|Home | About | Journals | Submit | Contact Us | Français|
At the final step of homologous recombination, Holliday Junction-containing joint molecules (JMs) are resolved to form crossover or noncrossover products. The enzymes responsible for JM resolution in vivo remain uncertain, but three distinct endonucleases capable of resolving JMs in vitro have been identified: Mus81-Mms4(EME1), Slx1–Slx4(BTBD12) and Yen1(GEN1). Using physical monitoring of recombination during budding yeast meiosis, we show that all three endonucleases are capable of promoting JM resolution in vivo. However, in mms4 slx4 yen1 triple mutants, JM resolution and crossing-over occur efficiently. Paradoxically, crossing-over in this background is strongly dependent on the Blooms helicase ortholog, Sgs1, a component of a well-characterized anti-crossover activity. Sgs1-dependent crossing-over, but not JM resolution per se, also requires XPG-family nuclease, Exo1, and the MutLγ complex, Mlh1–Mlh3. Thus, Sgs1, Exo1 and MutLγ together define a previously undescribed meiotic JM resolution pathway that produces the majority of crossovers in budding yeast and, by inference, in mammals.
Homologous recombination is an essential chromosome repair process that also facilitates chromosome segregation during meiosis (Hunter, 2006). The central reaction of recombination is formation of a joint-molecule intermediate (JM) via DNA strand-exchange between a broken chromosome and a homologous template. Following DNA synthesis to restore sequences that were lost or damaged at the site of the original lesion, JMs must be resolved to allow chromosome segregation.
Regulation of JM resolution is imperative for cells undergoing meiosis, in which hundreds of recombination events are induced by programmed formation of DNA double-strand-breaks (DSBs)(Hunter, 2006). Meiotic recombination produces a highly regulated distribution of crossovers, with each pair of homologs becoming connected by at least one exchange. Crossovers, in combination with sister-chromatid cohesion, tether maternal and paternal homologs to allow their stable biorientation on the spindle and, consequently, efficient disjunction at meiosis I.
The specific structure of a JM dictates whether resolution necessitates action of a DNA helicase, topoisomerase, endonuclease or a combination of these enzymes (Mimitou and Symington, 2009; Schwartz and Heyer). Displacement loops (D-loops; e.g. the Single-End-Invasion shown in Figure 1C), resulting from DNA strand-exchange of a single break end, can be unwound by helicases. In contrast, resolution of JMs connected by Holliday junctions (HJs; e.g. the double Holliday junction in Figure 1C) necessitates the action of endonucleases. Prokaryotic HJ resolving endonucleases have been known for over 20 years (West, 1997), but lack of sequence conservation has hindered discovery of their eukaryotic counterparts. However, over a decade of intense investigations have identified three distinct eukaryotic endonucleases with in vitro JM cleaving activities:
MUS81-EME1/Mms4 (Mus81-Mms4 in budding yeast) is an XPF-family endonuclease capable of cleaving a variety of branched structures including 3’-flaps, D-loops and nicked-HJs, favoring substrates with a nick or gap adjacent to the branch point that is ultimately incised (Schwartz and Heyer, 2011). Genetic studies provide compelling evidence that MUS81 enzymes process JMs in vivo although the exact substrate(s) remains uncertain (Boddy et al., 2001; Cromie et al., 2006; Jessop and Lichten, 2008; Oh et al., 2008).
SLX1 contains a UvrC-intron (URI)-endonuclease domain and C-terminal PHD-type zinc finger characteristic of the URI–YIG family of endonucleases (Dunin-Horkawicz et al., 2006). Nuclease activity depends on interaction with the scaffold protein, SLX4 (a.k.a. BTBD12), and recombinant SLX1–SLX4 complexes from budding yeast and human can cleave Y-junctions, 5’-flaps and HJs (Fricke and Brill, 2003; Munoz et al., 2009; Svendsen et al., 2009). In Drosophila and to a lesser degree C. elegans, meiotic crossing-over is facilitated by SLX4 homologs, MU312 and HIM18, respectively (Saito et al., 2009; Yildiz et al., 2002). However, the crossover function of MUS312 involves not SLX1, but the XPF-ERCC1-family nuclease, MEI-9-ERCC1 (Andersen et al., 2009).
GEN1/Yen1 (Yen1 in budding yeast) was identified as a member of the Rad2/XPG endonuclease family and, by biochemical criteria, as a HJ-resolving endonuclease (Furukawa et al., 2003; Ip et al., 2008; Ishikawa et al., 2004). GEN1/Yen1 is the only nuclear endonuclease that appears to meet criteria for a bona fide HJ-resolvase, resolving via concerted symmetrical cleavages analogous to the archetypal prokaryotic RuvC resolvase (Ip et al., 2008; Rass et al., 2010). Although GEN1/Yen1 do not play essential roles in resolving recombination intermediates, Yen1 has been shown to partially suppress the recombination and damage-sensitivity phenotypes of mus81 mutants (Agmon et al., 2011; Blanco et al., 2010; Ho et al., 2010; Tay and Wu, 2010).
