Recombinant levels of Escherichia coli K-12 mutants deficient in various replication, recombination, or repair genes.

Escherichia coli strains containing mutations in lexA, rep, uvrA, uvrD, uvrE, lig, polA, dam, or xthA were constructed and tested for conjugation and transduction proficiencies and ability to form Lac+ recombinants in an assay system utilizing a nontandem duplication of two partially deleted lactose operons (lacMS286phi80dIIlacBK1). lexA and rep mutants were as deficient (20% of wild type) as recB and recC strains in their ability to produce Lac+ progeny. All the other strains exhibited increased frequencies of Lac+ recombinant formation, compared with wild type, ranging from 2- to 13-fold. Some strains showed markedly increased conjugation proficiency (dam uvrD) compared to wild type, while others appeared deficient (polA107). Some differences in transduction proficiency were also observed. Analysis of the Lac+ recombinants formed by the various mutants indicated that they were identical to the recombinants formed by a wild-type strain. The results indicate that genetic recombination in E. coli is a highly regulated process involving multiple gene products.     

Repair of cross-linked DNA and survival of Escherichia coli treated with psoralen and light: effects of mutations influencing genetic recombination and DNA metabolism.

Repair of cross-linked DNA was studied in Escherichia coli strains carrying mutations affecting DNA metabolism. In wild-type cells, DNA strands cut during cross-link removal were rejoined during a subsequent incubation into high-molecular-weight molecules. This rejoining was dependent on gene products involved in genetic recombination. A close correlation was found relating recombination proficiency, the rate of strand rejoining, and formation of viable progeny after DNA cross-linking by treatment with psoralen and light. Wild-type cells and other mutants which were Rec+ (sbcB, recL, recL sbcB, recB recC sbcA, recB recC sbcB, xthA1, and xthA11) rejoined cut DNA strands at a rate of 0.8 +/- 0.1 min -1 at 37 degrees C and survived 53 to 71 cross-links per chromosome. recB, recC, recB recC, recF, or polA strains showed reduced rates of strand rejoining and survived 4 to 13 cross-links per chromosome. Recombination-deficient strains (recA, recB recC sbcB recF, recB recL) and lexA failed to rejoin DNA strands after crosslink removal and were unable to form colonies after treatments producing as few as one to two cross-links per chromosome. Strand rejoining occurred normally in cells with mutations affecting DNA replication (dnaA, danB, dnaG, and dnaE) under both permissive and nonpermissive conditions for chromosome replication. In a polA polB dnaE strain strand rejoining occurred at 32 degree C but not at 42 degree C, indicating that some DNA synthesis was required for formation of intact recombinant molecules.     

Spontaneous and cisplatin-induced recombination in Escherichia coli

To measure cisplatin (cis-diaminodichloroplatinum(II))-induced recombination, we have used a qualitative intrachromosomal assay utilizing duplicate inactive lac operons containing non-overlapping deletions and selection for Lac+ recombinants. The two operons are separated by one Mb and conversion of one of them yields the Lac+ phenotype. Lac+ formation for both spontaneous and cisplatin-induced recombination requires the products of the recA, recBC, ruvA, ruvB, ruvC, priA and polA genes. Inactivation of the recF, recO, recR and recJ genes decreased cisplatin-induced, but not spontaneous, recombination. The dependence on PriA and RecBC suggests that recombination is induced following stalling or collapse of replication forks at DNA lesions to form double strand breaks. The lack of recombination induction by trans-DDP suggests that the recombinogenic lesions for cisplatin are purine-purine intrastrand crosslinks.     

Genetic Analysis of Mutations Indirectly Suppressing recB and recC Mutations

Mutations in sbcB inactivate exonuclease I and suppress the UV-sensitive, mitomycin-sensitive, recombination-deficient phenotypes associated with recB and recC mutations. Mapping experiments have located sbcB about 0.4 minutes from the his operon at 38.0 on the standard map of E. coli. This places sbcB between supD and his. A four-point cross shows that sbcB lies between P2 attH and his. P2 eduction deleting the his operon beginning with P2 attH also deletes sbcB and produces the expected exonuclease I deficiency and suppression of recB(-). The occurrence of chemical-mutagen-induced and spontaneous mutations indirectly suppressing recB(-) and recC(-) is examined. Three lines of strains produce only sbcA mutations while only sbcB mutations occur in a fourth line. Explanations for this behavior are proposed in light of the ability of the first three lines to express sbcB mutations which they inherit by transduction.     

