Modulation of RecFORQ- and RecA-Mediated Homologous Recombination in Escherichia coli by Isoforms of Translation Initiation Factor IF2

ABSTRACT

Homologous recombination (HR) is critically important for chromosomal replication, as well as DNA damage repair in all life forms. In Escherichia coli, the process of HR comprises (i) two parallel presynaptic pathways that are mediated, respectively, by proteins RecB/C/D and RecF/O/R/Q; (ii) a synaptic step mediated by RecA that leads to generation of Holliday junctions (HJs); and (iii) postsynaptic steps mediated sequentially by HJ-acting proteins RuvA/B/C followed by proteins PriA/B/C of replication restart. Combined loss of RuvA/B/C and a DNA helicase UvrD is synthetically lethal, which is attributed to toxicity caused by accumulated HJs since viability in these double mutant strains is restored by removal of the presynaptic or synaptic proteins RecF/O/R/Q or RecA, respectively. Here we show that, as in ΔuvrD strains, ruv mutations confer synthetic lethality in cells deficient for transcription termination factor Rho, and that loss of RecFORQ presynaptic pathway proteins or of RecA suppresses this lethality. Furthermore, loss of IF2-1 (which is one of three isoforms [IF2-1, IF2-2, and IF2-3] of the essential translation initiation factor IF2 that are synthesized from three in-frame initiation codons in infB) also suppressed uvrD-ruv and rho-ruv lethalities, whereas deficiency of IF2-2 and IF2-3 exacerbated the synthetic defects. Our results suggest that Rho deficiency is associated with an increased frequency of HR that is mediated by the RecFORQ pathway along with RecA. They also lend support to earlier reports that IF2 isoforms participate in DNA transactions, and we propose that they do so by modulation of HR functions.

IMPORTANCE The process of homologous recombination (HR) is important for maintenance of genome integrity in all cells. In Escherichia coli, the RecA protein is a critical participant in HR, which acts at a step common to and downstream of two HR pathways mediated by the RecBCD and RecFOR proteins, respectively. In this study, an isoform (IF2-1) of the translation initiation factor IF2 has been identified as a novel facilitator of RecA’s function in vivo during HR.

INTRODUCTION

Genetically first identified as a mechanism to mediate exchange of hereditary determinants between cells (1), the process of homologous recombination (HR) in bacteria has also been recognized to play a crucially important role in maintenance of genomic integrity, both during chromosomal replication and following DNA damage, just as is the case in archaea and eukaryotes (reviewed in references 2–12). In Escherichia coli, protein RecA is central to the synaptic step of HR: RecA binds a suitable single-strand (ss)-DNA substrate to form a nucleoprotein filament, which performs a homology search to enable strand invasion into a homologous duplex. This synaptic step is flanked by presynaptic and postsynaptic reactions, as described below.

The presynaptic reactions are designed to generate the ss-DNA substrate for RecA’s binding. Two alternative presynaptic pathways RecBCD and RecFOR (named after the principal proteins mediating them) exist to prepare substrates from, respectively, double-strand breaks (DSBs) and ss-gaps in DNA. Loss of either RecBCD or RecFOR pathway alone affects certain categories of DNA recombination and repair, whereas combined loss of both pathways confers a deficiency as severe as that with loss of RecA itself (3, 7).

The presence of RecA-bound nucleoprotein serves also as a trigger for an SOS response within the cell, by which several prophages are induced to enter lytic growth and genes of the LexA-repressed SOS regulon are transcriptionally activated. For the SOS response, RecA’s coprotease activity is stimulated following nucleoprotein assembly to facilitate auto-cleavage of LexA and prophage repressors (reviewed in reference 13).

The postsynaptic reactions act to generate discrete recombinant DNA molecules following RecA-catalyzed formation of recombinant intermediates and formation of Holliday junctions (HJs). The RuvAB helicase and RuvC resolvase are primary mediators at this step, which catalyze branch migration and resolution of HJs, respectively (3). HJ resolution can also be achieved in the absence of RuvABC, in E. coli cells harboring a mutation (rus-1) that activates expression of the protein Rus encoded by a cryptic prophage (14, 15); however, the physiological role, if any, of Rus in wild-type cells is unclear.