To date, the best-characterized eukaryotic JM-resolution pathway does not involve an endonuclease. Human BTR and budding yeast STR complexes comprise RecQ helicases, BLM/Sgs1, their cognate type-I topoisomerases, TOPIIIα/Top3, and specificity factors RMI1–RMI2/Rmi1. In isolation, Sgs1/BLM can migrate Holliday junctions and unwind D-loops (Bachrati et al., 2006; Cejka and Kowalczykowski, 2010; Karow et al., 2000; van Brabant et al., 2000), but the BTR/STR ensembles also perform a unique reaction to dissociate double Holliday junctions (dHJs) via convergent branch migration and decatenation yielding exclusively noncrossover products (Cejka et al., 2010; Singh et al., 2008; Wu and Hickson, 2003; Xu et al., 2008). Consistent with this activity, in vivo data indicate that BTR/STR is a prominent anti-crossover JM-resolving activity in the eukaryotic nucleus (Bzymek et al., 2010; Chu and Hickson, 2009; Hickson and Mankouri, 2011).
During meiosis in budding yeast, dHJ resolution and crossing-over require the polo-like kinase, Cdc5, and at least one resolvase, Mus81-Mms4, is activated by this kinase (Clyne et al., 2003; Matos et al., 2011; Sourirajan and Lichten, 2008). In contrast, noncrossovers are inferred to derive from relatively unstable non-dHJ intermediates, presumably D-loops, and their formation is Cdc5 independent (Allers and Lichten, 2001; Sourirajan and Lichten, 2008). These observations, together with the regulated distribution of meiotic crossovers, dictate that resolution of the two HJs of meiotic dHJs is coordinated to specify a crossover outcome.
The factors involved in crossover-specific dHJ-resolution and the mechanism of this reaction remain mysterious. However, mutation of a number of conserved genes causes reduced crossing-over without impacting the overall efficiency of DSB repair. These pro-crossover genes include the ZMMs, which encode a functionally diverse set of meiosis-specific proteins that appear to stabilize SEIs and promote their transition to dHJs (Hunter, 2006; Lynn et al., 2007). A distinct set of pro-crossover proteins is the DNA mismatch-repair components, Exo1 and the MutLγ complex, Mlh1–Mlh3 (Hunter, 2011; Kolas and Cohen, 2004). JM formation occurs normally in exo1 and mlh1/3 mutants, but crossing-over is defective, suggesting that Exo1 and MutLγ act at a late step of recombination to specify a crossover outcome (Zakharyevich et al., 2010). Consistent with a late function, MutLγ localizes specifically to future crossover sites in a number of organisms, including humans (Kolas and Cohen, 2004). Like Yen1, Exo1 is a member of the XPG/Rad2 nuclease superfamily but its nuclease activity is not required for crossing-over, although interaction with MutLγ is important (Keelagher et al., 2010; Zakharyevich et al., 2010). However, MutLγ is a putative endonculease and, as such, could participate directly in dHJ resolution (Hunter, 2011; Nishant et al., 2008).
In this study, we perform a comprehensive in vivo analysis of JM resolution during meiosis in budding yeast. We show that while Mus81-Mms4, Slx1–Slx4 and Yen1 can promote JM resolution in vivo, absence of all three nucleases has only a modest impact on JM resolution and formation of crossover and noncrossover products. Thus, the majority of meiotic JM resolution involves a previously undescribed pathway. We identify Sgs1 as being central to this pathway and show that Exo1 and MutLγ confer crossover-specific resolution. Finally, we show that, Mus81-Mms4, Slx1–Slx4, Yen1, Sgs1 and Exo1-MutLγ together account for essentially all JM resolution in vivo. These data reveal Sgs1 as a central regulator and mediator of meiotic JM resolution, with unanticipated roles in both noncrossover and crossover formation. Further, Exo1-MutLγ is implicated as a crossover-specific JM resolution factor.
We previously showed that cells lacking Sgs1 and Mus81-Mms4 die in meiosis because unresolved JMs persist into anaphase and impede chromosome segregation (“meiotic catastrophe”)(Jessop and Lichten, 2008; Oh et al., 2008). However, most JMs are eventually resolved and crossover levels reach ≥75% of wild-type levels, indicating the presence of at least one additional resolving activity.
To identify this activity (or activities), we screened null mutations of a number of candidate genes in the mms4 sgs1 mutant background. Lethality of the sgs1 mms4 mutant combination in mitotically cycling cells was circumvented using meiosis-specific conditional alleles in which the CLB2 promoter, which is strongly repressed during meiosis, replaces the native promoters of the SGS1 and MMS4 genes (Lee and Amon, 2003; Oh et al., 2008). Given that the pCLB2-SGS1 pCLB2-MMS4 combination (hereafter sgs1 mms4) is lethal in meiosis, recombination was monitored using a well-characterized DNA physical assay system at the HIS4LEU2 recombination hotspot (Figure 1)(Hunter and Kleckner, 2001; Schwacha and Kleckner, 1995). Critically, this approach allows us to monitor recombination in the entire, unbiased cell population and does not require cells to maintain viability during meiosis.
