|Home | About | Journals | Submit | Contact Us | Français|
The orf gene of bacteriophage λ, fused to a promoter, was placed in the galK locus of Escherichia coli K-12. Orf was found to suppress the recombination deficiency and sensitivity to UV radiation of mutants, in a Δ(recC ptr recB recD)::Ptac gam bet exo pae cI ΔrecG background, lacking recF, recO, recR, ruvAB, and ruvC functions. It also suppressed defects of these mutants in establishing replication of a pSC101-related plasmid. Compared to orf, the recA803 allele had only small effects on recF, recO, and recR mutant phenotypes and no effect on a ruvAB mutant. In a fully wild-type background with respect to known recombination and repair functions, orf partially suppressed the UV sensitivity of ruvAB and ruvC mutants.
Conjugative recombination in a recBC sbcB sbcC mutant of Escherichia coli depends upon the recF, recO, and recR genes, as well as other rec genes and the ruv genes, and is said to proceed via the RecF pathway (for a review, see reference 14). In the same cell, however, homologous recombination between bacteriophage λ chromosomes does not depend on recF, recO, or recR, even when the phage lacks its own recombination system, Red, and is thus dependent upon the host's recombination system (33). Sawitzke and Stahl (33) showed that λ's RecORF independence is due to its orf gene (previously called ninB). Further studies showed that orf encodes a protein and that plasmid-borne orf can act in trans to promote phage recombination, but not conjugative recombination, in the absence of RecORF (34).
Recombination between phage chromosomes via the Red pathway requires neither RecORF nor Orf, but Orf nonetheless is a participant (39). The result of its action is to focus crossovers near the initiating double-strand break in both the RecF (35) and Red (39) pathways.
An additional observation suggested that Orf can promote recombination events other than crossovers between phage chromosomes. In a cell in which the recC-ptr-recB-recD gene cluster is replaced by the λ Red system genes (gam, bet, and exo) and from which the recG gene is deleted, recombination between the bacterial chromosome and a 3.6-kbp linear double-stranded DNA (dsDNA) molecule is dependent upon recF, recO, and recR. However, in such a cell, ectopic expression from the bacterial chromosome of a segment of the λ chromosome, including the orf gene, partially decreases this dependence (29).
For this study, the orf gene was installed by itself in the galK locus. The expression of chromosomally encoded orf was found to have surprisingly pleiotropic effects on recombination, replication, and repair in E. coli. Orf suppressed the mutant phenotypes of not only recF, recO, and recR, but ruvAB and ruvC as well. In contrast, the recA803 allele, which partially suppresses the phenotypes of recF, recO, and recR mutations affecting RecF pathway recombination and repair (44, 45), had no comparable effect on the Red pathway.
The bacterial strains used for this study are described in Table Table1.1. Some of the constructions involved the replacement of chromosomal genes by means of λ Red-mediated recombination with linear DNA species introduced into the cell via electroporation.
The λ orf gene was inserted into a plasmid vector and then crossed into the galK locus, essentially as previously described for the rap and gfp genes (31). The primers used for orf amplification had the sequences 5′-GGAGAGGGAACATATGAAAAAACTAACC-3′ and 5′-ATATGCTGAGCTCCTTCAACCGGAGAA-3′. The orf-containing plasmid intermediate was designated pTP913.
A variant of the Δ(recC ptr recB recD)::Ptac gam bet exo pae cI822 substitution lacking red functions was constructed. Plasmid pTP822 contains recC-flanking sequences, a red-expressing cassette, PaeR7 restriction modification genes, the λ cI gene, and recD-flanking sequences (30). Parts of the red sequences were deleted by digestion of the plasmid with the HpaI restriction endonuclease and ligation in the presence of nonphosphorylated NotI linkers (AGCGGCCGCT) to make plasmid pTP964. The deleted sequences included codons for the C-terminal 63 amino acid residues of the Beta protein and the N-terminal 72 residues of the exonuclease.