Included in the postsynaptic phase of HR are steps of replication restart, by which products generated after resolution of HJs are assimilated into the circular bacterial chromosome (or plasmid). Proteins mediating replication restart include PriABC, DnaT, and Rep, which are proposed to act through several alternative and redundant pathways (16–21).

Control of HR in E. coli is achieved by autoregulation (through the SOS response [13]), as well as by several other factors (3, 22). One of the latter is UvrD, which is a 3′–5′ DNA helicase that also serves as an antirecombinase to disrupt RecA nucleoprotein filaments (23, 24). Loss of UvrD is associated with (i) a hyperrecombination phenotype mediated by the RecFORQ presynaptic pathway in HR assays, and (ii) synthetic lethality with ruv mutations, presumed to be the consequence of accumulation of trapped HJs leading to impedance of chromosomal replication (25–27). Both phenotypes are suppressed by loss of RecA, which also indicates that hyperrecombination is not essential for viability of uvrD mutants.

In this study we have identified that compromised function of transcription termination factor Rho also confers synthetic lethality with ruv mutations and that, just as in the case with uvrD-ruv mutants, abrogation of the RecFORQ pathway or of RecA rescues this lethality. Rho is an essential protein in E. coli that mediates the termination of transcripts (other than rRNAs and tRNAs) that are not being simultaneously translated (28–32). Rho’s function has been implicated in the maintenance of genomic integrity (33), and in avoidance of RNA–DNA hybrids or R-loops from nascent untranslated transcripts including from antisense RNAs (28, 29, 31, 34, 35).

Another finding being reported from this study is that both rho-ruv and uvrD-ruv synthetic lethalities were suppressed by mutations in infB, which encodes the translation initiation factor IF2 (36); we also show that these infB mutants are downregulated for HR functions. Studies of the past 15 years from the Nakai group had already identified an intriguing connection between IF2 on the one hand and DNA transactions including DNA damage repair on the other (37–39). IF2 is essential for E. coli viability (36), and its function in translation initiation is conserved across bacteria, archaea, and eukaryotes (40, 41); a mammalian mitochondrial IF2 homolog can restore viability to an E. coli IF2 knockout strain (42). Three IF2 isoforms are synthesized in E. coli (see Fig. S3A in the supplemental material), from in-frame initiation codons at positions 1,158, and 165 of the 890-codon-long infB ORF, which are herein designated as IF2-1, IF2-2, and IF2-3, respectively (43, 44). All isoforms are active for translation initiation, and any one of them is sufficient for viability (45, 46). At least two studies have shown that isoforms IF2-1 and IF2-2,3 (the latter designation is used for the mixture of IF2-2 and IF2-3, since they are only marginally different from one another) are expressed at more or less similar levels in E. coli (41, 46). The Nakai lab had demonstrated that isoforms IF2-1 and IF2-2,3 behave differently in vitro in assays for phage Mu transposition, and that they confer differential tolerance in vivo to genotoxic agents. They had proposed that the isoforms exert varying influences on different pathways of replication restart (37–39).

In the companion article (47), we show that infB mutants are also defective for two-ended DSB repair, which is mediated by RecA through the RecBCD-mediated presynaptic pathway. Based on these combined results, we propose that the IF2 isoforms differentially affect the efficiency of synapsis between DNA molecules during HR.

RESULTS

The genetic assays to identify lethality of mutants and their suppression.

The genetic (blue-white) assay to demonstrate lethality has been described earlier (28, 48–51). This method makes use of a single-copy-number-shelter plasmid encoding trimethoprim (Tp)-resistance and carrying lacZ⁺ as well as a functional copy of the gene(s) of interest, and whose partitioning into daughter cells during cell division is not stringently regulated. Consequently, when a strain with this plasmid, along with Δlac and a mutation in the gene of interest on the chromosome, is grown in medium not supplemented with Tp, plasmid-free cells that arise spontaneously in the culture (at around 5% to 20%) will be able to grow as white colonies on Xgal-supplemented plates if and only if the now unsheltered chromosomal mutation is not lethal; on the other hand, control blue colonies (formed from cells retaining the shelter plasmid) would be observed as a majority in all situations. In the studies below, we have employed the blue-white assay with a shelter plasmid carrying either rho⁺ or both rho⁺ and infB⁺ genes to examine lethality or synthetic lethality of rho, rho-ruv, and infB genes.

rho-ruv is synthetically lethal.

