Chi-Dependent Intramolecular Recombination in Escherichia coli

Corresponding author: Amikam Cohen, Department of Molecular Biology, The Hebrew University-Hadassah Medical School, Jerusalem 91010, Israel, amikamc{at}cc.huji.ac.il (E-mail).

Communicating editor: L. S. SYMINGTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Homologous recombination in Escherichia coli is enhanced by a cis-acting octamer sequence named Chi (5'-GCTGGTGG-3') that interacts with RecBCD. To gain insight into the mechanism of Chi-enhanced recombination, we recruited an experimental system that permits physical monitoring of intramolecular recombination by linear substrates released by in vivo restriction from infecting chimera phage. Recombination of the released substrates depended on recA, recBCD and cis-acting Chi octamers. Recombination proficiency was lowered by a xonA mutation and by mutations that inactivated the RuvABC and RecG resolution enzymes. Activity of Chi sites was influenced by their locations and by the number of Chi octamers at each site. A single Chi site stimulated recombination, but a combination of Chi sites on the two homologs was synergistic. These data suggest a role for Chi at both ends of the linear substrate. Chi was lost in all recombinational exchanges stimulated by a single Chi site. Exchanges in substrates with Chi sites on both homologs occurred in the interval between the sites as well as in the flanking interval. These observations suggest that the generation of circular products by intramolecular recombination involves Chi-dependent processing of one end by RecBCD and pairing of the processed end with its duplex homolog.

GENETIC recombination in wild-type Escherichia coli cells requires RecA, RecBCD, SSB and enzymes that resolve recombination intermediates (CLARK and MARGULIES 1965 ; HOWARD-FLANDERS and THERIOT 1966 ; EMMERSON 1968 ; GLASSBERG et al. 1979 ; LLOYD 1991 ). Recombination is stimulated by a cis-acting octamer (5'-GCTGGTGG-3') named Chi, which interacts with RecBCD in an orientation-dependent manner (reviewed in SMITH et al. 1984 ; MYERS and STAHL 1994 ). RecA catalyzes pairing of either 3'- or 5'-ending single-stranded DNA (ssDNA) with homologous double-stranded DNA (dsDNA) resulting in reciprocal strand exchange (reviewed in WEST 1992 ; KOWALCZYKOWSKI et al. 1994 ). RecBCD is a multifunctional heterotrimer, encoded by the recB, recC and recD genes. In vitro, it loads at DNA double-stranded ends and promotes DNA-dependent ATPase and ATP-dependent helicase activities. Its helicase activity is frequently associated with exonuclease and endonuclease activities (reviewed in WEST 1992 ; KOWALCZYKOWSKI et al. 1994 ).

Current hypotheses assume a role for Chi and RecBCD in the generation of ssDNA substrates for the homologous pairing reaction. This proposition is supported by in vitro studies that demonstrate a change in recBCD activity upon interaction with Chi (TAYLOR et al. 1985 ; TAYLOR and SMITH 1992 ; DIXON and KOWALCZYKOWSKI 1993 ; ANDERSON and KOWALCZYKOWSKI 1997 ). Both Chi-dependent and Chi-independent RecBCD activities are greatly influenced by assay conditions. Depending on the molar ratio of ATP to Mg²⁺, RecBCD activity at Chi may generate a nick at the 3' side of the sequence (as written above), attenuate RecBCD exonuclease activity or switch the polarity of its exonuclease activity from degrading the 3'-ending strand to degrading the 5'-ending strand (DIXON and KOWALCZYKOWSKI 1995 ; TAYLOR and SMITH 1995 ; ANDERSON and KOWALCZYKOWSKI 1997 ).

The mode of RecBCD function and the nature of the homologous pairing reaction that prevails in vivo are not known, and alternative models postulate different mechanisms for the presynaptic stage of recombination (THALER et al. 1988 ; ROSENBERG and HASTINGS 1991 ; MYERS and STAHL 1994 ; DIXON and KOWALCZYKOWSKI 1995 ; TAYLOR and SMITH 1995 ; ANDERSON and KOWALCZYKOWSKI 1997 ). In an attempt to resolve this problem and test predictions of the alternative models we recruited an experimental system that facilitates efficient and synchronous release of linear recombination substrates within E. coli cells (NUSSBAUM et al. 1992 ; SILBERSTEIN et al. 1993 ). We employed this system for physical monitoring of the recombining molecules and structural analysis of primary products. In this paper we report on the genetic requirements of intramolecular recombination by the linear substrates and the dependence of this mode of recombination on Chi elements and DNA double-stranded breaks. We also determined the distribution of recombination events along the homology and physically monitored the fate of the recombining substrates and the accumulation of the recombination products.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Bacterial strains and growth conditions:
E. coli strains (Table 1) were grown on LB medium (LURIA and BURROUS 1957 ). Plasmid-carrying strains were grown in media supplemented with the appropriate antibiotics (100 µg of ampicillin per milliliter, 20 µg of kanamycin per milliliter). Unless otherwise indicated, infected cultures harbored pMB4 (BETLACH et al. 1976 ). Infection protocol and growth conditions for cultures infected by phage precursors of the linear recombination substrates were as described previously (SILBERSTEIN et al. 1993 ).

Chimera phage and plasmids:
Chimera phage with intramolecular recombination substrates cloned between EcoRI sites (Table 2) were constructed by ligating linearized substrates to EMBL4 (FRISCHAUF et al. 1983 ) phage DNA arms as described previously (SILBERSTEIN et al. 1993 ). The orientation of the cloned substrate in the phage vector was verified by restriction analysis with HindIII endonuclease. Where indicated, oligomers containing one to three Chi octamers were inserted into the XhoI (Xho) or Bcl I (Bcl) sites on the recombination substrate (see Figure 1) or into both of them. The Chi octamers at Xho and Bcl were inserted into the luxA luxB fragment of pAP601 (NUSSBAUM and COHEN 1988 ) and pMS804 (SILBERSTEIN et al. 1993 ), respectively, and the respective fragments were assembled into intramolecular recombination substrates as described previously (SILBERSTEIN et al. 1993 ). To insert a single Chi octamer into the XhoI or Bcl I sites, complementary oligomers (A and B in Table 3) containing a Chi octamer, a Not I restriction site and the appropriate ssDNA overhangs were synthesized, annealed and ligated into the respective site. To insert a second Chi octamer, oligomers containing KpnI and Pst I restriction sites, a Chi octamer and Not I ssDNA overhangs (C in Table 3) were inserted into the Not I site of the plasmids containing the first Chi octamer. The third Chi octamer was inserted into the cloned KpnI site, using an oligomer containing a BamHI restriction site, two Chi octamers in opposite orientations and KpnI ssDNA overhangs. Chi octamers in all substrates depicted in Table 2 were in the functional orientation with respect to the adjacent EcoRI site on the chimera phage. The sequence and orientation of Chi in substrates containing a single Chi octamer were verified by sequence analysis. Chi orientation in all substrates was verified by PCR analysis (INNIS et al. 1990 ).

