Portal de Periódicos da CAPES

Recombination-dependent Repair of DNA Double-strand Breaks with Purified Proteins from Escherichia coli

1997; Elsevier BV; Volume: 272; Issue: 27 Linguagem: Inglês

10.1074/jbc.272.27.17091

1083-351X

P. Morel, Dmitry Cherny, S. Dusko Ehrlich, Era Cassuto,

CRISPR and Genetic Engineering

We have developed an in vitro system in which repair of DNA double-strand breaks is performed by purified proteins of Escherichia coli. A segment was deleted from a circular duplex DNA molecule by restriction at two sites. 3′ single-stranded overhangs were introduced at both ends of the remaining linear fragment. In a first step, RecA protein catalyzed the formation of a d-loop between one single-stranded tail and a homologous undeleted supercoiled DNA molecule. In a second step,E. coli DNA polymerase II or III used the 3′ end in thed-loop as a primer to copy the missing sequences of the linear substrate on one strand of the supercoiled template. Under proper conditions, the integrity of the deleted substrate was restored, as shown by analysis of the products by electrophoresis, restriction, and transformation. In this reaction, DNA synthesis is strictly dependent on recombination, and repair is achieved without formation of a Holliday junction. We have developed an in vitro system in which repair of DNA double-strand breaks is performed by purified proteins of Escherichia coli. A segment was deleted from a circular duplex DNA molecule by restriction at two sites. 3′ single-stranded overhangs were introduced at both ends of the remaining linear fragment. In a first step, RecA protein catalyzed the formation of a d-loop between one single-stranded tail and a homologous undeleted supercoiled DNA molecule. In a second step,E. coli DNA polymerase II or III used the 3′ end in thed-loop as a primer to copy the missing sequences of the linear substrate on one strand of the supercoiled template. Under proper conditions, the integrity of the deleted substrate was restored, as shown by analysis of the products by electrophoresis, restriction, and transformation. In this reaction, DNA synthesis is strictly dependent on recombination, and repair is achieved without formation of a Holliday junction. Double-strand breaks in DNA result essentially from exposure of the cells to ionizing radiations or defective repair of clustered single-strand lesions. In yeast, they also occur spontaneously during meiosis, particularly at homologous recombination hot spots (1Orr-Weaver T.L. Szostak J.W. Proc. Natl. Acad. Sci. U. S. A. 1983; 80: 4417-4421Crossref PubMed Scopus (304) Google Scholar). In the absence of an efficient repair system, the introduction of a double-strand break in a chromosome constitutes a lethal event.In mammalian cells the repair of double-strand breaks seems to be predominantly achieved by nonconservative mechanisms such as direct strand rejoining (2Morris T. Thacker J. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 1392-1396Crossref PubMed Scopus (126) Google Scholar, 3Phillips J.W. Morgan W.F. Mol. Cell. Biol. 1994; 14: 5794-5803Crossref PubMed Scopus (108) Google Scholar) or single-strand annealing (4Lin F.L. Sperle K. Sternberg N. Mol. Cell. Biol. 1984; 10: 103-112Crossref Scopus (85) Google Scholar), whereas in yeast or bacteria, homologous recombination appears to be the major way to overcome double-strand breaks. In Saccharomyces cerevisiae, most of the genes required for homologous recombination fall into the RAD52 epistatic group (5 and references therein). Mutations in these genes render cells sensitive to ionizing radiations (6Mortimer R.K. Rad. Res. 1958; 9: 3312-3316Crossref Scopus (106) Google Scholar) and to chemicals known to induce DNA breakage but do not significantly affect UV sensitivity. In bacteria, the repair of double-strand breaks is an inducible SOS function and requires the presence of another DNA duplex with the same base sequence as that of the broken helix (7Krasin F. Hutchinson F. J. Mol. Biol. 1977; 116: 81-98Crossref PubMed Scopus (152) Google Scholar). Rapidly growing Escherichia coli cells have multiple copies of most of their genome and exhibit increased resistance to killing by x-rays, showing that breaks are repaired efficiently (8Pollard E.C. Fluke D.J. Kazanis D. Mol. & Gen. Genet. 