Cultures of triple mutant strains (sgs1 mms4 plus candidate mutations) were induced to undergo synchronous meiosis and analyzed for defective JM resolution and decreased crossing-over. To preserve JM structures, interstrand DNA crosslinks were introduced using psoralen and long-wave UV light. Genomic DNA was then extracted, digested with XhoI and analyzed by gel electrophoresis and Southern blot hybridization. Each hybridizing signal was quantified using a phosphorimager.
XhoI polymorphisms between parental “Mom” and “Dad” chromosomes yield DNA fragments diagnostic for DSBs, JMs and crossovers (Figure 1A). DSBs and crossovers are analyzed using one-dimensional gels (Figure 1A, B). JMs are analyzed using native/native two-dimensional (2D) gels, which reveal the branched nature of these intermediates and facilitate their accurate quantification. Three types of JM can be detected: SEIs, dHJs and multi-chromatid JMs (mcJMs) comprising three or four interconnected chromatids (Figure 1 A, C, D)(Bell and Byers, 1983; Hunter and Kleckner, 2001; Oh et al., 2007; Schwacha and Kleckner, 1995). Noncrossover gene-conversion products are monitored by virtue of a BamHI/NgoMIV restriction-site polymorphism located directly at the site of DSB formation (Figure 1A). Double digestion with XhoI and NgoMIV produces fragments that are diagnostic for both crossovers and noncrossovers (Figure 1A, E).
Candidate meiotic JM resolving factors were selected based on several criteria including known biochemical properties, genetic and physical interactions with recombination factors, expression profile in meiosis and known meiotic phenotypes. Figure 2 summarizes initial analysis of five candidate factors in the sgs1 mms4 background. In each case, we analyzed levels of JMs and crossovers at 0, 9, 13 and 24 hours after the induction of meiosis. Figure 2 shows final JM and crossover levels at 24 hrs.
In wild-type cells, JM resolution is efficient and ~19% of chromosomes undergo crossing-over (Figure 2A–D). In contrast, in sgs1 mms4 cells, a subset of JMs remains permanently unresolved even after 24 hrs and crossover levels plateau at a slightly reduced level of ~17% (Figure 2A–D).
Expression of Yen1 is induced at late times during meiosis (Chu and Herskowitz, 1998) and the recent study of Matos et al. (2011) reveals that it is activated via dephosphorylation, but only after the first meiotic division has ensued. In an sgs1 mms4 yen1 triple mutant, high levels of JMs remain permanently unresolved (7% of DNA versus 1.8% in sgs1 mms4 cells; Figure 2A,B). Correspondingly, crossovers are significantly reduced relative to sgs1 mms4 cells (9.2% vs. 17.1%, respectively; Figure 2C,D). These data show that in the sgs1 mms4 background, Yen1 resolves a subset of JMs to produce crossovers. However, close to half the crossovers detected in sgs1 mms4 cells can form independently of Yen1 indicating the presence of at least one additional JM resolving activity.
Slx4 functions as a scaffold for assembly and activation of two nuclease complexes in budding yeast: Slx1–Slx4, which catalyzes HJ cleavage in vitro, and Rad1–Rad10-Saw1 that cleaves 3’-flaps formed during single-strand annealing (see Introduction)(Fricke and Brill, 2003; Lyndaker and Alani, 2009). Analogous to yen1, both slx4 and slx1 mutations greatly enhance the JM resolution defect of sgs1 mms4 cells (Figure 2A,B). However, crossover levels are less severely reduced in the sgs1 mms4 slx4 and sgs1 mms4 slx1 strains relative to sgs1 mms4 yen1 (Figure 2D). In contrast to SLX1, mutation of RAD1 has no discernable affect on JM resolution or crossing-over (data not shown). These observations directly implicate Slx1–Slx4 in processing a subset of JMs in the sgs1 mms4 background.
Exo1 and MutLγ act in a common pathway to promote crossing-over (Introduction)(Hunter, 2011; Zakharyevich et al., 2010). In contrast to yen1, slx4 and slx1, the exo1 and mlh3 mutations do not enhance the resolution defect of sgs1 mms4 cells (Figure 2A,B). In fact, final JM levels are lower in sgs1 mms4 exo1 and sgs1 mms4 mlh3 cells relative to sgs1 mms4, suggesting that absence of Exo1-MutLγ may allow for more efficient resolution of JMs by other factors such as Yen1 and Slx1–Slx4. Although JM resolution remains relatively efficient, crossing-over is significantly reduced relative to sgs1 mms4 cells, consistent with the known pro-crossover function of Exo1-MutLγ (Figure 2C,D). Thus, exo1 and mlh3 define a distinct class of mutations that reduce crossing-over in the sgs1 mms4 background, but do not confer significant blocks to JM resolution.
