.When SOS repair activity is induced by UV

radiation, the mutations caused by the radia-tion should include some that are not due toerror-prone repair of UV damage, but to themutator activity of the SOS repair system inundamaged portions of the DNA. The magni-tude of the mutator effect in thermally post-treated tif- populations (238) is such that thefraction ofUV-induced mutations caused by theSOS mutator activity does not exceed 10% un-der the conditions usually used, at least withrespect to the Trp- to Trp+ mutations scored.This maximum estimate includes mutationsthat may be caused by SOS mutator activity

during repair replication steps of the otherwiseerror-proof repair pathways (e.g., Fig. 1B andC) when these pathways operate in SOS-in-duced cells.A mutator effect of thermally induced SOS

repair activity has also been reported in a lig'smutant in which DNA ligase is inactive at hightemperatures (157). Since prophage is ther-moinducible in lig's lysogens (81), SOS repairactivity should be expressed at 420C as well.However, the data in the paper cited are notsufficient to establish that bacterial mutabilityis increased by heat treatment of this mutant.The increase reported is an apparent elevationof the frequency of mutants among survivors ofmore or less lethal heat treatments. Under con-ditions promoting filamentous growth, as heattreatment of lig's is known to do (72a), seem-ingly large mutagenic effects can often betraced to a methodological artifact (3, 227)which, if recognized, can be satisfactorily con-trolled (see Appendix).

If SOS repair activity is mutagenic in cellscontaining no structural damage in their DNA,one might expect that all agents capable ofinducing prophage and other SOS functionsshould be at least somewhat mutagenic. In fact,a high correlation exists in E. coli between themutagenicity of various treatments and theirability to induce prophage (127). SOS-inducingagents should be divisible into two distinctclasses with respect to their mutagenic potency:potent mutagens, like UV radiation, which notonly induce SOS repair activity but also pro-duce numerous sites in the DNA requiring itserror-prone action, and weak mutagens, whichinduce SOS repair activity by arresting DNAreplication but do not otherwise damage thegenetic material. SOS-inducing agents of thelatter type should exert only the relativelyweak mutagenic effect seen in dnaB's unirra-diated populations incubated at 4200 for 1 h,and then returned to the permissive tempera-ture (237). Thymine starvation, nalidixic acidinhibition, and temperature elevation in ligtsmutants should act as similarly weak muta-gens, since these treatments arrest DNA repli-cation but probably do not introduce any struc-tural damage requiring SOS repair activity,except perhaps at the replication fork(s). SinceSOS activity decays with a half-life of about 30min, the mutator effect should operate exclu-sively or at least primarily during the firstround of DNA replication after termination ofthese inhibitory treatments. Valid demonstra-tions of thymineless mutagenesis (those inwhich an absolute increase in the number ofmutant colonies is obtained after starvationtreatments causing little or no thymineless

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


886 WITKIN

death) usually show the expected relativelyweak effect (e.g., references 29, 116). Extremelyhigh mutation yields, calculated from data ob-tained after long periods of starvation whichcause much thymineless death, have been re-ported, but in these cases the frequency of mu-tants in thymine-starved populations may havebeen grossly overestimated (see Appendix).Sublethal starvation for thymine in strainWP2. thyA , a derivative ofE. coli B/r, causes aweak but genuine mutagenic effect, which isabsent in its lexA - and recA - derivatives, andenhances UV mutability at low fluences as ef-fectively as does thermal post-treatment of tifor dnaB's strains (E. Witkin, manuscript inpreparation). These observations confirm theconclusion (29) that UV radiation and thyminestarvation share a common pathway of muta-genesis. Similar weak mutagenesis and strongenhancement of low-fluence UV mutagenesishave been observed after sublethal treatmentwith nalidixic acid (E. Witkin, unpublisheddata). Both thymine starvation and nalidixicacid treatment, as well as ionizing radiation,which is also virtually nonmutagenic in lexA-strains (29), appear to be effective inducers ofSOS repair activity. The unusual kind of thy-mineless mutagenesis described by Bresler etal. (24), however, which operates only underspecial conditions and is independent of lexA+and recA+ gene products, is clearly not due toinducible SOS repair activity.

Mutations have been produced in untreatedphage by irradiating the host with UV radia-tion before infection (109, 115a) or by pretreat-ing the host with N-methyl-N'-nitro-N-nitroso-guanidine (126). These observations have beeninterpreted by Ichikawa-Ryo and Kondo (109)as expression of an inducible misreplicationsystem in the host that elevates the yield ofmutations occurring during the replication ofuntreated phage. These authors, however,doubt that the induced misrepair system is thesame one that is responsible for UV mutagene-sis of either bacterial or phage DNA, since UVirradiation of a lexA - mutant caused some mu-tagenesis of the untreated phage, although notas much as that observed in an irradiated lexA +host. It seems possible (as suggested by M.Volkert) that a low level of SOS repair activitymay be expressed in some lexA- strains, too lowto be readily detectable in assays for UV muta-genesis or W-reactivation, but detectable in themuch more sensitive assay system provided byunirradiated phage. If so, the same SOS repairsystem induced in E. coli could be responsiblefor bacterial UV mutagenesis, for mutagenicW-reactivation and for the mutator effect on

undamaged bacterial DNA and on untreatedphage infecting mutagen-treated hosts.

PLASMIDS AND SOS REPAIRIn Salmonella typhimurium, certain plas-

mids, including colIb and the drug-resistancefactor R-Utrecht, have been found to enhanceUV resistance and increase UV mutagenesis inwild-type cells (65, 107, 136). These effects havegenerally been interpreted as indicative of plas-mid-borne genes that determine or enhance er-ror-prone repair of UV-damaged hosts. Expres-sion of this effect by the plasmid R-Utrechtdepends upon the presence in Salmonella of agene product that appears to be similar to therecA+ product of E. coli (137). In the Salmo-nella tester strains developed by Ames et al.(2), some R factors enhance the mutagenic ef-fect of many carcinogens in recA + but not inrecA - hosts (146). In recA - strains, these car-cinogens are nonmutagenic with or without theR factor. An R factor introduced into E. coli K-12 enhances UV resistance and spontaneousmutability in wild-type hosts, but not in recA -or lexA - mutants (G. Walker, personal commu-nication). It is possible that some plasmidscarry genes specifying a unique SOS repairactivity, responsive to host regulation, or genesthat amplify the expression of the host's owninducible repair system.

MECHANISM OF SOS REPAIRAlthough current evidence is consistent with

the possibility that a single SOS repair activityis responsible for bacterial and phage UV mu-tagenesis, there is no proof that only one type oferror-prone DNA repair is induced by UV ra-diation in E. coli. The possible mechanism(s) ofSOS repair of phage and bacterial DNA willtherefore be discussed separately.

Mechanism of SOS Repair in BacteriophagesDevoret et al. (55) have reviewed the evi-

dence that mutagenic W-reactivation of bacte-riophage X is independent of both excision re-pair and recombinational repair and that itdepends upon a novel error-prone repair activ-ity induced or activated in the host cell by UVradiation and .other SOS-inducing treatments.The requirement for new protein synthesisafter the inducing treatment (52, 167a) suggestsinduction rather than activation of this error-prone repair activity, or of one or more of itscomponents.

Proof that at least one type of UV-inducibleSOS repair is an error-prone DNA polymeriz-ing activity has come recently from work done

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



by Radman and his collaborators with the sin-gle-stranded DNA phage qX174 and with syn-thetic template-primers (M. Radman, personalcommunication, data summarized in M. Rad-man, P. Caillet-Fauquet, M. Defais, and G.Villani, 1976, p. 537-545. In Screening tests inchemical carcinogenesis. R. Montesano, H.Bartsch, and L. Tomatis [ed.], IARC Publica-tions, no. 12, Lyon). When UV-irradiated4X174 infects untreated E. coli, the phageDNA remains largely unreplicated (i.e., singlestranded), and the amount of phage DNA syn-thesis that does occur is consistent with theinterpretation that replication is arrested bythe first pyrimidine dimer encountered in thetemplate (7). It appears, therefore, that none ofthe constitutive DNA polymerases can copyUV-irradiated DNA past pyrimidine dimers.However, if the host is exposed before infectionto UV radiation, under optimal SOS-inducingconditions, much of the single-stranded qX174DNA is converted to the covalently closed dou-ble-stranded replicative form, and the extent ofthe UV stimulation of phage DNA synthesis iscorrelated with the extent of W-reactivation.When primed 4X174 DNA is used as an in vitrotemplate with Kornberg's clear extracts of E.coli (128), DNA synthesis is largely preventedby prior UV irradiation of the phage DNA. Theextent of the inhibition and its photoreversibil-ity by PRE indicate that pyrimidine dimers areblocks to this in vitro DNA synthesis. The rea-son for blockage by dimers ofDNA synthesis byDNA polymerase I appears to be the "proof-reading" 3'-5'-exonuclease activity of this en-zyme, which increases two to three orders ofmagnitude, relative to the polymerizing activ-ity, when the 4X174 DNA template is exposedto a UV fluence of 500 J/m2 (G. Villani and M.Radman, personal communication). Because ofthis result, and because they have failed toidentify a new induced DNA polymerase, Vil-lani and Radman have suggested that an induc-ible inhibitor ofthe proofreading activity of oneor more constitutive DNA polymerases may bethe mutagenic effector. In the absence of thisinhibitor, nucleotides inserted opposite pyrimi-dine dimers or other noninstructive lesionswould be promptly removed. In its presence,the stable insertion of one or more "wrong"nucleotides would permit replication to con-tinue, albeit with a high probability of muta-tion. There is considerable evidence (66, 128)that spontaneous mutation rates are grosslyaffected by changes in the relative efficiency ofpolymerizing and proofreading activities inDNA polymerases. Thus, an induced inhibitorof proofreading activity would be expected to

cause transient mutator activity during repli-cation of undamaged DNA, as well as to pro-mote mutagenesis at the site of any noninstruc-tive lesion.

In experiments using a synthetic template-primer, poly(dT)-oligo(dA), Villani and Rad-man (personal communication) have developedan assay for inducible SOS repair activity.When the template-primer is UV-irradiated,crude extracts from "SOS-" cultures (tif-1 cellsincubated at 32°C) fail to promote DNA synthe-sis. When crude extracts from "SOS" cultures(tif-1 cells incubated at 42°C for 45 min in me-dium containing adenine) are used, DNA syn-thesis occurs on the UV-irradiated templates,and massive misincorporation (uptake ofdGTPand dCTP) is observed. Some misincorporationoccurs when the "SOS" crude extract is usedeven with unirradiated template-primer, corre-sponding to the mutator effect of SOS activityin undamaged E. coli cells. The in vitro misin-corporation promoted by crude extracts fromheated tif-1 mutants is recA+ dependent, as isSOS repair activity in vivo. The in vitro assayprovides a basis for possible purification of theprotein(s) responsible for SOS repair activity.

Mechanism of SOS Repair in BacteriaSince SOS-inducing treatments induce an er-

ror-prone DNA-polymerizing activity that is ca-pable of promoting replication past pyrimidinein 4X174 DNA and in synthetic template-primer, the simplest assumption is that thesame inducible activity is responsible for error-prone repair of UV-damaged bacterial DNA aswell. The same activity (or another activity ofthe same kind) could effect error-prone repairsynthesis to fill gaps such as those shown inFig. 4. However, other possibilities (236), in-cluding a minor inducible pathway of recombi-national repair, have not been ruled out.Bridges et al. (31) have concluded that DNApolymerase III is involved in bacterial UV mu-tagenesis, and that its ability to effect error-prone repair may depend upon an induciblefactor. The experiments upon which this con-clusion was based were done with a polCts de-rivative ofE. coli WP2M, in which DNA polym-erase III is temperature sensitive. UV-inducedmutations in this strain lose their photoreversi-bility progressively at the permissive tempera-ture (34°C), but not when the irradiated bacte-ria are transferred to the restrictive tempera-ture (42°C) after an initial postirradiation incu-bation for 15 min at 34°C. Bridges et al. equatethe LOP with the occurrence of error-pronepostreplication repair. Their expectation, if er-ror-prone gap-filling does not require active

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


888 WITKIN

DNA polymerase III, was that some LOP wouldoccur at the restrictive temperature, reflectingerror-prone repair of gaps formed during the 15min between UV irradiation and temperatureelevation. Since LOP stopped abruptly as soonas the temperature was raised to 420C, theyconcluded that error-prone gap-filling requiresactive DNA polymerase III, and proposed thatits ability to insert bases opposite pyrimidinedimers in the template may depend upon aninducible factor. These results are suggestive,and, if substantiated by further evidence, mayimplicate DNA polymerase III, as modified byan inducible factor (e.g., a proofreading inhibi-tor), as the agent of SOS repair of bacterialDNA.Analysis of the molecular basis of UV muta-

genesis in E. coli and some of its phages revealsthat base transitions, primarily from guanine-cytosine (GO) to adenine-thymine (AT) (3, 40a,167b), are efficiently induced. In one study ofUV mutagenesis in a single-stranded phage,the mutations induced were predominantly Cto T transitions (l0la). If polymerization pastnoninstructive UV photoproducts is the pri-mary mechanism ofUV mutagenesis in E. coli,the limited information available at the molec-ular level suggests that photoproducts involv-ing C (not necessarily pyrimidine dimers) maybe the most common premutational lesions, atleast in wild-type strains in which most pyrimi-dine dimers are excised. Even if such lesionsare C-containing dimers, requiring two nucleo-tide insertions by SOS activity without tem-plate instruction, most would produce muta-tions meeting the operational definition of sin-gle-base transitions, owing to the degeneracy ofthe genetic code, the probability that either orboth of the uninstructed insertions will be cor-rect, and the probability that only one of twoamino acids that might be affected will matter.