Plasmid pTP980, bearing a deletion-substitution of recC-ptr-recB-recD, was made by replacing DNA sequences between the BglII and SacI sites in pTP822 with the tetR and tetA genes of transposon Tn10. The tet genes were amplified by PCR with primers 5′-CTGCTGAGATCTCTCGACATCTTGGTTACCGT-3′ and 5′-GCAGCATCTAGACGCGGAATAACATCATTTGG-3′. The plasmid and PCR product were digested with BglII and SacI and then ligated together.
Deletion-substitutions of the recA and ruvC genes were made by electroporation of Red-expressing bacteria with PCR-generated linear DNA species consisting of the tetR and tetA genes of transposon Tn10 flanked by 40 bases matching sequences in and near the ends of the coding sequences of the target genes. The primers used for PCRs were as follows: for recA, 5′-ATTACCCGGCATGACAGGAGTAAAAATGGCTATCGACGAA CTCGACATCTTGGTTACCGT-3′ and 5′-GACCCTTGTGTATCAAACAAGACGATTAAAAATCTTCGTT CGCGGAATAACATCATTTGG-3′; and for ruvC, 5′-CTAAACAGCAAAACGGAGACGCGTGATGGCTATTATTCTCCTCGACATCTTGGTTACCGT-3′ and 5′-GAACTGACCGAGGCGGTATAACTTAACGCAGTCGCCCTCTCGCGGAATAACATCATTTGG-3′.
The recG and sulA genes were deleted by means of a variation of the ampicillin enrichment technique described by Murphy et al. (25). The targeted genes were first replaced with a gfp-cat cassette, which makes the resulting strain fluorescent and chloramphenicol resistant. The gfp-cat cassette was subsequently replaced by means of Red-mediated recombination with a linear DNA species consisting only of sequences flanking the target gene (recG6202 allele) (25). Recombinants lacking gfp-cat were enriched in the resulting mixed culture as previously described, except that their growth was arrested with 10 μg of chloramphenicol/ml before the addition of ampicillin to kill the growing cells. Only one cycle of enrichment was required, as the recombinants were readily detected among large numbers of nonrecombinant colonies by their lack of fluorescence under long-wavelength UV radiation. The gfp-cat cassette contains the gfp gene from pGreenTIR (21) fused to the strong, lacI-controllable promoter PA1/04 (11, 38). Its putative DNA sequence and details of its use will be provided upon request.
The reduced-genome E. coli strain MDS12 (9) was used to stabilize the Δ(recC ptr recB recD)::Ptac gam bet exo pae cI822 ruvC genotype, which has a tendency to acquire suppressor mutations in the AB1157 background (31). The suppressor mutations presumably activate rusA (18, 19). The cryptic prophage encoding rusA is deleted from MDS12, and a Δ(recC ptr recB recD)::Ptac gam bet exo pae cI822 ruvC derivative of MDS12 did not appear to acquire suppressors.
The recA803 allele was introduced via P1 transduction into strains bearing Δ(srl recA)306::Tn10, with selection for growth on sorbitol-containing minimal medium. The presence of the recA803 substitution V37M (17) was verified in all strains employed for this study. A segment of the chromosome containing the recA gene was PCR amplified, and the dsDNA product was sequenced by automated fluorescent dideoxynucleotide chain termination methods at the University of Massachusetts Medical School Nucleic Acids Facility (data not shown).
The lac::cat819 substitution borne by λ lac::cat819 nin5 (30) replaces bp 23,135 to 33,498 of the λ chromosome, including attP, int, xis, exo, bet, and gam; the nin5 deletion removes a number of other genes, including orf and rap. The phage thus lacks all of its known recombination functions. Infection of a host bearing the PaeR7 restriction modification system by unmodified λ lac::cat819 nin5 results in intracellular cutting of the phage chromosome at the boundaries of the lac::cat819 substitution, releasing it as a linear dsDNA species. λGB2, a hybrid between λ and the spectinomycin and streptomycin resistance-conferring plasmid pGB2 (5) was constructed by crossing wild-type λ with plasmid pTP933 and by isolating recombinant phages as described previously (28). In the hybrid, pGB2 sequences replace bp 23,135 to 33,498 of the λ chromosome. Plasmid pTP933 was constructed by ligating XhoI-digested pTP819 (30) with SalI-digested pGB2. The resulting XhoI-SalI joints are not cuttable by XhoI or its isoschizomer PaeR7; therefore, λGB2 is not cut by PaeR7 in the infected cell. Phages were propagated and titrated on strain KM32.