View larger version (55K): In this window In a new window Download PPT slide   Figure 1. A substrate prototype for Chi-dependent intramolecular recombination. In vivo restriction of the infecting phage DNA by EcoRI releases a linear fragment with a direct terminal repeat. For determination of recombinant frequency and the distribution of recombinational exchanges, fragments with pACYC184 origin of replication (ori) were used. For physical monitoring of recombination, the SacII-XbaI fragment containing the plasmid replication origin was deleted. The location of Chi insertion sites (Xho and Bcl ), relevant restriction sites and genes are indicated. Homology is designated by parallel boxes, phage arms (not to scale) by broken lines and mutations in the luxA luxB genes by triangles. The homology to the probe used in the hybridization experiments (probe) and the genetic intervals defined by the Chi sites (left, center, right) are indicated. The orientation of the inserted Chi octamers (5'-GCTGGTGG-3') is indicated by 3' to 5' arrows; sequences located "upstream" and "downstream" of Chi are defined here and elsewhere in the paper as sequences on the 3' and the 5' sides of the Chi octamer (5'-GCTGGTGG-3'), respectively.

In experiments designed for the determination of recombinant frequency and isolation of plasmid recombination products, substrates with pACYC184 (CHANG and COHEN 1978 ) origin of replication were used (Ori⁺ substrates). For physical monitoring of recombination by Southern blot hybridization, substrates devoid of a replication origin (Ori^- substrates) were used. All phage harboring Ori⁺ substrates were isogenic to MS805 and all phage harboring Ori^- substrates were isogenic to ZS820 (SILBERSTEIN et al. 1993 ). Ori^- substrates were derived from the appropriate plasmids and inserted into the phage vector by the methodology described previously (SILBERSTEIN et al. 1993 ). Maintenance and propagation of chimera phages harboring intramolecular recombination substrates were described previously (SILBERSTEIN et al. 1993 ). The cloned fragment in RF936 has a single copy of the luxA luxB genes. RF936 was constructed by deleting pACYC184 replication origin from pRF934 and ligating the linear EcoRI fragment to EMBL4 arms as described (SILBERSTEIN et al. 1993 ). pRF934 is an isogenic derivative of pMS804 (SILBERSTEIN et al. 1993 ) except that it contains a single EcoRI restriction site in the inactivated NdeI site of the luxA gene as in pZS824 (SILBERSTEIN et al. 1995 ) and Chi octamer triplets at the XhoI and Bcl I sites. AN612 has been described previously (NUSSBAUM et al. 1992 ).

Determination of recombinant frequency and structural analysis of plasmid recombination products:
To select for Kan^R recombinants, cells were infected by the appropriate chimera phage, and samples taken at 120 min following infection were plated on kanamycin-supplemented medium as described (NUSSBAUM et al. 1992 ). To minimize the occurrence of intermolecular recombination events or multiple recombination events within the same infected cell, the multiplicity of infection (MOI) was 0.2. Recombinant frequency was defined as the ratio of Kan^R recombinants per infected cell. Chi activity was defined as the recombinant frequency in cultures infected by phage harboring Chi⁺ substrates divided by the frequency in cultures infected by phage harboring Chi⁰ substrates. To determine the distribution of recombinational exchange along the homology, isolated colonies were inoculated into kanamycin-supplemented liquid medium. Plasmid DNA preparations of overnight cultures were made by the rapid boiling method (HOLMES and QUIGLEY 1981 ) and subjected to restriction endonuclease analysis using Bgl II, HindIII, and a combination of Not I and Sal I endonucleases (see Figure 1 for the location of the corresponding restriction sites). Recombinant frequency at each genetic interval was calculated by multiplying total recombinant frequency by the fraction of recombinational exchange occurring within the interval. Recombinant frequency at each interval was divided by the physical length of the interval and is presented as "recombinant frequency/bp."

Physical monitoring of recombination:
Cells were infected at an MOI of 2. Total DNA preparations of samples taken at the indicated times following infection were digested by Sal I endonuclease and subjected to Southern blot hybridization as described (SILBERSTEIN et al. 1993 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Chi-dependent recombination:
Linear intramolecular recombination substrates were cloned between EcoRI sites on phage vectors, delivered by infection into E. coli cells, lysogenic for (ind^-) and released from the chimera phage by in vivo restriction (NUSSBAUM et al. 1992 ). The cloned 13.6 kb substrate (Figure 1) had a direct terminal repeat of 3.5 kb, consisting of Vibrio fischeri luxA luxB genes, with each copy of the repeated sequence mutated at a different site (NUSSBAUM and COHEN 1988 ). An intramolecular recombinational exchange within these substrates would yield circular products with a single copy of the luxA luxB genes. To facilitate selection for infected cells harboring circular recombination products, the linear substrates also contained a pACYC184 origin of replication and a kan gene (Figure 1). To investigate the dependence of intramolecular recombination on Chi elements, oligomers containing one to three Chi octamers in the functional orientation with respect to the EcoRI-induced break (STAHL et al. 1980 ), were inserted into either XhoI, Bcl II or both sites. These sites are located at a distance of 837 and 1565 nucleotides from the EcoRI-induced DNA break, respectively (Figure 1). The effect of these insertions on recombinant frequency was determined by scoring Kan⁺ cells in cultures infected by the respective chimera phage. Infection of wild-type E. coli cells by chimera phage harboring substrates devoid of Chi octamers (Chi⁰ substrates) did not yield a detectable number of Kan⁺ recombinants (<1 x 10^-5 recombinants/infected cell). This low frequency is compared with recombinant frequencies of about 6% and 15% in infected recB recC sbcA and recD mutants, respectively (Table 4). Insertion of Chi octamers at either one of the two sites, or at both of them, led to an increase in recombinant frequency (Figure 2). A maximal recombinant frequency of 10% was achieved with substrates having three Chi octamers at each site. Recombination also depended on EcoRI-induced DNA breaks. The frequency of Kan⁺ recombinants in infected cells that did not express EcoRI was 1000-fold lower than that in infected cells that expressed EcoRI from a resident pMB4 plasmid (Table 4).