The analysis above reveals that Yen1 and Slx1–Slx4 can resolve JMs in vivo. Previous analysis has shown that a third structure-selective nuclease, Mus81-Mms4, also promotes meiotic JM resolution in vivo (Jessop and Lichten, 2008; Matos et al., 2011; Oh et al., 2008). To determine the collective contribution of these three nucleases, we analyzed an mms4 slx4 yen1 triple mutant (Figure 3). Monitoring of meiotic divisions by DAPI staining shows that chromosome segregation fails in this strain (Figure 3A). Thus, one or more of the three resolving nucleases are essential for chromosome separation.
Recombination phenotypes of mms4 slx4 yen1 cells were analyzed by DNA physical assays (Figure 3B–G). Meiotic recombination is surprisingly efficient, with noncrossovers forming at wild-type levels (Figure 3B,C) and crossovers reaching ~70% of wild-type levels (13.5% versus 19.1%; Figure 3D,E). Consistent with efficient product formation, JMs are resolved efficiently in mms4 slx4 yen1 triple mutants (Figure 3F,G). However, a significant level of unresolved JMs remains (0.45% versus 0.13% in wild-type), which is the likely cause of segregation failure in mms4 slx4 yen1 triple mutants. Surprisingly, however, we can conclude that the three JM-resolving nucleases, Mus81-Mms4, Slx1–Slx4 and Yen1, are dispensable for the majority of meiotic JM resolution.
To further explore the functions of Yen1 and Slx1–Slx4 we performed a detailed analysis of meiotic recombination in yen1, slx1 and slx4 single mutants (Figure 4). Cultures of mutant and wild-type strains were sporulated in parallel and recombination was analyzed by physical and genetic assays. Despite the robust resolution activities revealed for Yen1 and Slx1–Slx4 in the sgs1 mms4 background, yen1, slx1 and slx4 single mutants show little defect in meiotic recombination. In all three strains, overall timing and levels of DSBs, JMs, crossovers and meiotic divisions are very similar to those observed for wild-type strains (Figure 4A,B,C).
In the yen1 mutant culture, a small subset of JMs disappears with a delay of 1–2 hrs, paralleled by delayed appearance of a fraction of crossovers and slightly delayed divisions. The significance of these mildly biphasic kinetics is unclear because they are within the range of typical day-to-day variations (Cha et al., 2000). However, it remains possible that Yen1 facilitates timely resolution of a subset of JMs in wild-type cells (Matos et al., 2011).
Similar inferences can be made for slx1 and slx4 mutants. JM curves for both slx1 and slx4 time-course experiments have shoulders that suggest slightly delayed turnover of some intermediates (Figure 4A,B,C). Overall, however, meiotic recombination occurs efficiently and crossovers reach wild-type levels.
Efficient meiotic recombination in yen1, slx1 and slx4 mutants was confirmed by genetic methods (Figure 4D,E; Supplementary Figure S1 and Tables S1 and S2). Tetrad analysis was used to calculate genetic map distances for two intervals on chromosome 3. One interval, URA3-HIS4LEU2 flanks the HIS4LEU2 DSB hotspot and the adjacent interval, HIS4LEU2-MAT, spans the centromere. In yen1 strains, map distances in both intervals were indistinguishable from those calculated for wild-type tetrads (Figure 4D). Moreover, spore viability was identical to wild type (95% for both wild-type and yen1 tetrads; Figure 4E). Thus, meiosis occurs normally without Yen1.
Consistent with physical analysis of crossing-over (above), slx1 and slx4 mutations resulted in wild-type map distances for the interval encompassing the HIS4LEU2 hotspot (Figure 4D and Supplemental Table S1). However, the HIS4LEU2-MAT interval showed significantly increased recombination suggesting a role in negatively regulating crossing-over in some intervals. To further investigate this possibility, we analyzed crossing-over in eight linked intervals that span the length of chromosome III (Supplemental Figure S1)(Zakharyevich et al., 2010). In all intervals, map distances for wild-type and slx1 were statistically indistinguishable (although a slight reduction in crossing-over was suggested for two intervals). Moreover, high spore viabilities of slx1 and slx4 mutants indicate that Slx1–Slx4, like Yen1, is not essential for meiosis (Figure 4E).
To address whether Yen1 and Slx1–Slx4 have redundant functions, we also analyzed cells lacking both enzymes (Figure 4D,E). Map distances and spore viability of the slx4 yen1 double mutant were indistinguishable from an slx4 single mutant indicating that the two mutations do not interact.
