Sgs1 and Mph1 Helicases Enforce the Recombination Execution Checkpoint During DNA Double-Strand Break Repair in Saccharomyces cerevisiae

Abstract

We have previously shown that a recombination execution checkpoint (REC) regulates the choice of the homologous recombination pathway used to repair a given DNA double-strand break (DSB) based on the homology status of the DSB ends. If the two DSB ends are synapsed with closely-positioned and correctly-oriented homologous donors, repair proceeds rapidly by the gene conversion (GC) pathway. If, however, homology to only one of the ends is present, or if homologies to the two ends are situated far away from each other or in the wrong orientation, REC blocks the rapid initiation of new DNA synthesis from the synapsed end(s) and repair is carried out by the break-induced replication (BIR) machinery after a long pause. Here we report that the simultaneous deletion of two 3′→5′ helicases, Sgs1 and Mph1, largely abolishes the REC-mediated lag normally observed during the repair of large gaps and BIR substrates, which now get repaired nearly as rapidly and efficiently as GC substrates. Deletion of SGS1 and MPH1 also produces a nearly additive increase in the efficiency of both BIR and long gap repair; this increase is epistatic to that seen upon Rad51 overexpression. However, Rad51 overexpression fails to mimic the acceleration in repair kinetics that is produced by sgs1Δ mph1Δ double deletion.

DNA double-strand breaks (DSBs) are potentially lethal lesions that can be repaired either by nonhomologous end joining (NHEJ) or by homologous recombination (HR) (Krogh and Symington 2004; Haber 2008; Haber 2013). While NHEJ involves simple religation of the DSB ends with little or no homology, HR requires the presence of intact homologous sequences to serve as a template for repair. During HR-mediated repair, the DSB ends are resected to produce 3′-ended single-stranded DNA tails, which get coated with the Rad51 recombinase protein to form Rad51 nucleoprotein filaments. These Rad51 filaments then search for and strand invade homologous sequences to form a three-strand displacement loop (D-loop), which is followed by extension of the invading strand by new DNA synthesis using the paired homologous sequence as a template. When both DSB ends share homology with a sister chromatid, a homologous chromosome or an ectopically placed donor; repair occurs by gene conversion (GC), resulting in the synthesis of a short patch of new DNA at the recipient locus using the homologous donor as a template. New DNA synthesis is initiated within 30 min after strand invasion (White and Haber 1990; Sugawara et al. 2003; Hicks et al. 2011) and does not require components of the lagging strand DNA-synthesis machinery such as Polα and primase (Wang et al. 2004). In mitotic cells of budding yeast, GC proceeds primarily by the synthesis-dependent strand annealing (SDSA) mechanism, in which the newly-synthesized strands dissociate from the donor template and are returned to the recipient locus. GC is thus distinct from normal semiconservative replication in that all the newly-copied DNA is found in the recipient (Ira et al. 2006).

However, if there is homology to only one DSB end, repair occurs by another HR pathway called break-induced replication (BIR) (McEachern and Haber 2006; Llorente et al. 2008; Malkova and Ira 2013). Here, strand invasion by the homologous end results in the establishment of a migrating D-loop that can copy all sequences distal to the site of homology, resulting in a nonreciprocal translocation (Donnianni and Symington 2013; Saini et al. 2013; Wilson et al. 2013). Sequences on the other side of the break, lacking homology, are lost by degradation.

Compared to GC, BIR is less efficient; it is kinetically slower and new DNA synthesis does not initiate until ∼3–4 hr after strand invasion (Malkova et al. 2005; Lydeard et al. 2007; Jain et al. 2009). BIR requires leading and lagging strand DNA synthesis and essentially all of the DNA replication factors including Polα/primase, Cdc7, Cdt1, and the Cdc45-MCM-GINS helicase complex (Lydeard et al. 2007; Lydeard et al. 2010b); only the components specifically needed for origin-dependent DNA replication (Cdc6 and the ORC proteins) are dispensable for BIR. Additionally, while Polδ and Polε are redundantly required for GC (Holmes and Haber 1999; Wang et al. 2004), Polδ alone is required for the initiation of BIR whereas Polε becomes important for the completion of repair (Lydeard et al. 2007).