As mentioned above, transcription termination factor Rho is essential for E. coli viability (28, 33). In this study, a rho mutant was obtained with an opal (TGA) chain-terminating missense substitution in codon 136, which was shown to be viable in strains carrying the E. coli K-12 version of the prfB gene encoding release factor 2 (RF2) but not in those carrying the E. coli B version (Fig. S1A i–ii); there is evidence that the former but not the latter permits a low frequency of stop-codon readthrough (52). This was validated by Western blotting analysis, which showed, for the rho-136^opal mutant, bands corresponding to both full-length (faint) and truncated Rho polypeptides (Fig. S1B). Another viable opal mutant in codon 157 of rho has been reported recently in E. coli K-12 (53).

The rho-136^opal mutation was synthetically lethal, with disruptions of the ruv genes (ΔruvA, ΔruvC, or ΔruvABC) encoding the HJ enzymes RuvAB or RuvC, and the phenotype was manifested on both rich and defined media (see, for example, Fig. 1A ii–iii, Fig. 2 i, and Fig. S2 i). On the other hand, rho-136^opal was not lethal with recA (Fig. 1A iv). These results are consistent with those from an earlier report that ruv, but not recA, mutants are hypersensitive to the Rho inhibitor bicyclomycin (33).

FIG 1.

Synthetic lethality of rho ruv mutants, and their suppression by infB-161^ochre or rec mutations. Mutant designations rho and ruv refer to alleles rho-136^opal and ΔruvABC::Cm, respectively. All strain numbers mentioned are prefixed with GJ. (A) Blue-white screening assays, on defined medium at 30°, for strains carrying rho⁺ infB⁺ shelter plasmid pHYD5212. Representative images are shown; for each of the subpanels, relevant chromosomal genotypes/features are given on top while the numbers beneath indicate the percentage of white colonies to total (minimum of 500 colonies counted). Examples of white colonies are marked by yellow arrows. Strains employed for the different subpanels were pHYD5212 derivatives of the following: i, 15441; ii, 15447; iii, 19150; iv, 19845; v, 15460; vi, 15471; vii, 15491; viii, 15487; ix, 15497; and x, 15446. (B) Dilution-spotting assay, on minimal A supplemented with D-glucose (Glu) or Ara as indicated, of strains whose relevant genotypes/features are shown at left. Strains for different rows were (from top): 19379, 19380, 19381 (this strain was grown in Ara-supplemented medium before dilutions were spotted), and 19382.

FIG 2.

Modulation of rho ruv synthetic lethality by differential expression of IF2 isoforms. Blue-white screening assays were performed on defined medium at 30°, with strains carrying rho⁺ infB⁺ shelter plasmid pHYD5212. General notations are as described in the legend to Fig. 1A. All strains were ΔinfB, with the exception of that for subpanel v. Growth medium for subpanels iv, v, and viii was supplemented with IPTG. Strains employed for the different subpanels were pHYD5212 derivatives of the following (all strain numbers are prefixed with GJ): i, 15498; ii, 15500; iii, 15499; iv, 15458; v, 15457; vi, 19134; vii, 19131; and viii, 19132.

Synthetic rho-ruv lethality was suppressed by the rus-1 mutation (Fig. S2 ii), which activates expression of the Rus resolvase from a cryptic prophage and thereby alleviates several ruv phenotypes (14, 15). The lethality was likewise suppressed by the rpoB*35 mutation (54–56) (Fig. S2 iii), as also by ectopic expression of the (phage T4-encoded) R-loop helicase UvsW but not its active site mutant version UvsW-K141R (57, 58) (Fig. S2 iv–v). Both rpoB*35 and UvsW alleviate the deleterious effects of Rho deficiency and can rescue Δrho lethality (28, 29, 33).

Loss of RecA, or of RecFOR pathway components, suppresses rho-ruv lethality.

Given the observations that the rho-136^opal mutation was lethal with ruv but not with recA, the possibility was considered that excessive (and unnecessary) HR triggered in the presence of the rho mutation was leading to accumulation of HJ intermediates in the absence of RuvABC, resulting in cell death. Indeed, we could show that rho-ruv lethality is also rescued by ΔrecA (Fig. 1A v, and Fig. S2 vii). An alternative explanation, that death was on account of excessive RecA-dependent SOS induction in rho-ruv mutants, was excluded since lethality was not rescued by the lexA3 mutation (which encodes a LexA variant that is noncleavable by RecA, and hence also abolishes SOS induction [13]) (Fig. S2 viii).