View larger version (24K): In this window In a new window Download PPT slide   Figure 2. Dependence of recombinant frequency on the location of Chi sites and the number of Chi octamers at each site. A derivative of AC227 (rec+), containing pMB4 (EcoRI+) was infected by chimera phage, harboring intramolecular recombination substrates with the indicated number of Chi octamers at Xho (), Bcl (diagonal boxes), or at both insertion sites (). The frequency of Kan+ recombinants was determined as described in MATERIALS AND METHODS. The sum of the frequencies by substrates having Chi octamers at a single site (, Xho + Bcl) is presented.

Presumably, a Chi site at one end is sufficient to promote RecBCD-mediated generation of a recombinogenic single-stranded end, and this end may pair with the duplex homolog. However, with the substrate depicted in Figure 1, Chi may have a role at both ends: if pairing is by strand invasion, Chi may stabilize the noninvading end by modulating RecBCD exonuclease activity. Furthermore, with these substrates, a circular product may also be generated by a single-strand annealing (SSA) mechanism that depends on processing of both ends (SILBERSTEIN et al. 1993 ). To test whether a Chi site on both ends is required, we determined recombinant frequencies with sets of substrates having Chi octamers at a single site (substrates Xho and Bcl) or at both sites (substrates XhoBcl) (Figure 2). Insertion of Chi octamers at either one of the two sites enhanced recombination, and recombinant frequency increased with the number of Chi octamers at the site. Chi octamers at one site (Xho) gave a higher recombinant frequency than at the other site (Bcl). Recombinant frequencies of substrates having Chi octamers at both sites (XhoBcl) were consistently higher than the sum of the frequencies by substrates having Chi octamers at a single site (Xho + Bcl). Thus, a single Chi site was sufficient to stimulate intramolecular recombination, but a combination of two Chi sites was synergistic. These observations, the distribution of recombinational exchanges (see Figure 4), and the dependence on Holliday junction resolution systems (see Figure 6D) argue against an SSA recombination mechanism (see DISCUSSION). Assuming that pairing is by strand invasion, these observations suggest that Chi elements play a role both at the invading and at the noninvading ends.

View larger version (21K): In this window In a new window Download PPT slide   Figure 3. Dependence of recombinant production on MOI. A derivative of AC227 containing pMB4 was infected at the indicated MOI by chimera phage harboring intramolecular recombination substrates with three Chi octamers at each insertion site (RF951) and the number of Kan+ recombinants per 1000 cells was determined (). Theoretical curves (– – –) representing the percentage of cells infected by one or more (n 1) or by two or more (n 2) phage particles are presented as a reference. These theoretical curves were computed under the assumption that the number of phage DNA molecules entering a cell follows a Poisson distribution.

View larger version (47K): In this window In a new window Download PPT slide   Figure 4. The distribution of recombinational exchange along the homology. Substrates having one to three Chi octamers at each site were delivered to lysogenic cells as described. Recombinant frequencies were determined by scoring KanR cells in the infected cultures, and the distribution of recombinational exchanges along the homology was determined by restriction analysis of plasmid recombination products. Exchange frequencies at each interval were normalized for the length of the interval. Error bars represent standard deviation. The distribution of exchange frequencies along the homology in recD mutants (inset) is presented as a reference.

View larger version (39K): In this window In a new window Download PPT slide   Figure 5. Physical monitoring of Chi-dependent intramolecular recombination. Substrates were released from infecting chimera phage by in vivo restriction, and samples were taken at the indicated times after infection. Sal I-digested total cellular DNA preparations were subjected to Southern blot hybridization with the probe depicted in Figure 1. The number of Chi octamers at each site on the recombining substrates (Figure 1) is indicated. The locations of Sal I digestion products of circular and linear phage DNA (upper and lower DNA bands, respectively), EcoRI-restricted phage DNA (substrate) and the circular recombination product (product) are indicated. The predicted migration of a Sal I product of a hypothetical intermediate (a Sal I-Xho fragment) is indicated.

View larger version (26K): In this window In a new window Download PPT slide   Figure 6. The effect of the number of Chi octamers (A) and host genotype (B–D) on recombination kinetics. The kinetics of product formation in infected wild-type cells and in the indicated mutants was determined by phosphoimaging analysis of Southern blot hybridization experiments like the ones depicted in Figure 5 and is expressed as arbitrary units of radioactivity. The experiments in (A) were with wild-type cells (AC227), and the number of Chi octamers at each site on the delivered substrate is indicated. Substrates with three Chi octamers at each site (RF953) were used in most experiments depicted in (B) to (D). A phage harboring a Chi0 (0) substrate (ZS820) was used where indicated. Except where indicated, infected cells harbored pMB4. To monitor recombination in cells that do not express EcoRI, the infected cells harbored pBR322 (wild-type EcoRI-). The effect on recombination kinetics of recA and recBCD mutations and the dependence on in vivo restriction is depicted in (B). The effect of mutations inactivating single-strand-specific exonucleases (xonA and recJ ) on recombination kinetics is depicted in (C) and the effect of mutations that inactivate Holliday junction resolution systems (recG and ruvC) is depicted in (D). Experimental procedure is described in text.

We considered the possibility that the difference in Chi activity between substrates having Chi octamers at Xho and Bcl may be related to the orientation of the cloned substrate in the chimeric phage. This may have occurred if sequences on the phage vector, in substrates generated by partial restriction, contributed in any way to recombinant formation. To check for this possibility, we cloned substrates with a single and a double Chi octamer at the Bcl site in the vector in both orientations (Table 2) and compared recombinant frequencies in cultures infected by the two sets of chimeric phages (Table 5). Recombinant frequency was not affected by the orientation of the substrate in the vector. In both orientations, a single Chi octamer at Bcl had no detectable activity, and the activity of two Chi octamers at Bcl was similar to that of a single Chi octamer at Xho. We note that while a single Chi octamer at Bcl had no activity, it enhanced recombination when combined with a single Chi octamer at Xho.