1981; 184: 421-429Crossref PubMed Scopus (11) Google Scholar). The repair process is clearly conservative as the sequences of the participating DNA molecules are recovered.Several models have been proposed for the repair of double-strand breaks by homologous recombination (9Resnick M.A. J. Theor. Biol. 1976; 59: 97-106Crossref PubMed Scopus (345) Google Scholar, 10Szostak J.W. Orr-Weaver T.L. Rothstein R.J. Cell. 1983; 33: 25-35Abstract Full Text PDF PubMed Scopus (1745) Google Scholar, 11Thaler D.S. Stahl F.W. Annu. Rev. Genet. 1988; 22: 168-197Crossref Scopus (120) Google Scholar). The common feature is the exonucleolytic degradation on each side of the break, which exposes 3′ single-stranded (ss) 1The abbreviations used are: ss, single-stranded; SSB protein, single-strand binding protein; Pol, polymerase; RF, replicative form; wt, wild type; bp, base pair(s); ATPγS, adenosine 5′-O-(thiotriphosphate). 1The abbreviations used are: ss, single-stranded; SSB protein, single-strand binding protein; Pol, polymerase; RF, replicative form; wt, wild type; bp, base pair(s); ATPγS, adenosine 5′-O-(thiotriphosphate). tails. One 3′ end invades the donor duplex to form a d-loop and primes DNA synthesis on one strand of the duplex while the other strand is displaced. According to Szostak et al. (10Szostak J.W. Orr-Weaver T.L. Rothstein R.J. Cell. 1983; 33: 25-35Abstract Full Text PDF PubMed Scopus (1745) Google Scholar), as synthesis progresses, the displaced strand anneals to the noninvading 3′ ss end and is then used as template for repair synthesis, leading to the formation of two Holliday junctions. Resolution can occur with or without crossing over, and all products carry newly synthesized DNA. However, a number of genetic data obtained in a variety of organisms cannot be accounted for by a such a symmetrical mechanism. Belmaaza and Chartrand (12Belmaaza A. Chartrand P. Mutat. Res. 1994; 314: 199-208Crossref PubMed Scopus (92) Google Scholar) proposed an alternative model in which the newly synthesized strand is released from the d-loop and anneals to the noninvading 3′ ss tail. In this case, only one strand of the donor duplex serves as template for DNA synthesis, and repair occurs without formation of a Holliday junction. This mechanism is strongly reminiscent of a reaction that occurs in T4-infected cells, which converts recombination intermediates into replication forks (13Mosig G. Karam J.D. Molecular Biology of Bacteriophage T4. American Society for Microbiology, Washington, D. C.1994: 54-82Google Scholar), and was substantiated by the biochemical studies of Formosa and Alberts (14Formosa T. Alberts B.M. Cell. 1986; 47: 793-806Abstract Full Text PDF PubMed Scopus (198) Google Scholar). The in vitro reaction was strictly dependent on (i) the presence of the UvsX and gene 32 proteins (the T4 equivalents ofE. coli RecA and SSB protein); (ii) T4 DNA polymerase and its accessory proteins; (iii) homologous ss and duplex DNA substrates, and was greatly stimulated by the T4 dda helicase.The enzymes used in the T4 in vitro system have functional equivalents in E. coli, and there is little doubt that precise repair of double-strand breaks requires the cooperation of recombination and replication. In this report, we describe the reconstitution of a repair reaction with a mixture of purified E. coli recombination and replication proteins. The results support a mechanism in which RecA protein promotes one-sided invasion of a duplex donor by a linear homologous DNA bearing 3′ ss overhangs. The invading end serves as a primer for DNA synthesis by DNA polymerase II or III, which copy one strand of the duplex template. Repair is achieved without formation of a Holliday junction.RESULTSThe strategy developed to test for recombination-dependent repair is illustrated in Figs.1 and 2. The major difference between phage M13 wt and its derivative M13mp19 is the lac insert carried by the latter. In addition, the BamHI site at nucleotide 2220 in M13 wt has been moved to nucleotide 