We then tested which of the two pre-synaptic pathways putatively feeds into the process of excessive HR in rho-ruv mutants. The synthetic lethality was suppressed by mutations in recO or recR (of the RecFOR pathway) (Fig. 1A vi–vii), but not by mutation in recB (of the RecBC pathway) (Fig. 1A viii). RecQ helicase activity is implicated in RecFOR pathway function (3, 7), and ΔrecQ also was a suppressor of rho-ruv lethality (Fig. 1A ix).

Suppression of rho-ruv lethality by loss of IF2-1.

A new suppressor of rho-ruv lethality on both defined and rich media (Fig. 1A x, and Fig. S2 vi) was obtained and characterized. It was mapped to the infB-nusA locus, and was shown by DNA sequencing to be an ochre (TAA) nonsense codon mutation at position 161 of the infB ORF. One would expect, from its location in the infB ORF, that this mutation abrogates synthesis of the IF2-1 and IF2-2 isoforms, which was confirmed by Western blotting (Fig. S3C, lane 6).

By itself, the infB-161^ochre mutation was lethal (Fig. S3B i), but it could be rescued by rho mutation (Fig. S3B ii). We interpret these findings as indicative of the notions (i) that nonsense substitution at codon 161 in infB induces Rho-mediated premature transcription termination within the gene, and therefore (ii) that synthesis of the lone IF2-3 isoform to ensure viability is itself contingent on relief of transcriptional polarity conferred by rho mutation.

Thus, viability of a triple mutant rho ruv infB-161^ochre is based on mutual suppression (i) of infB-161^ochre lethality by rho, and (ii) of rho ruv lethality by infB-161^ochre. This inference was supported by findings from experiments in which expression of an ectopically located rho⁺ gene was placed under the control of an L-arabinose (Ara)-inducible promoter (48) (Fig. 1B): that a rho-ruv derivative (row 3) is viable only on Ara-supplemented medium, whereas an infB-161^ochre-rho mutant (row 2) as well as the triple mutant infB-161^ochre-rho-ruv (row 4) are viable only on medium not supplemented with Ara. The infB-rho mutant was inhibited for growth at 22° (Fig. S3D), consistent with the known requirement for IF2 at low temperatures (59).

Suppression of rho-ruv lethality by ectopic infB constructs.

Based on the suppressor characterization results above, we surmised that it is the loss of isoforms IF2-1 or IF2-2 (or both together) that confers suppression of rho-ruv lethality. Accordingly, we then tested two other sets of ectopic chromosomal infB constructs for their ability to rescue rho-ruv lethality, in derivatives carrying a deletion of the native infB locus. (In the descriptions below, the designations infB⁺ and ΔinfB are used for the wild-type and deletion alleles, respectively, at the native chromosomal location.)

In one set of three ectopic constructs, which has been described earlier by Nakai and coworkers (38, 39), IF2 expression remains under the control of the natural cis regulatory elements for infB, but not all isoforms are expressed from the different constructs (designation used in this study for each of them given in parentheses): that encodes only IF2-1, but not IF2-2 or IF2-3 (ΔIF2-2,3); that encodes both IF2-2 and IF2-3, but not IF2-1 (ΔIF2-1); and that encodes all three isoforms [IF2(wt)]. The constructs were validated for expression of the different IF2 isoforms by Western blotting analysis (Fig. S3C, lanes 2 to 5 and lane 9). (Earlier studies [41, 46] have established that the isoforms IF2-1 and IF2-2,3 are present in nearly equimolar amounts, and hence the finding from our Western blot data of a higher band intensity for the former may perhaps reflect preferential recognition of epitopes from the protein's N-terminal domain by the anti-IF2 antibody preparation used in this work.) Upon testing in the rho-ruv strains, our results showed that the ΔIF2-1 construct, but not IF2(wt) or ΔIF2-2,3, could suppress rho-ruv synthetic lethality (Fig. 2 and Fig. S4, compare in each case iii with i and ii). These findings are consistent with that of rho-ruv suppression by infB-161^ochre, since the latter also fails to express isoform IF2-1.