A single DNA molecule, released in vivo by EcoRI restriction, can presumably undergo intramolecular recombination to yield a circular product. To test whether the release of a single molecule is sufficient for recombinant production, we determined the dependence of recombinant production on MOI (Figure 3). It is clear that the observed dependence on MOI is better approximated by the theoretical curve corresponding to a single hit being sufficient for recombination than that requiring at least two hits.

Genetic requirements:
The RecBCD pathway, as defined in transduction and conjugational recombination, depends on functional recA and recBC genes (CLARK 1973 ). To test whether our recombination system has similar requirements, we infected the appropriate mutants by the chimeric phage precursor of the Chi⁺ substrate and scored Kan⁺ recombinants. Kan⁺ recombinants were not detected in infected recA or recBC mutants (<10^-5 recombinants/infected cell) (Table 4). The 10⁴-fold decrease in recombination proficiency caused by the recBC mutations was more severe than that in conjugation or transduction in which recombination proficiency of recBC mutants is 0.1–1.0% of that in recBC⁺ cells (CLARK 1973 ; MAHAJAN 1988 ). As expected, production of Kan⁺ recombinants in recD mutants was independent of the presence of Chi octamers on the substrate (Table 4). The effect of host genotype on recombination kinetics is presented below (see Figure 6).

The distribution of Chi-stimulated recombinational exchanges:
The distribution of recombinational exchanges in substrates with Chi octamers inserted at either Xho, Bcl or both sites (see Figure 1) was investigated. Kan⁺ clones were selected by plating infected cultures on kanamycin-supplemented medium, and kan plasmids of the selected clones were subjected to restriction analysis. All plasmid products, generated by intramolecular recombination of linear substrates with Chi octamers at a single site, lost the Chi octamers in recombination (100 independent plasmids products analyzed). This loss of Chi is consistent with models that postulate that Chi stimulates recombination downstream of its 5' end (THALER et al. 1988 ). However, it does not contradict the finding that some Chi-mediated exchanges in phage crosses are upstream to the 3' end of Chi (CHENG and SMITH 1989 ), since in the experiments reported here, Chi was inserted within heterologies of 30–76 nucleotides. These short heterologies may have prevented recombinational exchanges at the 3' end of Chi.

The distribution of genetic exchanges in substrates with Chi octamers at both the Xho and Bcl sites was investigated. The two Chi sites define three genetic intervals: a central interval between the Chi insertion sites and two flanking intervals between each site and the adjacent end (see Figure 1). Restriction analysis of plasmid recombination products indicated that genetic exchanges in substrates with Chi octamers at both sites occurred in all three intervals (Figure 4). Exchange frequencies in the central interval were higher than those in the flanking intervals. This distribution of recombination events is different from that in recD mutants (insert in Figure 4), where similar frequencies were measured in all three intervals.

Physical monitoring of recombination:
In an attempt to follow the fate of linear recombination substrates, we cloned substrates devoid of plasmid replication origins in phage vectors and delivered them by infection into (ind^-) lysogens expressing EcoRI from a resident plasmid (pMB4). Total cellular DNA preparations of samples taken at time intervals after infection were digested by Sal I endonuclease and analyzed by Southern blot hybridization as described previously (SILBERSTEIN et al. 1993 ; Figure 5). Three fragments with homology to the probe (see Figure 1) were observed in samples taken immediately following infection. These fragments corresponded in molecular length to the expected Sal I products of linear phage DNA that escaped EcoRI restriction (lower phage band), circular phage DNA (upper phage band) and EcoRI-restricted phage DNA (substrate). Incomplete restriction by the endogenous EcoRI endonuclease would generate two molecular species. Digestion of these products by Sal I endonuclease resulted in two molecules that are identical to the Sal I products of linear and fully restricted phage DNA, respectively.

Recombination products were not detected by hybridization in cells infected by phage harboring Chi⁰ substrates at any timepoint. Insertion of Chi octamers at the Xho and Bcl sites led to the generation of circular recombination products in the infected cells. These circular products were first detected 15 min following infection, and their amount increased thereafter (Figure 6A). Increasing the number of Chi octamers at each site from one to two enhanced the rate of circular products formation. However, a third Chi octamer at each site did not affect recombination kinetics (Figure 6A). Consistent with results presented in Table 4, the generation of circular products depended on in vivo restriction by EcoRI and on functional recA and recB recC genes. Recombination in recD mutants was independent of Chi and occurred at a higher rate than in recD⁺ cells (Figure 6B).

Another class of enzymes that affect recombinant frequency in some systems are ssDNA-specific exonucleases (FENG and HAYS 1995 ; MIESEL and ROTH 1996 ; RAZAVY et al. 1996 ). xonA and recJ products degrade ss-DNA from the 3' and the 5' ends, respectively. Recombination rate was lowered by a xonA mutation but was not affected by a recJ mutation. Recombination rate in xonA recJ double mutants was consistently lower than that in single xonA mutants (four independent experiments). Unlike recombination in recD⁺ cells, recombination in recD mutants depended on recJ activity (R. FRIEDMAN-OHANA, unpublished results).

The linear substrates used in this study may generate circular products by either one of two recombination mechanisms: One involves strand exchange and the other SSA (reviewed in KOLODNER et al. 1994 ; EGGLESTON and WEST 1996 ). Unlike the strand exchange pathway, the SSA pathway does not generate Holliday junction intermediates and therefore should not depend on their enzymatic resolution. To test the dependence on proteins that resolve Holliday junctions, we determined recombination kinetics in recG and ruvC mutants. Single recG or ruvC mutations had a small effect on recombination proficiency, but a double recG ruvC lowered the recombination rate by about 10-fold (Figure 7C). This result is consistent with a mechanism that involves generation and resolution of Holliday intermediates. The residual activity in ruvC recG double mutants may indicate that a small fraction of the recombination events occur via a pathway that does not depend on the formation and resolution of Holliday junctions. Alternatively, Holliday junctions were formed, but resolved by a mechanism that is independent of RecG and RuvABC.