IMPLICATIONS OF SOS REPAIR FORCARCINOGENESIS

It is not known whether DNA damage initi-ates the induction of an SOS-like cluster offunctions, including an error-prone repair ac-tivity, in eukaryotic cells. In bacteria, most ofthe agents that are effective SOS inducers arealso known to be carcinogenic in mammals.This is true not only for such long-known muta-gen-carcinogens as ionizing radiations, UVlight, and alkylating agents. Aflatoxin B1, apotent carcinogen, induces X prophage and ismutagenic for X in E. coli K-12 (82). The carcin-ogen 4-nitroquinoline-1-oxide is UV mimetic inE. coli, producing a type of DNA damage thatis biologically equivalent to pyrimidine dimer

(110). In the Salmonella test system (2), mostmammalian carcinogens are mutagenic, andmuch of this mutagenesis requires the productof a Salmonella equivalent of the E. coli recA +gene (146). The mutagenicity of many carcino-gens in bacteria may therefore depend upon theinduction of SOS repair activity. This, ofcourse, does not imply that inducible muta-genic DNA polymerase activity is involved inmammalian carcinogenesis. Speculation to thiseffect may be useful, however, if it stimulates adeliberate search for induction of similar activ-ity in normal mammalian cells after treatmentwith carcinogens.Mutagenic DNA polymerases, which pro-

mote misincorporation of"wrong" bases in vitroin assays using synthetic template-primers,have been found in human cancer cells (204).The enzyme TDT, which polymerizes DNA byrandom "end-addition" without template in-struction (15), has also been detected in humanleukemic cells (48, 145, 187, 205). TDT is nor-mally found only in thymus, where it may actas a somatic mutator in generating diversity ofantigenic responsiveness in T cells (5). Thepresence of TDT or other error-prone DNA po-lymerase does not imply a causative role forthese enzymes in carcinogenesis, although er-rors in DNA replication have been proposed aspossibly important factors in initiation of ma-lignant transformation (e.g., references 133,163, 187). The response of bacteria to carcino-gens suggests that mammalian cells, as well,may react to DNA-damaging agents by induc-ing mutagenic DNA polymerase activity as atransient repair function.Mammalian cells are capable of repairing

photochemical damage to their DNA (46), aswell as DNA damage caused by chemical car-cinogens and mutagens (177). UV radiationproduces pyrimidine dimers in the DNA ofmammalian cells (212), and most mammaliancells, including normal human cells, are capa-ble of excising them (178). A human hereditarydisease, xeroderma pigmentosum (XP), is asso-ciated with various deficiencies in DNA repaircapacity (44, 45, 47). The clinical symptoms ofXP include extreme sensitivity to UV radiationand susceptibility to development of multipleskin carcinomas, usually the cause of earlydeath. In one type of XP, caused by an autoso-mal recessive mutation, ability to perform thefirst step of excision repair (incision) is absent(196), as in Uvr- mutants ofE. coli. Skin fibro-blast cultures taken from this type of XP pa-tient are not only much more sensitive to UVradiation than normal skin fibroblasts, butthey are also UV hypermutable at low UV flu-ences (139) as are Uvr- mutants of E. coli.

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



There is no direct evidence that the cancerproneness of XP patients is a consequence oftheir inability to excise pyrimidine dimers.However, since unexcised dimers cause hyper-inducibility for SOS functions in E. coli, itseems possible that the induction of an error-prone repair activity in response to unexcisedUV damage may account for the hypermutabil-ity of XP cells, and may also contribute to theirhigh rate of carcinogenesis. Pyrimidine dimersare known to be causative agents of neoplastictransformation in fish (95). Carcinogens induceprophage inE. coli (156a), as well as mutagenicDNA polymerase activity, and Uvr- mutantsare hyperinducible for both functions. Ifcompa-rable SOS-type activities are induced by carcin-ogens in mammalian cells, the induction of la-tent virus and of error-prone DNA polymeraseactivity should occur simultaneously, andeither or both could contribute to carcinogene-sis.

There are some indications that UV and Xradiation induce repair functions in mamma-lian cells. UV- or X-irradiated human virusesare reactivated (as in W-reactivation of bacte-riophage) when the mammalian host cells arealso exposed to X or UV radiation (13, 14, 96,135). It is not known whether this radiation-induced reactivation is accompanied by viralmutagenesis. A UV-inducible enhancement ofpostreplicative repair, inducible also by N-ace-toxy-acetylaminofluorene, has recently beendemonstrated in Chinese hamster cells (51a).

In some mammalian cell lines, including nor-mal human cells, the DNA made shortly afterUV irradiation is produced in segments of lowmolecular weight, which are elongated andjoined with further incubation to producestrands of normal molecular weight (131). Theprocess involved in this postreplication repairappears to be primarily a form ofrepair replica-tion, since recombinational exchanges have notbeen detected. The DNA synthesized on UV-irradiated templates at long times after irradia-tion (e.g., 20 h) is produced in strands ofnormalmolecular weight, even when the UV photo-products causing the initial discontinuities arestill present in the template strands (36, 132,149). Although other explanations are possible,the long time required to develop the capacityto synthesize DNA continuously on damagedtemplates could indicate induction of a type ofrepair activity (possibly error prone) that isrepressed in undamaged mammalian cells.

REGULATION OF SOS REPAIR ANDOTHER UV-INDUCIBLE FUNCTIONSThe functions activated by DNA damage con-

stitute a regulatory unit that is unusual, per-

haps unique, in E. coli. Most bacterial controlsystems regulate blocks of genes, either linked(operons) or unlinked (regulons), all of whichcode products active in the same metabolicpathway, or in closely related pathways (6,179). Regulatory molecules tend to be sub-strates, end products, or other metabolitesbearing some direct biochemical relation to thepathway itself. SOS functions, at least concep-tually, are more like the metabolically diversebut teleonomically (i.e., adaptively) relatedgroups of functions that are simultaneously ac-tivated during eukaryotic development and dif-ferentiation (34, 100). The regulation of SOSfunctions is the answer from E. coli to a typi-cally eukaryotic question: how to turn on agroup of genes simultaneously when their coor-dinate expression is beneficial under particularconditions, although their products may act inunrelated pathways. An understanding of SOSregulation may therefore be of interest in con-sidering the evolution of eukaryotic control sys-tems.

Goldthwait and Jacob (80) proposed that Xrepressor is inactivated, in lysogens treated.with prophage-inducing agents, by a precursorof DNA synthesis (possibly a derivative of ade-nine), which accumulates when DNA replica-tion is interrupted. Hertman and Luria (97),who found that recA- mutations prevent UVinduction of X prophage, suggested that exten-sive DNA degradation (a characteristic ofrecA- mutants) produces breakdown productsthat antagonize the ability of a DNA precursor-inducer to inactivate the phage repressor. Tom-izawa and Ogawa (210) concluded that the pro-duction of X inducer is a complex biochemicalpathway, requiring postirradiation protein syn-thesis.To explain the similar requirements for in-

duction of prophage in lysogenic strains and offilamentous growth of E. coli B after UV irra-diation, Witkin (226) proposed that a bacterialgene coding a septum-inhibiting protein is gov-erned by a repressor similar enough to X repres-sor to respond to the same inducer, and that thecommon inducer is synthesized when DNA rep-lication is interrupted. As additional bacterialfunctions were identified as belonging to thesame regulatory system, and as their commonrequirement for the recA + lexA + genotype wasrecognized, this model was further developed(215, 233, 235, 237, 240) to assume that all SOSfunctions have evolved repressors similarenough to respond to the same inducer, andthat the recA + and lexA + products are neces-sary for their induction and/or expression. Thetif-1 mutation was believed to cause constitu-tive synthesis at 420C of an intermediate in the

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


890 WITKIN

induction pathway, thereby bypassing the re-quirement for interruption ofDNA synthesis asthe initiating signal.

X repressor, which binds specifically to XDNA (173), is reported to bind selectively tonicked DNA, an observation that has led to thehypothesis that SOS functions are induced byagents that introduce single-strand nicks intoDNA, thereby providing sites for competitivebinding of SOS repressors (206). This modeldoes not account for thermal inducibility ofSOSfunctions in tif mutants, in which DNA isintact at the inducing temperature, or for thefailure of SOS induction in recA - and lexA -mutants when their DNA is nicked. Selectivebinding of SOS repressors to nicked DNA mayenhance induction by "soaking up" excess re-pressor molecules, but it seems inadequate toaccount for the primary derepression of SOSfunctions.The first comprehensive model of SOS regu-

lation, specifying the roles of recA and lexAgene products, was proposed by Gudas and Par-dee (89), based largely upon their study of theinduction of protein X (112, 113) in variousstrains. As recently elaborated by Gudas (88a),this model assumes that the lexA + gene productis a repressor of protein X and possibly of otherSOS functions, which may be linked in a singleoperon or scattered throughout the chromo-some. Inactivation of the lexA-coded repressorresults, directly or indirectly, in the expressionof all SOS functions. In wild-type strains, SOS-inducing conditions initiate induction by caus-ing the accumulation of breakdown products ofDNA degradation, one of which interacts withthe lexA +-coded repressor so as to sensitize it toinactivation by an antirepressor (possibly aprotease) coded by the recA gene. The possibil-ity that the recA gene may code a protease wasproposed earlier by Roberts and Roberts (180),who found that X repressor is proteolyticallycleaved following prophage-inducing treat-ments. According to the Gudas-Pardee model,the tif-1 mutation is interpreted either asa temperature-sensitive operator-constitutivemutation at the binding site of the lexA-codedrepressor or as a mutation affecting the activityof the recA gene product. Gudas (88a) considersthe latter possibility more probable, and it isconsistent with the tight linkage between tif-1and recA mutations (40). The recA product, asaltered by a tif-1 mutation, would presumablyfunction as an antirepressor at 420C, inactivat-ing the lexA + product without requiring theparticipation of a metabolite of DNA degrada-tion.A valuable aspect of this model is the pro-

posal that a DNA breakdown product may act

as an inducer in SOS regulation, a possibilitythat is supported by the failure of nalidixicacid to induce protein X synthesis in recB orrecC mutants lacking exonuclease V activity,although bleomycin, an antibiotic that pro-motes DNA degradation directly, induces pro-tein X in these strains (89). The properties ofthe dam-3 mutant (142), which lacks an ade-nine methylase and degrades its DNA chroni-cally, are also consistent with this idea, sincedam-3 strains express at least some SOS func-tions constitutively. However, there is no sim-ple correlation between DNA degradation andthermoinducibility of prophage in differentdnats mutants (188).The Gudas-Pardee model is consistent with

the dominance of lexA- mutations (159) inmerodiploids with the wild-type allele. It alsoaccounts satisfactorily for the synthesis of pro-tein X in a number ofdifferent mutant strains.However, the assumption that the lexA + geneproduct functions as a repressor of SOS func-tions, or that it controls a product required forthe induction of these functions, is not consist-ent with some other facts. A major problem isX prophage, which codes its own repressor, yetrequires the lexA+ genotype for induction bySOS-inducing treatments. Gudas (88a) ex-plains this by proposing that X induction re-quires the participation of a protein or smallmetabolite synthesized via the lexA+-con-trolled SOS induction pathway or after itsexpression. However, in tif-1 X lysogens, theinactivation of X repressor begins without alag as soon as the temperature is raised to 42°Cand proceeds without requiring new proteinsynthesis (221). Thus, in tif-1 strains all pro-teins required for thermal inactivation of Xrepressor appear to have been constitutivelysynthesized before the temperature elevation.Nevertheless, the lexA + gene product isstrictly required for thermal induction of Xprophage in tif-1 strains, and this requirementis for the actual inactivation of X repressorrather than for any subsequent step leading tocell lysis. This has been demonstrated byshowing rapid lysis of tif-1 lexA- cells lysogen-ized with a X mutant (XcI857) which codes atemperature-sensitive repressor, whereas thesame strain lysogenized with wild-type X doesnot induce its prophage at 420C (E. Witkin,unpublished observation). Many lexA- muta-tions severely depress and delay UV inductionof X prophage, and this effect, too, is exerted atthe level of X repressor inactivation (59). TheGudas-Pardee model, as well as another model(162) in which the lexA+ product is assignedthe role of repressor of some SOS functions,must assume that this gene product acts in an

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from


UV MUTAGENESIS AND DNA REPAIR IN E. COLJ 891

entirely different capacity to influence X pro-phage induction. Although this is not impossi-ble, it seems simpler to assume that the lexA+gene product participates in the regulation ofall the functions it controls in the same way.Another set of facts that must be explained

by any model for SOS regulation is the diver-sity of patterns of expression of different SOSfunctions caused by mutations that probablymap within the lexA locus, such as tsl- (158,160) and rnm- (215). For example, in tsl- mu-tants, filamentous growth (160) and synthesisof protein X (89) are expressed constitutivelyat 420C, but X prophage is not thermally induc-ible, and UV mutability (i.e., SOS repair ac-tivity) is greatly reduced compared to the wildtype (158). In rnm- mutants, no UV mutabil-ity is detectable, but exonuclease V activityafter UV irradiation is controlled to a nearlynormal extent (215), and synthesis of proteinX is noninducible (D. Spencer, personal com-munication). In rnm- mutants, synthesis ofthe inhibitor of exonuclease V is constitutive(M. Volkert, personal communication).Models assuming that the lexA gene codes thecommon repressor of SOS functions could ac-count for such "split" phenotypes only by as-suming numerous nonidentical operator sitesthat respond differentially to an altered re-pressor. The assumption that all SOS func-tions belong to a single operon (89) cannotaccount for them in any obvious way.A different type of model is based on Clark's

suggestion (41) that the lexA gene codes aregulator of the recA gene, an attempt to ac-count for the extensive copleiotropic effects ofmutations in these two genes. Mount (per-sonal communication) has developed a modelin which the quantitative level of the recA +gene product (assumed to act as an antirepres-sor that destroys SOS repressors under SOS-inducing conditions) is one of the critical fac-tors determining whether or not a strain canexpress SOS functions normally. The lexA +gene product is assumed to act as a regulatorofrecA + synthesis, resulting in a level ofrecA +product that permits normal SOS induction inwild-type strains. Mutations in the lexA genethat result in a reduced level of recA+ productcause noninducibility of all SOS functions (theLexA phenotype). Certain other mutations inthe lexA gene (spr-) are assumed to increasethe synthesis of the recA + product to an abnor-mally high level, and thereby cause constitu-tive expression of SOS functions even at 300C,if other requirements for induction are pro-vided by a tif-1 sfi- genotype. Any model inwhich the 1exA product regulates synthesis ofthe recA product, and in which the quantita-

tive level of recA product is a critical factor inSOS induction, can explain all-or-none nonin-ducibility or all-or-none constitutive expres-sion of SOS functions. Such models, however,cannot easily explain the "split" phenotypes oftsl- or rnm- mutants (described above), or thequite different pattern of expression of SOSfunctions seen in lex-113 mutants (58). For anyone lexA allele, there can be only one level ofrecA + product produced under its control,whether that level be normal, abnormallyhigh or abnormally low. Yet tsl- and rnm-mutants are noninducible for some SOS func-tions, whereas they express others normally orare constitutive for still others under the sameconditions. In lex-113 mutants, the phenotypeis like lexA- in all respects, except that fila-mentous growth is expressed constitutively.One would have to make some complicatedassumptions to explain how various lexA al-leles (which can only either raise or lower theamount of recA product synthesized accordingto this model) can produce such a variety ofpatterns of "split" phenotypes with respect toSOS functions. It should be emphasized, how-ever, that tsl-, rnm-, and lex-113 mutationshave not been definitively mapped within thelexA gene, although the linkage data (espe-cially for tsl- and rnm-) are consistent withthis possibility.Figure 6 illustrates another way of inter-

preting SOS regulation. The essential featuresof the model are as follows. (i) Each SOS func-tion expresses the activity of a separate op-eron, controlled by its own repressor. (ii) SOSrepressors are similar enough to bind thelexA+ gene product when bound to their re-spective operator sites, but are not identical.(iii) In undamaged wild-type cells, the lexA+product is bound to all SOS repressors,thereby preventing their inactivation by con-stitutive antirepressors. In Fig. 6, the antire-pressors are shown as cellular proteases, onthe assumption that proteolytic cleavage is theprimary mechanism of SOS repressor inacti-vation. In that case, the lexA+ product mayserve to block protease-sensitive sites on theSOS repressors, or it may act more positivelyas a protease inhibitor. Whatever the primarymechanism of derepression, the role of thelexA+ gene product is protection ofSOS repres-sors against inactivation in the absence of anSOS induction signal. (iv) The recA-tif com-plex (so designated because the tif-1 mutationmaps in the recA region and may affect thesame protein) is inactive in the absence of anSOS signal, in the sense that it is unable toremove the lexA+ product from its bindingwith SOS repressors. In the inactive state, the

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


BACTERIOL. REV.