The second set of ectopic chromosomal constructs were prepared in this work and were designed for differential expression of IF2 isoforms from an isopropyl-β-d-thiogalactoside (IPTG)-inducible heterologous P_trc promoter (designation used for each given in parentheses): that expresses both IF2-2 and IF2-3, but not IF2-1 (P_trc::IF2-2,3); and that expresses IF2-3 alone (P_trc::IF2-3) (see Fig. S3C, lanes 8 and 7, respectively, for Western blot confirmation). Both constructs could suppress rho-ruv lethality (Fig. 2 iv and Fig. S4 iv–v), indicating once again that lethality suppression occurs when isoform IF2-1 is absent from the cells. Similar results were obtained irrespective of which ruv mutation (ΔruvA, ΔruvC, or ΔruvABC) was used in the experiments (Fig. 2 iv; and Fig. S4iv-v). With induced IF2-3 expression from the IPTG-regulated construct, suppression was obtained even in derivatives carrying the native infB⁺ locus, but the colony growth was less robust (Fig. 2 v). These results indicate that it is an imbalance between levels of isoforms IF2-1 (low) and IF2-2,3 (high) that determines suppression of rho-ruv lethality.

That this suppression is not because of a restoration (or bypass) of Rho or Ruv function in these strains was demonstrated as follows. Neither loss of IF2-1 nor overexpression of IF2-2,3 reversed the Gal⁺ phenotype that is associated with relief of premature transcription termination by rho mutation in a galEp3 strain (34) (Fig. S1C, note similarity across subpanels ii–iv). Furthermore, lethality caused by runaway replication of plasmid pACYC184 in rho mutants (34) was not suppressed by overexpression of IF2-2,3 (Fig. S1A, subpanels iii–v). Finally, the UV-sensitivity known for ruv mutants (3, 60) was not suppressed (and indeed was aggravated) upon loss of IF2-1 (Fig. S5A).

Loss of IF2-2,3 exacerbates rho-ruv defect.

The following experiments lent additional support to the notion that rho-ruv lethality is modulated by the balance between isoforms IF2-1 and IF2-2,3. Derivatives with the missense rho-4 mutation (encoding Rho-A243E [34]; see Fig. S1B lane 2 for Western blot) may be expected to be less compromised for Rho function than those with rho-136^opal, and a test for intracellular R-loop prevalence with monoclonal antibody S9.6 (29, 61) showed this to be so (Fig. S1D). Unlike rho-136^opal, the rho-4 mutation was not synthetically lethal with ruv in the Nakai IF2(wt) strain expressing all three IF2 isoforms (Fig. 2 vi). However, this combination (rho-4 ruv) conferred lethality in the ΔIF2-2,3 derivative (lacking IF2-2,3) (Fig. 2 vii), and this lethality was rescued by IPTG-induced expression of IF2-2,3 (Fig. 2 viii). Thus, absence of IF2-1 and of IF2-2,3 exerts the apparently opposite phenotypes of, respectively, alleviating and exacerbating the sickness of rho-ruv mutants.

Synthetic lethality of uvrD-ruv is also suppressed by loss of IF2-1.

Several groups have earlier reported synthetic uvrD-ruv lethality (25–27), which is suppressed by recA and recFORQ but not recBC (that is, very similar to our findings with rho-ruv lethality). The mechanism invoked has been that of excessive and unnecessary HR in the uvrD mutant, leading to accumulation of toxic recombination intermediates in the absence of RuvABC.

To test whether IF2 isoforms affect uvrD-ruv lethality, we expressed (from a doxycycline [Dox]-inducible promoter) a dominant negative RuvC protein designated as RDG that binds and traps HJs (60); these experiments were done in a set of ΔuvrD ΔinfB derivatives each carrying one of the ectopically integrated Nakai constructs for different IF2 isoforms. Viability of the IF2(wt) and ΔIF2-2,3 strain derivatives was reduced 10³- and 10⁴-fold, respectively, upon Dox addition, whereas the ΔIF2-1 derivative was only minimally affected (Fig. 3). A recA derivative of the IF2(wt) strain survived Dox, as too did the control uvrD⁺ recA⁺ strain (Fig. 3, last and first rows, respectively). These results confirm that uvrD-ruv is synthetically lethal (more severely so in the absence of IF2-2,3), and that this lethality is rescued upon loss of RecA or of IF2-1.