6264 in M13mp19 (Fig. 1). Treatment of M13mp19 RF I with XmnI produces two fragments of 4961 and 2288 bp (Fig. 2 A, a andb). The smaller fragment was discarded and the larger fragment subjected to controlled treatment with λ exonuclease, an enzyme specific for the 5′ ends of duplex DNA. Thus, 3′ ss tails were introduced on both strands of this fragment (Fig. 2 A,c). We expected that in the presence of RecA, 3′ ss tails would invade supercoiled M13 wt RF I DNA to form a d-loop (23Shibata T. Das Gupta C. Cunningham R.P. Radding C.M. Proc. Natl. Acad. Sci U. S. A. 1980; 77: 2606-2610Crossref PubMed Scopus (64) Google Scholar) and that the invading end would serve as a primer for DNA synthesis on the intact M13 wt template (Fig. 2 A,d). If this were the case, elongation of the newly synthesized strand might provide a complete copy of the 2288-bp fragment, which, except for a single nucleotide change, is identical in M13 and M13mp19. In principle, if no Holliday junction is formed, displacement or unwinding of the elongated strand from its template and annealing to the noninvading 3′ ss tail may lead to conservative repair of the deleted M13mp19 DNA (Fig. 2 B, b1,b2, and b3). Alternatively, if synthesis takes place on both strands of the donor duplex, resolution will be needed to release the repaired products.Figure 1Partial restriction map of M13 wt and M13mp19. Only the restriction sites relevant to this work are indicated. kb, kilobases.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Figure 2Experimental system. Formation of joint molecules (panel A) and invasion-primed synthesis (panel B). Panel A: a, schematic representation of the structure of M13mp19 (7249 bp). The two possible sites of invasion, XmnI-1 and XmnI-2, are shown.BglII and PacI are indicated for orientation. Here invasion is shown at XmnI-1, but it can occur atXmnI-2 with equal probability. The lac insert is absent in M13 wt. b, fragments produced by digestion withXmnI. Arrows correspond to the 3′ end.c, treatment of the 4961-bp fragment with λ exonuclease followed by incubation with M13 wt RFI (6404 bp, only the relevant region is shown), RecA, and SSB (d) to produce joint molecules. Panel B: three stages of invasion-primed synthesis are illustrated in a1, b1, andc1. In each case, the expected result of strand displacement and, when applicable, annealing to the noninvading 3′ ss tail are shown in a2, b2, and c2. In contrast witha2, the intermediates in b2 and c2 can in principle sustain repair synthesis and ligation, leading to the formation of b3 and c3. In all cases, the M13 wt template should be recovered intact. See “Results” for more details.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Formation and Structure of Joint MoleculesTo determine the requirements for optimal formation of joint molecules, we used a32P-labeled M13mp19 large XmnI fragment with 3′ ss tails. Analysis by electrophoresis showed that the formation and yield of joint molecules depended on several factors (Fig.3). As expected, the reaction required the presence of RecA, SSB, and ATP or ATPγS (adenosine 5′-O-thiotriphosphate), as well as homology between the linear substrate and the circular template (no joint molecules were formed if M13 wt RF I was replaced by φX174 RF I). Superhelicity of the template was also required, since either relaxation with topoisomerase I or nicking of the template (not shown) with gpII protein, the product of M13 gene II (24Greenstein D. Horiuchi K. J. Mol. Biol. 1987; 197: 157-174Crossref PubMed Scopus (26) Google Scholar), considerably lowered the yield of joint molecules. In addition, the reaction was strictly dependent on the presence of 3′ ss tails on the linear substrate, as the undigested fragment (not shown) or the fragment carrying 5′ ss tails originating from digestion of the 4961-bp XmnI fragment with exonuclease III did not produce joint molecules. This is consistent with the 5′ to 3′ polarity of the binding of RecA to ssDNA (25Register III, J.C. Griffith J. J. Biol. Chem. 1985; 260: 12308-12312Abstract Full Text PDF PubMed Google Scholar, 26Cassuto E. Howard-Flanders P. Nucleic Acids Res. 1986; 14: 