I. SOS FUNCTIONS EPESD (rcAA+hArA +adaoPsd cells, lid-i at 30)

rerirm pat~ sewum SOS repair Protein Emxoncleoueditor protein X inhibitor

k\ haI 2 TWAK I2 PTEm 3 PSOTEAEO 4 PRISE 5

AexA+ pvdsd WAS ed SWS represeors protects thewr proteine-sesiivesites opint cleva by coruttive p.iamse

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

etuupmeo "~)

2. SOS FUNCTIONS IDUua. recA~tf om"ple activated by f v

UV, X-adiation MMC, etc. octw

+r*CAkTtif

DN J 0 it S activator (4*"D activaed

b. rucA+-td9- complex activatd by complexincublion at 42*C withiW dwnapto DNA (no SOS activiator required)

X proho

,/ M'POEAKI 1 POIASE I

Kactivated complexerAW prnP ct,wn

from SOS rew ss

AAAA X Propho

ibenaKmait ~~fragnmetsors of repressorffter

cleavag byprat*ee

FIG. 6. Model to account for the regulation of SOS functions in E. coli (see text for explanation andalternative versions).

recA-tif complex may be free, as shown in Fig.6, or bound to the lexA + product. In eithercase, activation of this complex, enabling it toremove the lexA + product from SOS repres-sors, occurs in tifr strains only when an SOSactivator is produced as a consequence of ar-rested DNA replication. The activator may bea precursor ofDNA replication (80) or a break-down product of DNA degradation (89), or (assuggested by H. Ginsberg) activation may beaccomplished under some conditions by a pre-cursor and under other conditions by a break-down product. In the tif-1 mutant, activationof the recA-tif complex occurs when the tem-perature is raised to 420C without requiring anactivator, or at lower temperatures when ade-nine is present (80, 123, 234). Activation of thecomplex permits its binding to the lexA + prod-uct unless it is already so bound before activa-tion. In either case, the binding of activatedrecA-tifcomplex to the lexA + product results inthe release of the lexA + product from SOS

repressors, thus exposing their protease-sensi-tive sites to proteolytic cleavage (or permit-ting their inactivation by some other constitu-tive antirepressor).The model in Fig. 6 explains currently

known facts satisfactorily and generates anumber of testable predictions. Like anymodel offered when knowledge is incomplete,it may require drastic revision or rejection asadditional facts become known, and it is pro-posed here as a working hypothesis. The domi-nance oflexA - mutations in merodiploids withlexA+, regardless of which allele is located onthe F' factor (159), is readily explained. Muta-tions causing the noninducible phenotype forall SOS functions (the LexA- phenotype), ac-cording to this model, are those eliminatingthe ability of the lexA + product to bind theactivated recA-tif complex or to be releasedfrom SOS repressors when bound to the acti-vated complex. In merodiploids, any lexA +product bound to an SOS repressor would be

892 WITKIN


http://mm

br.asm.org/

Dow

nloaded from



removed under SOS-inducing conditions, andcould be replaced by a lexA- product, whichwould remain irreversibly bound to that re-

pressor. If the inactivation of SOS repressors

and/or the expression of SOS functions is slowcompared with the rate of binding of SOS re-

pressors by the lexA product, all SOS repres-

sors would ultimately become saturated withthe irreversibly bound product in a mero-

diploid.Mutations that map in the recA region but

have a Lex- phenotype (noninducibility forSOS functions but normal recombination abil-ity), such as zab- (40) or lexB- (11), may berecA alleles affecting a part of the recA gene

product not essential for genetic recombina-tion. Both of these mutations are recessivewith respect to UV sensitivity in merodiploidswith the wild-type allele (J. George, personalcommunication). This is consistent with themodel in Fig. 6, since the noninducible pheno-type in these mutants is ascribed to changes inthe rec-tif complex, such as to prevent eitherits activation or its interaction with the lexA+product. Presence of the normal allele shouldrestore inducibility in the merodiploid. The tif-1 mutation, on the other hand, should be dom-inant, or at least partly dominant. Althoughmost recA - mutations would be expected to berecessive, partial dominance of some recA al-leles would be consistent with the model, e.g.,if the mutation caused unactivated recA prod-uct to bind irreversibly to the lexA + product.

This model can explain the variety of "split"phenotypes for SOS functions caused by muta-tions that are tightly linked to lexA- muta-tions, and that may affect the same protein.These include tsl-, rnm-, and possibly lex-113(see above). Additional examples of "split"phenotypes in derivatives of lexA- strains se-

lected for increased UV resistance have beenreported (28, 191). A single mutation in thelexA gene could cause a variety of differentpatterns of expression of SOS functions, ac-

cording to the model in Fig. 6. If the mutationpermits normal binding of the altered lexAproduct to a particular SOS repressor, but pre-

vents its response (when so bound) to an acti-vated recA-tif complex, the mutant will benoninducible for that SOS function. If thesame altered lexA product is unable to bindone or more other SOS repressors, the mutantwill express the corresponding function(s) con-

stitutively, since the repressor(s) will be ex-

posed continuously to inactivation by antire-pressor (e.g., cleavage by protease). The mu-

tant may exhibit normal inducibility of stillother SOS functions, if the altered lexA prod-uct binds normally to some SOS repressors

and responds normally to activated recA-tifcomplex when so bound. Assuming nonidenti-cal repressors, changes at different siteswithin the lexA gene could thus cause a multi-plicity of different patterns of constitutivity,inducibility, and noninducibility of the var-ious SOS functions.A prediction generated by the model in Fig.

6 is that lexA mutations that cause "split"phenotypes for SOS functions should cause acorresponding pattern of "split" dominance,such that any function that is noninducible inthe mutant should be dominant in merodip-loids with lexA +, whereas any function that isexpressed constitutively in the mutant shouldbe recessive. Insofar as the dominance/reces-siveness of individual components of "split"phenotypes has been determined, the resultsagree with this prediction. In tslh mutants,low UV mutability (i.e., poor inducibility ofSOS repair activity) is dominant, but constitu-tive filamentous growth at 420C is recessive inmerodiploids with the wild-type allele (158). Inthe lex-113 mutant (which is closely linked tolexA but not unambiguously located withinthe lexA locus) the constitutive filamentousgrowth of the mutant is recessive in merodip-loids with the wild-type allele (J. Donch, per-sonal communication), but its UV sensitivityand UV nonmutability (i.e., noninducibility ofSOS repair activity) are dominant (86). Theconstitutive filamentous growth of the lex-113mutant is suppressed by a lexA - mutation(59), an observation that will add further sup-port to the model if these mutations prove tobe allelic.Another prediction based on this model is

that the phenotype of a deletion mutant lackingthe lexA gene should be constitutive for all SOSfunctions, whereas the phenotype of a recAdeletion mutant should be noninducible for allSOS functions, assuming that such mutantsare viable. If recessiveness is interpreted asindicative of loss of function of the type onemight expect of a deletion mutant, the reces-siveness of noninducible zab- and lexB- muta-tions indicates that the positive action of a func-tional recA allele is necessary for SOS induc-tion, whereas the recessiveness of the constitu-tive filamentous growth in tsl- (and possibly inlex-113) mutants indicates that the positive ac-tion of a functional lexA allele prevents induc-tion.The model in Fig. 6 can also explain the

unique phenotype of the double mutant tsl-recA- (108, 158a, 161, 162), which is differentwith respect to some SOS functions from eitherof the single mutants. The double mutant isunlike any other recA- strain in its UV resist-

VOL. 410, 1976


http://mm

br.asm.org/

Dow

nloaded from


894 WITKIN

ance and ability to perform some error-freepostreplication repair (in Uvr+ strains) (158a,161, 162), and in its constitutive synthesis ofprotein X (88a) and filamentous growth at 420C(161, 162). The tsl- recA- strain differs from thetsl- recA + mutant in its inability to controlDNA degradation after UV irradiation (i.e., itsfailure to induce the inhibitor of exonuclease V)and in its inability to produce any UV-inducedmutations (158, 158a). The phenotype of thedouble mutant suggests an interaction betweenthe recA and lexA gene products such that mu-tations in each can affect the role ofthe other inSOS induction. An interaction of this type isexpected in the model in Fig. 6. Devoret (per-sonal communication) has proposed a model inwhich the recA and lexA gene products form acomplex which acts as a "vise" or clamp, detect-ing SOS repressors after SOS-inducing treat-ments and positioning them so as to permittheir inactivation by an antirepressor. Such amodel (if one assumes different repressors forSOS functions) could also account for "split"phenotypes and for unique phenotypes in dou-ble mutants, but could not easily account forthe apparently invariable dominance of lexA-alleles, unless it is assumed that complete inac-tivation of the lexA product is lethal.A final aspect of the model in Fig. 6 is that it

assigns the primary repressor inactivation toproteases coded by genes outside the recA andlexA loci. The proteases are shown as differentfor each SOS repressor. This degree of diversityis not a necessary feature of the model, al-though more than one antirepressor must bepostulated. If there were only one SOS antire-pressor, there should be a third locus (the locuscoding the antirepressor) in which mutationscould eliminate inducibility of all SOS func-tions. Since no such mutations have been foundunlinked to recA and lexA, it is assumed thatmore than one protease is required to inactivateall SOS repressors.A variation of this model, which fits the facts

equally well, could operate if the lexA+ productfunctions actively as a protease inhibitor,rather than as a passive protector of protease-sensitive sites on SOS repressors. In that case,the lexA + product need not be assumed to inter-act with SOS repressors, but could be boundinstead to the antirepressor proteases in theuninduced state, and its inactivation by acti-vated recA-tif complex could result in therelease of proteases and in the cleavage of SOSrepressors. "Split" phenotypes exhibited bytslh, rnm- and other mutants believed to alterthe lexA product would then be due to specific-ity of interaction of the altered gene productwith different proteases. Although not neces-

sarily requiring that every SOS repressor besusceptible to cleavage by a different protease,this version of the model would require thatmore than one protease have the capacity tocleave SOS suppressors.A puzzling set of observations concerns the

need for new protein synthesis after variousSOS-inducing treatments in order to inactivateX repressor in various lysogenic strains. Thepresence of chloramphenicol during incubationat 420C does not prevent repressor inactivationin tif-1 (221) or dnaBts (166) lysogens, nor doeschloramphenicol, added immediately aftergamma irradiation, prevent repressor inactiva-tion in wild-type lysogens (221). After treat-ment with MMC or UV radiation, however, Xrepressor is not inactivated in the presence ofchloramphenicol (198, 210), indicating that newprotein synthesis is needed to accomplish de-repression of prophage when these inducers areused. These results imply that different SOS-inducing treatments are not alike in the man-ner in which they generate an effective SOS-inducing signal. Clarification of this questionInay be crucial for evaluation ofSOS regulationmodels.

PROTEASES AND THE EXPRESSION OFSOS FUNCTIONS

The initial basis for considering protease ac-tivity as a possible primary SOS repressor-inac-tivating event was the proteolytic cleavage of Xrepressor that occurs when X prophage is in-duced, although it is not known whether thiscleavage is the cause or the consequence ofrepressor inactivation (180). The specificity ofproteases and their importance in both bacte-rial and mammalian cell economy are receivingincreasing attention (79, 79a, 150, 178a, 207).A strong reason for invoking protease activ-

ity as a major factor in modulating the expres-sion ofSOS functions, as well as in their induc-tion, is the possibility that the ion gene, muta-tions in which cause extreme filamentousgrowth in response to SOS-inducing treatments(1, 104), may code a protease or a product thatregulates protease activity. The Ion gene is ap-parently identical with the gene degT (199).Mutations in lonldegT greatly increase the sta-bility of specific polypeptide fragments thatwould otherwise be rapidly degraded, as well asthe stability of some other proteins (79, 79a).Mutations designated degR have been reportedto alter the stability of different polypeptides,suggesting a unique pattern of specificity (2a),although it is apparently not yet certainwhether degT and degR are in fact distinct loci(79a). The products of the deg gene(s) have notbeen identified, but they could be proteases,

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



protease inhibitors, or products that otherwiseregulate selectively the rate of degradation ofvarious proteins. The pleiotropy of lon/degTmutations includes (among others) effects onthe rate of transcription of the gal operon (138),on the synthesis of capsular polysaccharides(37), and on the capacity to be lysogenized by Xphage (217). All of these effects could be theconsequence of altered stability of a variety ofmetabolically unrelated gene products due to alonldegT mutation affecting (directly or indi-rectly) the specificity and activity of a particu-lar protease or protease inhibitor. Sfi- muta-tions (suppressors of filamentous growth),which eliminate the filamentous growth ex-pressed at 4200 by tif-1 mutants (75, 238), mayalso affect the structure or activity of proteasesor protease inhibitors, imposing a new patternof stability upon numerous proteins. Independ-ent Sfi mutations which suppress the filamen-tous growth of the ion- wild-type strain E. coliB may or may not also alter sensitivity to peni-cillin, and at least one such mutation (responsi-ble for the nonfilamentous phenotype of strainB/r) radically changes the kinetics of cell divi-sion and death in the stationary phase ofgrowth (222).A change in the half-life of most proteins in