In Saccharomyces cerevisiae, when both ends of a DSB share homology to the donor, GC nearly always outcompetes BIR (Malkova et al. 2005). Although GC is the more commonly-used mode of DSB repair, BIR is required for the maintenance of telomeres in telomerase-deficient cells (Lundblad and Blackburn 1993; Le et al. 1999; Teng et al. 2000; McEachern and Haber 2006; Lydeard et al. 2007). A single homologous end can also arise when a DSB occurs in the vicinity of a dispersed, repeated sequence such as a transposable element that is present at other chromosome locations (Malkova et al. 2001; VanHulle et al. 2007).

Correct channeling of the DSBs is crucial to ensure that repair occurs with high fidelity. Independent initiation of BIR events from DSB ends that could repair by GC will produce nonreciprocal translocations if the ends are repaired by using two nonallelic, repeated donor sequences; or cause aneuploidy if one end is repaired using a sister chromatid while the other end used a homologous chromosome for repair. We previously showed evidence for the existence of a recombination execution checkpoint (REC) that acts prior to the actual initiation of the repair synthesis, apparently by sensing the presence and topology of the engaged ends (Jain et al. 2009). When DSB ends are engaged with closely-positioned and correctly-oriented homologous donors, a break is rapidly repaired by GC. This topological requirement to allow rapid repair is fulfilled even if the two ends are derived from two separate DSBs (Jain et al. 2016). However, if homology to only one end is present, or if the two ends engage far from each other, or if the homologies are arranged close together but in the wrong orientation; REC blocks the rapid initiation of repair by GC and repair occurs by BIR after a long pause (Jain et al. 2009). The delayed initiation of BIR may provide more time for the DSB ends to dissociate from their initial site of synapsis and continue searching for homology in the vicinity of each other, so that repair can proceed via the more conservative GC pathway. We have previously estimated that even GC repair involves, on average, four encounters between the broken end and the donor before the synapsis is irreversibly converted to a repair event (Coïc et al. 2011).

One indication of how REC monitors the location and orientation of the DSB ends came from the observation that Sgs1, a 3′→5′ helicase of the evolutionarily-conserved RecQ family, modulates the REC-mediated switch from GC to BIR (Jain et al. 2009). Whereas a 1.2-kb gap is rapidly repaired by GC, a 5-kb gap is repaired more slowly, apparently by Pol32-dependent BIR. However, deletion of Sgs1 accelerates the repair kinetics of the 5-kb gap to resemble repair of the 1.2-kb gap. This observation led us to postulate that capture of both DSB ends within a continuous D-loop may be required to signal repair by GC, and that Sgs1 might limit the enlargement of the D-loops formed upon strand invasion by the DSB ends, thereby preventing the GC machinery from repairing a 5-kb gap. However, sgs1Δ did not alter the repair kinetics of much larger gaps, e.g., one of 18 kb, although it increased the efficiency of BIR (Jain et al. 2009; Lydeard et al. 2010a).

Here we report that Mph1, another well-conserved 3′→5′ helicase, also regulates the REC-mediated choice between GC and BIR. Moreover, double deletion of SGS1 and MPH1 virtually eliminates the REC, resulting in a mechanistic shift in the repair of long gaps and BIR substrates, and much faster initiation of new DNA synthesis.