Bacterial strains were tested for recombination proficiency in groups of four. Single colonies were used to inoculate Luria-Bertani (LB) medium (1% tryptone, 0.5% yeast extract, and 0.5% NaCl, with 1 mM added NaOH). Cultures were incubated motionless for 18 to 24 h at 37°C. Cells were harvested by centrifugation for 10 min at 4,300 × g at 4°C, resuspended with 0.05 to 0.1 volume of TM (10 mM Tris-HCl [pH 7.4], 10 mM MgSO4), and kept on ice. The absorbance values at 600 nm of the resuspended cells were determined, and all were adjusted to the same value, that of the least dense culture or 4.0, whichever was lower. The actual adjusted values of A600 were in the range of 2.5 to 4.0. Dilutions of the adjusted resuspended cells were counted in a Petroff-Hauser chamber for total cell counts and were plated on LB agar (LB medium with 1.5% agar) for the determination of total viable titers. Cells (100 μl) were added to the phage mixture (10 μl) and incubated at 37°C for 15 min. The phage mixture consisted of λ lac::cat819 nin5 and λGB2, at approximately 109/ml and 107/ml, respectively, in TM plus 0.01% gelatin, resulting in nominal multiplicities of infection of approximately 0.1 and 0.001, respectively. Both phages have wild-type λ early regulatory systems and thus are prevented from expressing their genes, or replicating, by the cI repressor present in the infected cells. Mixtures were subsequently diluted 100-fold into TM with CHCl3 for the determination of unadsorbed phage titers as well as into prewarmed LB medium for 1 h of further incubation, with aeration, at 37°C, after which they were plated on LB agar containing 10 μg of chloramphenicol/ml and on LB agar containing 20 μg of spectinomycin/ml. Measurements of viability (viable titer divided by total cell count), numbers of chloramphenicol-resistant colonies per infected viable cell, and numbers of spectinomycin-resistant colonies per infected viable cell were recorded. Each infection was done in triplicate. Measured efficiencies of phage adsorption ranged from 76% to over 99%.
Bacterial strains were tested for sensitivity to UV radiation in groups of four. Dilutions of cells prepared and titrated as described above were spotted, in 20-μl portions, onto LB agar plates which had been previously incubated for 1 to 2 h uncovered at 37°C to promote quick drying of the spots. Three identical plates, each spotted with dilutions of four bacterial strains, were simultaneously irradiated with a set dose of short-wave UV in a Stratalinker apparatus (Stratagene). The parallel array of cylindrical UV lamps in this apparatus was expected to produce a uniform intensity of irradiation; in practice, a good agreement of colony counts among simultaneously irradiated plates was obtained (see Tables Tables22 and and6).6). The irradiated plates were removed from the apparatus under dim fluorescent light, wrapped in aluminum foil, and incubated at 37°C.
Orf is deleterious to E. coli, as previously noted by Sawitzke and Stahl (34). Expression of the orf gene from a multicopy plasmid (a pBR322 derivative) directed by the moderate, lac repressor-controllable promoter Pmac was found to be lethal to wild-type E. coli, even in a strain bearing F′ lacIQ. A nonlethal level of expression was obtained by supplying an additional lac repressor in trans from a compatible plasmid. In the chromosome, the galK::Pmac orf insertion conferred a slow-growth, small-colony phenotype on both wild-type and Δ(recC ptr recB recD)::Ptac gam bet exo pae cI strains. This growth retardation phenotype was exacerbated by the addition to the medium of IPTG (isopropyl-β-d-thiogalactopyranoside), which induces an elevated level of expression from Pmac in the strains used for this study, all of which bear the wild-type lacI gene. Otherwise identical constructs containing gfp or rap in place of orf (31) did not exhibit this phenotype (data not shown).