Genetic and Physical Analysis of Plasmid Recombination in Recb Recc Sbcb and Recb Recc Sbca Escherichia Coli K-12 Mutants

The effect of mutations in known recombination genes (recA, recB, recC, recE, recF, recJ, recN, recO, recQ and ruv) on intramolecular recombination of plasmids was studied in recB recC sbcB and recB recC sbcA Escherichia coli mutants. The rate of recombination of circular dimer plasmids was at least 1000-fold higher in recB recC sbcB or recB recC sbcA mutants as compared to wild-type cells. The rate was decreased by mutations in recA, recF, recJ, recO, ruv or mutS in recB recC sbcB mutants, and by mutations in recE, recN, recO, recQ, ruv or mutS in recB recC sbcA mutants. In addition to measuring the recombination rate of circular dimer plasmids, the recombination-mediated transformation of linear dimer plasmids was also studied. Linear dimer plasmids transformed recB recC sbcB and recB recC sbcA mutants 20- to 40-fold more efficiently than wild-type cells. The transformation efficiency of linear dimer plasmids in recB recC sbcB mutants was decreased by mutations in recA, recF, recJ, recO, recQ or lexA (lexA3). In recB recC sbcA mutants the transformation efficiency of linear dimers was decreased only by a recE mutation. Physical analysis of linear dimer- or circular dimer-transformed recB recC sbcB mutants revealed that all transformants contained recombinant monomer genotypes. This suggests that recombination in recB recC sbcB cells is very efficient.     

Suppressors of Recb Mutations in Salmonella Typhimurium

Using a screen that directly assesses transductional proficiency, we have isolated suppressors of recB mutations in Salmonella typhimurium. The alleles of sbcB reported here are phenotypically distinct from those isolated in Escherichia coli in that they restore recombination proficiency (Rec(+)), resistance to ultraviolet light (UV(R)), and mitomycin C resistance (MC(R)) in the absence of an accompanying sbcCD mutation. In addition the sbcB alleles reported here are co-dominant to sbcB(+). We have also isolated insertion and deletion mutants of the sbcB locus. These null mutations suppress only the UV(S) phenotype of recB mutants. We have also isolated sbcCD mutations, which map near proC. These sbcCD mutations increase the viability, recombination proficiency and MC(R) of both the transductional recombination suppressors (sbcB1 & sbcB6) and the sbcB null mutations. S. typhimurium recB sbcB1 sbcCD8 strains are 15-fold more recombination proficient than wild-type strains. The increase in transductants in these strains is accompanied by a loss of abortive transductants suggesting that these fragments are accessible to the mutant recombination apparatus. Using tandem duplications, we have constructed sbcB merodiploids and found that, in a recB mutant sbcCD(+) genetic background, the sbcB(+) allele is dominant to sbcB1 for transductional recombination but co-dominant for UV(R) and MC(R). However, in a recB sbcCD8 genetic background, the sbcB1 mutation is co-dominant to sbcB(+) for all phenotypes. Our results lead us to suggest that the SbcB and SbcCD proteins have roles in RecBCD-dependent recombination.     

Recombinant plasmid associated cell aggregation and high-frequency conjugation of Streptococcus lactis ML3

Lactose-positive (Lac+) transconjugants resulting from matings between Streptococcus lactic ML3 and S. lactis LM2301 possess a single plasmid of approximately 60 megadaltons (Mdal) which is nearly twice the size of the lactose plasmid of the donor. The majority of these Lac+ transconjugants aggregated in broth and were able to transfer lactose-fermenting ability at a frequency higher than 10(-1) per donor on milk agar plates or in broth. Lac+ transconjugants which did not clump conjugated at a much lower frequency. Lactose-negative derivatives of Lac+ clumping transconjugants did not aggregate in broth and were missing the 60-Mdal plasmid. The ability to aggregates in broth was very unstable. Strains could lose the ability to clump but retain lactose-fermenting ability. The majority of these Lac+ nonclumping derivatives of clumping transconjugants contained a plasmid of approximately 33 Mdal, the size of the lactose plasmid of the original donor ML3. These strains transferred lactose-fermenting ability at a frequency of approximately 10(-6) per donor, resulting in both Lac+ clumping transconjugants which contained a 60-Mdal plasmid and Lac+ nonclumping transconjugants which possessed a 33-Mdal plasmid. Our results suggest that the genes responsible for cell aggregation and high-frequency conjugation are on the segment of deoxyribonucleic acid which recombined with the 33-Mdal lactose plasmid in S. lactis ML3.     