Materials and Methods

Strains and plasmids

The construction of the break-repair strain, YSJ119; gap-repair strains, YSJ130 (1.2-kb gap), YSJ133 (5-kb gap), YSJ135 (18-kb gap); and the BIR strain, YSJ131; and their sgs1∆ derivatives (YSJ190, YSJ191, YSJ192, YSJ230, and YSJ193, respectively) is described in (Jain et al. 2009). The mph1Δ and mph1Δ sgs1Δ strains were made by the standard PCR-based gene disruption method to obtain strains YSJ204 (YSJ119 mph1Δ), YSJ205 (YSJ130 mph1Δ), YSJ206 (YSJ133 mph1Δ), YSJ207 (YSJ131 mph1Δ), YSJ211 (YSJ135 mph1Δ), YSJ244 (YSJ191 mph1Δ), YSJ245 (YSJ193 mph1Δ), YSJ281 (YSJ192 mph1Δ), YSJ417 (YSJ190 mph1Δ), and YSJ418 (YSJ230 mph1Δ). Wild type (WT) and mutant strains were transformed with the Rad51 overexpressing plasmid pSJ5 (Lydeard et al. 2010a) to obtain strains YSJ178 (YSJ119 pSJ5), YSJ179 (YSJ130 pSJ5), YSJ180 (YSJ131 pSJ5), YSJ181 (YSJ133 pSJ5), YSJ182 (YSJ135 pSJ5), YSJ279 (YSJ245 pSJ5), YSJ282 (YSJ207 pSJ5), and YSJ283 (YSJ193 pSJ5). WT and mutant strains were transformed with the Pol30-89 expressing plasmid pTR1 [which was obtained by moving the pol30-89 allele from pBL248-89 (Eissenberg et al. 1997) into the pRS314 vector (Sikorski and Hieter 1989)], or just the empty pRS314 vector alone to obtain strains YAM007 (YSJ131 pTR1), YAM008 (YSJ207 pTR1), YAM009 (YSJ193 pTR1), YAM010 (YSJ245 pTR1), YAM011 (YSJ135 pTR1), YAM012 (YSJ418 pTR1), YAM013 (YSJ119 pTR1), YAM014 (YSJ417 pTR1), YAM015 (YSJ131 pRS314), YAM016 (YSJ207 pRS314), YAM017 (YSJ193 pRS314), YAM018 (YSJ245 pRS314), YAM019 (YSJ135 pRS314), YAM020 (YSJ418 pRS314), YAM021 (YSJ119 pRS314), and YAM022 (YSJ417 pRS314). Strains and plasmids are available upon request.

Viability measurements

Yeast cells were grown in YEP containing 2% raffinose to a density of ∼1 × 10⁷ cells/ml. Equal volumes of appropriate dilutions were plated on YEP containing 2% galactose (YEPGal; to induce the HO break) and YEP containing 2% dextrose (YEPD; no DSB control). Viability was determined from the ratio of CFUs able to survive the break (number of colonies that grew on YEPGal) to the total number of CFUs plated (number of colonies that appeared on YEPD). For the Pol30-89 experiments (Figure 5), Trp drop-out medium was used instead of YEP to ensure plasmid retention. Data represent mean ± SD (n ≥ 5), and P-values were calculated using the two-tailed, unpaired Student’s t-test.

Figure 5.

Repair of long gaps and BIR substrates becomes largely pol30-89-independent when both Sgs1 and Mph1 are absent. (A–C) Viabilities of the indicated strains. Data represent mean ± SD (n ≥ 3). Asterisks denote that the viability of the strain carrying the pol30-89 allele is significantly different from its empty vector control. * indicates P < 0.05 and ** denotes P < 0.005. Ratio of the viability obtained upon transformation with the empty vector to vector harboring the pol30-89 allele is indicated at the bottom for each strain.

HO induction for kinetic analysis of DSB repair

Yeast cells were grown in YEP containing 2% raffinose to a density of ∼1 × 10⁷ cells/ml and HO endonuclease was induced by adding galactose to a final concentration of 2%. Samples were collected for DNA analysis just prior to and at different time points following addition of galactose, as described before (Holmes and Haber 1999).

PCR-based primer extension assay

Equal amounts of genomic DNA isolated from samples collected at different time points were amplified within linear range as described before (White and Haber 1990). A pair of primers 500-bp upstream of the U2 donor and 800-bp downstream of leu2::HOcs was used to analyze the repair kinetics. The PCR reactions were run on agarose gels and the repair product was quantified using Bio-Rad (Hercules, CA) Quantity One software. For the break-repair strain, YSJ119, the primer pair used to study U2 repair gives a band at 0 hr time points as well; however, that band is ∼120-bp bigger than the one obtained after repair [owing to the presence of the HO cleavage site (HOcs)], and was not taken into account in our analysis. PCR signal from an independent locus (LDB16) was used to normalize for input DNA. The ratio of test and reference signals obtained from the DNA of a repaired colony was set to 100%.

Colony PCR

A small quantity of cells taken from colonies growing on YEPGal (from the viability assay described above) were heated at 100° for 10 min and the lysates were amplified with primers upstream of the LE donor to assay for the loss of the distal end of chromosome (Chr) 5.