View larger version (16K): In this window In a new window Download PPT slide   Figure 7. The effect of Chi octamers on the longevity of linear fragments that could circularize by recombination (A) and on recombinationally inert fragments (B). Cells were infected by chimera phage harboring intramolecular recombination substrates (A) or linear fragments with no terminal or internal repeats (B). The persistence of the EcoRI-generated fragments, following infection by phage harboring recombination substrates with the indicated number of Chi octamers at each site, was determined by phosphoimaging analysis of Southern blot hybridization experiments like the ones depicted in Figure 5. The persistence of the linear fragments is expressed as the percent of radioactivity associated with the "substrate" band at time zero.

We did not observe molecular intermediates predicted by models that postulate attenuation of RecBCD exonuclease activity at Chi. Possible reasons for our inability to detect these putative intermediates are discussed below.

The longevity of the linear substrates in the infected cells was monitored by hybridization, as seen in Figure 5. The density of the substrate band decreased with time (Figure 7A). However, the persistence of linear fragment did not increase because of the inserted Chi octamers. Furthermore, the persistence of the Chi⁰ substrate was consistently higher than that of substrates with Chi octamers at both insertion sites (12 independent experiments).

Linear DNA fragments, generated in vivo by terminase cleavage of cosmid DNA, were degraded in E. coli cells in a RecBCD-dependent fashion and Chi sites increased the survival of these fragments (KUZMINOV et al. 1994 ). Since Chi octamers had an opposite effect on the persistence of the EcoRI-generated linear recombination substrates (Figure 7A), we considered the possibility that this difference is the result of the engagement of Chi-specific end-processing products in homologous pairing. An alternative explanation is that the observed difference reflects an intrinsic difference between the two methodologies for generation of linear DNA fragments within E. coli cells. To test for the latter possibility, we infected cells that express EcoRI with chimeric phage harboring DNA fragments that cannot serve as intramolecular recombination substrates. The effect of Chi octamers on the longevity of the EcoRI-generated fragments was ascertained by comparing Southern blot hybridization patterns of DNA preparations from cells infected by chimeric phage harboring Chi⁺ and Chi⁰ DNA fragments. A fragment with triple Chi octamers in the functional orientation at a distance of 382 and 973 nucleotides from the substrate's ends (RF936) was degraded more slowly than a Chi⁰ substrate (AN612) (Figure 7B). This observation supports the proposition that the apparent failure of Chi octamers to stabilize EcoRI-generated linear intramolecular recombination substrate is related to the efficient engagement of Chi-specific intermediates in homologous pairing.

As indicated above, the substrate hybridization band (see Figure 5) contains Sal I digestion products of phage DNA that has been fully or partially restricted in vivo by EcoRI. To monitor the persistence of each one of these two molecular species, we subjected total nondigested DNA preparations of samples taken at timepoints after RF953 infection to Southern blot hybridization analysis (Figure 8) and quantitated the corresponding hybridization bands (Figure 9). The density of both hybridization bands decreased with time. However, the density of the band representing partially restricted phage DNA decreased more rapidly than that of the band representing fully restricted phage DNA. This difference suggests that part of the partially restricted phage DNA observed at time zero is further restricted to yield fully restricted phage DNA. This is consistent with the observed increase in density of the band representing fully restricted phage DNA between 0 and 5 min after infection (Figure 9).

View larger version (43K): In this window In a new window Download PPT slide   Figure 8. Southern blot hybridization analysis of nondigested DNA preparations from cells infected by a phage harboring a Chi+ substrate (A). Experimental procedure is as described in the legend to Figure 5, except that the DNA preparations were not digested prior to the Southern blot hybridization analysis. The infecting phage (RF953) harbored a substrate with three Chi octamers at each site. The locations of circular and linear phage DNA, phage DNA that had been partially or fully restricted, covalently closed (ccc) and nicked (oc) forms of the circular recombination products are indicated.

View larger version (14K): In this window In a new window Download PPT slide   Figure 9. The persistence of fully and partially EcoRI-restricted phage DNA. The appropriate hybridization bands in the experiment depicted in Figure 8 were scanned by phosphoimaging analysis. The persistence of the fully restricted () and partially restricted () phage DNA products is presented as the percent of radioactivity associated with the corresponding bands (see Figure 8) at time zero.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The effect of Chi octamers on intramolecular recombination of linear substrates with direct terminal repeats was investigated. Linear substrates, released by in vivo restriction of infecting chimera phage DNA, recombined in wild-type E. coli cells by a mechanism that depended on recA, recBCD and Chi sites. These requirements indicate that recombination of these substrates proceeded by the RecBCD pathway, perceived first for transductional and conjugational recombination (CLARK 1973 ). Recombination also depended on DNA double-stranded breaks, since in vivo restriction increased recombinant frequency by about 1000-fold (Table 4).

The synchrony and high efficiency of the experimental system facilitated physical monitoring of recombination by Southern blot hybridization (Figure 5). The generation of recombination products depended on in vivo restriction of the chimeric phage, the presence of Chi octamers on the substrate and functional recA and recBCD genes. The similarity of these requirements to the requirements for Kan⁺ recombinants formation by the Ori⁺ substrates (Table 4) suggests a common recombination mechanism. Chi-specific intermediates, predicted by models that postulate attenuation of RecBCD exonuclease at Chi and observed in an in vitro system (DIXON and KOWALCZYKOWSKI 1993 ) were not detected by Southern blot hybridization. This may be the result of one or more of the following reasons: (1) the putative intermediates were rapidly engaged in homologous pairing and therefore did not accumulate, (2) the intermediates were substrates for other E. coli nucleases (MYERS et al. 1995 ) and/or (3) the intermediates did accumulate but were not detectable by Southern blot hybridization. The latter may be the case if the putative intermediates were not of a discrete length or if they comigrated with another band in electrophoresis. Comigration of Chi-specific products with the substrate band is predicted by the "nick at Chi" model (SMITH et al. 1984 ).

