(:opyright 0 199 I by the Society of America

Control of Large Chromosomal Duplications in Escherichia coli by the Mismatch Repair System

Marie-Agnes Petit,* Joan Dimpfl,* Miroslav Radmant and * “Division of and , University of Calfornia, Berkeley, Calfornia 94720, and ‘Institut , F-75251 Paris-Cedex 05, France Manuscript received May 23, 199 1 Accepted for publication July 2, 1991

ABSTRACT Excessive recombination between repeated, interspersed, and divergedDNA sequences is a poten- tial source of genomic instability. We have investigated the possibility that a mechanism exists to suppress genetic exchange between these quasi-homologous (homeologous) sequences. We examined the role of the general mismatch repair system of Escherichia coli because previous work has shown that the mismatch repair pathway functions as a barrier to interspecies recombination between E. coli and Salmonella typhimurium. The formation of large duplications by homeologous recombination in E. coli was increased some tenfold by mutations in the mutL and mutS genes that encode themismatch recognition proteins. These findings indicate that the mismatch recognition proteins act to prevent excessive intrachromosomal exchanges. We conclude that mismatch repair proteins serve as general controllers of the fidelity of genetic inheritance, acting to suppress chromosomal rearrangements as well as point mutations.

SCHERICHIA coli and Salmonellatyphimurium creased genetic variation for an endangered bacterial E carry large duplications of genomic segments at population (DIMPFLand ECHOLS1989). Based on the frequenciesthat can be as high as of abacterial SOS work, we were led to explore mechanisms that population (ANDERSONand ROTH 1977; PETESand might normally suppress formation of large duplica- HILL1988). These duplications are thought to arise tions and other chromosomal rearrangements. by an unequal recombination event between homol- A property of the rrn and rhs gene families is a ogous sequences dispersed in the genome. Two ex- slight divergence in DNA sequence(homeologous amples of such “gene families” are therrn for sequences) (HILL, HARVEYand GRAY1990). We were ribosomal RNA andthe rhs sequences (ANDERSON intrigued with the possible regulatory significance of and ROTH 1977; PETESand HILL 1988). Seven rrn this property because the recombinational barrier be- loci are present in the E. coli chromosome.More tween the homeologous DNA sequences of E. coli and recently, up to five rhs loci have been described in E. S. typhimurium can be largely eliminated by mutations coli; the rhs sequences lack a known genetic function in the mismatch repair system (RAYSSIGUIER,THALER (SADOSKYet al. 1989; FEULNERet al. 1990). In addition and RADMAN1989). Thus, mismatch repair proteins to duplication, rrn operons have been defined as sites might function aspart of a generalmechanism to limit for inversion, deletion, and transposition of chromo- intrachromosomal as well as interchromosomalre- somal segments (HILLet al. 1977; HILLand HARNISH combination between duplicated, diverged DNA se- 1981, 1982). quences. Chromosomalrearrangements, especially duplica- To study the recombination events leading to du- tions, can serve as valuable sources for environmental plication, we have used a bacterial strain that allows or evolutionary adaptation. For example, duplication quantitative measurementof duplications between the in Salmonella appears to aid growth on limiting car- rhsA and rhsB sequences some 140 kbp apart in the bon source, probably by a gene dosage effect (SONTI genome (LIN, CAPAGEand HILL1984). The rhsA and and ROTH 1989).However, some geneticre- rhsB sequences share a 3.7-kbp region of substantial arrangements, especially deletions, can be highly del- homology, as judged by hybridization data, but differ eterious. Recent work has shown that the frequency by 22 mismatches among the 990 bp alreadyavailable of large duplications in E. coli increases tenfold when for sequence comparison (SADOSKYet al. 1989; FEUL- the SOS response is constitutively activated (DIMPFL NER et al. 1990; C. W. HILL,personal communication). and ECHOLS1989). This observation indicates that The formation of largeduplications was increased duplication formation might be subject to some form some tenfold by mutations inactivating the mutL and of cellular regulation; the SOS response might allow mutS genes of the mismatch repair pathway. Thus, morefrequent rearrangements as a source of in- mismatch repairproteins are likely to function in

Genetics 129 327-332 (October, 1991) 328 M.-A. Petit et al.