Data availability

The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article.

Results

Mph1 exerts an effect opposite to that of Sgs1 during short gap repair

We had previously constructed a series of haploid yeast strains to study competition between GC and BIR. These strains harbor a HOcs within a LEU2 gene (leu2::HOcs) inserted at the can1 locus on Chr 5. Homologies to the LE and U2 ends of the HO break (∼700-bp 5′ and ∼1-kb 3′ segments, respectively, of the LEU2 gene) are present on the left arm of Chr 3. The strains differ from each other by the distance separating the LE and U2 donors on Chr 3 (Jain et al. 2009; Figure 1A). A strain carrying the entire LEU2 gene as donor serves as the break-repair (GC) control (YSJ119), and a strain carrying only the U2 donor serves as the BIR control (YSJ131). In the gap-repair strains, repair can occur either by GC (resulting in copying of all the Chr 3 sequences between the LE and U2 donors on to Chr 5) or by BIR (leading to duplication of Chr 3 sequences distal to the U2 donor, and a concomitant loss of the nonessential distal left end of Chr 5). We showed previously that increasing the gap between the donors from 1.2 kb (YSJ130) to 5 kb (YSJ133) or 18 kb (YJS135) caused the ends to lose sight of each other resulting in a shift from the quick and efficient GC to the slower and less efficient Pol32-dependent BIR pathway of repair (Figure 1; data from Jain et al. 2009).

Figure 1.

Mph1 modulates the efficiency and kinetics of gap repair. (A) Schematic representation of the strains used to study competition between GC and BIR (Jain et al. 2009). leu2::HOcs is inserted at the can1 locus on Chr 5. Homology to right end of the break, the U2 end, is present upstream of the SPS22 locus ∼41 kb from the left end of Chr 3. Homology to left end of the break, the LE end, is present at indicated distances from the U2 donor. The endogenous MAT, HML, HMR, and LEU2 sequences are deleted. (B) Viability of the indicated WT (black) and mph1∆ (red) strains. Data represent mean ± SD (n ≥ 5). * denotes that viabilities of WT and mph1Δ strains are significantly different (P ≤ 0.005). (C and D) Kinetics of repair measured from the U2 end in the indicated WT and mph1∆ strains as determined by a quantitative PCR-based primer extension assay using primers shown schematically in (A). The amount of PCR product obtained from a repaired colony was set to 100%. Data represent mean of three PCR reactions from each of two or three independent time courses ± SD. The WT data has been published in Jain et al. (2009).

As reported previously, deleting Sgs1 increased the efficiency of BIR and accelerated the kinetics of 5-kb gap repair (Figure 2, A and D; data from Jain et al. 2009). We have now tested the Mph1 helicase for a possible role in modulating gap repair. Mph1 has been implicated in promoting HR-mediated, error-free bypass of DNA lesions (Scheller et al. 2000; Schürer et al. 2004; Panico et al. 2009). Mph1, as well as its Schizosaccharomyces pombe ortholog Fml1, have been shown to restrict crossover formation during GC by channeling repair into the SDSA pathway (Prakash et al. 2009; Lorenz et al. 2012). It has also been implicated in the generation of Rad51-dependent regressed fork intermediates (Chen et al. 2009; Mankouri et al. 2009). Its human ortholog FANCM has been shown to promote branch migration of Holliday junctions and replication forks in vitro (Gari et al. 2008). However, Mph1 suppresses BIR (Luke-Glaser and Luke 2012; Stafa et al. 2014) and its overexpression increases gross chromosomal rearrangements, apparently by inhibiting HR (Banerjee et al. 2008). Mph1 has also been shown to play a role in processing Okazaki fragments during replication (Kang et al. 2009). Consistent with previous studies examining the role of Mph1 in ectopic GC (Prakash et al. 2009), deleting Mph1 did not affect the efficiency of break repair (when the donors are not separated by a gap) (Figure 1B). However, it significantly reduced the efficiency of 1.2-kb gap repair from ∼56 to ∼38% (Figure 1B). Intriguingly, mph1Δ did not affect the repair efficiency of a 5-kb gap but increased the efficiency of repairing an 18-kb gap, as well as a BIR substrate. Deletion of MPH1 resulted in a small shift in the kinetics of break repair, and it significantly slowed down the initiation of new DNA synthesis during 1.2-kb gap repair, as determined by a PCR-based primer extension assay (Figure 1C). However, the kinetics of 5-kb and 18-kb gap repair as well as BIR remained largely unaltered (Figure 1D). Although we observed an apparent small shift in the kinetics of 18-kb gap repair and BIR in the absence of Mph1, this difference disappeared when the primer extension data were normalized to the amount of product obtained 15 hr after HO induction (Figure 2F). Therefore, the apparent shift in the kinetics of 18-kb gap repair (as well as BIR) could be attributed solely to the increased efficiency of repair in an mph1Δ background.