Linkage distortion following conjugational transfer of sbcC+ to recBC sbcBC strains of Escherichia coli.

Conjugational recombination in Escherichia coli depends normally on RecBCD enzyme, a multifunctional nuclease and DNA helicase produced by the recB, recC, and recD genes. However, recombination can proceed efficiently without RecBCD in recB or recC strains carrying additional mutations in both the sbcB and sbcC genes. Recombination in these strains, sometimes referred to as the RecF pathway, requires gene products that are not essential in the RecBCD-dependent process predominating in the wild type. It has also been reported to produce a different spectrum of recombinant genotypes in crosses with Hfr donors. However, the sbcC+ gene was unknowingly transferred to the recipient strain in some of these crosses, and this may have affected the outcome. This possibility was examined by conducting parallel crosses with Hfr donors that were either wild type or mutant for sbcC. Transfer of sbcC+ from an Hfr donor is shown to alter the frequency of recombinant genotypes recovered. There is a severe reduction in progeny that inherit donor markers linked to the sbcC+ allele and an increase in the incidence of multiple exchanges. Colonies of mixed genotype for one or more of the unselected proximal markers are also much more prevalent. Since the yield of recombinants is lower than normal, these changes are attributed to the reduced viability of recombinants that inherit sbcC+ from the Hfr donor. When the Hfr donor used is also mutant for sbcC, the yield of recombinants is greater and the frequencies of the different genotypes recovered are similar to those obtained in crosses with a rec+ sbc+ recipient, in which transfer of sbcC+ has no apparent effect. Earlier studies are re-examined in light of these findings. It is concluded that, while recombination in recBC sbcBC strains involves different enzymes, the underlying molecular mechanism is essentially the same as that in the wild type.     

Exonucleases I, III, and V are required for stability of ColE1-related plasmids in Escherichia coli.

The stability of two ColE1-related plasmids (pRSF2124 and pMB9) was examined in strains of Escherichia coli multiply deficient in exonucleases I (sbcB), III (xthA), or V (recB recC). Any combination of exonuclease I, III, and V deficiency resulted in dramatically decreased stability of both pRSF2124 and pMB9. Inactivation of the RecF pathway by introducing either recF or recJ mutations to the recB recC subcB background resulted in nearly wild-type levels of stability for both plasmids. In contrast, the introduction of uvrD3 uvr-257, uvrE100, or recL152 into the recB21 recC22 sbcB15 strain did not affect plasmid stability. Furthermore, the amount of plasmid DNA recovered from pRSF2124 or pMB9 transformants of a xthA1 sbcB15 strain was strikingly reduced relative to that of a wild-type control. Taken together, these results suggest that some aspect of DNA repair is required for stable maintenance of ColE1-related plasmids in E. coli.     

Mechanism of conjugation and recombination in bacteria. XXI. Role of F factor genes in post-conjugational recombination in Escherichia coli K-12

Temperature sensitive dnaA recipient crossed at the restrictive temperature with HfrH, free from contaminating F+ cells, forms recombinants almost as proficiently as at the permissive temperature. The merozygotes are able to synthesize DNA at 42 degrees C, although the recipient and donor cells do not incorporate 3H-thymine. A substantial fraction of Lac+ recombinants, irrespective of the mating temperature, is temperature resistant (42 C-R); 15% from among those mated at 35 C and 30% from those mated at 42 C. The presence of dnaA mutation in these Lac+ 42 C-R recombinants was ascertained by co-transduction with ilv. Cell division at 42 C is inhibited in the Lac 42 C-R recombinants by acridine orange. The presence of F factor DNA in these recombinants was demonstrated directly by DNA: DNA hybridization. Suppression of dnaA mutation in Lac+ 42 degrees C-R recombinants and their sensitivity to acridine orange at 42 degrees C suggests that at least part of the F factor is integrated into the recombinant chromosome. A large fraction of the Lac+ 42 degrees C-R recombinants (up to 80%) is sensitive to male phage R17 and fertile. In crosses with HfrC there is a marked decrease of recombination frequency at 42 degrees C in the dnaA recipient. The fraction of Lac+ 42 degrees C-R recombinants is low (up to 10%) and the 42 degrees C-R recombinants are neither sensitive to male phage nor fertile. The results are discussed on the basis of the previously proposed model of post-conjugational recombination.     