MATERIALSAND METHODS for details). A schematic map of the CHI 504 chromosome, before and after duplication, is presented in Figure 1. Duplication fre- quency was measured in strain CH 1504 and its deriv- atives harboring a mutL, mutS, mutH or uvrD muta- tion. Each of these mutations inactivates the general mismatch repair pathway (MODRICH 1989; RADMAN FIGURE1 .-Chromosome map of strains used to detect duplica- 1989). The results of duplication assays are presented tion events. The structureof the duplicated chromosome is depicted in Table 2. In a mut+ context, the duplication fre- on the right side of the figure, with the arrows representing the quency was 1.1 X 1 0-4, which agrees with previously rhsA and rhsB sequences. reported data (LIN, CAPACEand HILL 1984; DIMPFL limiting intrachromosomal exchanges. Additional ex- and ECHOLS1989). In the absence of an active mutL periments indicate that the recombination events in- or mutS gene product, duplication frequency was 10- hibited by the MutL and MutS mismatchproteins are 15-fold higher (Table 2, lines 2 and 3). In contrast, RecBC-dependent. the mutH and uvrD mutants did not show an increase in duplication frequency (Table 2, lines 4 and 5). MATERIALS AND METHODS To verify the presence of a duplication in the large colonies, a segregation experiment was performed on Bacterial strains: Bacterial strains are listed in Table 1 several purified colonies. Duplication mutations are with their genotypes. New strains were prepared using trans- distinguished by their instability when the selective duction by phage P1 of transposon insertions in the gene of interest. pressure for them is removed, presumably because of Assay of gly& duplications: The genetic test allowing homologous recombination within the duplicated re- the detection of duplications has been previously described gion (CAMPBELL1965; ANDERSONand ROTH 1977). (LIN, CAPACEand HILL 1984; DIMPFLand ECHOLS1989). The results of the segregation analysis are presented The duplication assay relies on the glySL cold-sensitive mu- tant allele of the glycyl-tRNA synthetase gene, in conjunc- in Table 3. Among 1 1 large colonies tested for each tion with trpA36glyU(sup). The trpA36 mutation is effectively strain, the mutL, mutS and mutH strains were found suppressed by the glyU(sup) allele at 37". However, at 20" to give 100% with unstable phenotype. In addition, suppression of trpA36 is inefficient because ofpoor charging the percentage of segregation was indistinguishable of the suppressor tRNA, leading to poor growth in the between the wild type and the mutL or mutS deriva- absence of tryptophan. In a glySL duplication, the glycine- inserting missense suppressors are charged more efficiently tives. This observation indicates that the frequency of because of a gene dosage effect, leading to better growth homologous recombination between duplicated re- and large colony formation in the absence of tryptophan. gions responsible for segregation was approximately Southern hybridization has shown that this selection is spe- the same for the three strains. Thus the enhanced cific for duplication of the rhsA-rhsB region, which includes duplication formation in the mismatch-defective mu- the glySL gene (LIN, CAPACEand HILL 1984). Duplication frequency was deduced from the ratioof large colonies over tants appears to bespecific for mismatched DNA total colonies (20 plates scored per experiment). Asmis- sequences (although our experiments do not demon- match-deficient strains are known to favor transposon exci- strate that the same sequences are recombined in the sion (LUNDBLADand KLECKNER1984), several large colonies mismatch mutants and wild-type strains). The uvrD from each experiment were checked forthe antibiotic marker linked with the mismatch mutation; 45 among 45 strain gave 0/11 unstable phenotypes; therefore, du- had retained the antibiotic-resistant phenotype. plication frequency might be overestimated in the Segregation analysis of duplications: These measure- uvrD strain. ments were based on techniques developed for Salmonella From the data of Table 2 and Table 3, we infer (ANDERSON,MILLER and ROTH 1976). To remove selective that the MutL and MutS proteins are likely to have pressure favoring glySL duplications, individual large colo- nies were grown at 37 " in LB broth for10 to 15 generations, an active role in limiting the frequency of large dupli- and then plated on the original selective medium without cations. The MutL and MutS proteins are involved in tryptophan at 20" for 4 days. If segregation occurred, a the initial recognition of mismatched heteroduplex mixed population of large and smallcolonies could be DNA(GRILLEY et al. 1989; MODRICH 1989). MutS detected, among which the proportion of small colonies binds to the mismatch, and MutL associateswith represented the proportion of segregants. MutS. The MutH protein nicks at an unmodified GATC methylation sequence, and the UvrD helicase RESULTS (helicase 11) presumablyevicts the unwanted DNA Effect of mismatch repair mutations on duplica- strand (MODRICH 1989). Our data indicate a special tionformation: Inthe duplication test strain role for the mismatch recognition proteins. CH 1504,duplications of the rhsA-rhsB region contain- The genetic analysis reported in Table 2 reflects ing the glrS, gene are recognized as large colonies two processes: the formation of duplications by hom- after growth under low temperature conditions (see eologous recombination and the resolution of dupli- Mismatch Control of Duplications 329

TABLE 1 E. coli K-12strains

Strain designation Relevant genotype Source or reference Construction CH 1504 trp A36 glyU(sup) glySL LIN, CAPAGEand HILL(1984) JCD 1000 Same as CH1504 but recA730 DIMPFLand ECHOLS(1 989) srlC30O:TnlO sulA::Tn5 pyrD JCD1400 Same as CH 1504 but This laboratory P1 transduction of mutL::TnlO mutL::TnlO from GW3733 into CH1504 JCDl402 Same as CH 1504 but This laboratory P 1transduction of mutS::TnlO mutS::TnlO from GW3731 into CH1504 JCD1404 Same as CH1504 butmutH::Tn5 This laboratory P 1transduction of mutH::Tn5 from GW3732 into CH 1504 JCD1406 Same as CH1504 butuvrD::Tn5 This laboratory P 1transduction of uurD::Tn5 from GW7303 into CH 1504 JCDl408 Same as CH1504 butmutL::Tn5 This laboratory P1 transduction of mutL::Tn5 from NK7084 into CH 1504 JCD1410 Same as CH 1504 but mutS::Tn5 This laboratory P1 transduction of mutS::Tn5 from NK7085 into CH1504 JCD1412 Same as JCD1408 but This laboratory P1 transduction of rccB:TnlO recB::TnlO from N2101 into JCD1408 JCD1414 Same as JCD 1410 but This laboratory P1 transduction of rccB:TnlO recB::TnlO from N2101 into JCD1410 JCD1416 Same as CH 1504 butrecB::TnlO This laboratory P1 transduction of recB::TnlO from N2101 into CH1504 JCD1418 Same as JCD1400 butrecF::Tn3 This laboratory P1 transduction of recF:Tn3 from JCDl300 into JCD 1400 JCD1420 Same as JCD1402 but rccF::Tn3 This laboratory P1 transduction of recF::Tn3 from JCD 1310 into CH 1402 JCD1310 Same as CH 1504 but recF::Tn3 This laboratory P1 transduction of recF:Tn3 from JC 10990 into CH 1504 JCD1422 Same as JCDlOOO mutL::TnlO This laboratory P 1 transduction of mutL::TnlO from GW3733 into a TetS revertant of JCDl 000 JCDl424 Same as JCDlOOO mutS::TnlO This laboratory P1 transduction of mutS::TnlO from GW3731 into a TetS revertant of JCD 1000 NK7084 mutL103::Tn5 LUNDBLADand KLECKNER(1 984) NK7085 mutS104::TnS LUNDBLADand KLECKNER(1984) N2101 rccB268::TnlO LLOYD,BUCKMAN and BENSON 987)(1 JC 10990 rccF332::Tn3 J. CURK GW3733 mutL218::TnlO PANG,LUNDBERG and WALKER(1 985) GW3731 mulS215::TnlO G. WALKER GW3732 mutH471::Tn5 G. WALKER GW7303 uvrD260::Tn5 G. WALKER cations by homologous recombination. Thusthe If this alternative were true, segregation frequency MutL and MutS proteins might destabilize the dupli- should be significantly reduced in mutL or mu6 mu- cated state rather than inhibit duplication formation. tants compared to wild type. The data in Table 3 330 M.-A. Petit et al.