Linear substrates with direct terminal repeats may yield circular products by either one of two distinct intramolecular recombination mechanisms. In recombination by the strand exchange mechanism, one end is processed to a single-stranded overhang that invades the homologous dsDNA sequence. Conversely, in recombination by the SSA mechanism, resectioning of both ends generates complementary single-stranded overhangs and annealing of these strands yields a circular deletion product (reviewed in KOLODNER et al. 1994 ; EGGLESTON and WEST 1996 ). The following evidence argues for the strand exchange mechanism. First, Chi is presumably essential for RecBCD-mediated formation of a recombinogenic end. The observation that Chi octamers induced recombination when present at only one site (Figure 2; Table 5) suggests that processing of only one end is sufficient. Second, the observation that in recombination of substrates with two Chi sites about half of the recombinational exchanges occur in the flanking intervals (Figure 4) also argues against a mechanism that requires RecBCD- and Chi-dependent processing of both ends. If the loss of Chi in recombination is associated with end processing or with pairing of a processed end, both Chi sites are expected to be lost in recombination by an SSA mechanism. Third, Holliday junction intermediates are expected in the strand exchange pathway but not in recombination by the SSA mechanism. The dependence of recombination on ruvC and recG (Figure 6D) is consistent with a mechanism that involves formation and resolution of Holliday junctions.

While the genetic requirements are in general agreement with the requirements for recombination by the RecBCD pathway, this recombination system is more sensitive than others to some genetic deficiencies. For example, recBC mutations, which lower conjugation and transduction proficiencies by about 100-fold (CLARK 1973 ; MAHAJAN 1988 ), lowered recombinant frequency in the investigated system by more than 10,000-fold, and a xonA mutation, which by itself has little effect on transduction or recombination (MIESEL and ROTH 1996 ; RAZAVY et al. 1996 ), lowered the recombination rate by about 10-fold. The difference in the requirement for Chi octamers is discussed below. The high sensitivity of this recombination assay may be related to the short length of homology (about 3 kb) of the recombining sequences. It has been suggested that the high sensitivity of "short homology" assays stems from the requirement to complete the pairing event before the recombining sequences are degraded (MIESEL and ROTH 1996 ).

The requirement for xonA activity in phage DNA recombination and in the short homology transduction assay was interpreted as evidence for a role for exonuclease I in the generation of homologous pairing substrates with 5'-ended strands (MIESEL and ROTH 1996 ; RAZAVY et al. 1996 ). This does not seem to be the case in our system since exonuclease I deficiency impairs a recombination pathway that involves pairing of a 3'-ending strand (data not shown). It seems more likely that exonuclease I role in recombination is enhancing the rate of strand exchange by degrading the displaced strand (KONFORTI and DAVIS 1992 ; BEDALE et al. 1993 ; MIESEL and ROTH 1996 ; RAZAVY et al. 1996 ).

Enhancement of recombination by the inserted Chi octamers was strongly influenced by the location of the insertion site and the number of inserted Chi octamers. The observed difference in Chi activity between Chi octamers inserted at Xho and Bcl (Figure 2; Table 5) was not related to the orientation of the cloned substrate in the chimeric phage, since inversion of the substrate in the phage vector did not affect Chi activity (Table 5). We also considered the possibility that the difference in Chi activity between the two insertion sites was the result of a difference in the distribution of Chi-like sequences (SCHULTZ et al. 1981 ; CHENG and SMITH 1984 , CHENG and SMITH 1987 ) along the homology. Sequence analysis of the linear substrate revealed the presence of a single Chi-like sequence (5'-CCTGGTGG-3'), in the functional orientation with respect to the adjacent end, at a distance of 332 nucleotides upstream to Xho (R. FRIEDMAN-OHANA, unpublished results). This sequence had no residual Chi activity in phage recombination (SCHULTZ et al. 1981 ). The possibility that Chi-like sequences that differ from the Chi octamer by two nucleotides (two sequences on each side) contribute to the difference in Chi activity between Xho and Bcl cannot be ruled out. Xho and Bcl are located at a distance of 837 and 1565 nucleotides from the substrate's ends, and the length of homology downstream of these sites is 2700 and 1977 nucleotides, respectively. The possibility that the differences between the distances from the respective ends and/or the lengths of homology downstream of Chi affect Chi activity at the two sites is being investigated.

The observation that multiple Chi octamers are required for maximal Chi activity is consistent with the finding that interaction of RecBCD with a single Chi octamer is only about 50% efficient (YAGIL and STROMAS 1985 ; STAHL et al. 1990 ; TAYLOR and SMITH 1992 ). Recombination efficiency, measured by determining the rate of product accumulation at the first 2 hr after infection, increased by three- to six-fold as the number of Chi octamers at each site increased from one to two, but an additional Chi octamer at each site did not affect recombination kinetics (Figure 6A). Conversely, addition of a third Chi octamer at each insertion site tripled recombinant frequency (Figure 2). The effect of a third Chi octamer on recombinant frequency, but not on recombination kinetics in the first 2 hr after infection, suggests that this octamer may contribute to the persistence of the recombination process. This may occur whether increasing the number of Chi octamers contributes to the stability of the restriction-generated substrate or to its continuous production. Stabilization of linear DNA fragments by multiple Chi octamers has been reported (DABERT et al. 1992 ; KUZMINOV et al. 1994 ) and is also apparent from the data depicted in Figure 7B. A mechanism that may provide linear recombination substrates continuously in the experimental system described here is suggested by the following observation: phage DNA that escaped restriction replicates by a mechanism that generates linear multimers, and the newly synthesized multimers are only partially resistant to in vivo EcoRI restriction. This mode of replication depends on recA and the presence of multiple Chi elements on the chimeric phage. It is independent of phage or plasmid replication origin (I. KARUNKER, R. FRIEDMAN-OHANA and A. COHEN, unpublished results). It is conceivable that linear substrates, generated by partial restriction of the linear multimers, may contribute to recombinant production at a later period following infection.

Chi increases the persistence of linear DNA that cannot undergo intramolecular recombination (KUZMINOV et al. 1994 ; Figure 7B), but not of linear fragments that could circularize by recombination (Figure 7A). One interpretation of this structure-dependent difference is that Chi-specific intermediates that comigrate with the linear substrate are converted into circular recombination products. Alternatively, recombination-mediated pairing of the processed ends with circular phage DNA that escaped restriction may prime synthesis of phage DNA linear multimer, and restriction of the multimer would yield "linear substrates" as described above. If linear multimer synthesis is primed by an intermolecular recombination event, the engagement of the processed end in intramolecular recombination may reduce the efficiency of this process. This explanation does not account for the observed difference in longevity between Chi⁺ and Chi⁰ substrates of intramolecular recombination (Figure 7A).