Figure 2.

Deletion of Mph1 and Sgs1 eliminates the REC-mediated delay in the initiation of long gap repair. (A) Viability of the indicated strains. Data represent mean ± SD (n ≥ 5). * denotes that viabilities of the mph1Δ sgs1Δ double-mutant strains are significantly different from the mph1Δ single mutants (P ≤ 0.005). (B–E) Repair kinetics of the indicated break-repair, 1.2-kb, 5-kb, and 18-kb gap-repair strains, respectively, as determined by a quantitative PCR-based primer extension assay using primers shown schematically in Figure 1A. The amount of PCR product obtained from a repaired colony was set to 100%. (F) Data from (E) plotted after normalizing the amount of product obtained at 15 hr to 100% for each strain. In all cases, the kinetic data represent mean of three PCR reactions from each of two or three independent time courses ± SD. Data from WT and sgs1Δ single mutants have been published in Jain et al. (2009).

As postulated above, if coordination between the DSB ends occurs by strand invasion/annealing within the same D-loop, then these data suggest that Mph1 could potentially facilitate DSB end coordination by promoting the extension of the D-loops formed upon strand invasion by the broken ends. Therefore, in the absence of Mph1, even ends that are synapsed close together (only 1.2 kb apart) may behave as though they are separated by a larger distance. If Mph1 indeed promotes D-loop extension, it may act only over short distances, because mph1Δ did not compromise the already slow and inefficient repair of 5-kb or 18-kb gaps (Figure 1, B and D). In contrast to Mph1, Sgs1 seems to restrict the enlargement of the D-loops (Jain et al. 2009).

Double deletion of SGS1 and MPH1 eliminates REC-mediated delay in the initiation of long gap repair and BIR

Since Sgs1 and Mph1 appeared to play opposing roles in the regulation of 1.2-kb and 5-kb gap repair, we made double deletions to examine their functional interactions. Deleting Sgs1 suppressed the mph1Δ defect of break repair as well as the 1.2-kb gap repair, as both the efficiency (1.2-kb gap repair; Figure 2A) and kinetics of repair (break repair and 1.2-kb gap repair; Figure 2, B and C) were restored to WT levels. Conversely, mph1Δ did not suppress the sgs1Δ-dependent acceleration of repair kinetics in the 5-kb gap-repair strain (Figure 2D). Strikingly, while neither of the single mutations affected the kinetics of the 18-kb gap repair, sgs1Δ mph1Δ double deletion resulted in a marked acceleration in the appearance of the primer-extension product (Figure 2, E and F).

We next examined the effect of sgs1Δ mph1Δ double deletion on BIR. Consistent with the previous reports (Jain et al. 2009; Lydeard et al. 2010a; Luke-Glaser and Luke 2012; Stafa et al. 2014), single deletion of either Sgs1 or Mph1 increased the efficiency of BIR by ∼twofold (Figure 3A). In addition, we found that the simultaneous deletion of both Sgs1 and Mph1 increased the efficiency of BIR by nearly fourfold (Figure 3A). Thus, akin to their additive/synergistic roles in suppression of heteroallelic recombination and crossover formation in mitotic cells (Prakash et al. 2009; Sebesta et al. 2011; Mitchel et al. 2013), Mph1 and Sgs1 seem to act in parallel pathways to inhibit BIR. The sgs1Δ mph1Δ double deletion also resulted in a dramatic acceleration in the initiation of repair of the BIR substrate (Figure 3, B and C). Thus, the simultaneous deletion of both Sgs1 and Mph1 eliminates the REC-mediated kinetic differences between the different gap-repair and BIR strains such that the repair kinetics in all cases is comparable to the repair of a 1.2-kb gap in the WT background (Figure 3, B and C).