TABLE 2 TABLE 3 Effect of the mismatch repairgenes on the duplication of the Segregation of rhsA-rhsB duplications rhsA-rhsB region No. of unsta- Duplication Mismatch repairble colonies/ Percent segregation Mismatch repair frequency Straingenotype nametotal of unstablecolonies Straingenotype nanle (X 104)" CH 1504 + 12/12 26 (25) CH 1504 + 1.1 (k0.5) JCD 1400 mutL 11/11 39 (k8) JCD1400 mutL 16.0 (k6) JCD 1402 mutS 11/11 20 (212) JCD 1402 mutS 11 .0 (k3) JCD1404 mutH 11/11 41 (k6) JCD 1404 mutH 0.3 (f0.2) JCD 1406 uvrD 0/11 0 JCDl406 uvrD 0.6 (k0.3) '' Average and standard deviation are given for three separate TABLE 5 c.xperiments. Additive effect on duplication frequency of mismatch repair TABLE 4 deficiency and SOS activation Effect of recombination mutationson duplication frequency Mismatch repair Duplication frequency Strainnamegenotype sos (X 104)" Mismatch repairRecombination Duplication frequency Stwinnamegenotype genotype (x 1 04y JCDlOOO + Constitutive 9 (k6) JCD 1422 mutL Constitutive 26 (k8) <:H 1504 + + 1.1 (k0.5) JCD1424 mutS Constitutive 28 (23) JCD 1408 mutL + 10.0 (f6) J<:D14 10 mutS + 6.4 (k4) a Average and standard deviation are given from three separate JCD~~I'L WL recB 0.9 (f1.5) experiments. JC I) 14 14 mutS recB 0.5 (k0.5) quency: In a recent study, another route to enhanced + recB JCD1416 0.3 (kO.1) duplication formation was described; the same dupli- JCD1418 mutL recF 100.0 (k7.0) JCD1420 mutS recF 17.0 (k12) cation studied here became some tenfold more fre- JCDl3lO + recF 1.6 (kl.1) quent when the SOS regulon was derepressed by an

" Average and standard deviation are given from three or four SOS-constitutive mutant of recA (DIMPFLand ECHOLS separate experinlents. 1989). We investigated the combined effect of a mutL or mutS defect and SOS constitutive expression (Table show that the segregation frequency was not reduced 5); the resultantstrains exhibited a 25-30-fold in- for mutL or mutS strains compared to wild type. We crease in duplication frequency. Therefore, the two believe therefore that this alternative explanation is mutational routes to enhanced duplication formation unlikely. We conclude that the MutL and MutS pro- appear to be additive. In additional experiments, we teins probably act to reduce formation of large dupli- found that the mutL mutation conferred a ten-fold cations. increase in duplication formation in a strain carrying Recombination genes affecting enhanced dupli- a noninducible allele of the LexA repressor of SOS cation formation: We wanted to determine which (data not shown). The SOS pathway to induced du- recombination pathway of E. coli was stimulated by plication depends on both RecF and RecBC (DIMPFL the mutL and mutS mutations. For this purpose, we and ECHOLS1989; and our unpublished data). This introducedadditional mutations inactivating the dependence on RecF indicates that SOS-inducible du- RecBC or RecF pathways (CLARK 1973;SMITH 1987) plications involve at least some differences in mecha- (Table 4). Duplication formation in mutL and mutS nism from those stimulated by a defect in mismatch strains was reduced by a factor of 10 when a recB repair. mutation was introduced. Therefore, mutL and mutS gene products seem to interferewith a recombination DISCUSSION event promoted by the RecBC pathway of E. coli. In contrast,introduction of a recF mutationdid not Role of mismatch recognition proteins