Kinetics of recA function in conjugational recombinant formation.

Summary Genetic recombination was studied in F^- strains of E. coli carrying a mutation (recA200) that confers a thermosensitive Rec^- phenotype. Recombination during Hfr matings at 35C was monitored by raising the temperature of incubation to 42C at various intervals so that only merozygotes that had completed those functions dependent on the activity of the recA gene product could form recombinant progeny. The results indicated that no more than 1–2% of the merozygotes present while mating was in progress were able to form recombinant colonies at 42C. Separation of mating pairs reduced the yield of recombinants obtained at 35C by 50 to 200-fold if plating on agar medium was delayed for 15–30min by continuing incubation in broth medium. recA200 merozygotes that were also recB21 sbcB15 proved relatively stable when plating was delayed in this manner, which suggested that Hfr DNA is prone to exonuclease inactivation in recA200 merozygotes after mating pairs have separated. Post-mating incubation in high salt medium or on agar plates promoted the recovery of recombinants at 35C. However, the majority of recA200 merozygotes did not acquire the ability to form recombinant colonies at 42C under these more stable conditions until mating pairs had been separated and incubation continued at 35C for 40–60 min. It was concluded that recA200 strains are partially defective for recombination even at low temperature but that terminating mating promotes the recovery of recombinants. A mechanism involving the stimulation of RecA activity by mating pair separation is postulated to account for the efficient recovery of recombinants from HfrxF^- recA200 crosses at 35C.     

Physical and biochemical characterization of cloned sbcB and xonA mutations from Escherichia coli K-12.

In Escherichia coli K-12, sbcB/xonA is the structural gene for exonuclease I, an enzyme that hydrolyzes single-stranded DNA to mononucleotides in the 3'-to-5' direction. This enzyme has been implicated in the DNA repair and recombination pathways mediated by the recB and recC gene products (exonuclease V). We have cloned several sbcB/xonA mutant alleles in bacterial plasmids and have partially characterized the cloned genes and their protein products. Two of the mutations (xonA2 and xonA6) retain no detectable exonucleolytic activity on single-stranded DNA. The xonA6 allele was shown to harbor an insertion of an IS30-related genetic element near the 3' end of the gene. Two other mutations, sbcB15 and xonA8, exhibited significantly reduced levels of exonuclease I activity as compared to the cloned wild-type gene. A correlation was observed between levels of exonuclease I activity and the ability of the sbcB/xonA mutations to suppress UV sensitivity in recB and recC strains. Also, recombinant plasmids bearing either the sbcB15 or xonA6 allele exhibited a high degree of instability during growth of their bacterial hosts. The results suggest that the sbcB/xonA gene product is a bi- or multifunctional protein that interacts with single-stranded DNA and possibly with other proteins in the suppression of genetic recombination and DNA-repair deficiencies in recB and recC mutants.     

Plasmid control of recombination in E. coli K 12

Summary The recombination proficiency of three recipient strains of Escherichia coli K 12 carrying different plasmids was investigated by conjugal mating with Hfr Cavalli. Some plasmids (e.g. R1drd 19, R6K) caused a marked reduction in the yield of recombinants formed in crosses with Hfr but did not reduce the ability of host strains to accept plasmid F104. The effect of plasmids on recombination was host-dependent. In Hfr crosses with AB1157 (R1-19) used as a recipient the linkage between selected and unselected proximal markers of the donor was sharply decreased. Plasmid R1-19 also decreased the yield of recombinants formed by recF, recL, and recB recC sbcA mutants, showed no effect on the recombination proficiency of recB recC sbcB mutant, and increased the recombination proficiency of recB, recB recC sbcB recF, and recB recC sbcB recL mutants. An ATP-dependent exonuclease activity was found in all tested recB recC mutants carrying plasmid R1-19, while this plasmid did not affect the activity of exonuclease I in strain AB1157 and its rec ^– derivatives. The same plasmid was also found to protect different rec ^– derivatives of the strain AB1157 against the lethal action of UV light. We suppose that a new ATP-dependent exonuclease determined by R1-19 plays a role in both repair and recombination of the host through the substitution of or competition with the exoV coded for by the genes recB and recC.     