The above data indicate that Yen1 is not required for JM resolution and crossing-over in wild-type cells. However, Yen1 clearly does resolve JMs in the sgs1 mms4 double mutant. To understand the epistatic relationships between Yen1 and the other putative JM resolution pathways, we compared final crossover levels at the HIS4LEU2 locus in wild-type, exo1, sgs1, mms4 and slx4 single mutants; and exo1 yen1, sgs1 yen1, mms4 yen1 and slx4 yen1 double mutants (Figure 4F). Crossover levels in sgs1 and sgs1 yen1 strains were identical indicating that the sgs1 mutation does not create a condition in which Yen1 is required for crossing-over. Similarly, crossover levels in slx4 and slx4 yen1 strains were indistinguishable, consistent with the genetic analysis, above. Although exo1 mutation reduces crossing-over by 2-fold, absence of Yen1 does not cause an additional decrease in crossing-over. In contrast, in the mms4 background, yen1 mutation caused an additional reduction of 21%, for a total crossover reduction of ~39% in the mms4 yen1 double mutant (Figure 4F). Thus, specifically in the absence of Mus81-Mms4, Yen1 makes a significant contribution to crossing-over. Moreover, mms4 yen1, but no other yen1 double mutant combination, causes meiotic catastrophe resulting in dead spores (Figure 4H).
Segregation failure in mms4 yen1 cells predicts a defect in JM resolution. 2D gel analysis confirms that mms4 yen1 cells fail to resolve a small subset of JMs (Figure 4I,J). Thus, Yen1 is a cryptic resolvase in wild-type cells, whose activity is only manifested in the absence of Mus81-Mms4. These data extend results of recent studies showing that yen1 mutation enhances the recombination defects of mms4/mus81 mutants (Agmon et al., 2011; Blanco et al., 2010; Ho et al., 2010; Matos et al., 2011; Tay and Wu, 2010).
Slx1–Slx4, like Yen1, has clear JM resolving activity in the mms4 sgs1 background. Interaction between slx4 and mutations in the other resolvase genes was also investigated by analyzing crossover levels in various double mutant strains (Figure 4G). In no case did slx4 mutation cause an additional reduction in crossing-over relative to the corresponding single mutants. However, in sgs1 slx4 cells, but no other double mutant involving slx4, we observed meiotic catastrophe accompanied by unresolved JMs (Figure 4H,I,J; analogous phenotypes are seen in sgs1 slx1 cells – Figure 4H and data not shown).
Taken together, these data indicate that Slx1–Slx4 is essential for resolution of a subset of JMs formed specifically when Sgs1 is absent. Despite significant levels of unresolved JMs, sgs1 slx4 cells do not have reduced levels of crossover or noncrossover products (Figure 4G and data not shown). A possible explanation for this discordance is that Slx1–Slx4 plays an early function in sgs1 mutants to prevent formation of extra JMs that would otherwise become dependent on Sgs1 for their resolution.
The question remains, which activities are responsible for the efficient JM resolution and product formation observed in mms4 slx4 yen1 triple mutants? The data above, together with our previous analysis of sgs1 mms4 mutants (Oh et al., 2008), points to Sgs1 as a likely candidate. This inference was confirmed by the striking phenotypes of an mms4 slx4 yen1 sgs1 quadruple mutant (Figure 5A–F). Both crossovers and noncrossovers are diminished in this strain and extremely high levels of JMs remain unresolved. Crossovers in mms4 slx4 yen1 sgs1 cells are reduced to just 10% of wild-type levels, a more than 6-fold reduction relative to the mms4 slx4 yen1 triple mutant (Figure 5A,B). Noncrossovers are reduced 3.6-fold relative to levels detected in both wild-type and mms4 slx4 yen1 triple mutant strains (Figure 5C,D). Diminished product formation is a consequence of unresolved JMs, which persist at ~13% of hybridizing DNA, almost twice levels seen in sgs1 mms4 yen1, sgs1 mms4 slx1 and sgs1 mms4 slx4 triple mutant strains (compare Figure 5F and Figure 2B).
Given that sgs1 single mutants resolve JMs efficiently and form near wild-type levels of crossovers and noncrossovers (Jessop and Lichten, 2008; Jessop et al., 2006; Oh et al., 2007), three key inferences can be made from the recombination phenotypes of the mms4 slx4 yen1 sgs1 strain. First, structure-selective endonucleases, Mus81-Mms4, Slx1–Slx4 and Yen1, mediate most JM resolution when Sgs1 is absent. Second, compensatory activities of these three nucleases have obscured the principal role of Sgs1 in JM resolution in wild-type cells. Third, the Sgs1-dependent pathway of JM processing produces both crossovers and noncrossovers.
The pro-crossover role of Sgs1, revealed above, is both unanticipated and paradoxical given that Sgs1 is a well-characterized anti-crossover factor. We reasoned, therefore, that Sgs1-dependent crossing-over requires an additional JM resolution activity. Exo1 and the MutLγ complex are good candidates for such an activity as their mutation reduces crossing-over without altering JM levels, suggesting a late role in specifying crossover-specific resolution (Zakharyevich et al., 2010).