Figure 3.

Mph1 and Sgs1 act independently to suppress BIR. (A and D) Viabilities of the indicated strains. Data represent mean ± SD (n ≥ 4). * in (A) denotes that the viabilities of mph1Δ and sgs1Δ single-mutant strains are significantly different from the WT strain (P ≤ 0.005). ** denotes that the viabilities of the indicated strains are significantly different from WT as well as mph1Δ and sgs1Δ single-mutant strains (P ≤ 0.005). * in (D) denotes that the viabilities of the indicated Rad51 overexpressing strains are significantly different from the parent strains (P ≤ 0.005). (B) Kinetics of the indicated mph1Δ and sgs1Δ double-mutant strains as determined by a quantitative PCR-based primer extension assay using primers shown schematically in Figure 1A. Kinetics of WT break-repair and BIR strains (dotted lines) are shown for comparison. (C) Kinetic data of mph1Δ and sgs1Δ double-mutant strains from (B) plotted after normalizing the amount of product obtained at 15 hr to 100% in each strain. Kinetic data represent mean of three PCR reactions from each of two or three independent time courses ± SD. Data from WT and sgs1Δ single mutants have been published in Jain et al. (2009).

It is possible that the improved efficiency of BIR in the absence of Sgs1 and Mph1 could be due to the increased stability of the D-loop intermediates. In this regard, both Sgs1 and Mph1 have been shown to disrupt the strand-invasion intermediates (Bachrati et al. 2006; Prakash et al. 2009). The human ortholog of Sgs1 has also been shown to disrupt Rad51 filament formation in vitro (Bugreev et al. 2007). Sgs1 might additionally affect the stability of the Rad51 filament near the DSB by stimulating 5′→3′ resection to generate more single-stranded DNA substrate for Rad51 binding further away from the break (Mimitou and Symington 2009; Lydeard et al. 2010a). Consistent with the role of Sgs1 and Mph1 in modulating D-loop formation, we found that Rad51 overexpression also improved the efficiency of BIR as well as long gap repair (Figure 3D; also Lydeard et al. 2010a). There was no further increase in the efficiency of BIR when Rad51 was overexpressed with sgs1Δ and mph1Δ single or double deletions (Figure 3A). These data suggest that Sgs1 and Mph1 act in parallel pathways to suppress BIR, presumably by destabilizing the Rad51 filament and/or the strand-invasion intermediates.

We next examined the effect of overexpressing Rad51 on the kinetics of repair. Interestingly, while overexpression increased the efficiency of gap repair and BIR (Figure 3D; also Lydeard et al. 2010a), it did not accelerate the kinetics of either the 18-kb gap repair (Figure 2, E and F) or BIR (Lydeard et al. 2010a). This is in stark contrast to the sgs1Δ mph1Δ double deletion, which increased the efficiency as well as the kinetics of BIR and 18-kb gap repair (Figure 2, A and F, and Figure 3C). These data argue that in addition to regulating the stability/topology of the D-loops, Sgs1 and Mph1 must play a more direct role in imposing the REC, either by slowing down BIR and/or by preventing the GC machinery from loading inappropriately at single-end invasion intermediates.

Mechanistic shift in the repair of long gap repair and BIR substrates in the absence of both Sgs1 and Mph1