Lactose and melibiose metabolism in Erwinia chrysanthemi.

A Lac+ mutant of Erwinia chrysanthemi was isolated from the Lac- wild type on lactose agar. beta-Galactosidase was expressed independently of lactose transport in both the mutant and the wild type, and neither strain expressed thiogalactoside transacetylase. Lactose transport and alpha-galactosidase, constitutive in the Lac+ strain, were coordinately induced in the Lac- strain by melibiose and raffinose but not by isopropyl-beta-D-thiogalactopyranoside or thiomethyl-beta-D-galactopyranoside. Melibiose was a strong inhibitor of both the melibiose- and the raffinose-induced lactose permeases, whereas raffinose was a strong inhibitor of only the raffinose-induced lactose permease.     

Participation of rec genes of Escherichia coli K 12 in W-reactivation of UV-irradiated phage lambda.

The effect of the recombinational deficiency on W-reactivation of UV-damaged phage lambda was explored. In this paper we show that W-reactivation is reduced by the recB21 and recF143 mutations after bleomycin (BM) and UV treatment. Combination of these mutations in the recB21recF143 double mutant blocks W-reactivation completely after BM induction, but leaves residual W-reactivation ability after UV-irradiation, which is abolished by the introduction of uvrB deficiency (delta(uvrB-chlA]. W-reactivation has been rendered constitutive in recB21C22sbcB15, but the efficiency of reactivation remained virtually constant over the range of BM and UV doses, indicating the role of the RecBC(D) enzyme in W-reactivation.     

recA (Srf) suppression of recF deficiency in the postreplication repair of UV-irradiated Escherichia coli K-12.

The mechanism by which recA (Srf) mutations (recA2020 and recA801) suppress the deficiency in postreplication repair shown by recF mutants of Escherichia coli was studied in UV-irradiated uvrB and uvrA recB recC sbcB cells. The recA (Srf) mutations partially suppressed the UV radiation sensitivity of uvrB recF, uvrB recF recB, and uvrA recB recC sbcB recF cells, and they partially restored the ability of uvrB recF and uvrA recB recC sbcB recF cells to repair DNA daughter-strand gaps. In addition, the recA (Srf) mutations suppressed the recF deficiency in the repair of DNA double-strand breaks in UV-irradiated uvrA recB recC sbcB recF cells. The recA2020 and recA801 mutations do not appear to affect the synthesis of UV radiation-induced proteins, nor do they appear to produce an altered RecA protein, as detected by two-dimensional gel electrophoresis. These results are consistent with the suggestion (M. R. Volkert and M. A. Hartke, J. Bacteriol. 157:498-506, 1984) that the recA (Srf) mutations do not act by affecting the induction of SOS responses; rather, they allow the RecA protein to participate in the recF-dependent postreplication repair processes without the need of the RecF protein.     

Involvement of RecB-mediated (but not RecF-mediated) repair of DNA double-strand breaks in the gamma-radiation production of long deletions in Escherichia coli.

Experiments were designed to determine the association between the repair of gamma-radiation-induced DNA double-strand breaks (DSB) and the induction of 700-1000 bp long deletions (Lac(-)----Lac+), base substitutions (leuB19----Leu+), and frameshifts (trpE9777----Trp+) in Escherichia coli K-12. Over the range of 2.5-20 krad, deletions were induced with linear kinetics, as has been shown for the induction of DSB, while the induction kinetics of base substitutions and frameshifts were curvilinear. Like the repair of DSB, deletion induction showed an absolute requirement for an intact recB gene as well as a dependency on the type of preirradiation growth medium; these requirements were not seen for base substitutions or frameshifts. In addition, about 80% of the spontaneous deletions were absent in the recB21 strain. A recC1001 mutation, which confers a 'hyper-Rec' phenotype, increased the rate of gamma-radiation-induced deletions as well as the low-dose production of base substitutions and frameshifts. A recF143 mutation increased the yield of gamma-radiation-induced deletions without increasing base substitutions or frameshifts. A mutS mutation markedly enhanced the gamma-radiation induction of frameshifts, and had a slight effect on base substitutions, but did not affect the induction of deletions. Resistance to gamma-irradiation and the capacity to repair DSB (albeit at about half the normal rate) were restored to the radiosensitive recB21 strain by the addition of the sbcB21 and sbcC201 mutations. However, the radioresistant recB sbcBC strain, which is recombination proficient via the RecF pathway, was still grossly deficient in the ability to produce deletions. A model for deletion induction as a by-product of the recB-dependent (Chi-dependent) repair of gamma-radiation-induced DSB is discussed, as is the inability to detect deletions in cells that use only the recF-dependent (Chi-independent) mechanism to repair DSB.     

