|Home | About | Journals | Submit | Contact Us | Français|
Faithful chromosome segregation in meiosis is crucial to form viable, healthy offspring and in most species, it requires programmed recombination between homologous chromosomes. In fission yeast, meiotic recombination is initiated by Rec12 (Spo11 homolog) and generates single Holliday junction (HJ) intermediates, which are resolved by the Mus81–Eme1 endonuclease to generate crossovers and thereby allow proper chromosome segregation. Although Mus81 contains the active site for HJ resolution, the regulation of Mus81–Eme1 is unclear. In cells lacking Nse5–Nse6 of the Smc5–Smc6 genome stability complex, we observe persistent meiotic recombination intermediates (DNA joint molecules) resembling HJs that accumulate in mus81Δ cells. Elimination of Rec12 nearly completely rescues the meiotic defects of nse6Δ and mus81Δ single mutants and partially rescues nse6Δ mus81Δ double mutants, indicating that these factors act after DNA double-strand break formation. Likewise, expression of the bacterial HJ resolvase RusA partially rescues the defects of nse6Δ, mus81Δ and nse6Δ mus81Δ mitotic cells, as well as the meiotic defects of nse6Δ and mus81Δ cells. Partial rescue likely reflects the accumulation of structures other than HJs, such as hemicatenanes, and an additional role for Nse5–Nse6 most prominent during mitotic growth. Our results indicate a regulatory role for the Smc5–Smc6 complex in HJ resolution via Mus81–Eme1.
The repair of DNA damage by homologous recombination (HR) is central to the faithful propagation of chromosomes during both the mitotic and meiotic cycles. During mitotic growth, exogenous or endogenous genotoxic agents can induce DNA double-strand breaks (DSBs), which may also arise from replication fork mishaps [reviewed in (1)]. The HR-based repair of these insidious forms of DNA damage is executed by a relatively well-defined and largely overlapping set of proteins. Broken DNA ends are first resected to generate a single-stranded (ss) 3′-overhang, which acts as a platform for the subsequent homology search and invasion steps of HR [reviewed in (2)]. The ss DNA-binding protein complex replication protein A initially coats the 3′-overhang, which is subsequently displaced by loading of the RecA homolog Rad51, also called Rhp51 in fission yeast, as well as Dmc1 during meiosis (3–5). The Rad51 nucleoprotein filament invades a homologous duplex, forming a displacement loop (D-loop) that is used to prime repair synthesis and, at a broken fork, to restart replication. Meiotically induced Rec12 (Spo11 homolog) and several partner proteins form DSBs during fission yeast meiosis, to form crossovers important for the proper segregation of homologs at the first meiotic division (6,7). Genomic regions that are hotspots for meiotic DSB formation by Rec12 require Rad51 for DSB repair and recombination; in regions with lower DSB levels, both Rad51 and Dmc1, a meiosis-specific paralog, are required (3–5).
Following D-loop formation, there are various postulated pathways for the completion of DSB repair. A major pathway, thought to be essential for crossover formation, requires the processing of four-way DNA junctions called Holliday junctions (HJs (8–11)). The nature of the initiating lesion and whether the cell is in mitosis or meiosis influence the mode of HJ processing. Closely spaced double HJs may be dissolved by the concerted action of a RecQ-related helicase and a topoisomerase partner (e.g. Sgs1–Top3 in budding yeast (12)). Notably, in fission yeast, the restart of broken replication forks critically depends on Mus81–Eme1 but not on the dissolution activity of Rqh1 (Sgs1 homolog)–Top3 (13). This likely reflects the formation of single, not double, HJs (or an HJ precursor) during restart of replication (12), because single HJs cannot be dissolved by a helicase and topoisomerase. Endonucleolytic HJ resolution facilitates crossing over between homologous chromosomes (homologs) and dominates during meiosis (8,14,15), when single HJs accumulate in mus81 mutants (8,15). Dissolution of double HJs without associated crossover seems to be the major pathway in mitotic cells of some species (12) and seems to play a role during meiosis in budding yeast (16).