The ability of the Orf protein to substitute for various E. coli recombination and repair functions was tested by the construction of isogenic galK::orf and galK::gfp strains and comparisons of the effects of null mutations in rec or ruv genes in the two backgrounds.
The first set of experiments was done with strains in which RecBCD had been replaced by Red and from which recG was deleted. The reason for employing a background lacking RecG was that it exhibits elevated levels of Red-mediated recombination (29). Efficient recombination between the chromosomes of these strains and double-stranded linear DNA molecules depends upon the red, recA, recF, recO, recR, recQ, ruvAB, and ruvC genes (24, 29, 30). The recombining linear dsDNA molecule was released by intracellular PaeR7 restriction endonuclease digestion of the chromosome of infecting λ lac::cat nin5, which was prevented from replicating or expressing lytic genes by the host-encoded cI repressor (30). The efficiencies of formation of chromosomal lacZ::cat recombinants by the wild-type and mutant strains are shown in Table Table2.2. Orf effects (the ratios of recombinants of the galK::orf strain to recombinants of the corresponding galK::gfp strain) are shown in Table Table3.3. Orf strongly stimulated recombination in recF, recO, and recR mutant cells, to nearly the level seen for the wild type. In addition, and unexpectedly, Orf significantly stimulated recombination in ruvAB and ruvC mutants. Orf had little or no effect on recombination in the wild type or in recA, recQ, or red mutants.
The viability of a Δ(recC ptr recB recD)::Ptac gam bet exo pae cI822 strain is significantly reduced by the elimination of recF, recO, or recR. This viability defect is partially suppressed by deletion of the sulA gene (29). The remaining viability defects of these strains are more prominent in logarithmically growing cells than in stationary-phase cells (data not shown). To minimize the role of viability effects in experiments on recombination, we deleted sulA from all Red-substituted strains used for this study; in addition, we used saturated or near-saturated cultures grown without active aeration for infections. Even so, some mutants exhibited reduced viability—as much as sevenfold in the case of recF (Table (Table2).2). As indicated in Table Table3,3, Orf partially compensated for the viability defects of the recF, recR, and ruvC mutants; less activity was seen in the cases of recO and ruvAB mutants, but this may not be significant. Orf made no contribution to the viability of recA, recQ, and red mutants.
Some of the mutant strains have a reduced ability to become spectinomycin resistant after infection with λGB2 (Table (Table2).2). This phage, a hybrid between λ and the pSC101-based plasmid vector pGB2, converts an immune (λ repressor-expressing) wild-type host to spectinomycin resistance with nearly 100% efficiency. The ruvAB strain exhibited an especially marked defect in conversion; smaller defects were seen in the recF, recO, recR, and ruvC mutants. The mutants exhibited no defect in plaque formation when they were infected with heteroimmune λ strains (data not shown). This observation indicates that phage-borne genes are efficiently taken up and expressed in the mutant strains and therefore that the mutants probably are defective in replication from the pSC101 origin. Consistent with this interpretation, we observed that many of the ruvAB cells, but not wild-type cells, infected with λGB2 formed microcolonies on spectinomycin-containing plates (data not shown). Orf restored the efficiencies of conversion of all of the defective mutants to nearly the wild-type level (Table (Table33).
The “wild-type” strain used for these experiments lacks the recG gene, and as a consequence, is UV sensitive relative to the true wild type. As shown in Table Table2,2, the deletion of additional rec or ruv genes increased the UV sensitivity to various degrees, as was previously reported (29). Orf greatly decreased the UV sensitivity of recF, recO, recR, ruvAB, and ruvC mutants, slightly decreased the UV sensitivity of the recA mutant, and had little or no effect on the UV sensitivity of the wild type or the red mutant (Table (Table33).