A single Chi site is sufficient to stimulate intramolecular recombination, but a combination of two is synergistic (Figure 2). Assuming that Chi elements at only one site are involved in each recombinational exchange, this synergistic effect suggests a role for the Chi octamers at the other site. It is likely that a Chi site on the noninvasive end enhances recombination by attenuating RecBCD nuclease activity and thereby increases the survival of the linear restriction product (DABERT et al. 1992 ; KUZMINOV et al. 1994 ). However, since multiple Chi sites also enhance phage DNA linear multimer synthesis, we cannot rule out the possibility that recombination enhancement by Chi octamers at the nonpairing end is an outcome of continuous generation of recombination substrates by the mechanism described above.

Whereas in a recD mutant, genetic exchanges were evenly distributed along the short homology, in wild-type cells, exchange frequencies were higher in the interval between the Chi sites than in the flanking intervals (Figure 4). Since the central interval is downstream of both Chi sites and each flanking interval is downstream of only one, this observation is consistent with the proposal that, in this system, all recombination events occur downstream of Chi. Exchange frequencies in the central interval approximated the sum of the frequencies in the flanking intervals. This suggests that in most recombination events at least part of the interval distal to the pairing end is available for strand exchange. If RecBCD degrades DNA from the substrate's end up to the Chi site, this distribution would indicate that in most recombining substrates, only one end interacted with RecBCD.

Chi activity is defined by the ratio of recombinant frequency in a given genetic region containing Chi, compared to that in the same region without Chi (STAHL and STAHL 1977 ). The maximal Chi activity in the experimental system presented here is about 100-fold higher than in gene replacement by a short linear DNA fragment (DABERT and SMITH 1997 ), and several hundredfold higher than in RecBCD-mediated phage recombination (LAM et al. 1974 ; STAHL et al. 1980 ). These marked differences in Chi activity probably reflect the relatively low efficiency of Chi-dependent recombination in the gene replacement system and the high background of RecBCD-mediated Chi-independent recombination of DNA. The difference in the efficiency of Chi-dependent recombination between the intramolecular recombination and the gene replacement systems may be the result of the difference in the configuration of the respective recombination substrates. While in intramolecular recombination only one recombinogenic end is required for recombinant formation, two ends may be required for gene replacement. It is also conceivable that the search for homology by the short substrates is more efficient in intramolecular recombination than in the replacement of a gene on the bacterial chromosome.

	ACKNOWLEDGMENTS

We thank JOHN CLARK, ROBERT LLOYD, SUSAN ROSENBERG and WILFRIED WACKERNAGEL for bacterial strains and plasmids; GERRY SMITH for communicating unpublished results; NORMAN GROVER, STEVE KOWALCZYKOWSKI, ANDREI KUZMINOV, SUSAN ROSENBERG, GERRY SMITH, FRANK STAHL and WILFRIED WACKERNAGEL for helpful scientific discussions; and SUE AMUNDSEN, LAURA BENJAMIN, GERRY SMITH, FRANK STAHL and anonymous reviewers for critical review of the manuscript. This work was supported by grants from the Israel Science Foundation (141/93-3) and the UK-Israel Science and Technology Research Fund (Project 023).

Manuscript received June 13, 1997; Accepted for publication October 17, 1997.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDERSON, D. G. and S. C. KOWALCZYKOWSKI, 1997  The recombination hot spot is a regulatory element that switches the polarity of DNA degradation by the RecBCD enzyme. Genes Dev. 11:571-581[Abstract/Free Full Text].

BACHMANN, B. J., 1987 Derivation and genotype of some mutant derivatives of Escherichia coli K-12, pp. 2460–2488 in Escherichia coli and Salmonella: Cellular and Molecular Biology, edited by F. C. NEIDHARDT, J. L. INGRAHAM, K. B. LOW, B. MAGASANIK, M. SCHAECHTER et al. ASM Press, Washington, DC.

BEDALE, W. A., R. B. INMAN, and M. M. COX, 1993  A reverse DNA strand exchange mediated by recA protein and exonuclease I. J. Biol. Chem. 268:15004-15016[Abstract/Free Full Text].

BETLACH, M. C., V. HERSCHFIELD, L. CHAN, W. BROWN, and H. GOODMAN et al., 1976  A restriction endonuclease analysis of the bacterial plasmid controlling EcoRI restriction and modification of DNA. FASEB J. 35:2037-2043.

CHANG, A. C. Y. and S. N. COHEN, 1978  Construction and characterization of amplifiable multicopy cloning vehicle derived from P15A, the cryptic miniplasmid. J. Bacteriol. 134:1141-1151[Abstract/Free Full Text].

CHENG, K. C. and G. R. SMITH, 1984  Recombinational hotspot activity of Chi-like sequences. J. Mol. Biol. 180:371-377[Medline].

CHENG, K. C. and G. R. SMITH, 1987  Cutting of Chi-like sequences by the recBCD enzyme of E. coli.. J. Mol. Biol. 194:747-750[Medline].

CHENG, K. C. and G. R. SMITH, 1989  Distribution of Chi-stimulated recombinational exchanges and heteroduplex endpoints in phage lambda. Genetics 123:5-17[Abstract/Free Full Text].

CLARK, A. J., 1973  Recombination-deficient mutants of E. coli and other bacteria. Annu. Rev. Genet. 7:67-86[Medline].

CLARK, A. J. and A. D. MARGULIES, 1965  Isolation and characterization of recombination-deficient mutants of E. coli K-12. Proc. Natl. Acad. Sci. USA 53:451-459[Free Full Text].

DABERT, P. and G. R. SMITH, 1997  Gene replacement with linear DNA fragments in wild-type Escherichia coli: enhancement by Chi sites. Genetics 145:877-889[Abstract].

DABERT, P., S. D. EHRLICH, and A. GRUSS, 1992  Chi sequence protects against RecBCD degradation of DNA in vivo.. Proc. Natl. Acad. Sci. USA 89:12073-12077[Abstract/Free Full Text].

DIXON, D. A. and S. C. KOWALCZYKOWSKI, 1993  The recombination hotspot, Chi, is a regulatory sequence that acts by attenuating the nuclease activity of the Escherichia coli recBCD enzyme. Cell 73:87-96[Medline].