HJs arising through HR-mediated DSB repair can be resolved by endonucleolytic activities such as Mus81–Eme1 ((8,15–19)), Slx1–Slx4 (20,21)) and Yen1 (19,22). Recently, it was shown in the budding yeast Saccharomyces cerevisiae that Mus81–Mms4 and Yen1 endonucleases collaborate during meiosis to resolve HJs and that their activity is carefully regulated during the cell cycle (19). Mms4 (Eme1 homolog) is phosphorylated and then hyperactivated during meiosis I, whereas Yen1 activity is inhibited until meiosis II.
Mus81–Eme1 is critical in the fission yeast Schizosaccharomyces pombe where it is the only known complex involved in meiotic HJ resolution (8,15,23). A Yen1 ortholog has not been identified in the S. pombe genome, and Slx1–Slx4 has no detectable defect in meiotic recombination (our unpublished data). The Mus81–Eme1 heterodimer is required for HR repair at stalled or broken replication forks and is also critical during mitotic (13,24,25) and meiotic HR (15,26). In vitro, Mus81–Eme1 can cleave various substrates that mimic stalled replication forks and nicked and intact HJs (15,27,28). Mus81–Eme1 not only has a binding and cleavage preference for nicked HJs but also has a robust cleavage activity on intact HJs (29). S. pombe meiotic cells deleted for mus81 accumulate single HJs, as shown by 2D gel analysis and electron microscopy (8). The accumulation of HJs in mus81Δ cells results in severe meiotic defects (15) and sensitivity to the topoisomerase I poison camptothecin (CPT (30)); these phenotypes can be partially suppressed by expression of the Escherichia coli HJ resolvase RusA. Therefore, RusA has been used extensively, and is widely accepted, to identify the structures accumulated in various DNA repair mutants (15,26,30–32).
As described above, the last decade has seen major advances in the identification of activities most proximally involved in the processing of joint molecules (JMs) formed during HR. However, currently, little is known about how JM processing enzymes are recruited to their substrates, and whether their activities are facilitated by additional chromatin-associated factors. Interestingly, in this regard, others and we have identified and characterized the Smc5–Smc6 holocomplex, which seems to play multiple roles in HR (33). The octameric Smc5–Smc6 complex is structurally related to the cohesin and condensin complexes but, uniquely, contains two subunits, Nse1 and Nse2, which exhibit E3 ligase activity for ubiquitin and SUMO (Small Ubiquitin-like Modifier), respectively (33,34). In addition to Smc5, Smc6, Nse1 and Nse2, the complex contains a melanoma antigen-domain protein Nse3, a kleisin-like protein Nse4, and two armadillo/Huntington, Elongation Factor 3, PR65A, TOR repeat proteins, Nse5 and Nse6. Notably, unlike in budding yeast (33), nse5Δ and nse6Δ mutants are viable and display indistinguishable sensitivities to all tested DNA-damaging agents (35). In addition, Nse5 and Nse6 form a stoichiometric heterodimer when purified from insect cells (35) or bacterial cells (our unpublished data), and furthermore, nse5Δ nse6Δ double mutants are no more sensitive to UV irradiation than either single mutant (35). Thus, Nse5 and Nse6 function as an obligate heterodimer.
Reflecting key HR roles of the complex, Smc5–Smc6 in budding yeast and human cells is loaded near an enzymatically induced DNA DSB and is important for HR-mediated repair of the break (36–38). In addition, smc5–smc6 mutation causes hypersensitivity to ionizing radiation (IR)-induced DSBs, which is not additive when combined with a rad51 deletion; smc5–smc6 mutants, like rad51Δ mutants, fail to restore chromosome integrity following IR (35,39–43). Intriguingly, the Smc5–Smc6 complex has been implicated in the processing of HR intermediates or suppression of their formation or both (reviewed in (33,44)). For example, smc5–smc6 hypomorphs, including nse2 E3 SUMO ligase-deficient cells, accumulate Rad51-dependent HR intermediates following replication stress, indicating a concerted action of the entire complex in the productive completion of HR (35,45–54).
What is the physical nature of the recombination structures that form when Smc5–Smc6 function is compromised? In budding yeast, a heterogeneous group of hemicatenane and converged replication fork structures arise during replication stress in Smc5–Smc6-deficient mitotic cells (45,47,48,55). These DNA structures appear similar to those that accumulate in sgs1Δ mutant cells, and thus, it has been proposed that Sgs1 acts in concert with Smc5–Smc6 and sumoylation in the removal of such linkages during vegetative growth (45).