1149-1157Crossref PubMed Scopus (16) Google Scholar) and of RecA-catalyzed branch migration (27Cox M.M. Lehman I.R. Proc. Natl. Acad. Sci. U. S. A. 1981; 78: 6018-6022Crossref PubMed Scopus (137) Google Scholar).Figure 3Requirements for the formation of joint molecules (JM). The 32P-labeled 4961-bpXmnI fragment from M13mp19, digested with λ exonuclease (3′ ss tails) or exonuclease III (5′ ss tails), migrates slightly slower than the M13mp19 RF I marker. Lane 1 contains the 4961-bp fragment with 5′ ss tails. All other lanes contain the 4961-bp fragment with 3′ ss tails. Lane 2, no RecA. Lane 3, no SSB. Lane 4, complete reaction. Lane 5, complete reaction with topoisomerase I. Lane 6, φX174 RFI DNA instead of M13 wt. Lane 7, M13wt/linear substrate (w/w) = 0.3. Lane 8, complete reaction with gyrase. Lane 9, complete reaction with ATPγS instead of ATP. Lane 10, no ATP.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Aside from these absolute requirements, the yield of joint molecules was strongly influenced by the total concentration of DNA and proteins and by the relative concentration of the substrates. The length of the ss tails seemed to have no effect within the available range (250–700 nucleotides), nor did the addition of DNA gyrase. Under the optimal conditions described under “Materials and Methods,” a substantial fraction of the input radioactivity was found in joint molecules (Fig.3, lane 4).Joint molecules were analyzed by electron microscopy after removal of free proteins. Out of several hundred DNA molecules examined, close to 30% of the linear molecules were joined to circular molecules, whereas no joint molecules were found in controls lacking RecA. As expected, all joint molecules resulted from one-sided invasion of the M13 wt template (see Fig. 4 for representative examples), since RecA-promoted invasion of a supercoiled DNA molecule by ssDNA induces unwinding of the duplex (28Cunningham R.P. Shibata T. Das Gupta C. Radding C.M. Nature. 1979; 281: 191-195Crossref PubMed Scopus (78) Google Scholar), thus preventing invasion at another site. Most ss tails were covered by RecA, but the coverage appeared to be somewhat irregular, possibly because of the presence of ATP (instead of ATPγS) in the reaction (29Egelman E.H. Stasiak A. J. Mol. Biol. 1986; 191: 677-697Crossref PubMed Scopus (188) Google Scholar).Figure 4Electron microscopic visualization of joint molecules. Grids were prepared and examined as described under “Materials and Methods.” Examples of joint molecules are shown inpanels A, B, and C. The input linear and supercoiled substrates are shown in panel D. Thick arrows indicate the position of the joint, thin arrowsthe 3′ ss overhangs. The scale bar corresponds to 200 nm. For clarity, the linear and circular DNAs in theinsets are drawn, respectively, with a thin andthick line.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Invasion-primed DNA SynthesisIn preliminary control experiments (not shown), the linear and circular starting substrates were incubated separately with Pol I, Pol II, or Pol III under the conditions described under “Materials and Methods” for invasion-primed DNA synthesis, or together without RecA. With Pol II and Pol III, no incorporation was observed with either substrate in the presence of RecA, although traces of radioactivity, supposedly due to self-priming, could be detected in the absence of RecA at the position of the linear substrate. However, in controls with Pol I, incorporation probably due to nick translation and/or strand displacement produced species migrating from the original positions of the substrates to the position of joint molecules and above. To minimize such events in further experiments with Pol I, joint molecules made with a3H-labeled large XmnI fragment with 3′ ss tails were purified as described under “Materials and Methods.” This step completely removed the free linear substrate, but a substantial amount of M13 RFI was still present in the preparation.The ability of the invading 3′ ss tails to serve as primers for DNA synthesis was tested with the three DNA polymerases in the presence or absence of gyrase and helicase II (UvrD). T4 DNA ligase was present in all reactions, as were RecA and SSB, the former to favor annealing of potential complementary single strands, the latter because it stimulates Pol II and Pol III activity. Analysis of the products on agarose gels showed that with all three polymerases, radioactivity was found at the position of joint molecules after 5 min of incubation (Fig. 5, top lanes). With Pol I, however, incorporation also took place at and above the position of the starting substrates (lanes a–c), as observed in the controls. After a 30-min incubation (Fig. 5, bottom lanes), radioactivity at the position of joint molecules decreased in reactions with Pol II and Pol III (but not with Pol I), and a fraction of the incorporated label was found at the position of M13mp19 RF II DNA. In the presence of gyrase, radioactivity was also found at the position of M13mp19 RF I DNA. Gyrase or UvrD stimulated the formation of M13mp19-like species, but stimulation was not increased when both proteins were present in the same reaction.Since M13mp19-like species could also result from contamination of any of the enzymes used in these experiments by an activity capable of resolving Holliday junctions, we incubated a synthetic junction with Pol II with or without UvrD or gyrase and with Pol III, under the exact conditions used for invasion-primed synthesis. As a positive control, the synthetic junction was also incubated with T4 endonuclease VII (21Shah R. Bennett R.J. West S.C. Cell. 1994; 79: 853-864Abstract Full Text PDF PubMed Scopus (135) Google Scholar,30Mizuuchi K. Kemper B. Hays J. Weisberg R.A. Cell. 1982; 29: 357-365Abstract Full Text PDF PubMed Scopus (185) Google Scholar). As shown in Fig. 6, the synthetic junction was resistant to all treatments with the exception of T4 endonuclease VII.Figure 6M13mp19-like species (forms I and II) are not due to contamination by a Holliday junction resolvase. Synthetic Holliday junctions (1 ng) were incubated under the conditions described in Fig. 5, lanes d–g. Incubation was for 30 min at 37 °C. Lane a, Holliday junctions, no addition;lane b, with Pol II; lane c, with Pol II and UvrD; lane d, with Pol II and gyrase; lane e, with Pol III; lane f, with T4 endonuclease VII (10 units).View Large Image Figure ViewerDownload Hi-res image Download (PPT)Restriction Analysis of M13mp19-like DNATo verify the identity of the RF I and RF II M13mp19-like species, invasion-primed synthesis reactions containing gyrase were set up as in Fig. 5,lanes f and i, and electrophoresed on agarose gel. Bands at the position indicated by the RF I and RF II markers were cut out and the DNA electroeluted and cleaved with XmnI orBamHI. The former enzyme should produce two bands from M13mp19 DNA, the invading 4961-bp fragment and the 2288-bp fragment originally deleted (Figs. 1 and 2). These fragments were observed in all cases (Fig. 7, lanes e–h).Figure 7Restriction analysis of the products of invasion-primed DNA synthesis catalyzed by Pol II and Pol III. For details, see “Results.” Primed synthesis reactions were carried out with Pol II and Pol III in the presence of gyrase, as in Fig. 5,lanes f and i, and electrophoresed. Bands at the position indicated by the M13mp19 RF I and RF II markers were excised. The DNA was electroeluted and digested with BamHI orXmnI and the digests electrophoresed. A BstEII digest of λ DNA was run as a size marker. Lanes a–d,BamHI. Lanes e–h, XmnI. Lanes a, b, e, and f, Pol II reaction.Lanes c, d, g, and h, Pol III reaction. Lanes a, c, e, andg, digests of RF II-like DNA. Lanes b,d, f, and h, digests of RF I-like DNA. In lanes e and g, the extra fragment at the position of linear M13mp19 DNA (7249 bp) is presumably due to cleavage at only one of the two XmnI sites. Markers on theleft indicate the position of the DNA eluted from the excised bands prior to digestion.View Large Image Figure ViewerDownload Hi-res image Download (PPT)BamHI cuts both M13mp19 and M13 wt at one location, the former within the lac insert, the latter within the region corresponding to the 2288-bp XmnI fragment (Fig. 1). Since the newly acquired 2288-bp fragment of M13mp19 must be a copy