The accelerated repair kinetics of sgs1Δ mph1Δ gap-repair and BIR strains could either be attributed to a more rapid initiation of the BIR pathway itself or it could indicate a mechanistic shift from the slower BIR to the quicker GC-like mode of repair. To distinguish between these possibilities, we carried out several tests, including an analysis of repair products and the effect of impairing BIR-specific repair with a pol30-89 mutation. First, we examined ∼100 repaired colonies from WT and sgs1Δ mph1Δ strains for the retention (indicative of GC) or loss (indicative of BIR) of the distal nonessential end of Chr 5. We have previously shown that increasing the gap between the donors results in an increase in the proportion of colonies that repair the break by BIR (Jain et al. 2009). We note that during long gap repair, independently initiated BIR events from the widely-separated LE and U2 ends may resolve to yield GC-like outcomes, potentially resulting in a gross underestimation of the actual frequency of BIR. Nevertheless, if double deletion of SGS1 and MPH1 merely accelerates the kinetics of BIR, the proportion of colonies that lose the distal end of Chr 5 should not change relative to the WT background. Conversely, if there is a switch from the BIR to GC mode of repair, the proportion of colonies that lose the distal end might be expected to decrease; however an increase in the proportion of GC events that are accompanied by crossing over (CO) could also result in the loss of the distal end of Chr 5 (Figure 4A). If repair occurred in the G₂ phase of the cell cycle such that one of the chromatids repaired by GC leading to a CO while the other repaired by GC without a CO, the chromatids could segregate in a manner that would give rise to one daughter cell which has lost the distal end of Chr 5, and another one, which lacks the left end of Chr 3 (Figure 4A, panel ii). Since the left end of Chr 3 is essential for cell survival, only the daughter that has lost the distal nonessential end of Chr 5 would be viable and grow into a BIR-like colony. In the absence of SGS1 and MPH1 we observed a significant increase in the proportion of repaired colonies that had lost the distal end of Chr 5; even for break repair, this proportion increased from ∼1% in WT to ∼25% in the sgs1Δ mph1Δ background (Figure 4B). Since both Sgs1 and Mph1 are known to suppress mitotic crossovers (Ira et al. 2003; Prakash et al. 2009), the increase in the proportion of colonies lacking the distal end of Chr 5 could be a result of the much higher frequency of CO-associated GC events in the absence of these helicases. To test this, we analyzed the repaired colonies by Southern blot. As expected, all of the WT colonies analyzed gave products consistent with repair by GC without an associated CO (Figure 4C). Conversely, 3 out of 12 (25%) colonies derived from the sgs1Δ mph1Δ break-repair strain harbored products of a CO as well as a noncrossover (NCO) event, and a similar proportion of colonies yielded a BIR-like outcome (Figure 4D). It is likely that these two populations represent the alternate segregation outcomes of GC-mediated repair outlined in Figure 4A, in which case the colonies lacking the distal end of Chr 5 would not have arisen from a true BIR event but from the enhanced level of CO associated with GC. Thus, it is difficult to assess whether the elevated repair seen in sgs1Δ mph1Δ arose from increased BIR or GC.

Figure 4.

Analysis of repaired colonies. (A) Schematic representation of GC-mediated repair in G₂ resulting in a BIR-like outcome. Top part shows Chr 5 (red) harboring the leu2::HOcs cassette, and Chr 3 (blue) harboring the donor LEU2. If the DSB on Chr 5 is repaired via GC in the G₂ phase of the cell cycle such that repair on one chromatid is associated with a noncrossover (NCO) and repair on the other with a crossover (CO), the chromatids can randomly segregate as shown in (i) or (ii). In one scenario (i) the products of NCO cosegregate to one daughter while the products of CO segregate to the other daughter such that the repaired colony harbors products of both NCO and CO. In the other scenario (ii) each of the daughters receives an NCO and a CO product. However, the daughter that lacks the distal end of Chr 3 lacks some essential genes and is therefore inviable while the other daughter that lacks the distal nonessential end of Chr 5 grows into a colony harboring a BIR-like repaired product. Vertical lines indicate the positions of the DraI enzyme and the numbers in parentheses indicate the band sizes obtained with a leu2 probe for the various repair outcomes on a Southern blot. (B) Proportion of repaired colonies from the indicated genotypes lacking the distal end of Chr 5, as determined by colony PCR using a primer pair upstream of the LE donor on Chr 3. At least 100 colonies were analyzed in each case. The WT data have been published in Jain et al. (2009). (C and D) Southern blot analysis of the repaired WT (C) and sgs1Δ mph1Δ (D) break-repair colonies. Genomic DNA from repaired colonies was digested with DraI and probed with a leu2 probe.