Here, we identify a critical role for fission yeast Nse5–Nse6 in chromosome disjunction during meiosis. We use both genetic and physical assays to determine the underlying cause of meiotic failure in nse6Δ cells, the stage at which Nse5–Nse6 function is required, and the DNA structure(s) that prevent(s) chromosome segregation at the first meiotic division (MI) in nse6Δ cells. Our analyses demonstrate that Nse5–Nse6 acts after programmed meiotic DSB formation to drive the timely resolution of JMs whose persistence in nse6Δ cells causes the profound MI defects. The JMs observed in nse6Δ cells resemble the HJs that accumulate in mus81Δ cells, providing the first direct evidence for a role of the Smc5–Smc6 complex in HJ resolution. Although Nse5–Nse6 and Mus81 have some distinct roles, both proteins are essential to avoid an accumulation of JMs that impede meiosis. Here, we define Nse5–Nse6 as a novel regulator of nuclear division, which may act directly in the HJ resolution mechanism or by attracting the Smc5–Smc6 complex to its needed sites of action—mitotic or meiotic HJs.
Unless otherwise indicated, all NBY strains are ura4-D18 and leu1-32. pREP1-RusA, pREP1-RusA-D70N and pREP1 were described in (15). pE119, containing LEU2 and GST, was used as a control vector (57). The minichromosome assay using Ch16-MHH, Ch16-MGH and pREP81X-HO:LEU2 has been described in (58). Mutations other than mating type and commonly used auxotrophies are described in previous studies: mus81::kanMX6 (15), nse6::kanMX6 (35), prh1::hphMX6 (this study), rad51::ura4+ (59), dmc1::hphMX6 (this study), rec12-152::LEU2 (60), mbs1-24 and mbs1-25 (61), bub-1243, vtc4-1104 (9), lacO::lys1 lacI-GFP::arg3+ (62) and nse6::natMX6 (the kanMX6 marker in strain NBY835 (35) was switched to natMX6 as described in (64)). In Table 1 ‘:’ indicates marked by the following gene, and ‘::’ indicates that the preceding gene is replaced by the following marker.
Polymerase chain reaction (PCR) was used to create the hphMX6 cassette flanked by genomic DNA, to enable integration into the middle of the 1.2 kb intergenic region between the prh1 and SPAC2G11.10c loci. The hphMX6 cassette from pCR2.1 hphMX6 (63) was amplified in two steps, using a combination of the primers shown in Table 2, as described in (63,64). A stable transformant was obtained with the hphMX6 cassette integrated between prh1 and SPAC2G11.10c loci, approximately 70kb to the right of the mbs1 meiotic DSB hotspot on chromosome I (61,65).
PCR was used to create the hphMX6 cassette flanked by genomic DNA, to replace the dmc1 gene. The hphMX6 cassette from pCR2.1 hphMX6 (63,64) was amplified in two steps, using a combination of the primers shown in Table 2, as described in (63,64).
Cells were cultured in repressive (+ thiamine; B1) medium (EMM2), and a sample was plated onto non-selective medium (EMM2) for the 0 h time point. Cells were then cultured in repressive or de-repressive (−thiamine) media for 48 h and plated onto non-selective media. Colonies were counted after 3 days growth at 30°C. Replica plating onto media that contained hygromycin (or kanamycin for the rad51Δ mutant), or lacked either adenine or histidine, allowed calculation of the frequency of marker loss. Data are means and standard error of the mean from three independent assays, as described in (58).
Cells from equal volumes (10µl) of each parental haploid culture were mixed, washed, resuspended in water and plated onto supplemented sporulation agar plates (66). After 2 days at 25°C, the cell–ascus mixture was observed by microscopy or processed further for viability assays. Meiotic 4′,6-diamidino-2-phenylindole (DAPI) staining was performed essentially as described in (15). Zygotic asci were fixed in methanol (−80°C) for 10 min and then washed three times with phosphate-buffered saline (1× PBS). The fixed asci were treated with Zymolyase 100T (0.1mg/ml) for 10 min at 37°C. Asci were collected by centrifugation and resuspended in 0.1% Triton X-100 for 2 min at room temperature. Asci were then washed three times (in 1× PBS) and resuspended in a drop of 1× PBS containing 0.5μg/ml DAPI. Asci were photographed using a Nikon Eclipse E800 microscope equipped with a Photometrics Quantix CCD camera. Images were acquired with IPlab Spectrum software (Signal Analytics Corporation).