DIXON, D. A. and S. C. KOWALCZYKOWSKI, 1995  Role of the Escherichia coli recombination hotspot, , in RecABCD-dependent homologous pairing. J. Biol. Chem. 270:16360-16370[Abstract/Free Full Text].

EGGLESTON, A. K. and S. C. WEST, 1996  Exchanging partners: recombination in E. coli.. Trends Genet. 12:20-26[Medline].

EMMERSON, P. T., 1968  Recombination-deficient mutants of Escherichia coli K-12 that map between thyA and argA.. Genetics 60:19-30[Free Full Text].

FENG, W. Y. and J. B. HAYS, 1995  DNA structures generated during recombination initiated by mismatch repair of UV-irradiated non-replicating phage DNA in Escherichia coli: requirements for helicase, exonucleases, and RecF and RecBCD functions. Genetics 140:1175-1186[Abstract].

FRISCHAUF, A. M., H. LEHRACH, A. POUSTKA, and N. MURRAY, 1983  Lambda replacement vectors carrying polylinker sequence. J. Mol. Biol. 170:827-842[Medline].

GLASSBERG, J., R. R. MEYER, and A. KORNBERG, 1979  Mutants of single-strand binding protein of Escherichia coli: genetic and physiological characterization. J. Bacteriol. 140:14-19[Abstract/Free Full Text].

HOLMES, D. S. and M. QUIGLEY, 1981  A rapid boiling method for the preparation of bacterial plasmids. Anal. Biochem. 114:193-198[Medline].

HOWARD-FLANDERS, P. and L. THERIOT, 1966  Mutants of Escherichia coli K12 defective in DNA repair and genetic recombination. Genetics 53:1137-1150[Free Full Text].

INNIS, M. A., D. H. GELFAND, J. J. SNINSKY and T. J. WHITE (Editors), 1990 PCR Protocols. Academic Press, San Diego.

KOLODNER, R., S. D. HALL, and C. LUISI-DELUCA, 1994  Homologous pairing proteins encoded by the Escherichia coli recE and recT genes. Mol. Microbiol. 11:23-30[Medline].

KONFORTI, B. B. and R. W. DAVIS, 1992  ATP hydrolysis and the displaced strand are two factors that determine the polarity of recA-promoted DNA strand exchange. J. Mol. Biol. 227:38-53[Medline].

KOWALCZYKOWSKI, S. C., D. A. DIXON, A. K. EGGLESTON, S. D. LAUDER, and W. M. REHRAUER, 1994  Biochemistry of homologous recombination in Escherichia coli.. Microbiol. Rev. 58:401-465[Abstract/Free Full Text].

KUZMINOV, A., E. SCHABTACH, and F. W. STAHL, 1994  sites in combination with RecA protein increase the survival of linear DNA in Escherichia coli by inactivating exoV activity of RecBCD nuclease. EMBO J. 13:2764-2776[Medline].

LAM, S. T., M. M. STAHL, K. D. MCMILLIN, and F. W. STAHL, 1974  Rec-mediated recombinational hot spot activity in bacteriophage lambda. II. A mutation which causes hot spot activity. Genetics 77:425-433[Abstract/Free Full Text].

LLOYD, R. G., 1991  Conjugational recombination in resolvase-deficient ruvC mutants of Escherichia coli K-12 depends on recG.. J. Bacteriol. 173:5414-5418[Abstract/Free Full Text].

LURIA, S. E. and J. W. BURROUS, 1957  Hybridization between Escherichia coli and Shigella.. J. Bacteriol. 74:461-476[Free Full Text].

MAHAJAN, S. K., 1988 Pathways of homologous recombination in Escherichia coli, pp. 87–140 in Genetic Recombination, edited by R. KUCHERLAPATI and G. R. SMITH. American Society for Microbiology, Washington, DC.

MIESEL, L. and J. R. ROTH, 1996  Evidence that SbcB and RecF pathway functions contribute to RecBCD-dependent transductional recombination. J. Bacteriol. 178:3146-3155[Abstract/Free Full Text].

MYERS, R. S. and F. W. STAHL, 1994  and the RecBCD enzyme of Escherichia coli.. Annu. Rev. Genet. 28:49-70[Medline].

MYERS, R. S., A. KUZMINOV, and F. W. STAHL, 1995  The recombination hot spot activates RecBCD recombination by converting Escherichia coli to a recD mutant phenocopy. Proc. Natl. Acad. Sci USA 92:6244-6248[Abstract/Free Full Text].

NUSSBAUM, A. and A. COHEN, 1988  The use of a bioluminescence gene reporter for the investigation of Red-dependent and Gam-dependent plasmid recombination in Escherichia coli K-12. J. Mol. Biol. 203:391-402[Medline].

NUSSBAUM, A., M. SHALIT, and A. COHEN, 1992  Restriction-stimulated homologous recombination of plasmids by the RecE pathway of Escherichia coli.. Genetics 130:37-49[Abstract].

RAZAVY, H., S. K. SZIGETY, and S. M. ROSENBERG, 1996  Evidence for both 3' and 5' single-strand DNA ends in intermediates in chi-stimulated recombination in vivo.. Genetics 142:333-339[Abstract].

RINKEN, R., B. THOMS, and W. WACKERNAGEL, 1992  Evidence that recBc-dependent degradation of duplex DNA in Escherichia coli recD mutants involves DNA unwinding. J. Bacteriol. 174:5424-5429[Abstract/Free Full Text].

ROSENBERG, S. M. and P. J. HASTINGS, 1991  The split-end model for homologous recombination at double-strand breaks and at Chi. Biochimie 73:385-397[Medline].

SCHULTZ, D. W., J. SWINDLE, and G. R. SMITH, 1981  Clustering of mutations inactivating a Chi recombinational hotspot. J. Mol. Biol. 146:275-286[Medline].

SHURVINTON, C. E., R. G. LLOYD, F. E. BENSON, and P. V. ATTFIELD, 1984  Genetic analysis and molecular cloning of the Escherichia coli ruv genes. Mol. Gen. Genet. 194:322-329[Medline].

SILBERSTEIN, Z., M. SHALIT, and A. COHEN, 1993  Heteroduplex strand-specificity in restriction-stimulated recombination by the RecE pathway of Escherichia coli.. Genetics 133:439-448[Abstract].

SILBERSTEIN, Z., Y. TZFATI