RecBC enzyme activity is required for far-UV induced respiration shutoff in Escherichia coli K12

Shutoff of respiration is one of a number of recA+ lexA+ dependent (SOS) responses caused by far ultraviolet (245 nm) radiation (UV) damage of DNA in Escherichia coli cells. Thus far no rec/lex response has been shown to require the recB recC gene product, the RecBC enzyme. We report in this paper that UV-induced respiration shutoff did not occur in either of these radiation-sensitive derivatives of K12 strain AB1157 nor in the recB recC double mutant. The sbcB gene product is exonuclease I and it has been reported that the triple mutant strain recB recC sbcB has near normal recombination efficiency and resistance to UV. The sbcB strain shut off its respiration after UV but the triple mutant did not show UV-induced respiration shutoff; the shutoff and death responses were uncoupled. We concluded that respiration shutoff requires RecBC enzyme activity. The RecBC enzyme has ATP-dependent double-strand exonuclease activity, helicase activity and several other activities. We tested a recBC+ (double dagger) mutant strain (recC 1010) that had normal recombination efficiency and resistance to UV but which possessed no ATP-dependent double-strand exonuclease activity. This strain did not shut off its respiration. The presence or absence of other RecBC enzyme activities in this mutant is not known. These results support the hypothesis that ATP-dependent double-strand exonuclease activity is necessary for UV-induced respiration shutoff.     

Factors which equalize the representation of genome segments in recombinant libraries

Genomic segments which contain inverted repetitions longer than 300 bp are frequently lost from recombinant libraries grown on rec+ hosts. We have found that 9% of phage lambda clones that contain 15-20-kb insertions of human or Drosophila DNA are inhibited on rec+ hosts and as a result will become under-represented in amplified genomic libraries. We have therefore examined several factors of both host and vector origin which affect the fidelity of representation of genomic sequences in recombinant DNA libraries constructed in bacteriophage lambda vectors. This loss may be diminished if the vector carries either a chi element or a functional gam gene. The most successful approach, however, involves using a host with mutations in recB, recC, and sbcB, or in recD. We have shown that recombinant clones which require such mutant hosts for growth are somewhat more likely to contain DNA derived from loci in the genome which are polymorphic than are clones recovered on conventional hosts.     

Recombination between Homologies in Direct and Inverse Orientation in the Chromosome of Salmonella: Intervals Which Are Nonpermissive for Inversion Formation

Sequences placed in inverse order at particular chromosome sites (permissive) recombine to generate an inversion; the same sequences, placed at other sites (nonpermissive) interact recombinationally but do not form the expected inversion recombinants. We have investigated the events that occur between sequences at nonpermissive sites. Genetically marked lac operons in inverse order were placed at nonpermissive sites in a single chromosome and Lac+ recombinants were selected. No inversions were formed. The Lac+ recombinants recovered include double-recombinant types in which information appears to have undergone a nonreciprocal information exchange; one mutant copy is repaired with no alteration of the other copy. Recombination within the lac operon is stimulated more than 100-fold by the presence of extensive homology (antenna sequences) outside of the region for which recombination is selected. Sequences placed in direct order at the ends of the same noninvertible chromosome segment recombine to form all the expected recombinant types including those in which a reciprocal exchange has generated a duplication. All the detected recombinant types can be accounted for by recombination between sister chromosomes. These results are discussed in terms of two alternative models. One explanation of the failure to detect inversion of some intervals is that particular inversions are lethal, despite the fact that no essential sequences are disrupted. Another explanation is that chromosome topology prevents sequences at nonpermissive sites in a single chromosome from engaging in the direct interaction required for inversion formation, but allows the sister strand exchanges that can generate the recombinant observed.