To address the role of Exo1-MutLγ in Sgs1-dependent crossing-over, we analyzed an mms4 slx4 yen1 mlh3 quadruple mutant (Figure 5A–F). Crossing-over in this strain was diminished to the same extent as seen in mms4 slx4 yen1 sgs1 cells indicating that Sgs1-dependent crossovers do indeed require Exo1-MutLγ (Figure 5A,B). However, in stark contrast to mms4 slx4 yen1 sgs1 cells, noncrossovers form at wild-type levels and JM resolution is efficient (Figure 5C,D,E and F). These data show that Exo1-MutLγ is specifically required for Sgs1-dependent crossing-over but, unlike Sgs1, is not important for noncrossovers or for efficient JM resolution per se. We assume that in the absence of Exo1-MutLγ, crossover precursors (presumably dHJs) undergo Sgs1-dependent disassembly to yield noncrossovers.
We recently showed that the nuclease activity of Exo1 is not required for its pro-crossover function, pointing to MutLγ as the nuclease component of the Exo1-MutLγ ensemble (Zakharyevich et al., 2010). In support of this inference, we found that mutation of a conserved nuclease motif found in Mlh3 (mlh3-D523N) mimicked the effects of an mlh3 null mutation in the mms4 slx4 yen1 triple mutant background (Figure 5A–F)(Kadyrov et al., 2006; Nishant et al., 2008).
Taken together, epistasis analysis implies that two resolution factors, Exo1-MutLγ (facilitated by Sgs1) and Mus81-Mms4, are responsible for most crossovers in wild-type cells. In mms4 mutants, compensatory action of Yen1 obscures the true contribution of Mus81-Mms4. Therefore, to assess the combined contributions of Exo1-MutLγ and Mus81-Mms4 to meiotic crossing-over, we compared exo1, mms4 yen1 and exo1 mms4 yen1 strains (Figure 5G,H). Physical assays indicate that exo1 mutation reduces crossing-over by 49%, mms4 yen1 by 39%, and in the exo1 mms4 yen1 triple mutant crossovers are reduced by 86%. Thus, Exo1-MutLγ and Mus81-Mms4 make independent contributions to crossing-over and account for the vast majority of crossovers in wild-type cells.
Our analysis demonstrates that five factors, Exo1-MutLγ, Mus81-Mms4, Slx1–Slx4, Sgs1 and Yen1, can process JMs in vivo. We asked whether these five factors account for all meiotic JM resolution activity in vivo by analyzing meiotic recombination in an mlh3 mms4 slx4 sgs1 yen1 quintuple mutant (Figure 6). Strikingly, crossing-over is abolished in this strain and noncrossovers are reduced by 4-fold (Figure 6A,B,C and D). Consistent with diminished product formation, JMs remain unresolved and accumulate to very high levels (Figure 6E,F). In fact, the level of unresolved JMs in mlh3 mms4 slx4 sgs1 yen1 cells is significantly higher than that detected in cells lacking Ndt80, a meiosis-specific transcription factor required for JM resolution and exit from the pachytene stage of meiotic prophase (Figure 6F)(Allers and Lichten, 2001; Xu et al., 1995). Both the levels and profile of JMs are most similar to those reported previously for sgs1 ndt80 cells, with high levels of multi-chromatid JM species (Oh et al., 2007). However, unlike ndt80 and sgs1 ndt80 strains and despite massive levels of unresolved JMs, mlh3 mms4 slx4 sgs1 yen1 cells progress through the meiotic program and undergo meiotic catastrophe (not shown).
The features of meiotic JM resolution revealed here explain the failure of genetic screens to directly identify meiotic JM resolving enzymes. First, multiple compensatory activities minimize the impact of inactivating any single resolvase; and second, failure to resolve even a minority of JMs in a timely fashion is lethal because unresolved JMs do not trigger a checkpoint response. These features dictate that unambiguous identification of factors responsible for JM resolution in vivo could only be achieved via direct analysis of endogenous JM resolution in strains lacking multiple candidate resolvase genes. By applying this approach to meiotic cells, we have shown that five distinct factors account for essentially all JM resolution in vivo and have illuminated the pathways of meiotic JM resolution, as summarized in Figure 7.
Whenever Sgs1 is present, JMs are resolved efficiently and Mus81-Mms4, Yen1 and Slx1–Slx4 are largely dispensable. Conversely, when Sgs1 is absent, JM resolution remains efficient but becomes dependent on Mus81-Mms4 and Slx1–Slx4 (and, specifically when Mus81-Mms4 is inactivated, on Yen1). In addition, sgs1 mutants form higher levels of JMs such that crossovers and noncrossovers arise at approximately wild-type levels. Consequently, the principal role of Sgs1 for both crossover and noncrossover pathways has been obscured, until now. Similar conclusions are reached by De Muyt et al., who further show that noncrossovers, which normally form independently of polo-like kinase Cdc5, become Cdc5-dependent in sgs1 mutants (De Muyt, 2012). Thus, in wild-type cells, Sgs1 mediates noncrossover formation via a resolvase independent pathway (Figure 7).