The recA803 allele was tested in the same way as Orf, by the construction of isogenic strains bearing the wild-type and 803 alleles of recA and comparisons of the effects of knocking out rec or ruv genes in the two backgrounds. The data in Tables Tables44 and and55 show that the recA803 allele had little or no effect on the ability of recF, recO, recR, and ruvAB mutants to recombine. The recA803 allele modestly decreased the UV sensitivity of the recF, recO, and recR mutants but had almost no effect on that of the ruvAB mutant. The recA803 allele additionally had little effect on the ability of the ruvAB mutant to convert to spectinomycin resistance after infection with λGB2. The effect of recA803 on this phenotype in the other mutants was not meaningfully measurable; the experiments were done in the MDS12 background, in which the phenotype is less pronounced than in the AB1157 background.
The inability of RecA803 to bypass the need for RecORF raised the question of whether RecA is directly involved in Red-mediated recombination. It might, alternatively, be required only for the production of an SOS function which participates directly. To test this idea, the recombination proficiencies of lexA::Tn5 and lexA::Tn5 recAΔ::tet strains with λ lac::cat819 nin5 were compared. These strains (TP657 and TP851) produced 0.24% ± 0.036% and 0.004% ± 0.0004% recombinants, respectively, per infected viable cell (compare with the recombination figures shown in Table Table2).2). The RecA dependence of recombination in the constitutively SOS-induced lexA::Tn5 strain, while not as strong as that of the wild type (60-fold versus 700-fold), indicates that RecA has a role other than helping the induction of SOS functions.
The large effects of Orf on the UV sensitivity of recF, recO, recR, ruvAB, and ruvC mutants in the Δ(recC ptr recB recD)::Ptac gam bet exo pae cI822 ΔrecG genetic background raised the question of whether Orf would have similar effects in a genetic background more closely related to the wild type with regard to known recombination and repair functions. To test this, isogenic galK::orf and galK::gfp variants of strain AB1157 were constructed, followed by corresponding pairs of mutants lacking rec or ruv genes. The data in Table Table66 show that Orf significantly reduced the UV sensitivity of the ruvAB and ruvC mutants; this Orf effect is further illustrated by the survival curves in Fig. Fig.1.1. Orf additionally modestly increased the survival of recF, recO, recR, recG, and recA strains after UV irradiation (Table (Table66).
The Orf protein can partially substitute for RecORF in Red-mediated recombination between the bacterial chromosome and short linear dsDNA molecules, in promoting the replication of a pSC101-derived plasmid, and in recovery of a cell from UV-induced damage. These findings are superficially inconsistent with those of a previous study, which suggested a more limited capability of Orf to substitute for RecORF (34). However, in the previous study, orf was expressed from a plasmid. As noted by the investigators, in a recBC sbcB sbcC background, RecORF and Orf both might be expected to influence the tendency of a plasmid to produce potentially lethal linear multimers (10). Moreover, there is indirect evidence that plasmid linear multimers can inhibit recombination, perhaps by competing for a limited supply of recombination proteins (24).
The ability of Orf to substitute for RecORF has suggested all along that Orf may perform the same mechanistic function as RecORF (33). The term “RecORF” (also appearing in the literature as “RecFOR”) recognizes that the three polypeptide products of the recF, recO, and recR genes are thought to function as a physically interacting complex in the same step(s) in recombination and gap repair (7, 22, 45). Genetic (44) and biochemical (4, 22, 41, 42) evidence indicates that the key RecORF-promoted step is the loading of RecA protein onto single-stranded DNA (ssDNA), displacing the E. coli single-stranded DNA-binding protein (SSB).