E. coli may not cause a strikingly obviouschange in the phenotype, as long as the proteindoes not become so unstable that it is essen-tially unable to perform its function. Alteredstability of the proteins synthesized in responseto an SOS induction signal, however, coulddrastically alter the expression of at least someSOS functions. SOS operons are transcribedonly during a limited period oftime while SOS-inducing conditons prevail, and rapid degrada-tion of their products would, a priori, be ex-pected. This expectation is consistent with the30-min half-life of SOS repair activity (52, 238)and with the existence of a class of unstableproteins in bacteria with half-lives of 20 to 60min (79a). When UV radiation is the inducingagent, transcription ofthese operons is possibleonly until DNA repair is completed and normalDNA replication resumes. The length of thefilaments produced after exposure of E. coli B(a ion mutant) to UV radiation is strictly pro-portional to the amount of protein synthesisthat is allowed to occur before DNA repair iscompleted, an observation that led to the hy-pothesis that the duration ofthe delay in septa-tion is quantitatively related to the amount of aseptum-inhibiting protein synthesized duringthe limited period of derepression (226). Georgeet al. (75) have pointed out that Ion mutationscould cause extreme filamentous growth in re-sponse to SOS-inducing agents either by stabi-

lizing the postulated septum inhibitor (if lonidegT codes a protease) or by increasing the rateof its synthesis. Mutations affecting proteaseactivity and/or specificity could mimic regula-tory mutations by affecting the stability of reg-ulatory proteins. The apparent regulatory roleof the ion product in the galactose operon (138)may exemplify the difficulty of distinguishingbetween mutations directly affecting regula-tory protein structure and those affecting theirmetabolism (6).The limited and often very brief period dur-

ing which SOS operons are derepressed makesthe expression of SOS functions particularlysusceptible to mutations that increase or de-crease rates of transcription, rates of transla-tion, stability ofmRNA or stability of proteins.To the extent that such mutations act pleiotrop-ically on specific spectra of substrates, theymay amplify or diminish the expression of dif-ferent SOS functions in a variety of patterns.Mutations at any ofthese levels, perhaps exem-plified by ion and sfi mutations, may not only'fine-tune" the expression of SOS functions butcould also represent a means of generatingasexually, via a single mutation, a rich diver-sity of phenotypic variation. Their importancein evolution should not be underestimated.

CONCLUSIONSUV-induced mutations in E. coli and some of

its phages are caused primarily by inaccuraterepair of UV-damaged DNA. Most of the UVdamage produced in the DNA of surviving bac-teria is repaired by relatively "error-proof"mechanisms (e.g., photoreactivation, "short-patch" excision repair, the major pathways ofrecombinational postreplication repair), whichdo not contribute substantially to UV mutagen-esis. Some kinds of DNA damage (probablysingle-strand gaps not repairable by any consti-tutive accurate mechanism) are targets for theactivity of inducible "error-prone" repair activ-ity ("SOS" repair), which is entirely responsiblefor UV mutagenesis inE. coli and in X bacterio-phage. SOS repair may participate in two mi-nor DNA repair pathways: a "long-patch" exci-sion repair and chloramphenicol-sensitive post-replication repair.SOS repair activity is repressed in undam-

aged wild-type cells, but is induced in responseto UV radiation and other agents (includingmany mutagens and carcinogens) that damageDNA or interrupt its replication. Such treat-ments generate a regulatory signal (the "SOS"signal), which initiates a complex inductionprocess culminating in the derepression of agroup of metabolically diverse but coordinatelyregulated functions ("SOS" functions), all of

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


896 WITKIN

which presumably promote the survival of thedamaged cell or that of its phages. Includedamong SOS functions, in addition to error-prone DNA repair activity, are prophage induc-tion, cell division delay (sometimes leading tofilamentous growth), inhibition of the DNA-degrading activity of exonuclease V and "aber-rant" reinitiation ofDNA synthesis at the chro-mosomal origin. Expression of these and otherinducible SOS functions requires the recA + andlexA + gene products and new protein synthesisduring or after the inducing treatment. Thenature of the SOS induction signal is not yetunderstood, nor is the mode of regulation ofSOS functions.At least one kind of SOS repair activity in-

duced in E. coli has been identified as a DNA-polymerizing activity that is distinguishablefrom the activities of the constitutive DNA po-lymerases I, II, and III by its ability to polymer-ize DNA past pyrimidine dimers in the tem-plate strand, with a high probability of error.UV mutagenesis in E. coli and X bacteriophagemay be due primarily to the insertion of"wrong" bases by this UV-inducible SOS repairactivity as it replicates DNA past noninstruc-tive UV photoproducts, which are not necessar-ily pyrimidine dimers. The inducible compo-nent of this SOS repair system need not be anew DNA polymerase, but may be a factor thatinhibits the 3'-5'-exonuclease "proofreading"activity of one or more of the constitutive DNApolymerases, or one that relaxes their templatedependence in some other way. It is possiblethat more than one type of SOS repair activityis induced in E. coli. In that case, some or all ofthe UV mutagenesis occurring in bacterial andin X phage DNA could be due to error-pronerepair by a minor inducible pathway of recom-binational repair, although there is no evidencethat such a system is induced.Anomalous induction of SOS repair activity

in undamaged cells containing intact DNA ca-pable of replication results in a mutator effect,i.e., greatly increased generalized spontaneousmutation rates. SOS repair activity in crudeextracts of UV-irradiated E. coli also promotesmisincorporation of bases during replication ofunirradiated synthetic template-primer DNA.Both in vivo and in vitro, however, the pres-ence of UV photoproducts in the DNA greatlyincreases the mutagenic effect of SOS repairactivity.The kinetics of induction and decay of SOS

repair activity are similar, whether the error-prone activity is assayed by its capacity to en-hance bacterial UV mutagenesis or to promoteX UV mutagenesis in UV-irradiated hosts. Inboth types of assay, the half-life of SOS repair

activity is about 30 min. These observations areconsistent with the possibility that the sameinducible error-prone DNA polymerase activityis responsible for both bacterial and phage UVmutagenesis.The induction of at least some SOS functions

by UV radiation requires the absorption of twophotons. UV-induced mutations, however, maybe caused either by single UV photoproducts orby the cooperative interaction of two UV photo-products. All or nearly all detectable UV-in-duced mutations occur before the second postir-radiation DNA replication. In wild-typestrains, some UV-induced mutations occur be-fore DNA replication via SOS repair of gapsresulting from excision of UV photoproducts;others occur during or after the first postirra-diation DNA replication via SOS repair of dam-age (probably gaps in the daughter strands)caused by unexcised UV photoproducts passingthrough a replication fork.There is some evidence that mutagens and

carcinogens induce DNA repair activities inmammalian cells. It is not known whetherthese activities are error-prone or whether theyare induced in coordination with other SOS-likefunctions in resonse to DNA damage. If so, atleast two such functions (induction of latentvirus and of error-prone DNA repair activity)could contribute to carcinogenesis.

APPENDIX: THE CELLDENSITY ARTIFACT

Many studies of bacterial mutagenesis areflawed by an artifact that has not been gener-ally recognized. The problem stems from thecommon assumption that percentage of sur-vival, after termination of a more or less lethaltreatment, is independent of the cell density atwhich the bacteria are subsequently plated.The "cell density artifact" can be illustrated

by describing a protocol widely used to measurethe mutagenic effect of UV radiation in anauxotrophic (e.g., Trp-) strain, scoring muta-tions to prototrophy (Trp+). This kind of sys-tem, when used properly (3), allows accuratecalculation of the frequency of induced muta-tions per survivor. A key feature ofthe auxotro-phy to prototrophy system is the use of thesame medium for assaying auxotrophic sur-vival and for selecting prototrophic mutants.The medium used (SEM) is a minimal agarpartially enriched with nutrient broth (1 to 5%,vol/vol). When a highly diluted portion of aTrp- culture (100 to 300 cells) is plated on SEM,the tryptophane in the supplement allows thissmall number of cells to undergo many celldivisions, so that each cell forms a small butvisible colony before its growth is arrested by

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



exhaustion of the medium for the requiredgrowth factor. When a vary large population ofTrp- cells (107 to 108) is plated on the samemedium, the required amino acid is used upafter relatively few divisions, and the auxo-trophic population forms a confluent but barelyvisible lawn. Any Trp+ mutants either presentin the inoculum or arising during the divisionson the plate can continue to grow to form largecolonies. This system was originally developedby Demerec and Cahn (54a).The use ofSEM medium in the auxotrophy to

prototrophy system automatically satisfiesthree major requirements of sound methodol-ogy for bacterial mutagenesis. (i) Survival andmutation are scored on the same medium, thuseliminating any medium effect on survival as apossible source of error. (ii) The mutagen-treated population undergoes a few cell divi-sions before stringent selection for the proto-trophic phenotype begins, permitting pheno-typic expression of any newly induced muta-tions. This requirement is not met when muta-gen-treated auxotrophs are plated directly onminimal agar, since the phenotype of potentialmutants remains auxotrophic until new jnfor-mation in the DNA can be transferred to thegene product. (iii) The treated population isallowed to undergo the divisions required forphenotypic expression while immobilized onsolid medium, thus preserving the essential 1:1relation between the number of mutant colo-nies counted (above the number produced byunirradiated controls) and the number of muta-tions induced. Allowing phenotypic expressionto occur during a period of liquid cultivationlong enough to guarantee complete expressionpermits division of induced mutants beforeplating, and makes accurate determination ofmutation frequency impossible.

Before UV irradiation and after each incre-ment of radiation exposure, both diluted andundiluted aliquots of the Trp- cell suspensionare plated on SEM agar. The diluted samplesare used to calculate percent survival for eachUV fluence. The undiluted samples are used toenumerate Trp+ colonies after 2 to 3 days ofincubation. The frequency of UV-induced mu-tations is then calculated, after subtracting thenumber of Trp+ colonies on plates seeded withunirradiated controls. (The number of Trp+ col-onies on control plates depends primarily uponthe final size of the plated population, andtherefore is constant over a wide range of inocu-lum sizes, at a given level of nutrient brothenrichment.) Any excess of Trp+ colonies overthe number on control plates represents thenumber of Trp+ mutations induced by UV ra-diation in the surviving fraction of the plated

population. It is usually taken for granted thatpercentage of survival, as indicated by dilutedplatings, is the same as percentage of survivalin the undiluted populations plated to selectTrp+ mutants. The assumption that percentageof survival is independent of cell density on thepostirradiation plating medium has beenshown to be invalid for UV-treated populationsof the lon- strain B, but is valid for one of itsnonfilamentous derivatives, strain B10/r (227),and is also valid for strain B/r over a widerange of cell densities (E. Witkin, unpublishedobservation).Under some conditions of postirradiation cul-

tivation, strain B, like lon- mutants of K-12,forms long filamentous cells and is conse-quently UV sensitive (104, 222). Under differ-ent postirradiation conditions, however, thesebacteria divide normally and are UV resistant(lb, lc, 2a, 226). Survival after UV irradiation,or after treatment with other SOS-inducingagents, is not irreversibly determined in lon-strains at the time the treatment is terminated,but can be influenced subsequently over severalorders of magnitude by manipulating the com-position of the medium, the temperature of in-cubation, or the cell density. Filamentousgrowth itself appears to be lethal unless celldivision is resumed before a certain "criticallength" is reached (117a), and is the cause oftheUV sensitivity ofE. coli B and of lon- mutantsof K-12 to most SOS-inducing agents. High celldensity is one ofthe factors (among others) thatcan prevent extreme (i.e., lethal) filamentousgrowth by promoting early resumption of nor-mal cell division and radiation resistance inlon- strains, a phenomenon described as"neighbor restoration" (1, 53a) or "crowding re-covery" (227). Thus, the "cell density artifact"applies to any strain treated with an agent thatcauses extensive filamentous growth under theconditions used to determine percentage of sur-vival, but not under the conditions used to se-lect mutants. Under these conditions, grossoverestimation of the frequency of induced mu-tations per survivor results from extrapolatingpercentage of survival from diluted to undi-luted platings. The actual size of the viablepopulation from which the observed number ofinduced mutations is arising may be 100 timeslarger than estimated, if filamentous growthkills 99% of the cells plated at low density, butkills none of the cells plated at high density.The cell density artifact due to filamentous

growth can sometimes apply to lon+ strains,such as K-12, 15TAU, and others that are asUV-resistant as strain B/r. E. coli K-12 wildtype does not form long filaments after UVirradiation, and the cell density artifact proba-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


898 WITKIN

bly does not affect studies ofUV mutagenesis inthis or other lon+ strains. However, lon+ UV-resistant strains are capable of producing longfilamentous cells, and may well be subject tothe cell density artifact, after relatively pro-longed starvation for thymine (243a), or afterincubation for 1 h or more at 420C in tif-1 (39,122), dnaBts (99a), or lig's (72a) derivatives.Thus, lon+ K-12 and similar strains appear tobe capable of undergoing "filamentous death"after some SOS-inducing treatments. In con-trast, strain B/r does not form filaments whenits dnaB's or tif-1 derivatives are incubated at420C, and seems to be immune to the cell den-sity artifact under any condition examined (E.Witkin, unpublished observations).

Microscopic observation of treated bacteria,incubated for several hours at relatively highand low cell densities after an SOS-inducingtreatment, can often indicate whether the celldensity artifact is likely to operate. The prob-lem can be avoided entirely by the use of amutation scored without selection, such as mu-tations from Lac+ to Lac- detected on indicatormedium, since survival and mutations arescored on the same plates at a single cell den-sity. This kind of system is slow and expensive,however, since large numbers of plates are re-quired to obtain accurate mutation frequencydata. The use of selected mutants (auxotrophyto prototrophy, streptomycin sensitivity to re-sistance) can provide reliable mutation fre-quency data in strains not subject to filamen-tous growth after the treatment used. Even instrains capable of undergoing lethal filamen-tous growth at low cell density after a particu-lar treatment, accurate induced mutation fre-quencies can be obtained by limiting the treat-ment to levels still permitting 100% survival asdetermined by plating dilutions, or by plate-wash assays (227, 234) of cells harvested fromdensely populated plates after an incubationlong enough to allow DNA replication but notcell division. Many SOS-inducing treatmentsare demonstrably mutagenic even in this highsurvival range, at least in strains exhibiting a"shoulder" in their survival curves. Inducedmutations scored in the 100% survival range oftreatment can be assumed to have arisen fromthe whole plated population (frequency per sur-vivor = frequency per bacterium plated). Thus,in lon+ K-12 and similar strains, mutagenesisby treatments such as thymine starvation or bytemperature elevation in dnaBts, ligs or tif-1derivatives is likely to be highly overestimated,if indeed it is demonstrated at all, unless treat-ment is limited to levels that are sublethal insurvival assays.