As an alternate approach to distinguish between the GC and BIR modes of repair, we employed the pol30-89 allele of PCNA, which has a dominant-negative effect on BIR but does not affect GC (Lydeard et al. 2010b). We reasoned that if the sgs1Δ mph1Δ double deletion merely accelerates the kinetics of BIR, then pol30-89 would compromise the repair efficiency of the gap-repair and BIR strains. However, if the cells switched from BIR to GC in the absence of both Sgs1 and Mph1, the efficiency of repair would remain unaffected upon addition of pol30-89. We note that even though pol32Δ has a much more profound effect on BIR (> 20-fold reduction) (Lydeard et al. 2007; Jain et al. 2009) compared to the three- to fourfold reduction imparted by the pol30-89 allele (Lydeard et al. 2010b), we could not use pol32Δ for this analysis because sgs1Δ and pol32Δ are synthetic lethal (Hanna et al. 2007).

As expected, pol30-89 did not affect the repair efficiency of WT or sgs1Δ mph1Δ break-repair strains (Figure 5A). Furthermore, pol30-89, like pol32Δ (Jain et al. 2009), severely impaired WT cells with the 18-kb or BIR substrates, which use BIR as their principal mode of repair. However, pol30-89 had very little effect on the repair efficiencies of these strains in the sgs1Δ mph1Δ background (Figure 5, B and C). These data suggest that, in the absence of both Sgs1 and Mph1, REC is eliminated and there is a mechanistic shift in the repair of long gaps and BIR substrates such that their repair is no longer impaired by a mutation that preferentially affects BIR. In accordance with our kinetic data, this mechanistic shift occurs only if both Sgs1 and Mph1 are deleted because pol30-89 significantly affects the repair efficiency of sgs1Δ or mph1Δ single-mutant BIR strains (Figure 5C).

Discussion

The REC can be envisioned to work in two different ways. In one scenario, the engagement of DSB ends with their homologous donors in a GC-compatible configuration could generate a positive “GO” signal to rapidly recruit the GC machinery to carry out repair. In the second scenario, the inability of both the DSB ends to pair in a GC-compatible configuration could send a negative “STOP” signal to prevent the recruitment of the GC machinery to a single engaged end (or distantly engaged ends). The failure to initiate new DNA synthesis in the absence of a GO signal or due to the generation of a STOP signal presumably leads to repeated cycles of Rad51 filament dissociation and reinvasion in the attempt to find homologies to both DSB ends in the vicinity of each other so that repair can proceed via the more conservative GC pathway. However, after several failed attempts to find homologies in a GC-compatible configuration, the cells would resort to BIR to initiate repair from a single engaged end. According to the first scenario, if the REC is disrupted, even the ends that can be repaired by GC would fail to generate/perceive the GO signal, and therefore be inappropriately channeled into the slow and inefficient BIR pathway. Conversely, disruption of REC in the second scenario would result in a failure to generate the STOP signal if only one of the DSB ends is paired with its homologous donor, thereby channeling even the BIR substrates into the GC pathway. Our results support the creation of a STOP signal, since disruption of the REC by Sgs1 and Mph1 double deletion results in rapid, Pol30-89-independent repair of the BIR and gap-repair substrates. We propose that in the absence of REC, the repair machinery is able to initiate new DNA synthesis soon after a DSB end strand invades its homologous donor, irrespective of whether homology to the other end has been found. Although there is a dramatic acceleration of BIR and gap repair in the absence of Sgs1 and Mph1, their repair kinetics are still slower than the kinetics of break repair (Figure 3, B and C). The reason for this is not clear, although it is plausible that when both DSB ends are juxtaposed right next to each other, they might act cooperatively to modify the chromatin structure and load the repair synthesis machinery or prevent the D-loop from collapsing.

It will be worthwhile to test whether any other factors (besides Pol30-89) that are normally required for BIR, are also dispensable for repair in the absence of REC. It also remains to be determined what really triggers the REC and how Sgs1 and Mph1 act as sensors or mediators/effectors of this pathway. Since mutations in the human orthologs of these helicases have been known to cause Bloom’s syndrome and Fanconi Anemia, both of which are characterized by genomic instability and cancer predisposition, a deeper understanding of the REC could potentially provide useful insights into the etiology of these debilitating disorders in the future.