of the corresponding fragment in M13, repaired M13mp19 molecules should carry two BamHI sites, and cleavage at both sites should yield two fragments of 4044 and 3205 bp. Again, the expected fragments were seen in all cases (Fig. 7, lanes a–d), confirming the structure of the “repaired” RF I and RF II species. In all digests (lanes a–h), the higher radioactivity in the smaller fragment indicates that synthesis occurred predominantly in the originally deleted region of the M13mp19 chromosome, as expected from the model in Fig. 2 B.TransformationReconstitution in vitro of viable M13mp19 phage from the large XmnI fragment and M13 wt DNA was also examined by a transformation test. M13mp19 can be identified by its ability to perform α-complementation in alacI q lacZΔM15 host and to form blue plaques in the presence of 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside and isopropyl-1-thio-β-d-galactopyranoside, whereas M13 wt will form colorless plaques (22Langley K.E. Villarejo M.R. Fowler A.V. Zamenhof P.J. Zabin I. Proc. Natl. Acad. Sci. U. S. A. 1975; 72: 1254-1257Crossref PubMed Scopus (152) Google Scholar).Primed synthesis reactions (scaled up to a total volume of 50 μl) were carried out with Pol I, II, and III, with and without gyrase or UvrD and electrophoresed on agarose gel alongside M13mp19 RF I and RF II markers. Bands at the positions indicated by the markers were excised and the DNA eluted by a modification of the freeze-squeeze method (31Thuring R.W. Sanders J.B. Borst P.A. Anal. Biochem. 1975; 66: 213-220Crossref PubMed Scopus (351) Google Scholar) and introduced into competent JM109 (recA) cells as described under “Materials and Methods.” 93 to 97% of the total number of plaques (several hundred) issued from Pol II and Pol III reactions were blue. No blue plaques were seen for reactions with Pol I or when the cells were transformed with a mixture of the starting substrates incubated without RecA (Table I).Table IYield of repaired DNARF I bandRF II band%Pol I0Pol I +Gyrase00Pol I +UvrD0Pol II12Pol II +Gyrase413Pol II +UvrD23Pol III6Pol III +Gyrase29Pol III +UvrD13The yield was estimated on the basis of the following considerations. One joint molecule can produce at best one molecule of M13mp19 (size 7.25 kilobase pairs), or 64% of input (w/w). As the primed synthesis reactions contained 80 ng of joint molecules, 100% repair would yield 51 ng of M13mp19 DNA. In preliminary experiments, we found that recovery of elution of M13mp19 RF I or RF II DNA from excised gel bands was close to 40% and that after phenol extraction, ethanol precipitation, and dialysis, the transformation frequency of the eluted DNA was 8 × 104 plaque-forming units/mg in strain JM 109 (recA). Therefore 100% repair would correspond approximately to 20 ng of M13mp19 DNA or 1.6 × 103plaque-forming units. Results are the number of observed blue plaques/number expected for complete repair of joint molecules. Open table in a new tab DISCUSSIONIn this paper, we describe a recombination-dependentin vitro system for repair of double-strand breaks. In a first step, RecA promoted the formation of joint molecules between a covalently closed duplex DNA donor and a deleted linear homologous substrate bearing 3′ ss tails, one of which invaded the donor DNA. In a second step, Pol II and Pol III used the invading end as a primer to copy the deleted sequences on one strand of the template.As seen in Fig. 2 B, the precision of the repair process should depend on the extent of DNA polymerization. Complete repair can only occur if the end of the newly synthesized strand is complementary (and therefore can anneal) to sequences contained in the noninvading ss tail of the linear substrate (b2). In such a case, repair synthesis and ligase must carry out the final steps (b3). If invasion-primed synthesis does not proceed far enough (a1), the product will correspond to the original linear substrate with one extended ss tail (a2). If, on the contrary, primed synthesis proceeds past the noninvading ss tail, the product will be an open duplex circle with an extra whisker (c3).Analysis of the results of primed synthesis (Fig. 5) indicates that