ACKNOWLEDGMENTSMost of the work described in papers by the au-

thor and co-workers cited in this review was sup-ported by Public Health Service grant AI-10778 fromthe National Institute of Allergy and Infectious Dis-eases.

ADDENDUM IN PROOFS. Meyn, T. Rossman, and W. Troll (Proc. Natl.

Acad. Sci U.S.A., in press) have shown that tif-1-mediated thermal inactivation of X repressor is pre-vented by a protease inhibitor, antipain, which alsoblocks the induction and/or expression ofSOS repairactivity and filamentous growth. Antipain does notpromote uncontrolled DNA degradation after UVirradiation of wild-type cells, however, and there-fore does not seem to interfere with induction of theinhibitor of exonuclease V. These observations areconsistent with the model of SOS regulation in Fig.6, but do not support models in which a single pro-tease is the antirepressor for all SOS functions.

Studies of merodiploids tif-1 Itifr and tif-1 IrecA (J.George, personal communication) indicate that tif-1and tifr are codominant and that tif-1 and recAmutations probably affect the same gene product.These observations support the version of the SOSregulation model in Fig. 6 in which the wild-typerec-tif complex (probably a single protein) is boundto the lexA+ product in the uninduced state (seetext). Complete dominance of the tif-1 allele wouldbe expected if the recA-tif complex were free, asshown in Fig. 6.

LITERATURE CITED1. Adler, H. I., W. D. Fisher, A. A. Hardigree,

and G. S. Stapleton. 1966. Repair of radia-tion-induced damage to the cell divisionmechanism of Escherichia coli. J. Bacteriol.91:737-742.

la. Adler, H. I., and A. A. Hardigree. 1964. Anal-ysis of a gene controlling cell division andsensitivity to radiation in Escherichia coli. J.Bacteriol. 87:720-726.

lb. Adler, H. I., and A. A. Hardigree. 1965.Postirradiation growth, division and recov-ery in bacteria. Radiat. Res. 25:92-102.

ic. Alper, T., and N. E. Gillies. 1958. RestorationofEscherichia coli strain B after irradiation:its dependence on suboptimal growth condi-tions. J. Gen. Microbiol. 18:461-472.

2. Ames, B. N., W. E. Durston, E. Yamasaki,and F. D. Lee. 1973. Carcinogens are muta-gens: a simple test system combining liverhomogenates for activation and bacteria fordetection. Proc. Natl. Acad. Sci. U.S.A.70:2281-2285.

2a. Anderson, E. H. 1951. Heat reactivation ofultraviolet inactivated bacteria. J. Bacteriol.61:389-394.

2b. Apte, B. N., H. Rhodes, and D. Zipser. 1975.Mutation blocking the specific degradationof reinitiation polypeptides in E. coli. Nature(London) 257:329-331.

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



3. Auerbach, C. 1976. Mutation research. HalstedPress, New York.

4. Bachmann, B., K. B. Low, and A. L. Taylor.1976. Recalibrated linkage map of Esche-richia coli K-12. Bacteriol. Rev. 40:116-167.

5. Baltimore, D. 1974. Is terminal deoxynucleoti-dyl transferase a somatic mutagen in lym-phocytes? Nature (London) 248:409-411.

6. Beckwith, J., and P. Rossow. 1974. Analysis ofgenetic regulatory mechanisms. Annu. Rev.Genet. 8:1-13.

7. Benbow, R., A. Zuccarelli, and R. L. Sinshei-mer. 1974. A role for single-strand breaks inbacteriophage 4)X174 genetic recombination.J. Mol. Biol. 88:629-651.

8. Beukers, R., and W. Berends. 1960. Isolationand identification of the irradiation productof thymine. Biochim. Biophys. Acta 41:550-551.

9. Billen, D. 1969. Replication of the bacterialchromosome: location of new initiation sitesafter irradiation. J. Bacteriol. 97:1169-1175.

10. Blanco, M., and R. Devoret. 1973. Repairmechanisms involved in prophage reactiva-tion and UV reactivation of UV-irradiatedphage X. Mutat. Res. 17:293-305.

11. Blanco, M., A. Levine, and R. Devoret. 1975.lexB: a new gene governing radiation sensi-tivity and lysogenic induction in Escherichiacoli K12, p. 379-382. In P. Hanawalt and R.B. Setlow (ed.), Molecular mechanisms forthe repair of DNA. Plenum Press, NewYork.

12. Bockrath, R., and M. K. Cheung. 1975. Therole of nutrient broth supplementation inUV mutagenesis of E. coli. Mutat. Res.19:23.

13. Bockstahler, L. E., C. D. Lytle, J. E. Stafford,and K. F. Haynes. 1976. Ultraviolet en-hanced reactivation of a human virus: effectof delayed infection. Mutat. Res. 35:189-198.

14. Bockstahler, L. E., and C. D. Lytle. 1971. X-ray enhanced reactivation of UV-irradiatedhuman virus. J. Virol. 8:601-602.

15. Bollum, F. J. 1974. Terminal deoxynucleotidyltransferase, p. 145-171. In P. D. Boyer (ed.),The enzymes, vol. 1, 3rd ed. Academic PressInc., New York.

16. Bonura, T., and K. C. Smith. 1975. Quantita-tive evidence for enzymatically inducedDNA double-strand breaks as lethal lesionsin UV-irradiatedpol+ andpolI strains ofE.coli K-12. Photochem. Photobiol. 22:243-248.

17. Bonura, T., and K. C. Smith. 1975. Enzymaticproduction of deoxyribonucleic acid double-strand breaks after ultraviolet irradiation ofEscherichia coli K-12. J. Bacteriol. 121:511-517.

18. Borek, E., and A. Ryan. 1958. The transfer ofirradiation-elicited induction in a lysogenicorganism. Proc. Natl. Acad. Sci. U.S.A.44:374-377.

19. Borek, E., and A. Ryan. 1973. Lysogenic in-duction. In J. Davidson and W. Cohn (ed.),

Progess in nucleic acid research and molecu-lar biology, vol. 13. Academic Press Inc.,New York.

20. Boyce, R. P., and P. Howard-Flanders. 1964.Release of ultraviolet light-induced thyminedimers from DNA inE. coli K-12. Proc. Natl.Acad. Sci. U.S.A. 51:293-300.

21. Boyle, J. M., M. C. Paterson, and R. B. Set-low. 1970. Excision-repair properties of anEscherichia coli mutant deficient in DNApolymerase. Nature (London) 226:708-710.

22. Braun, A., and L. Grossman. 1974. An endonu-clease from Escherichia coli that acts prefer-entially on UV-irradiated DNA and is absentfrom the uvrA and uvrB mutants. Proc. Natl.Acad. Sci. U.S.A. 71:1838-1842.

23. Bresler, S. E. 1975. Theory of misrepair muta-genesis. Mutat. Res. 29:467-474.

24. Bresler, S. E., M. I. Mosevitsky, and L. G.Vyacheslavov. 1973. Mutations as possiblereplication errors in bacteria growing underconditions of thymine deficiency. Mutat.Res. 19:281-293.

25. Bridges, B. A. 1966. A note on the mechanismof UV mutagenesis in Escherichia coli. Mu-tat. Res. 3:273-279.

26. Bridges, B. A. 1969. Mechanisms of radiationmutagenesis in cellular and subcellular sys-tems. Annu. Rev. Nucl. Sci. 19:139-178.

27. Bridges, B. A., R. E. Dennis, and R. J. Mun-son. 1967. Differential induction and repairof ultraviolet damage leading to true rever-sions and external suppressor mutations ofan ochre codon in Escherichia coli B/rWP2.Genetics 57:897-908.

28. Bridges, B. A., W. J. H. Gray, M. H. L. Green,M. A. Rothwell, and S. G. Sedgwick. 1973.Genetic and physiological separation of therepair and mutagenic functions of the exrAgene inEscherichia coli. Genetics 73(Suppl.):123-129.

29. Bridges, B. A., J. Law, and R. J. Munson.1968. Mutagenesis inEscherichia coli. 11. Ev-idence for a common pathway for mutagene-sis by ultraviolet light, ionizing radiationand thymine deprivation. Mol. Gen. Genet.103:266-273.

30. Bridges, B. A., and R. Mottershead. 1971.RecA+-dependent mutagenesis occurring be-fore DNA replication in UV- and y-irradi-ated Escherichia coli. Mutat. Res. 13:1-8.

31. Bridges, B. A., R. P. Mottershead, and S. G.Sedgwick. 1976. Mutagenic DNA repair inEscherichia coli. HI. Requirement for a func-tion of DNA polymerase III in ultraviolet-light mutagenesis. Mol. Gen. Genet. 144:53-58.

32. Bridges, B. A., and R. J. Munson. 1966. Exci-sion-repair of DNA damage in an auxo-trophic strain of E. coli. Biochem. Biophys.Res. Commun. 22:268-273.

33. Bridges, B. A., M. A. Rothwell, and M. H. L.Green. 1973. Repair processes and dose-re-sponse curves in ultraviolet mutagenesis of

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


900 WITKIN

bacteria. An. Acad. Bras. Cienc. 45(Suppl):203-209.

34. Britten, R. J., and E. H. Davidson. 1969. Generegulation for higher cells: a theory. Science165:349-357.

35. Brooks, K., and A. J. Clark. 1967. Behavior ofX bacteriophage in a recombination deficientstrain ofEscherichia coli. J. Virol. 1:283-293.

36. Buhl, S. N., R. B. Setlow, and J. D. Regan.1973. Recovery of the ability to synthesizeDNA in segments of normal size at longtimes after ultraviolet irradiation of humancells. Biophys. J. 13:1265-1275.

37. Bush, J. W., and A. Markovitz. 1973. The ge-netic basis for mucoidy and radiation sensi-tivity in capR (lon) mutants ofE. coli K-12.Genetics 74:215-225.

39. Castellazzi, M., J. George, and G. Buttin.1972. Prophage induction and cell division inE. coli. I. Further characterization of thethermosensitive mutation tif-1 whose expres-sion mimics the effect of UV irradiation.Mol. Gen. Genet. 119:139-152.

40. Castellazzi, M., J. George, and G. Buttin.1972. Prophage induction and cell division inE. coli. II. Linked (recA zab) and unlinked(lex) suppressors of tif-1-mediated inductionand filamentation. Mol. Gen. Genet.119:153-174.

40a. Cheung, M. K., and R. C. Bockrath. 1970. Onthe specificity of UV mutagenesis in E. coli.Mutat. Res. 10:521-523.

41. Clark, A. J. 1971. Toward a metabolic interpre-tation of genetic recombination ofE. coli andits phages. Annu. Rev. Microbiol. 25:437-464.

42. Clark, A. J. 1973. Recombination deficient mu-tants of E. coli and other bacteria. Annu.Rev. Genet. 7:67-86.

43. Clark, A. J., and A. D. Margulies. 1965. Isola-tion and characterization of recombination-deficient mutants ofE. coli K-12. Proc. Natl.Acad. Sci. U.S.A. 53:451-459.

44. Cleaver, J. E. 1968. Defective repair replicationof DNA in xeroderma pigmentosum. Nature(London) 218:652-656.

45. Cleaver, J. E. 1969. Xeroderma pigmentosum:a human disease in which an initial stage ofDNA repair is defective. Proc. Natl. Acad.Sci. U.S.A. 63:428-435.

46. Cleaver, J. E. 1974. Repair processes for photo-chemical damage in mammalian cells, p. 1-75. In J. T. Lett, H. Adler, and M. R. Zelle(ed.), Advances in radiation biology, vol. 4.Academic Press Inc., New York.

47. Cleaver, J. E., and D. Bootsma. 1975. Xero-derma pigmentosum: biochemical and ge-netic characteristics. Annu. Rev. Genet.9:19-38.

48. Coleman, M. S., J. J. Hutton, P. DeSimone,and F. J. Bollum. 1974. Terminal deoxynu-cleotidyl transferase in human leukemia.Proc. Natl. Acad. Sci. U.S.A. 71:4404-4408.

49. Copper, P. K., and P. C. Hanawalt. 1972. Het-erogeneity of patch size in repair replication

DNA in Escherichia coli. J. Mol. Biol. 67:1-10.

50. Cooper, P. K., and P. C. Hanawalt. 1972. Roleof DNA polymerase I and the rec system inexcision repair in Escherichia coli. Proc.Natl. Acad. Sci. U.S.A. 69:1156-1160.

51. d'Ambrosio, S. M., and R. B. Setlow. 1976.Enhancement of postreplicative repair inChinese hamster cells. Proc. Natl. Acad. Sci.U.S.A. 73:2396-2400.

52. Defais, M., P. Caillet-Fauquet, M. S. Fox, andM. Radman. 1976. Induction kinetics of mu-tagenic DNA repair activity in E. coli follow-ing ultraviolet irradiation. Mol. Gen. Genet.148:125-130.

53. Defais, M., P. Fauquet, M. Radman, and M.Errera. 1971. Ultraviolet reactivation andultraviolet mutagenesis of X in different ge-netic systems. Virology 43:495-503.

53a. Delaporte, B. 1956. La "restauration par vois-inage" chez des bacteries irradies par desrayons X. Ann. Inst. Pasteur 91:727-735.

54. De Lucia, P., and J. Cairns. 1969. Isolation ofan E. coli strain with a mutation affectingDNA polymerase. Nature (London) 224:1164-1166.

54a. Demerec, M., and E. Cahn. 1953. Studies ofmutability in nutritionally deficient strainsofEscherichia coli. J. Bacteriol. 65:27-36.

55. Devoret, R., M. Blanco, J. George, and M.Radman. 1975. Recovery of phage X fromultraviolet damage, p. 155-171. In P. Hana-walt and R. B. Setlow (ed.), Molecular mech-anisms for repair of DNA, part A. PlenumPress, New York.

56. Devoret, R., and J. George. 1967. Inductionindirecte du prophage X par le rayonnementultraviolet. Mutat. Res. 4:713-734.

57. Donch, J. J., and J. Greenberg. 1974. The effectof lex on UV sensitivity, filament formationand X induction in lon mutants of Esche-richia coli. Mol. Gen. Genet. 128:277-281.