FIG 3.

Synthetic lethality conferred by dominant negative ruvC mutation (RDG) in ΔuvrD mutant, and its suppression by loss of IF2-1. Dilution-spotting assay was performed on LB medium without (−) and with (+) Dox supplementation of RDG-bearing derivatives whose relevant genotypes/features are indicated at left; strains on all but the top two rows were also ΔinfB. Strains employed for different rows were (from top; all strain numbers are prefixed with GJ) 19127, 19161, 19801, 19802, 19803, 19843, and 19835.

HR frequency is oppositely affected by loss of IF2-1 and of IF2-2,3, and is elevated in rho mutants.

The results above had indicated that loss of IF2-1 phenocopies the loss of RecA or of the presynaptic RecFORQ pathway to confer suppression of rho-ruv and uvrD-ruv lethalities. We then tested the differential effects, if any, of IF2 isoforms on recovery of recombinants following HR, for which we employed several assays such as those of phage P1 transduction, interplasmidic recombination (that leads to reconstitution of a tetracycline (Tet)-resistance gene from two partially overlapping deletion alleles) (62–65), and the Konrad assay (that similarly entails reconstitution of an intact lacZ gene from a split pair of partially overlapping lacZ fragments located at distant sites on the chromosome [66, 67]). Recombination events in each of the assays above are RecA-dependent (3, 62, 66); interplasmidic recombination is mediated by the RecFOR presynaptic pathway (62, 64, 65), whereas P1 transduction and recombination in the Konrad assay are RecBCD-dependent (66). The frequency distribution of recombinants in the Konrad or interplasmid recombination assays across replicate cultures exhibits large variations (analogous to that of fluctuation tests [68] for mutation frequency determinations), whereas P1 transduction frequencies across replicate experiments exhibit a Gaussian distribution.

Across the different HR assays, loss of IF2-1 was associated with 30% to 70% reduction and that of IF2-2,3 with 110% to 225% elevation in recovery of recombinants, in comparison with values for the control strain expressing all three IF2 isoforms (Fig. 4A to C). The difference between parent and ΔIF2-1 was statistically significant in the P1 transduction assays (P = 0.007), and borderline so in the Konrad assay (P = 0.11). The difference between parent and ΔIF2-2,3 in the Konrad assay was also statistically significant (P = 0.01). A moderate reduction in P1 transduction frequency for the strain lacking IF2-1 has been reported earlier (38).

FIG 4.

Effect of IF2 isoforms on HR. Recombination frequency data are given for strains with the indicated genotypes/features in the Konrad (A), inter-plasmid recombination (B), and P1 transduction (C) assays, after normalization to the value for the cognate control strain (taken as 1, and shown at extreme left for each panel); the actual control strain values were 2.8*10⁻⁷/viable cell, 6*10⁻⁴/viable cell, and 1.5*10⁻⁵/phage, respectively. In panels A and B, values from all individual experiments are shown, and median values are given beside the denoted horizontal lines. Panel C depicts the mean value (given alongside each bar) and standard error for each strain. Horizontal lines at bottom of the panels depict the pairwise statistical comparisons that were performed between strains, and the corresponding P value is given beneath each line. In all panels, strains whose designations include IF2(wt), ΔIF2-1, or ΔIF2-2,3 were also ΔinfB. Strains used were (from left, all strain numbers mentioned are prefixed with GJ unless otherwise indicated) as follows: in panel A, SK707, 19171, 19186, 19162, 19184, and 19165; in panel B, 19197, 19196, 19195, and 19847; and in panel C, 19193, 19194, and 15494.

Consistent with earlier published data (3, 62, 66), the frequency of recovery of recombinants in recA derivatives for each of the three assays was < 0.01 of that in the infB⁺ or IF2(wt) strains (hence are not depicted in the panels of Fig. 4).