the three kinds of products may be formed. Only the DNA species migrating at the position of M13mp19 RF I in the presence of gyrase reflect complete repair. Ligase and gyrase were present in sufficient concentration to close and supercoil an amount of nicked M13 DNA corresponding to several times the amount of DNA in the reaction mixtures, as determined in preliminary assays. We therefore assume that the M13mp19 RF II-like species that are refractory to ligation, hence to supercoiling, correspond to duplex circles with short ss gaps and/or ss tails (Fig. 2 B, b2, c2, andc3). It should be noted that species with long whiskers, which would migrate well above the M13mp19 RF II marker, will not be taken into account in this assay. Smears (or even bands) between the RF I and RF II markers seem to be produced by displacement from the joint molecules of newly synthesized strands too short to anneal to the noninvading ss tail. Radioactivity at and above the position of joint molecules is likely to reflect synthesis in progress or displacement failure.The results of restriction analysis and transformation experiments confirm the general structure of the repaired DNA but do not guarantee that the newly synthesized strand is a perfect copy of its template. The region deleted in the starting linear substrate is essential for phage viability. Gross rearrangements during invasion-primed synthesis would interfere with the life cycle of the phage and are therefore unlikely. But mismatches, for example, would not be detected by restriction analysis and would be eliminated in the cells upon transformation, as would other errors that can be repaired in the absence of RecA. Imperfect products, such as circles with short tails or gaps, could easily be trimmed or filled in the cells and produce viable phage.In vivo, repair synthesis is thought to be carried out by Pol I or Pol III, depending on the size of the repair patch, and in some cases by Pol II (32Kornberg A. Baker T.A. DNA Replication. 2nd Ed. W. H. Freeman and Co, New York1992: 113-169Google Scholar). In the present experiments, the results obtained with Pol I are open to several interpretations: the label found around the position of joint molecules could result from primed synthesis without release of the new strand, but also reflects incorporation at nicks, as observed in the controls. Moreover, because of its low processivity and accessory degradative activities, Pol I could also produce new strands too short to anneal to the noninvading 3′ ss tail. Whatever the reason, M13mp19-like forms were never detected in our experiments with Pol I, which confirms that the RF I and RF II species obtained with Pol II and Pol III are not due to nick translation or strand displacement. In addition, the results in Fig. 6eliminate a potential contamination by a Holliday junction resolvase. The activities of Pol II and Pol III in the repair of double-strand breaks are similar, although the former is apparently more efficient. Whether these differences among the three polymerases hold truein vivo remains to be determined.Both UvrD and gyrase increased the yield of M13mp19-like structures but do not seem to act in a concerted or cooperative way since stimulation by either protein was not improved by the addition of the other. The effect of UvrD is reminiscent of that of the T4 dda helicase which stimulates recombination-dependent DNA synthesis in vitro (14Formosa T. Alberts B.M. Cell. 1986; 47: 793-806Abstract Full Text PDF PubMed Scopus (198) Google Scholar). Interestingly, both dda and UvrD are known to stimulate, respectively, UvsX- and RecA-promoted strand exchange (16Morel P. Hejna J.A. Ehrlich S.D. Cassuto E. Nucleic Acids Res. 1993; 21: 3205-3209Crossref PubMed Scopus (68) Google Scholar,33Kodadek T. Alberts B.M. Nature. 1987; 326: 312-314Crossref PubMed Scopus (52) Google Scholar, 34Kodadek T. J. Biol. Chem. 1991; 266: 9712-9718Abstract Full Text PDF PubMed Google Scholar), and neither has been shown to play a direct role in replication. However, UvrD does stimulate DNA synthesis by Pol III at an artificial replication fork (35Kuhn B. Abdel-Monem M. Eur. J. Bioche