With respect to noncrossover formation, the most straightforward model is that Sgs1 helicase activity directly unwinds D-loops to facilitate synthesis-dependent strand-annealing (Figure 7). As suggested by genetic studies, dHJ dissolution via the STR complex may also account for some noncrossovers (Gilbertson and Stahl, 1996; Martini et al., 2011). It is also possible that other helicases implicated in noncrossover formation, such as Srs2 and FANCM homolog Mph1, act in concert with Sgs1 in noncrossover formation (Ira et al., 2003; Prakash et al., 2009). These helicases could be responsible for the residual noncrossovers detected in mlh3 mms4 sgs1 slx4 yen1 quintuple mutants (Figure 6).
The highly regulated distribution of meiotic crossovers predicts the existence of a crossover biased JM resolution factor. Our analysis shows that resolution mediated by Mus81-Mms4, Slx1–Slx4 and Yen1 produces a mixture of crossovers and noncrossovers. In sharp contrast, Exo1-MutLγ specifically promotes crossover-biased resolution, functioning in conjunction with Sgs1.
A nuclease motif identified in human MutL homolog, Pms2 (ortholog of yeast Pms1), is conserved in Mlh3 proteins and its mutation in budding yeast MLH3 reduces crossing-over to the same extent as an mlh3Δ null mutation (Figure 5)(Kadyrov et al., 2006; Nishant et al., 2008). These data implicate Exo1-MutLγ as a major JM resolving nuclease during meiosis. Consistent with this inference, MutLγ localizes specifically to crossover sites and is required for crossing-over in a number of organisms (Hunter, 2006; Kolas and Cohen, 2004). Direct demonstration of inferred nuclease and JM resolution activities of MutLγ remain important goals for the future.
During mismatch repair, a MutS complex (Msh2–Msh6 or Msh2–Msh3) specifically binds the DNA mismatch and MutLα (human Mlh1-Pms2, yeast Mlh1-Pms1) subsequently nicks DNA to initiate mismatch excision (Kadyrov et al., 2006; Kadyrov et al., 2007; Pluciennik et al., 2010). Analogous to mismatch repair, meiotic crossing over involves a meiosis-specific MutS complex, MutSγ (Msh4 and Msh5) and MutLγ. In vitro and in vivo data are consistent with a model in which MutSγ complexes specifically bind JMs and subsequently stabilize them by acting as sliding clamps that embrace the recombining duplexes (Borner et al., 2004; Jessop et al., 2006; Oh et al., 2007; Snowden et al., 2004). We suggest that directional loading of MutSγ complexes during JM formation could subsequently impart resolution bias to the Exo1-MutLγ ensemble, although other polarity signals such as unligated nicks could also play a role (Stahl et al., 2004).
MutSγ and Exo1-MutLγ define the major crossover pathway in many species, including mammals. Previous studies showed that Sgs1 becomes a potent anti-crossover activity when this pathway is inactivated (Jessop et al., 2006; Oh et al., 2007; Zakharyevich et al., 2010). These and other observations led to the idea that MutSγ and Exo1-MutLγ protect crossover precursors from disruption by Sgs1. However, our observations of blocked JM resolution and diminished crossing-over in the mms4 slx4 yen1 sgs1 quadruple mutant (in contrast to mms4 slx4 yen1 cells) reveal an unanticipated pro-crossover function for Sgs1.
Exactly how Sgs1 promotes Exo1-MutLγ dependent crossing-over remains unclear. Notably, JM formation is radically altered in the absence of Sgs1, with decreased levels of two-chromatid interhomolog dHJs and greatly elevated levels of intersister-JMs and multi-chromatid-JMs (Oh et al., 2007). Thus, Sgs1 may facilitate formation of a specific JM structure (e.g. a dHJ with specific dimensions) that can be correctly recognized and processed by the MutSγ + Exo1-MutLγ ensemble. In this regard, it is possible that Sgs1 works in conjunction with MutSγ and other ZMM factors to stabilize strand-exchange intermediates and facilitate the orderly formation of dHJs (Oh et al., 2007). Another, nonexclusive, possibility is that Sgs1 play a more direct role in crossover-specific dHJ resolution mediated by Exo1-MutLγ. In this scenario, Sgs1 could function to target and/or activate the Exo1-MutLγ nuclease activity to HJs via a mechanism reminiscent of the canonical bacterial resolvase, RuvABC (West, 1997).
Consistent with recent studies, we show that loss of Yen1 impacts JM resolution only when Mus81-Mms4 is absent (Agmon et al., 2011; Blanco et al., 2010; Ho et al., 2010; Matos et al., 2011; Tay and Wu, 2010). However, this interaction does not reflect simple redundancy between Mus81-Mms4 and Yen1 because mus81/mms4 single mutants have clear defects in JM processing (De Los Santos et al., 2003; Kaliraman et al., 2001; Matos et al., 2011; Oh et al., 2008), whereas yen1 single mutants have little if any meiotic defect. Thus, while Yen1 can resolve some of the intermediates that would normally be processed by Mus81-Mms4 (or that aberrantly form in its absence), its activity is cryptic when Mus81-Mms4 is present.