in hom- reduceduplication formation in mutL and mutS eologous recombination: From the data reported in strains. this paper, we conclude that the mismatch recognition Our data on the RecBC pathway indicate that this proteins MutS and MutLact toprevent excessive recombinational route is highly efficient for intrach- intrachromosomalrecombination between duplica- romosomal exchanges in the absence of intervention ted, diverged (homeologous) DNA sequences. Inter- by the mismatch proteins. Thus, genomic stability may estingly, our dataon mismatch repairmutants are depend on the ability of the mismatch repair system highly similar to those obtained in E. coli-S. typhimu- to exert a negative controlon the activity of the rium crosses, in which mutS and mutL mutations were RecBC pathway in homeologous exchanges. by far themost effective in stimulating recombination Additivitywith SOS-enhanced duplication fre- (RAYSSIGUIER,THALER and RADMAN 1989).SHEN and Mismatch Control of Duplications 331

HUANG(1989) have previously foundthat a mutS betweenhomologous chromosomes. The likelihood mutation increased recombination involving homeo- of such a recombination-limiting system has also been logous sequences between X phageand a plasmid. inferred from another point of view: the much lower FEINSTEINand Low(1 986) have notedthat mutations frequency of duplication by homeologous recombi- in the mismatch repair genes confer increased recom- nation compared to the frequency of segregation by bination in Hfr X F- crosses; in theseexperiments homologous recombination (HILL,HARVEY and GRAY involving highly homologoussequences, the major 1990). Our experiments define the mismatch recog- effect was noted for a mutantuvrD gene. These results nition proteins as likely components of this recombi- indicatethat a somewhat different, but mismatch- national fidelity system. Eukaryotic cells appearto related mechanism can limit recombination in the case carry a generalmismatch repair system similar to that of very limited sequence divergence (the two point found in E. coli (HOLMES, CLARKand MODRICH 1990; mutations in the hisF gene studied in the cross). THOMAS,ROBERTS and KUNKEL199 1). This mismatch The intervention of the MutL and MutS proteins pathway might also prevent aberrant recombination in the recombination process might occurimmediately (RAYSSIGUIER,THALER and RADMAN 1989; BAILIS after the RecA-promoted strand exchange between and ROTHSTEIN1990). rhsA and rhsB. As judged by experiments in vitro, For eukaryotic organisms, recombination between RecA will catalyze heteroduplex formation through homologous chromosomes is almost exclusively lim- small sequenceheterologies (DAS GUPTAand RAD- ited to meiosis. The recombinational fidelity mecha- DING 1982). The consecutive mismatches would offer nism proposed here might participatein this limitation atarget for the MutL and MutS proteins, whose on mitotic recombination. The frequent occurrence activity could resultin elimination of the heteroduplex of homeologous sequences in eukaryotes could trigger intermediate, hence preventing its resolution in du- the mismatch-directed system to abort attempted mi- plication. The ability of the MutL and MutS proteins toticrecombination between homologous chromo- to detect a mispaired region of a chromosome seems somes; suppression of this pathway would then be a to be utilized in at least two ways. As