We also tested the effects of UvrD deficiency, alone or in combination with loss of IF2 isoforms, in the Konrad and inter-plasmid recombination assays. As expected (23, 66, 67), loss of UvrD conferred a hyperrecombination phenotype, with 2.3- to 4.5-fold increase of recombination frequency in each of the two assays, which was statistically significant (Fig. 4A and B; P = 0.006 and 0.02, respectively). On the other hand, HR frequency in a derivative that had lost both UvrD and IF2-1 was once again low and resembled that in a strain lacking IF2-1 alone (Fig. 4A, compare difference, statistically significant, between the values of columns 4 and 6; P = 0.007). Thus, loss of IF2-1 is epistatic to ΔuvrD.

Since the results above suggested that loss of IF2-1 is apparently associated with reduced RecA function in HR (see Discussion), we examined whether there is reduction in SOS induction (which is mediated by RecA’s coprotease activity, which is activated upon binding to ss-DNA [13]) as well in this situation. The results indicate that, as measured by sulA-lac expression (69), the SOS response (both basal as well as that following DNA damage with phleomycin) is not decreased and may in fact be modestly elevated in the ΔIF2-1 strain (Fig. S5B).

Our finding of rho-ruv synthetic lethality that is suppressed by recA also suggests, based on its parallels with uvrD-ruv lethality, that nonessential HR occurs at elevated frequency in the rho-136^opal mutant. Measurements of HR frequency, both by the Konrad assay and by conjugation, indicate that the rho mutant does exhibit a moderate increase in HR frequency (Fig. S1E).

DISCUSSION

The major findings of this study are (i) that Rho deficiency is synthetically lethal with ruv mutations in a manner that is similar to uvrD-ruv lethality (with trapped HJs, generated through the RecFORQ presynaptic pathway and RecA, being responsible for inviability in both cases); and (ii) that an imbalance of isoforms of the translation initiation factor IF2 tends to affect HR functions in the cells. Each of these is further discussed below.

Why is rho-ruv lethal?

Several features are shared between synthetic lethalities rho-ruv and uvrD-ruv, suggesting commonality of mechanisms in the two instances. Thus, both lethalities require proteins RecA and RecFORQ, and translation initiation factor isoform IF2-1.

In the case of uvrD-ruv, the model proposed by Rosenberg and colleagues (25) is that in the absence of UvrD, there is excessive but unnecessary HR through the RecFORQ pathway, which then renders RuvABC essential for resolving the ensuing HJ intermediates. Likewise for the rho mutant, we suggest that on account of an increased prevalence of R-loops with displaced ss-DNA, increased HR is triggered through the RecFOR pathway, thus necessitating RuvABC’s presence for viability (Fig. 5, panels a–c). In this model, UvsW expression and rpoB*35 are suppressors of rho-ruv lethality because they presumably act to reduce R-loop prevalence in Rho-deficient strains (28, 29, 33).

FIG 5.

Model to explain HJ formation (and hence need for RuvABC) in rho mutants. (a–b) Rho deficiency provokes R-loop generation from nascent untranslated transcripts, associated with RNA polymerase (RNAP) backtracking and arrest. (c) HR is initiated by invasion of ss-DNA of R-loop into a sister chromosome, mediated by RecFOR and RecA.

Opposing effects of IF2-1 and IF2-2,3 isoforms in HR pathways.

Early studies had established that isoforms IF2-1 and IF2-2,3 are together required for optimal growth at temperatures below 37° (46). Nakai and coworkers (38, 39) have reported that loss of IF2-1 or of IF2-2,3 is each associated with sensitivity to different kinds of DNA damage.

In our study, loss of IF2-1 or of IF2-2,3 was associated with suppression or aggravation, respectively, of rho-ruv and uvrD-ruv sickness. The model for opposing effects of these isoforms is supported also by the trends of our findings in HR assays. At a mechanistic level, however, it is unclear whether a particular phenotype is caused by absence of one IF2 isoform or exclusive presence of another.

How do IF2 isoforms influence HR?

The studies reported in this and in the companion article (47), taken together, have identified IF2 isoforms as novel players in HR, but the mechanisms by which they act for this purpose are unknown. The present study has shown that loss of IF2-1 (i) phenocopies loss of the RecFORQ presynaptic pathway and of RecA in suppressing rho-ruv and uvrD-ruv lethalities, and (ii) tends to reduce the recovery of recombinants following HR. In the companion article (47), it is shown that loss of IF2-1 confers profound sensitivity to two-ended DSBs, whose repair is mediated by the RecBCD presynaptic pathway along with RecA.