Although the biological significance of Yen1 remains unclear, our observations are reconciled by Matos et al. (2011) who showed that Yen1 is subject to phosphorylation-dependent inhibition, only becoming active at meiosis II where it is proposed to resolve persistent JMs. Thus, Yen1 rescues the viability of mus81/mms4 mutants via a temporally distinct wave of resolution activity after the first meiotic division has been attempted. Importantly, partial suppression of the mus81/mms4 crossover defect by Yen1 means that we have previously underestimated the contribution of Mus81-Mms4 to crossing-over. We infer that Mus81-Mms4 actually accounts for close to 40% of crossovers at the HIS4LEU2 locus, which is located on chromosome III, one of the smallest yeast chromosomes. Genome-wide, contribution of this pathway is likely to be lower because the proportion of Mus81-Mms4-dependent crossovers decreases with increasing chromosome size (De Los Santos et al., 2003).
We provide direct in vivo evidence that Slx1–Slx4 processes JMs specifically in the absence of Sgs1. However, absence of Slx1–Slx4 alone has little if any impact and its role in recombination in wild-type cells remains unclear. Yeast Slx4 interacts with a second nuclease, XPF-ERCC1 ortholog Rad1–Rad10, but we have not detected a JM resolution or recombination defect for rad1 or sgs1 rad1 mutants (not shown). This sharply contrasts Drosophila, in which Slx4 ortholog, MUS312, functions with XPF-ERCC1 ortholog MEI9-ERCC1 to define the major crossover pathway (Yildiz et al., 2002).
Dependency of meiotic crossing-over on the various JM resolving factors differs greatly between organisms. Fission yeast lack ZMM and Yen1 pathways and appear to rely exclusively on Mus81-Eme1 (Cromie et al., 2006; Smith et al., 2003). However, S. pombe Sgs1/BLM ortholog, Rqh1, is important both for crossover and noncrossover formation (Cromie et al., 2008), a commonality with budding yeast that could not have been appreciated until now. Similarly, Drosophila and C. elegans Sgs1/BLM orthologs, MUS309 and HIM-6, appear to play positive roles in meiotic crossing-over (McVey et al., 2007; Zetka and Rose, 1995). Parallels between metazoan RTEL1 and Sgs1 can also be made: RTEL1 is a FANCJ-related DNA helicase that can unwind D-loops and C. elegans rtel1-1 mutants appear to channel recombination intermediates into a MUS81-dependent resolution pathway that is normally cryptic in wild-type worms (Youds et al., 2010).
As described above, crossing-over in Drosophila requires Slx4/BTBD12 homolog, MUS312, and XPF-ERCC1 nuclease MEI9-ERCC1. Similar to flies, at least a fraction of crossovers in C. elegans require XPF-1 and BTBD12 ortholog, HIM-18 (Saito et al., 2009). However, in contrast to Drosophila, which lacks Msh4/Msh5 homologs, the MutSγ complex is essential for crossing-over in C. elegans, although no role for Exo1 or MutL proteins has been reported (Zalevsky et al., 1999).
In mouse and Arabidopsis, like budding yeast, a majority of crossovers requires MutLγ and a minority involves Mus81 (Berchowitz et al., 2007; Higgins et al., 2008a; Holloway et al., 2008). Phenotypes of a conditional Blm mutant mouse are consistent with JM resolution being dysregulated in the absence of BLM (Holloway et al., 2010), suggesting that JM metabolism in mammals may be most similar to the budding yeast scenario described here. Finally, in no organism has an essential role for Yen1/Gen1 in meiotic recombination been reported.
In conclusion, we have achieved the critical goal of identifying all factors that can contribute to meiotic JM resolution and crossing-over in vivo, revealing the hierarchies and epistatic relationships among the different resolvases. The overall picture that emerges from this study and those of Matos et al. (2011) and De Muyt et al. (2012) is one of spatial, temporal and regulatory specialization of the different resolving factors to achieve two biological imperatives: (1) efficient implementation of crossovers with a regulated distribution; and (2) timely and efficient JM resolution to allow chromosome segregation.
Full genotypes and details of strain constructions are described in Supplementary Table S3.
Detailed protocols for meiotic time courses and DNA physical assays have been described (Oh et al., 2009). Data points represent averages (±SEM) of two to six experiments.
The timing and efficiency of meiotic divisions and sporulation was performed as described in Oh et al (2008).
Thanks to Eric Alani for the mlh3-D523N::KanMX cassette, Chu-Chun Huang for help with tetrad analysis, Wolf Heyer for valuable input, and Arnaud Muyt and Michael Lichten for communication of unpublished data and stimulating discussions. This work was supported by grant GM074223. N.H. is an Early Career Scientist of the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.