That RecORF's main role in recombination is, in effect, to assist RecA was indicated by the isolation of srf (suppressor of recF) alleles of recA (44). The protein encoded by one such allele, RecA803, has been shown to compete more effectively than wild-type RecA with SSB for binding to ssDNA (16). Both a recA803 recBC sbcB sbcC strain and a Δ(recC ptr recB recD)::Ptac gam bet exo pae cI ΔrecG galK::orf strain are capable of recombining efficiently in the absence of RecORF. This observation led to the prediction that RecA803 might do for the Δ(recC ptr recB recD)::Ptac gam bet exo pae cI ΔrecG cell what Orf does: promote efficient RecORF-independent recombination. The inability of RecA803 to promote RecORF-independent recombination via the Red pathway (Table (Table5)5) raised the question of why RecA803 needs RecORF to load it onto ssDNA, given that RecA803 can displace SSB by itself. One interesting possible answer is that for Red recombination, RecA needs to displace the Beta protein, not SSB, from ssDNA.
The idea that a dsDNA end acted on by Red is converted to a Beta protein-coated 3′-ended ssDNA is consistent with the enzymology of λ exonuclease (Exo) (12), the interaction between Exo and Beta (32), and the DNA-binding properties of Beta (8, 23). The idea that RecA has to displace Beta to act on this DNA end has implications for our understanding of the Red pathway.
According to the model for Red recombination developed by Stahl and coworkers (37), the Red pathway is best understood as having two branches, with one proceeding via strand annealing and the other proceeding via strand invasion. The strand invasion mechanism requires RecA, while strand annealing is RecA independent. RecA-promoted strand invasion is necessary when only one of the two recombining partners has a free end and the other does not, as is the case in the recombination event monitored in this study involving the lac::cat dsDNA segment and the (uncut) bacterial chromosome.
The inference, outlined above, that RecA must displace Beta to form recombinants via the strand invasion pathway suggests that the progression of a double-strand-break repair-recombination event down one or the other branch of the Red pathway is kinetically regulated by the type of partner available to the linear DNA species (Fig. (Fig.2).2). The sequence of events in Red recombination, according to this view, would be as follows. (i) An end produced by a double-strand cut to a chromosome is acted on by the Exo-Beta complex. Exo digests the 5′-ended strand, and at the same time, deposits the Beta protein on the exposed 3′-ended single strand in a way which is analogous to the deposition of RecA protein on ssDNA by RecBCD after the interaction of the latter with a χ site (1, 43). (ii) If a complementary ssDNA is present in the cell, Beta promotes annealing, forming a recombinant. (iii) If a complementary ssDNA is not present and a dsDNA with shared sequences is present, recombination will still take place, but only after RecORF (or possibly Orf) removes Beta from the 3′-ended ssDNA and replaces it with RecA.
The idea of sequential and mutually exclusive actions by Beta and RecA suggests that strand annealing is the primary Red pathway while strand invasion is a salvage pathway, only taking place when strand annealing is blocked. In the salvage pathway, perhaps the only role of Red is to convert a dsDNA end into a 3′ overhang, and a large number of additional cellular recombination proteins may be needed to make a recombinant.
The sequential Beta-RecA hypothesis raises the following question: why is Beta required at all in the strand invasion pathway? Beta is specifically required for Red activity in both strand annealing and strand invasion events (26). Possibly, Beta modulates the exonuclease activity of Exo, which otherwise would destroy recombination intermediates.
The surprising finding of this study was that Orf complements null mutations in ruvAB and ruvC. In the study that originally characterized Orf functions, this activity could not have been detected, as the recombination event under investigation was Ruv independent (33). The Ruv-complementing activity was observable in a cell which was mutated only in ruvAB or ruvC, not in genes for other known repair-recombination functions (Fig. (Fig.1),1), implying that it is not a peculiarity of the highly engineered Δ(recC ptr recB recD)::Ptac gam bet exo pae cI ΔrecG genetic background in which most of the experiments were done.
The RuvA, RuvB, and RuvC proteins have been shown to act in concert to resolve branched DNA molecules which model recombination intermediates (46). RuvB is a helicase which drives branch migration and which is targeted to Holliday junctions by the RuvA protein (40). RuvC is an endonuclease (resolvase) which specifically cleaves Holliday junctions at symmetrically related strands (3). The phage λ-encoded Rap protein, which is also a junction-targeted endonuclease (36), can, like Orf, partially substitute for RuvC, but not for RuvAB or other E. coli recombination proteins (31).