58. Donch, J. J., and J. Greenberg. 1976. Suppres-sion of filamentation in a new lex mutant bya linked (lexA) mutation in Escherichia coli.Mutat. Res. 34:533-538.

59. Donch, J., J. Greenberg, and M. H. L. Green.1970. Repression of induction by u.v. of Xphage by exrA mutations in E. coli. Genet.Res. 15:87-97.

60. Doubleday, 0. P., B. A. Bridges, and M. H. L.Green. 1975. Mutagenic DNA repair inEsch-erichia coli. II. Factors affecting loss of pho-toreversibility of UV induced mutations.Mol. Gen. Genet. 140:221-230.

61. Doudney, C. 0. 1975. The two lesion hypothesisfor UV-induced mutation in relation to re-covery of capacity for DNA replication, p.389-392. In P. Hanawalt and R. B. Setlow(ed.), Molecular mechanisms for the repair ofDNA, part A. Plenum Press, New York.

62. Doudney, C. 0. 1976. Mutation in ultravioletlight-damaged microorganisms, p. 309-374.In. S. Y. Wang (ed.), Photochemistry andphotobiology of nucleic acids, vol. II. Aca-

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from



demic Press Inc., New York.63. Doudney, C. 0. 1976. Complexity of the ultra-

violet mutation frequency response curve inEscherichia coli B/r: SOS induction, one-le-sion and two-lesion mutagenesis. J. Bacte-riol. 128:815-826.

64. Doudney, C. O., and C. S. Young. 1962. Ultra-violet light induced mutation and deoxyribo-nucleic acid replication in bacteria. Genetics47:1125-1138.

65. Drabble, W. T., and B. A. D. Stocker. 1968. R(transmissible drug resistance) factors inSalmonella typhimurium: pattern of trans-duction by-phage P22 and ultraviolet-protec-tion effect. J. Gen. Microbiol. 53:109-123.

66. Drake, J. W. 1970. The molecular basis of mu-tation. Holden-Day, San Francisco.

67. Drake, J. W., and R. H. Baltz. 1976. The bio-chemistry of mutagenesis. Annu. Rev. Bio-chem. 45:11-37.

68. Emmerson, P. T. 1968. Recombination-defi-cient mutants of Escherichia coli that mapbetween thyA and argA. Genetics 60:19-30.

69. Eyfjord, J. E., M. H. L. Green, and B. A.Bridges. 1975. Mutagenic DNA repair inEscherichia coli: conditions for error-free fill-ing of daughter-strand gaps. J. Gen. Micro-biol. 91:369-375.

70. Fauquet, P., and M. Defais. 1972. UV reactiva-tion of phage X in a polA mutant of E. coli.Mutat. Res. 15:353-355.

71. Ganesan, A. K. 1974. Persistence of pyrimidinedimers during postreplication repair in ul-traviolet light-irradiated Escherichia coliK12. J. Mol. Biol. 87:103-119.

72. Ganesan, A. K., and P. C. Seawell. 1975. Theeffect of lexA and recF mutations on postrep-lication repair and DNA synthesis in Esche-richia coli K12. Mol. Gen. Genet. 141:189-206.

72a. Gellert, M., and M. L. Bullock. 1970. DNAligase mutants of Escherichia coli. Proc.Natl. Acad. Sci. U.S.A. 67:1580-1587.

73. George, D. L., and E. M. Witkin. 1974. Slowexcision repair in an mfd mutant of Esche-richia coli B/r. Mol. Gen. Genet. 133:283-291.

74. George, D. L., and E. M. Witkin. 1975. Ultra-violet light-induced responses of an mfd mu-tant of Escherichia coli B/r having a slowrate of dimer excision. Mutat. Res. 28:347-354.

75. George, J., M. Castellazzi, and G. Buttin.1975. Prophage induction and cell division inE. coli. III. Mutations sfiA and sfiB restoredivision in tif and Ion strains and permit theexpression of mutator properties of tif. Mol.Gen. Genet. 140:309-332.

76. George, J., and R. Devoret. 1971. Conjugaltransfer of UV-damaged F-prime sex factorsand indirect induction of prophage X. Mol.Gen. Genet. 111:103-119.

77. George, J., R. Devoret, and M. Radman. 1974.Indirect ultraviolet-reactivation of phage X.

Proc. Natl. Acad. Sci. U.S.A. 71:144-147.78. Glickman, B. 1974. The role of DNA polymer-

ase I in pyrimidine dimer excision and repairreplication in Escherichia coli K12 followingultraviolet irradiation. Biochim. Biophys.Acta 335:115-122.

79. Goldberg, A. L., and J. F. Dice. 1974. Intracel-lular protein degradation in mammalian andbacterial cells. Annu. Rev. Biochem. 43:835-869.

79a. Goldberg, A. L., and A. C. St. John. 1976.Intracellular protein degradation in mam-malian and bacterial cells: part 2. Annu.Rev. Biochem. 45:747-803.

80. Goldthwait, D., and F. Jacob. 1964. Sur lem6canisme de l'induction du d6veloppementdu prophage chez les bacteries lysogenes. C.R. Acad. Sci. Paris D 259:661-664.

81. Gottesman, M. M., M. L. Hicks, and M. Gel-lert. 1973. Genetics and function of DNA li-gase in Escherichia coli. J. Mol. Biol. 77:531-547.

82. Goze, A., A. Sarasin, Y. Moule, and R. Devo-ret. 1975. Induction and mutagenesis of pro-phage X in Escherichia coli K12 by metabo-lites of aflatoxin B.. Mutat. Res. 28:1-7.

83. Green, M. H. L., J. Greenberg, and J. Donch.1969. Effect of a recA gene on cell divisionand capsular polysaccharide production in alon strain of Escherichia coli. Genet. Res.14:159-162.

84. Greenberg, J., L. J. Berends, J. Donch, and M.H. L. Green. 1974. exrB: a malB-linked genein Escherichia coli B involved in sensitivityto radiation and filament formation. Genet.Res. 23:175-184.

85. Greenberg, J., L. Berends, J. Donch, and B.Johnson. 1975. Reversion studies with exrBin Escherichia coli. Genet. Res. 25:109-117.

86. Greenberg, J., J. Donch, and L. Berends. 1975.The dominance of exrB over exrB+ in hetero-diploids of Escherichia coli. Genet. Res.25:39-44.

87. Gross, J., and M. Gross. 1969. Genetic analysisof an E. coli strain with a mutation affectingDNA polymerase. Nature (London) 224:1166-1168.

88. Grossman, L., A. Braun, R. Feldberg, and I.Mahler. 1975. Enzymatic repair of DNA.Annu. Rev. Biochem. 44:19-43.

88a. Gudas, L. J. 1976. The induction of protein Xin DNA repair and cell division mutants ofEscherichia coli. J. Mol. Biol. 104:567-587.

89. Gudas, L. J., and A. B. Pardee. 1975. Model forthe regulation of Escherichia coli DNA re-pair functions. Proc. Natl. Acad. Sci. U.S.A.72:2330-2334.

90. Gudas, L. J., and A. B. Pardee. 1976. DNAsynthesis inhibition and the induction of pro-tein X in Escherichia coli. J. Mol. Biol.101:459-477.

91. Hanawalt, P. C. 1975. Molecular mechanismsinvolved in DNA repair. Genetics 79:179-197.

92. Hanawalt, P. 1975. Repair processes in diversesystems: overview, p. 503-506. In P. Hana-walt and R. B. Setlow (ed.), Molecular mech-

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


902 WITKIN

anisms for repair of DNA, part B. PlenumPress, New York.

93. Harm, H. 1976. Repair of UV-irradiated biolog-ical systems: photoreactivation, p. 219-263.In S. Y. Wang (ed.), Photochemistry andphotobiology of nucleic acids, vol. II. Aca-demic Press Inc., New York.

94. Harm, W., C. S. Rupert, and H. Harm. 1971.The study of photoenzymatic repair of UVlesions in DNA by flash photolysis, p. 279-324. In A. C. Giese (ed.), Photophysiology,vol. 6. Academic Press Inc., New York.

95. Hart, R. W., and R. B. Setlow. 1975. Directevidence that pyrimidine dimers in DNA re-sult in neoplastic transformation, p. 719-724.In P. Hanawalt and R. B. Setlow (ed.), Mo-lecular mechanisms for repair of DNA, partB. Plenum Press, New York.

96. Hellman, K. B., K. F. Haynes, and L. E. Bock-stahler. 1974. Radiation enhanced survivalof a human virus in normal and malignantrat cells. Proc. Exp. Biol. Med. 145:255-262.

97. Hertman, I., and S. E. Luria. 1967. Transduc-tion studies on the role of a rec+ gene inultraviolet induction of prophage lambda. J.Mol. Biol. 23:117-133.

98. Hill, R. F. 1965. Ultraviolet-induced lethalityand reversion to prototrophy in Escherichiacoli strains with normal and reduced darkrepair ability. Photochem. Photobiol. 4:563-568.

99. Hill, R. F., and E. R. Nestmann. 1973. Effect ofthe recC gene in Escherichia coli on frequen-cies of ultraviolet-induced mutants. Mutat.Res. 17:27-36.

99a. Hirota, Y., A. Ryter, and F. Jacob. 1968.Thermosensitive mutants of E. coli affectedin the processes ofDNA synthesis and cellu-lar division. Cold Spring Harbor Symp.Quant. Biol. 33:677-693.

100. Holliday, R., and J. E. Pugh. 1975. DNA modi-fication mechanisms and gene activity dur-ing development. Science 187:226-232.

101. Horii, Z. I., and A. J. Clark. 1973. Geneticanalysis of the RecF pathway to genetic re-combination in Escherichia coli K-12: isola-tion and characterization of mutants. J. Mol.Biol. 80:327-344.

lOla. Howard, B. D., and I. Tessman. 1964. Identi-fication of the altered bases in mutated sin-gle-stranded DNA. J. Mol. Biol. 9:372-375.

102. Howard-Flanders, P. 1975. Repair by geneticrecombination in bacteria: overview, p. 265-274. In P. Hanawalt and R. B. Setlow (ed.),Molecular mechanisms for repair of DNA,part A. Plenum Press Inc., New York.

103. Howard-Flanders, P., and R. P. Boyce. 1966.DNA repair and genetic recombination;studies on mutants of Escherichia coli defec-tive in these processes. Radiat. Res.6:(Suppl)156-184.

104. Howard-Flanders, P., E. Simson, and L. Ther-iot. 1964. A locus that controls filament for-mation and sensitivity to radiation in Esche-richia coli K-12. Genetics 49:237-246.

105. Howard-Flanders, P., W. D. Rupp, B. M. Wil-kins, and R. S. Cole. 1968. DNA replicationand recombination after UV irradiation.Cold Spring Harbor Symp. Quant. Biol.33:195-205.

106. Howard-Flanders, P., and L. Theriot. 1966.Mutants ofEscherichia coli K-12 defective inDNA repair and in genetic recombination.Genetics 53:1137-1150.

107. Howarth, S. 1966. Increase in frequency of ul-traviolet-induced mutation brought about bythe colicine factor colI in Salmonella typhi-murium. Mutat. Res. 3:129-134.

108. Hull, R. A. 1975. Effect of tsl mutations on ColEl expression in a recA strain ofE. coli K-12.J. Bacteriol. 123:775-776.

109. Ichikawa-Ryo, H., and S. Kondo. 1975. Indi-rect mutagenesis in phage lambda by ultra-violet preirradiation of host bacteria. J. Mol.Biol. 97:77-92.

110. Ikenaga, M., H. Ichikawa-Ryo, and S. Kondo.1975. The major cause of inactivation andmutation by 4-nitroquinoline-1-oxide inEscherichia coli: excisable 4NQO-purine ad-ducts. J. Mol. Biol. 92:341-356.

111. Inouye, M. 1971. Pleiotropic effect of the recAgene of Escherichia coli: uncoupling of celldivision from deoxyribonucleic acid replica-tion. J. Bacteriol. 106:539-542.

112. Inouye, M., and J. P. Guthrie. 1969. A muta-tion which changes a membrane protein ofE.coli. Proc. Natl. Acad. Sci. U.S.A. 64:957-961.

113. Inouye, M., and A. B. Pardee. 1970. Changes ofmembrane proteins and their relation toDNA synthesis and cell division ofE. coli. J.Biol. Chem. 245:5813-5819.

114. Iyer, V. N., and W. D. Rupp. 1971. Usefulnessof benzoylated napthoylated DEAE-celluloseto distinguish and fractionate double-stranded DNA bearing different extents ofsingle-stranded regions. Biochim. Biophys.Acta 228:117-126.

115. Jacob, F. 1950. Induction de la lyse et de laproduction de bacteriophages chez un Pseu-domonas pyocyanea lysogene. C. R. Acad.Sci. Paris D 231:1585-1587.

115a. Jacob, F. 1954. Mutation d'un bacteriophageinduite par l'irradiation des seules bacteries-h6tes avant l'infection: C. R. Acad. Sci. ParisD 238:732-734.

116. Kanazir, D. 1958. The apparent mutagenicityof thymine deficiency. Biochim. Biophys.Acta 30:20-23.

117. Kanner, L., and P. Hanawalt. 1970. Repairdeficiency in a bacterial mutant defective inDNA polymerase. Biochem. Biophys. Res.Commun. 39:149-155.

117a. Kantor, G. C., and R. A. Deering. 1966. Ul-traviolet radiation studies on filamentousEscherichia coli. J. Bacteriol. 92:1062-1069.

119. Kelley, R. B., M. R. Atkinson, J. A. Huber-man, and A. Kornberg. 1969. Excision of thy-mine dimers and other mismatched se-quences by DNA polymerase of Escherichia

BACTERIOL. REV.


http://mm

br.asm.org/

Dow

nloaded from


UV MUTAGENESIS AND DNA REPAIR IN E. COLu 903

coli. Nature (London) 224:495-501.120. Kelner, A. 1949. Effect of visible light on the

recovery of Streptomyces griseus conidiafrom ultraviolet irradiation injury. Proc.Natl. Acad. Sci. U.S.A. 35:73-79.

121. Kelner, A. 1949. Photoreactivation of ultravi-olet-irradiated Escherichia coli with specialreference to the dose-reduction principle andto ultraviolet-induced mutation. J. Bacte-riol. 58:511-522.

122. Kirby, E. P., F. Jacob, and D. A. Goldthwait.1967. Prophage induction and filament for-mation in a mutant strain of Escherichiacoli. Proc. Natl. Acad. Sci. U.S.A. 58:1903-1910.