shown previously, component of meiotic development. The mismatch the general mismatch repair pathway contributes to system could be the regulatory balance limiting re- replicational fidelity through repairof errors by DNA combination to the proper meiotic pathway. polymerase (MODRICH 1989;RADMAN 1989). Mis- We thank CHARLESHILL for unpublished information, RICHARD match repair proteins also contribute to recombina- EISNERfor editorial help, MARCIE ROSENBERCand MALINI RAJA- tional fidelity, by preventing the introduction of het- GOPALAN for advice about the manuscript, and GRAHAM WALKER, erologous sequences by horizontal transfer between NANCY KLECKNER,ROBERT LLOYD and MARGARETLIEB for strains. species (RAYSSIGUIER,THALER and RADMAN 1989) This work was supported by a grant from theNational Institutes of Health (CA41655). and by limiting the introductionof vertical divergence by large rearrangements (this work). LITERATURECITED Importance of controlled homeologous recombi- ANDERSON, R.P., C. G. MILLERand J. R. ROTH, 1976 Tandem nation between gene families: Recombination be- duplications of the histidine observed following gener- tween members of the rm gene family is another alized transduction in Salmonella typhimurium. J. Mol. Biol. 105: known possibility for making large rearrangementsof 201-218. the E. coli chromosome(HILL, HARVEYand GRAY ANDERSON,R. P., and J. R. ROTH, 1977 Tandem gene duplica- 1990). The rrn operons are known to contain se- tions in phage and bacteria. Annu. Rev. Microbiol. 31: 473- 505. quencedivergence that couldappear as mismatch BAILIS,A. M., and R. ROTHSTEIN,1990 A defect in mismatch protein targets in the course of the recombination repair in Saccharomyces cereuisiae stimulates ectopic recombina- process UINKS-ROBERTSONand NOMURA1987); for tion between homeologous genesby an excision repair depend- example,about 1% divergence was observed in a ent process. Genetics 126 535-547. CAMPBELL,A., 1965 The steric effect in lysogenization by bacte- sequence comparison between 16s RNA of rrnG and riophage lambda. 1. Lysogenization of a partially diploid strain rrnB (SHEN,SQUIRES and SQUIRES1982). In addition, of Escherichia coli K12. Virology 27: 329-339. the spacer tRNA sequences might serveas targets for CLARK,A. J., 1973 Reconlbinationdeficient mutantsof E. coli a special recognition system. For the single re- and other bacteria. Annu. Rev. Genet. 7: 67-86. arrangement investigated in this study, the occurrence DASGUPTA, C., and C.M. RADDING, 1982Lower fidelity of RecA protein catalysed homologous pairing with a super helical sub- can be as high as 0.1 % in the absence of mutL or muts strate. Nature 295: 71-73. gene products; underthese conditions, the cumulated DIMPFL,J., and H. ECHOLS,1989 Duplicationmutagenesis as an probability of a rearrangement between any two mem- SOS response in E. coli: enhanced duplication formation by a bers of agene family couldreach 5%. Thus, we constitutively-activated RecA. Genetics 123: 255-260. presume that long-term genomic stability requires a FEINSTEIN,S. I., and K.B. LOW, 1986 Hyper-recombining recip- ient strains in bacterial conjugation. Genetics 113: 13-33. general mechanism for negative control of intrachro- FEULNER, G.,J. A. GRAY,J. A. KIRSCHMAN, A.F. LEHNER,A. B. mosomal recombination and unequal crossing over SAWSKY,D. A. VLAZNY,J. ZHANG,S. ZHAOand C. W. HILL, 332 M.-A. Petit et al.