The model we propose is that absence of IF2-1 leads directly or indirectly to reduction in efficiency of a step in HR that is (i) downstream of (and common to) the RecBCD and RecFOR pathways, and (ii) upstream of the postsynaptic reactions mediated by RuvABC. Formation of the RecA nucleoprotein itself is apparently unaffected, since the SOS response (13) is not perturbed by loss of IF2-1. In a further elaboration of this model in the companion article (47), it is the strength of RecA-mediated synapsis between a pair of homologous DNA molecules that is postulated to be decreased in the absence of IF2-1.

The proposed model is consistent with our finding that IF2-1's role in HR is epistatic to that of UvrD. Deficiency of IF2-2,3 is proposed to have the opposite effect, of enhancing the efficiency of the same step in HR as that diminished by loss of IF2-1.

Our model that loss of IF2-1 compromises RecA function is different from that of Nakai and colleagues (38, 39), who proposed that IF2 isoforms differentially influence different replication restart pathways (which are postsynaptic, and downstream of RuvABC action). As explained above, our finding that loss of IF2-1 phenocopies the recA mutation in suppressing rho-ruv and uvrD-ruv lethalities can best be explained only by invoking a role for IF2-1 prior to the step of RuvABC action.

To test the models above, in vitro studies may be needed to examine whether IF2 isoforms act directly to modulate HR, and to determine their precise role(s) in the process. Regulation of HR and of recombinational repair functions is important in both prokaryotes and eukaryotes (11, 70), and factors previously identified for such regulation in E. coli include UvrD, mismatch repair proteins, DinI, and RecX (3, 22, 23, 71, 72).

MATERIALS AND METHODS

Growth media, bacterial strains, and plasmids.

The routine rich and defined growth media were, respectively, LB and minimal A with 0.2% glucose (73), and unless otherwise indicated, the growth temperature was 37°. Supplementation with Xgal and with antibiotics ampicillin (Amp), chloramphenicol (Cm), kanamycin (Kan), spectinomycin (Sp), Tet, and Tp were at the concentrations described earlier (48). Phleomycin supplementation was at 3 μg/mL. For induction of gene expression from the appropriate regulated promoters, Ara, Dox, and IPTG were added at 0.2%, 50 ng/mL, and 0.5 mM, respectively. E. coli strains used are listed in Table S1, with the following knockout (Kan^R insertion-deletion) alleles sourced from the collection of Baba et al. (74): dinF, ilvA, leuA, leuD, racC, recA, recB, recO, recQ, recR, ruvA, serA, thrA, uvrD, ybfP, and yihF; the ΔinfB knockout mutation has also been described earlier (42).

Plasmids described earlier include pBR322 (Tet^R Amp^R, ColE1 replicon) (75); pACYC184 (Tet^R Cm^R, p15A replicon) (76); pCL1920 (Sp^R, pSC101 replicon) (77); pMU575 (Tp^R, single-copy-number vector with lacZ⁺) (78); pHYD2411 (Tp^R, pMU575 derivative with rho⁺) (28); and pTrc99A (Amp^R, for IPTG-inducible expression of gene of interest) (79). Plasmids pKD13 (Kan^R Amp^R), pKD46 (Amp^R), and pCP20 (Cm^R Amp^R), for use in recombineering experiments and for Flp-mediated site-specific excision of FRT-flanked DNA segments, have been described by Datsenko and Wanner (80). Plasmids constructed in this study are described in the supplemental material.

Methods.

Procedures for P1 transduction (81), recombineering on the chromosome or plasmids (80), determination of UV tolerance (73), and R-loop detection with S9.6 monoclonal antibody (29) were as described. The Miller protocol was followed for β-galactosidase assays (73), and enzyme specific activity values are reported in the units defined therein. Protocols of Sambrook and Russell (82) were followed for recombinant DNA manipulations, PCR, and transformation. The Western blotting procedure, with rabbit polyclonal anti-IF2 or anti-Rho antisera (kind gifts from Umesh Varshney and Ranjan Sen, respectively), was essentially as described (50). Chromosomal integration, at the phage λatt site, of pTrc99A derivatives expressing IF2-2,3 or IF2-3, was achieved by the method of Boyd et al. (83). HR assay methods are described in the supplemental material, and include those based on conjugation (73), the Konrad assay (66), and interplasmid recombination (62, 63).