123. Kirby, E. P., W. L. Ruff, and D. A. Gold-thwait. 1972. Cell division and prophage in-duction in Escherichia coli: effects of pantoyllactone and various furan derivatives. J.Bacteriol. 111:447-453.

124. Kondo, S. 1969. Mutagenicity versus radiosen-sitivity in Escherichia coli, p. 126-127. Pro-ceedings of the XIIth International Congresson Genetics, vol. II.

125. Kondo, S. 1973. Evidence that mutations areinduced by errors in repair and replication.Genetics 73:109-122.

126. Kondo, S., and H. Ichikawa. 1973. Evidencethat pretreatment of Escherichia coli cellswith N-methyl-N'-nitro-N-nitrosoguanidineenhances mutability of subsequently infect-ing phage X. Mol. Gen. Genet. 126:319-324.

127. Kondo, S., H. Ichikawa, K. Iwo, and T. Kato.1970. Base-change mutagenesis and pro-phage induction in strains ofEscherichia coliwith different DNA repair capacities. Ge-netics 66:187-217.

128. Kornberg, A. 1974. DNA synthesis. W. H.Freeman and Co., San Francisco.

129. Latarjet, R. 1951. Induction, par les rayons X,de la production d'un bacteriophage chez B.megatherium lysogene. Ann. Inst. PasteurParis 81:389-393.

130. Lawrence, C. W., and R. Christensen. 1976.UV mutagenesis in radiation sensitivestrains of yeast. Genetics 82:207-232.

131. Lehmann, A. R. 1972. Postreplication repair ofDNA in ultraviolet-irradiated mammaliancells. J. Mol. Biol. 66:319-337.

132. Lehmann, A. R., and S. Kirk-Bell. 1972. Post-replication repair ofDNA in ultraviolet-irra-diated mammalian cells: no gaps in DNAsynthesized late after ultraviolet irradiation.Eur. J. Biochem. 31:438-445.

133. Loeb, L. A., C. F. Springgate, and N. Battula.1974. Errors in DNA replication as a basis ofmalignant changes. Cancer Res. 34:2311-2321.

134. Lwoff, A., L. Simonovitch, and N. Kjeldgaard.1950. Induction de la production de bacterio-phages chez une bacterie lysogene. Ann.Inst. Pasteur Paris 79:815-859.

135. Lytle, C. D., S. G. Benane, and L. E. Bock-stahler. 1974. Ultraviolet-enhanced reactiva-tion of herpes virus in human tumor cells.

Photochem. Photobiol. 20:91-94.136. MacPhee, D. G. 1973. Effect of an R factor and

caffeine on ultraviolet mutability on Salmo-nella typhimurium. Mutat. Res. 18:367-370.

137. MacPhee, D. G. 1973. Effect of rec mutationson the ultraviolet protecting and mutation-enhancing properties of the plasmid R-Utrecht in Salmonella typhimurium. Mutat.Res. 19:357-359.

138. Mackie, G., and D. B. Wilson. 1972. Regula-tion of the gal operon of Escherichia coli bythe capR gene. J. Biol. Chem. 247:2973-2978.

139. Maher, V. M., L. M. Ouellette, R. D. Curren,and J. J. McCormick. 1976. Frequency ofultraviolet light-induced mutations is higherin xeroderma pigmentosum variant cellsthan in normal human cells. Nature (Lon-don) 201:593-595.

140. Marcovich, H. 1956. Etude radiobiologique dusysteme lysogene d'Escherichia coli K12. 1.Rayons X. Ann. Inst. Pasteur Paris 90:303-319.

141. Marcovich, H. 1956. Etude de l'action des ray-ons ultraviolets sur le systeme lysogeneEscherichia coli K12(X), K12S, X. Ann. Inst.Pasteur Paris 91:511-522.

142. Marinus, M. G., and N. R. Morris. 1975. Pleio-tropic effects of a DNA adenine methylationmutation (dam-3) in Escherichia coli K12.Mutat. Res. 28:15-26.

143. Marsden, H. S., E. C. Pollard, W. Ginoza, andE. P. Randall. 1974. Involvement ofrecA andexr genes in the in vivo inhibition of therecBC nuclease. J. Bacteriol. 118:465-470.

144. Masker, W., P. Hanawalt, and H. Shizuya.role of DNA polymerase II in repairreplication in Escherichia coli. Nature (Lon-don) New Biol. 244:242-243.

145. McCaffrey, R., D. F. Smoler, and D. Balti-more. 1973. Terminal deoxynucleotidyltransferase in a case of childhood acute lym-phoblastic leukemia. Proc. Natl. Acad. Sci.U.S.A. 70:521-525.

146. McCann, J., N. E. Springarn, J. Kobori, andB. Ames. 1975. Detection of carcinogens asmutagens: bacterial tester strains with Rfactor plasmids. Proc. Natl. Acad. Sci.U.S.A. 72:979-983.

147. Melechen, N. E., and P. D. Skaar. 1962. Theprovocation of an early step of induction bythymine deprivation. Virology 16:21-29.

148. Mennigmann, H. D. 1972. Pyrimidine dimersas pre-mutational lesions in Escherichia coliWP2 Hcr-. Mol. Gen. Genet. 117:167-186.

149. Meyn, R. E., and R. M. Humphrey. 1971. De-oxyribonucleic acid synthesis in ultraviolet-light-irradiated Chinese hamster cells. Bio-phys. J. 11:295-301.

150. Miller, C. G. 1975. Peptidases and proteases ofEscherichia coli and Salmonella typhimu-rium. Annu. Rev. Microbiol. 29:485-504.

151. Miura, A., and J. Tomizawa. 1968. Studies onradiation-sensitive mutants of E. coli. 1I.Participation of the Rec system in inductionof mutation by ultraviolet irradiation. Mol.

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


BACTERIOL. REV.

Gen. Genet. 103:1-10.152. Monk, M. 1967. Observations on the mecha-

nism of indirect induction by mating withultraviolet colU donors. Mol. Gen. Genet.100:264-274.

153. Monk, M. 1969. Induction of phage X by trans-ferred irradiated coil DNA. Mol. Gen. Genet.106:14-24.

154. Monk, M., and J. Gross. 1971. Induction ofprophage X in a mutant ofE. coli K12 defec-tive in initiation of DNA replication at hightemperatures. Mol. Gen. Genet. 110:299-306.

155. Monk, M., M. Peacey, and J. D. Gross. 1971.Repair of damage induced by ultravioletlight in DNA polymerase-defective Esche-richia coli cells. J. Mol. Biol. 58:623-630.

156. Moody, E. E. M., K. B. Low, and D. W.Mount. 1973. Properties of strains of Esche-richia coli K12 carrying mutant lex and recalleles. Mol. Gen. Genet. 121:197-205.

156a. Moreau, P., A. Bailone, and R. Devoret. 1976.Prophage X induction in Escherichia colienvA uvrB: a highly sensitive test for poten-tial carcinogens. Proc. Natl. Acad. Sci.U.S.A., 73:3700-3704.

157. Morse, L. S., and C. Pauling. 1975. The induc-tion of error-prone repair as a consequence ofDNA ligase deficiency in Escherichia coli.Proc. Natl. Acad. Sci. U.S.A. 72:4645-4649.

158. Mount, D. W., and C. Kosel. 1975. Ultraviolet-induced mutation in UV-resistant, thermo-sensitive derivatives oflexA strains ofE. coliK-12. Mol. Gen. Genet. 136:95-106.

158a. Mount, D. W., C. Kosel, and A. Walker. 1976.Inducible error-free DNA repair in tsl recAmutants ofE. coli. Mol. Gen. Genet. 146:37-42.

159. Mount, D. W., K. B. Low, and S. J. Edmiston.1972. Dominant mutations (lex) in Esche-richia coli K-12 which affect radiation sensi-tivity and frequency of ultraviolet light-in-duced mutations. J. Bacteriol. 112:886-893.

160. Mount, D. W., A. C. Walker, and C. Kosel.1973. Suppression of lex mutations affectingdeoxyribonucleic acid repair in Escherichiacoli K-12 by closely linked thermosensitivemutations. J. Bacteriol. 116:950-956.

161. Mount, D. W., A. C. Walker, and C. Kosel.1975. Effect of tsl mutations in decreasingradiation sensitivity of a recA strain ofEscherichia coli K-12. J. Bacteriol. 121:1203-1207.

162. Mount, D. W., A. C. Walker, and C. Kosel.1975. Indirect suppression of radiation sensi-tivity of a recA strain of Escherichia coliK12, p. 383-388. In P. Hanawalt and R. B.Setlow (ed.), Molecular mechanisms for re-pair of DNA, part A. Plenum Press, NewYork.

163. Nelson, R. L., and H. S. Mason. 1972. Anexplicit hypothesis for chemical carcinogene-sis. J. Theor. Biol. 37:197-200.

164. Nishioka, H., and C. 0. Doudney. 1969. Differ-ent modes of loss of photoreversibility of mu-tation and lethal damage in ultraviolet-light

resistant and sensitive bacteria. Mutat. Res.8:215-228.

165. Nishioka, H., and C. 0. Doudney. 1970. Differ-ent modes of loss of photoreversibility of ul-traviolet light-induced true and suppressormutations to tryptophan independence in anauxotrophic strain of Escherichia coli. Mu-tat. Res. 9:349-358.

166. Noack, D., and S. Klaus. 1972. Inactivationkinetics of lambda phage repressors in a mu-tant ofE. coli temperature sensitive in DNAreplication. Mol. Gen. Genet. 115:216-224.

167. Novick, A., and L. Szilard. 1949. Experimentson light reactivation of ultraviolet-inacti-vated bacteria. Proc. Natl. Acad. Sci. U.S.A.35:591-600.

167a. Ono, J., and Y. Shimazu. 1966. Ultravioletreactivation of a bacteriophage containing asingle-stranded deoxyribonucleic acid as agenetic element. Virology 29:295-302.

167b. Osborn, M., S. Person, S. Phillips, and F.Funk. 1967: A determination of mutagenspecificity in bacteria using nonsense mu-tants of bacteriophage T4. J. Mol. Biol.26:437-447.

167c. Otsuji, N., M. Sekiguchi, T. Ijima, and Y.Takagi. 1959. Induction of phage formationin the lysogenic Escherichia coli by mitomy-cin C. Nature (London) 184:1079-1080.

168. Paterson, M. C., J. M. Boyle, and R. B. Set-low. 1971. Ultraviolet and X-ray-induced re-sponses of a deoxyribonucleic acid polymer-ase-deficient mutant of Escherichia coli. J.Bacteriol. 107:61-67.

169. Patrick, M. H., and R. 0. Rahn. 1976. Photo-chemistry of DNA and polynucleotides: pho-toproducts, p. 35-95. In S. Y. Wang (ed.),Photochemistry and photobiology of nucleicacids, vol. II. Academic Press Inc., NewYork.

170. Pettijohn, D., and P. Hanawalt. 1964. Evi-dence for repair replication of ultravioletdamaged DNA in bacteria. J. Mol. Biol.9:395-410.

171. Pollard, E., and E. P. Randall. 1973. Studieson the inducible inhibitor of radiation-in-duced DNA degradation of E. coli. Radiat.Res. 55:265-279.

172. Pritchard, R. H., and K. G. Lark. 1964. Induc-tion of replication by thymine starvation atthe chromosome origin in Escherichia coli. J.Mol. Biol. 9:288-307.

173. Ptashne, M. 1967. Specific binding of the Xphage repressor to X DNA. Nature (London)214:232-234.

174. Radman, M. 1974. Phenomenology of an induc-ible mutagenic DNA repair pathway inEscherichia coli: SOS repair hypothesis, p.128-142. In L. Prokash, F. Sherman, M.Miller, C. Lawrence, and H. W. Tabor (ed.),Molecular and environmental aspects of mu-tagenesis. Charles C Thomas Publisher,Springfield, Ill.

175. Radman, M. 1975. SOS repair hypothesis: phe-nomenology of an inducible DNA repair

904 WITKIN


http://mm

br.asm.org/

Dow

nloaded from


UV MUTAGENESIS AND DNA REPAIR IN E. COLu 905

which is accompanied by mutagenesis, p.355-367. In P. Hanawalt and R. B. Setlow(ed.), Molecular mechanisms for repair ofDNA, part A. Plenum Press, New York.

176. Radman, M., and R. Devoret. 1971. UV-reacti-vation of bacteriophage X in excision repair-deficient hosts: independence of red func-tions and attachment regions. Virology43:504-506.

177. Regan, J. D., and R. B. Setlow. 1974. Twoforms of repair in the DNA of human cellsdamaged by chemical carcinogens and muta-gens. Cancer Res. 34:3318-3325.

178. Regan, J. D., J. E. Trosko, and W. L. Carrier.1968. Evidence for excision of ultraviolet-in-duced pyrimidine dimers from the DNA ofhuman cells in vitro. Biophys. J. 8:319-325.

178a. Reich, E., D. B. Rifkin, and E. Shaw (ed.).1975. Proteases and biological control. ColdSpring Harbor Laboratory, Cold Spring Har-bor, N.Y.

179. Reznikoff, W. S. 1972. The operon revisited.Annu. Rev. Genet. 6:133-156.

180. Roberts, J. W., and C. W. Roberts. 1975. Prote-olytic cleavage of bacteriophage lambda re-pressor in induction. Proc. Natl. Acad. Sci.U.S.A. 72:147-151.

181. Rosner, J. L., L. R. Kass, and M. B. Yarmolin-sky. 1968. Parallel behavior of F and P1 incausing indirect induction of lysogenic bacte-ria. Cold Spring Harbor Symp. Quant. Biol.33:785-789.

182. Rothman, R. H., T. Kato, and A. J. Clark.1975. The beginning of an investigation ofthe role of recF in the pathways of metabo-lism of ultraviolet-irradiated DNA in Esche-richia coli, p. 283-291. In P. C. Hanawalt andR. B. Setlow (ed.), Molecular mechanismsfor repair of DNA, part A. Plenum Press,New York.

183. Rupert, C. S., H. Harm, and K. To. 1973. Theanatomy of direct repair. An. Acad. Bras.Cienc. 45(Suppl.):151-159.

184. Rupert, C. S. 1975. Enzymatic photoreactiva-tion: overview, p. 73-87. In P. Hanawalt andR. B. Setlow (ed.), Molecular mechanismsfor repair of DNA, part A. Plenum Press,New York.

185. Rupp, W. D., and P. Howard-Flanders. 1968.Discontinuities in the DNA synthesized inan excision-defective strain of Escherichiacoli following ultraviolet irradiation. J. Mol.Biol. 31:291-304.