1990 Structure of the rhsA locus from Escherichia coli K-12 LUNDRLAD,V., and N. KLECKNER, 1984 Mismatch rep.'llr muta- and comparison of rhsA with other members of the rhs multi- tions of Escherichia coli K12 enhance transposon excision. <;e- gene f:,unily.J. Bacteriol. 172: 446-456. netics 109: 3-1 9. (;RII.I.EY, M..K. M. WELCH, S. S. SU andP. MODRICH, MODRICH,P., 1989 Methyl-directedDNA ntismatch correction. 1989 Isolation andcharacterization of the Escherichia coli J. Biol. Chem. 264: 6597-6600. mutL gene product. J. Biol. Chem. 264: 1000-1004. PANG, P. P., A. S. LUNDBERGandG. C. WALKER, I III.I., <:. W., and B. W. HARNISH, 1981 Inversions between ribo- 1985 Identification andcharacterization of the mutL md son~alRNA genesof Escherichia coli. Proc. Natl.Acad. Sci. mutS gene products of Salmonella typhimurium LT2. J. Bacte- USA 78: 7069-7072. riol. 163: 1007-1015. I III.~.,<:. W., and B. W. HARNISH, 1982 Transposition of a chro- PETES, P. D.,and C. W.HILL, 1988 Recombinationbetween ~noson~alsegment bounded by redundant rRNA genes into repeated genesin microorganisms. Annu. Rev. Genet. 22: 147- other rRNA genes in Escherichia coli. J. Bacteriol. 149 449- 168. RADMAN,M., 1989 Mismatch repairand genetic reco~nbination, 457. pp. 169- 192 in GeneticRecombination, edited by K. KUCHER- I 111.1.. C. W., S. HARVEY andJ. A. GRAY,1990 Recombination LAPA'I'I and G. R. SMITH. American Society of Microbiology, bctween rRNA genes in Escherichia coli and Salmonella typhi- Washington, D.C. murium, pp. 335-340 in TheBacterial Chromosome, edited by RAYSSIGUIER,C., D. S. THALER and M. RADMAN,I989 The K. Drlica and M. Riley. American Society for Microbiology, barrier to recotnbination between Escherichia coli and Salmo- W;~shington,DC. nellatyphimurium is disrupted in mismatch-repair nuta ants. I III.I.,C. W., R. H. GRAFSTROM,B. W. HARNrsHand B. S. HILLMAN, Nature 342: 396-401. 1977 Tandem duplications resulting from recombination be- SADOSKY, A.B., A. DAVIDSON,R. J. LIN and C. W. HILI., 1989 rhs tween ribosomal RNA genes in Escherichia coli. J. Mol. Biol. gene family of Escherichia coli K-12. J. Bacteriol. 171: 636- 116: 407-428. 642. I IOI.MES, J., S. CLARKand P. MODRICH,1990 Strand-specific SHEN,P., and H. V. HUANG,1989 Effect of base pair mismatches mismatch correction in nuclear extractsof human andDrosoph- on recotnbination via the RecBCD pathway. Mol. Gen. Genet. ila melanogastercell lines. Proc. Natl. Acad. Sci. USA 87: 5837- 218: 358-360. 5841. SHEN,W. F., C. SQUIRESand C. I-. SQUIRES,1982 Nucleotide JINKS-ROBERTSON,S., and M. NOMURA,1987 Ribosomes and sequence of the rrnG ribosomal RNApromoter region of tRNA, pp. 1358-1385 in Escherichia coli and Salmonella typhi- Gscherichia coli. Nucleic Acids Res. 10: 3303-33 13. murium: Cellular and Molecular Biology, edited by F. C. NEID- SMITH,G. R., 1987 Mechanisnl and control of homologous reco~tb HARDT.American Society for Microbiology, Washington, D.C. bination in Escherichia coli. Annu. Rev. Genet. 21: 179-201. I.IN, K. J.. M. CAPAGE andC. W. HILL, 1984 A repetitive DNA SONTI,R. V., and J. K. ROIH, 1989 Role of gene duphcations in sequence, rhs, responsible forduplications within the Esche- the adaptation of Salmonella typhimurium to growth 011 lintiting richia coli K12 chromoson~e.J. Mol. Biol. 177: 1-18. carbon sources. Genetics 123: 19-28. I.I.OYII, K. G., C. BUCKMANand F. E. BENSON,1987 Genetic THOMAS, D. C., J. D. ROBERTS and T. A. KUNKEI., 1993 Heteroduplex repair in extracts of human HeLa cells. :ulalysis of conjugational recombination in Escherichia coli K 12 J. Biol. Chetlt. 266: 3744-3751. strains deficient in RecBCD enzyme. J. Gen. Microbiol. 133: 253 1-2538. Communicating editor: J. W. DRAKE