The pleiotropy of Orf action suggests that Orf does not directly replace the proteins whose functions it renders nearly unnecessary. It is readily conceivable that Orf might have the same functional activity as RecORF. If a single amino acid substitution in RecA, turning it into RecA803, is almost all that is needed to dispense with RecORF, there is no obvious reason why the same effect could not be achieved by even the small Orf protein, which has a monomer molecular mass of 16.6 kDa. It is harder to imagine that Orf could have the same enzymatic activities as RuvABC and is nearly inconceivable that it could mimic both RecORF and RuvABC.
One way in which Orf might indirectly suppress the phenotypes of strains lacking RecORF or RuvABC components is by inducing the expression of a set of cryptic genes which have RecORF- and RuvABC-like activities. Indeed, ruv mutations are known to be suppressed by mutational activation of the cryptic rusA gene (19). The ability of Orf to suppress a ruvC mutation is rusA independent, as it occurs in a genetic background from which rusA and all the cryptic prophage genes in its vicinity have been deleted (Table (Table4)4) (9), but the possibility that Orf activates some other set of genes cannot be ruled out.
A second way in which Orf might work is by modifying some other protein or multiprotein complex, making it RecBCD-like, i.e., capable of carrying out moderately efficient recombination in the absence of RecORF or RuvABC (see references 14 and 27 for reviews). This hypothetical mechanism of Orf action is constrained by three observations, as follows. (i) In the Δ(recC ptr recB recD)::Ptac gam bet exo pae cI ΔrecG galK::orf background, recombination and repair are independent of RecORF and RuvABC but are still highly dependent upon Red, despite the presence of Orf (Table (Table2).2). It follows from this observation that there is no Orf-modified protein complex in the cell, other than Exo-Beta, which can operate on dsDNA ends to promote efficient exchanges, and also that RecG cannot be the Orf target. (ii) Suppressing effects of Orf are also seen in cells in which RecBCD is present (Table (Table6).6). Therefore, having both RecBCD and Orf gives cells capabilities beyond those which result from having RecBCD without Orf. (iii) Orf does not suppress a recQ mutant (Table (Table3),3), leaving open the possibility that RecQ might be the Orf target. This possibility is made unlikely, however, by the observation that the slow growth phenotype conferred by Orf is not suppressed by a recQ null mutation, as would be expected if Orf acted only by modifying RecQ; the recQ mutation itself does not noticeably affect the growth rate (data not shown).
A third way in which Orf might work is by modulating the activity of SSB. This SSB modulation hypothesis provides a ready explanation for how Orf could help RecA displace SSB from ssDNA. Orf-modulated SSB may also help RecA to compete with Beta for ssDNA binding. The expression level-dependent growth slowing or stopping activity of Orf can also be readily understood as an effect of its interfering to various degrees with SSB's function in DNA replication (see reference 20 for a review). It is not obvious how SSB modulation could account for Orf's RuvABC-bypassing activity. The essential role of SSB in replication makes a precise evaluation of its role in recombination difficult. However, there is evidence indicating that SSB has an essential role in recombination as well as in replication (6). The current understanding of recombination mechanisms permits speculation that the resolution of recombination intermediates can occur by a number of different pathways (13, 47), but at least in a Δ(recC ptr recB recD)::Ptac gam bet exo pae cI ΔrecG strain, only RuvABC is fast enough to resolve them in such a way as to form recombinants efficiently. If RuvABC is not present, perhaps other fast processes resolve the intermediates without forming recombinants. If SSB has a critical role in any of these other processes, then modulation of its activity by Orf might slow them down or accelerate still other recombinant-forming processes.
I thank Kenan Murphy and Michael Volkert for helpful discussions; Michael Volkert, Tim Durfee, and Fred Blattner for strains; and Ashwini Nadkarni for excellent technical assistance.
This research was supported by grant MCB-0234991 from the National Science Foundation.