186. Rupp, W. D., C. E. Wilde III, D. L. Reno, andP. Howard-Flanders. 1971. Exchanges be-tween DNA strands in ultraviolet-irradiatedEscherichia coli. J. Mol. Biol. 61:25-44.

187. Sarin, P. S., and R. C. Gallo. 1974. Terminaldeoxynucleotidyl transferase in chronic mye-logenous leukemia. J. Biol. Chem. 249:8051-8053.

188. Schuster, H., D. Beyersmann, M. Miko-lajczyk, and M. Schlicht. 1973. Prophage in-duction by high temperature in thermosensi-tive dna mutants lysogenic for bacteriophage

lambda. J. Virol. 11:879-885.189. Sedgwick, S. G. 1975. Inducible error-prone re-

pair in Escherichia coli. Proc. Natl. Acad.Sci. U.S.A. 72:2753-2757.

190. Sedgwick, S. G. 1975. Genetic and kinetic evi-dence for different types of postreplicationrepair in Escherichia coli B. J. Bacteriol.123:154-161.

190a. Sedgwick, S. G. 1976. Misrepair of overlap-ping daughter-strand gaps as a possiblemechanism for UV induced mutagenesis inuvr strains of Escherichia coli: a generalmodel for induced mutagenesis by misrepair(SOS repair) of closely spaced DNA lesions.Mutat. Res., in press.

191. Sedgwick, S. G., and B. A. Bridges. 1972. Sur-vival, mutation and capacity to repair sin-gle-strand breaks after gamma-irradiationin different Exr strains of Escherichia coli.Mol. Gen. Genet. 119:93-102.

192. Sedgwick, S., and B. A. Bridges. 1974. Re-quirement for either DNA polymerase I orDNA polymerase m in post-replication re-pair in excision proficient Escherichia coli.Nature (London) 249:348-349.

193. Setlow, J. K. 1966. Photoreactivation. Radiat.Res. 6(Suppl.):141-155.

194. Setlow, J. K., and M. E. Boling. 1970. Ultravi-olet action spectra for mutation in Esche-richia coli. Mutat. Res. 9:437-442.

195. Setlow, R. B., and W. L. Carrier. 1964. Thedisappearance of thymine dimers from DNA:an error-correcting mechanism. Proc. Natl.Acad. Sci. U.S.A. 51:226-231.

196. Setlow, R. B., J. D. Regan, J. German, and W.L. Carrier. 1969. Evidence that xerodermapigmentosum cells do not perform to firststep in the repair of ultraviolet damage totheir DNA. Proc. Natl. Acad. Sci. U.S.A.64:1035-1041.

197. Setlow, R., and J. K. Setlow. 1972. Effects ofradiation on polynucleotides. Annu. Rev.Biophys. Bioeng. 1:293-346.

198. Shinagawa, K., and T. Itoh. 1973. Inactivationof DNA binding activity of repressor in ex-tracts of lambda lysogens treated with mito-mycin C. Mol. Gen. Genet. 126:103-110.

199. Shineberg, B., and D. Zipser. 1973. The longene and degradation of 3l-galactosidase non-sense fragments. J. Bacteriol. 116:1469-1471.

199a. Sicard, N., and R. Devoret. 1962. Effects de lacarence en thymine sur des souches lyso-genes K12T- et 15T-. C. R. Acad. Sci. ParisD 255:1417-1419.

200. Siegel, E. C. 1973. Ultraviolet-sensitive muta-tor strain of Escherichia coli K-12. J. Bacte-riol. 113:145-160.

201. Sinzinis, B. I., G. B. Smirnov, and A. S.Saenko. 1973. Repair deficiency in an Esche-richia coli UV-sensitive mutator strainuvr5O2. Biochem. Biophys. Res. Commun.53:309-317.

202. Smith, K. C. Ultraviolet radiation effects. InK. C. Smith (ed.), The science of photobiol-ogy, in press.

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


BACTERIOL. REV.

203. Smith, K. C., and D. H. C. Meun. 1970. Repairof radiation-induced damage in Escherichiacoli. I. Effect of rec mutations on postreplica-tion repair of damage due to ultraviolet ra-diation. J. Mol. Biol. 51:459-472.

204. Springgate, C., and L. A. Loeb. 1973. Muta-genic DNA polymerases in human leukemiccells. Proc. Natl. Acad. Sci. U.S.A. 70:245-249.

205. Srivastava, B. I. S., and J. Minowada. 1973.Terminal deoxynucleotidyl transferase ac-tivity in a cell line (molt-4) derived from theperipheral blood of a patient with acute lym-phoblastomic leukemia. Biochem. Biophys.Res. Commun. 51:529-535.

206. Sussman, R., and H. BenZeev. 1975. Proposedmechanism of bacteriophage lambda induc-tion: acquisition of binding sites for lambdarepressor by DNA of the host. Proc. Natl.Acad. Sci. U.S.A. 72:1973-1976.

207. Sutton, H. E., and R. P. Wagner. 1975. Muta-tion and enzyme functions in humans. Annu.Rev. Genet. 9:187-212.

208. Swenson, P. A., and R. L. Schenley. 1974.Respiration, growth and viability of repair-deficient mutants of Escherichia coli afterultraviolet irradiation. Int. J. Radiat. Biol.25:51-60.

209. Tait, R. C., A. L. Harris, and D. W. Smith.1974. DNA repair in Escherichia coli mu-tants deficient in DNA polymerases I, II and/or III. Proc. Natl. Acad. Sci. U.S.A. 71:675-679.

209a. Tessman, E. S., and T. Ozaki. 1960. The in-teraction of phage S13 with ultraviolet irra-diated host cells and properties of the ultra-violet irradiated phage. Virology 12:431-449.

210. Tomizawa, J., and T. Ogawa. 1967. Effect ofultraviolet irradiation on bacteriophagelambda immunity. J. Mol. Biol. 23:247-263.

211. Tomizawa, J., and H. Ogawa. 1972. Structuralgenes of ATP-dependent deoxyribonucleaseofE. coli. Nature (London) New Biol. 239:14-16.

212. Trosko, J. E., E. H. Y. Chu, and W. L. Carrier.1965. The induction of thymine dimers inultraviolet-irradiated mammalian cells. Ra-diat. Res. 24:667-672.

213. Van Sluis, C. A., I. E. Mattern, and M. C.Paterson. 1974. Properties of uvrE mutantsofEscherichia coli K12. I. Effects ofUV irra-diation on DNA metabolism. Mutat. Res.25:273-279.

215. Volkert, M., D. L. George, and E. M. Witkin.1976. Partial suppression of the LexA pheno-type by mutations (rnm) which restore ultra-violet resistance but not ultraviolet mutabil-ity to Escherichia coli B/r uvrA lexA. Mutat.Res. 36:17-28.

216. Wacker, A., H. Dellweg, and D. Jackerts. 1962.Thymin-dimerisierung und Uberlebensratebei Bakterien. J. Mol. Biol. 4:410-412.

217. Walker, J. R., C. L. Ussery, and J. S. Allen.1973. Bacterial cell division regulation: lyso-genization of conditional cell division lon

mutants ofEscherichia coli by bacteriophagelambda. J. Bacteriol. 113:1326-1332.

218. Webb, R. B., and M. M. Malina. 1967. Muta-genesis in E. coli by visible light. Science156:1104-1105.

219. Wechsler, J. A., and J. D. Gross. 1971. Esche-richia coli mutants temperature-sensitive forDNA synthesis. Mol. Gen. Genet. 113:273-284.

220. Weigle, J. J. 1953. Induction of mutation in abacterial virus. Proc. Natl. Acad. Sci. U.S.A.39:628-636.

221. West, S. C., K. A. Powell, and P. T. Emmer-son. 1975. recAl-dependent inactivation ofthe lambda repressor in Escherichia coli ly-sogens by y-radiation and by tif expression.Mol. Gen. Genet. 141:1-8.

222. Witkin, E. M. 1947. Genetics of resistance toradiation in Escherichia coli. Genetics32:221-248.

223. Witkin, E. M. 1956. Time, temperature andprotein synthesis: a study of ultraviolet-in-duced mutation in bacteria. Cold SpringHarbor Symp. Quant. Biol. 21:123-140.

224. Witkin, E. M. 1959. Post-irradiation metabo-lism and the timing of ultraviolet-inducedmutations in bacteria, p. 280-299. In Pro-ceedings of the X International Congress onGenetics, vol. I.

225. Witkin, E. M. 1966. Radiation-induced muta-tions and their repair. Science 152:1345-1353.

226. Witkin, E. M. 1967. The radiation sensitivity ofEscherichia coli B: a hypothesis relating fila-ment formation and prophage induction.Proc. Natl. Acad. Sci. U.S.A. 57:1275-1279.

227. Witkin, E. M. 1967. Mutation-proof and muta-tion-prone modes of survival in derivatives ofEscherichia coli B differing in sensitivity toultraviolet light. Brookhaven Symp. Biol.20:17-55.

228. Witkin, E. M. 1969. The role ofDNA repair andrecombination in mutagenesis, p. 225-245. InProceedings of XII International Congress onGenetics, vol. 3.

229. Witkin, E. M. 1969. The mutability towardultraviolet light of recombination-deficientstrains of Escherichia coli. Mutat. Res. 8:9-14.

230. Witkin, E. M. 1969. Ultraviolet-induced muta-tion and DNA repair. Annu. Rev. Microbiol.23:487-514.

231. Witkin, E. M. 1971. Ultraviolet mutagenesis instrains of E. coli deficient in DNA polymer-ase. Nature (London) New Biol. 229:81-82.

232. Witkin, E. M. 1972. Ultraviolet mutagenesis inrepair-deficient derivatives of Escherichiacoli B/r: uvrA recB and uvrA recC strains.Mutat. Res. 16:235-242.

233. Witkin, E. M. 1973. Ultraviolet mutagenesisin bacteria: the inducible nature of error-prone repair. An. Acad. Bras. Cienc.45(Suppl.):188-192.

234. Witkin, E. M. 1974. Thermal enhancement ofultraviolet mutability in a tif-1 uvrA deriva-

906 WITKIN


http://mm

br.asm.org/

Dow

nloaded from



tive of Escherichia coli B/r: evidence thatultraviolet mutagenesis depends upon an in-ducible function. Proc. Natl. Acad. Sci.U.S.A. 71:1930-1934.

235. Witkin, E. M. 1975. Elevated mutability ofpolA and uvrA poLA derivatives of Esche-richia coli B/r at sublethal doses of ultravi-olet light: evidence for an inducible error-prone repair system ("SOS repair") and itsanomalous expression in these strains. Ge-netics 79(Suppl.):199-213.

236. Witkin, E. M. 1975. Relationships among re-pair, mutagenesis and survival: overview, p.347-353. In P. Hanawalt and R. B. Setlow(ed.), Molecular mechanisms for repair ofDNA, part A. Plenum Press, New York.

237. Witkin, E. M. 1975. Thermal enhancement ofultraviolet mutability in a dnaB uvrA deriv-ative of Escherichia coli B/r: evidence forinducible error-prone repair, p. 369-378. InP. Hanawalt and R. B. Setlow (ed.), Molecu-lar mechanisms for repair of DNA, part A.Plenum Press, New York.

238. Witkin, E. M. 1975. Persistence and decay ofthermo-inducible error-prone repair activityin nonfilamentous derivatives of tif-1 Esche-richia coli B/r: the timing of some criticalevents in ultraviolet mutagenesis. Mol. Gen.Genet. 142:87-103.

239. Witkin, E. M., and E. L. Farquharson. 1969.Enhancement and diminution of ultravioletlight-initiated mutagenesis by posttreat-ment with caffeine inEscherichia coli, p. 36-49. In G. E. W. Wolstenholme and M.O'Connor (ed.), Ciba Foundation Symposiumon Mutation as Cellular Process. J. & A.Churchill, Ltd., London.

240. Witkin, E. M., and D. L. George. 1973. Ultra-violet mutagenesis in poIA and uvrA poIAderivatives of Escherichia coli B/r: evidencefor an inducible error-prone repair system.Genetics 73(Suppl.):91-108.

241. Witkin, E. M., and E. C. Parisi. 1974. Bromou-racil mutagenesis: mispairing or misrepair?

Mutat. Res. 25:407-409.242. Witkin, E. M., and I. E. Wermundsen. 1973.

Do ultraviolet-induced mutations to strepto-mycin resistance exhibit susceptibility tomutation frequency decline? Mutat. Res.19:261-264.

243. Worcel, A. 1970. Induction of chromosome re-initiation in a thermosensitive DNA mutantofE. coli. J. Mol. Biol. 52:371-386.

243a. Yoshinaga, K., M. Fusaya, and M. Shimo-mura. 1971. On the mechanism of thymine-less death in E. coli strain 15 TAU, p. 41-53.Reports of Faculty of Science, Shizuoka Uni-versity, vol. 6.

244. Youngs, D. A., and K. C. Smith. 1973. Involve-ment of DNA polymerase III in excision re-pair after ultraviolet irradiation. Nature(London) New Biol. 244:240-241.

245. Youngs, D. A., and K. C. Smith. 1973. Evi-dence for the control by exrA and polA genesof two branches of the uvr gene-dependentexcision repair pathway in Escherichia coliK-12. J. Bacteriol. 116:175-182.

246. Youngs, D. A., and K. C. Smith. 1973. X-raysensitivity and repair capacity of a polAlexrA strain of Escherichia coli K-12. J. Bac-teriol. 114:121-127.

247. Youngs, D. A., and K. C. Smith. 1976. Geneticcontrol of multiple pathways of post-replica-tional repair in uvrB strains of Escherichiacoli K-12. J. Bacteriol. 125:102-110.

248. Youngs, D. A., E. van der Scheuren, and K. C.Smith. 1974. Separate branches of the uvrgene-dependent excision repair process in ul-traviolet-irradiated Escherichia coli K-12cells; their dependence upon growth mediumand the polA, recA, recB, and exrA genes. J.Bacteriol. 117:717-725.

249. Zelle, M. R., J. E. Ogg, and A. Hollaender.1958. Photoreactivation of induced mutationand inactivation of Escherichia coli exposedto various wave lengths of monochromaticultraviolet radiation. J. Bacteriol. 75:190-198.

VOL. 40, 1976


http://mm

br.asm.org/

Dow

nloaded from


ultraviolet mutagenesis andinducible dna repair escherichia …ultraviolet mutagenesis andinducible...

Documents