For viability assays, the cell–ascus mixture was suspended in 1 ml of H2O and treated with 2% glusulase overnight at room temperature. Addition of ethanol to 30% for 1 h killed remaining vegetative cells, essentially as described in (67). To assess spore viability, spores were counted using a hemocytometer, diluted and plated onto the appropriate media. Spore viability was scored following incubation for 5–6 days at 30°C. For meiotic recombination assays, an appropriate dilution of spores was plated onto non-selective (YES) media and replica plated to appropriate test media: EMM2 (66)±adenine for ade7-152;±leucine for leu1-32;±lysine for lys4-95;±histidine for his4-239;±uracil for ura1-61 and YES±hygromycin (100μg/ml) for hphMX6.
pat1-114 strains were thermally induced for meiosis, and DNA was extracted from cells embedded in agarose plugs (68). The DNA was digested with appropriate restriction enzymes and analyzed by gel electrophoresis and Southern blot hybridization (8,69) and quantification (9).
The sensitivity of DNA to S1 nuclease (Roche) and to RuvC (gift of Ken Marians, Memorial Sloan-Kettering Cancer Center) was assayed by digesting the DNA, in plugs, with PvuII restriction enzyme, followed by washing twice with TE (Tris-EDTA; 10 mM Tris–HCl, pH 7.0, 1 mM ethylenediaminetetraacetic acid (EDTA) pH 8.0) and twice with nuclease S1 buffer (33 mM sodium acetate, 100 mM NaCl, 0.033 mM ZnSO4, pH 4.5 at room temperature) or RuvC buffer (50 mM Tris–acetate pH 7.5, 10 mM Mg(OAc)2, 1 mM DTT, 50μg of BSA per ml); duration for each wash was 20 min at room temperature. The plugs were incubated in 100μl of buffer with the indicated amount of S1 for 2 h at 4°C or RuvC for 4 h at 4°C, then for 1 h at 37°C (S1) or for 4 h at 55°C (RuvC) to facilitate branch migration of HJs embedded in agarose; the digestions were stopped by adding 2.5μl of 500 mM EDTA pH 8.0 and placing the incubation tubes on ice for 10min. 2D electrophoresis was performed, and recombination intermediates were detected at mbs1 as described in (8).
The budding yeast and human Smc5–Smc6 complexes have been implicated in the efficient repair of an enzymatically induced DNA DSB (36–38). To determine whether the fission yeast Nse5–Nse6 functionally interdependent heteromeric complex (35) plays an analogous role, we tested the DSB repair capacity of nse6Δ cells using a derivative of the non-essential minichromosome Ch16, which contains the MATa target site for the S. cerevisiae HOmothallic switching (HO) endonuclease (see schematic in Figure 1A, described in (58)). Given the relatively mild IR sensitivity of fission yeast nse6Δ cells versus the extreme sensitivity of rad51Δ cells (35), it was surprising that both nse6Δ and rad51Δ cells were highly defective in the HR-based repair of the HO-induced break. Unlike wild-type cells, which repaired the majority of the HO-induced DSBs by gene conversion, the nse6Δ and rad51Δ mutants instead lost nearly all of the HO cut minichromosomes (Figure 1B and C (58,70)). This result indicates that in fission yeast, Smc5–Smc6, is required for the repair of enzymatically induced DSBs.
To test for an additional role of Nse6 in mitotic DNA break repair, we determined the sensitivity of nse6Δ mutants to CPT, which stabilizes topoisomerase I–DNA covalent complexes and can induce fork breakage during replication (30). The nse6Δ mutant was sensitive to CPT, although not as sensitive as mus81Δ or the mus81Δ nse6Δ double mutant [(71) and Figure 1D]. Although the expression of E. coli RusA HJ resolvase is mildly toxic to wild-type cells, as observed previously (72), the CPT sensitivities of both nse6Δ and mus81Δ cells were suppressed by RusA expression (Figure 1D). Notably, the nse5Δ mutant has a CPT sensitivity similar to that of nse6Δ, which is also rescued by RusA expression (Figure 1E). This result indicates a role for Nse5–Nse6, like Mus81, in HJ resolution. The mus81Δ nse6Δ double mutant was also suppressed by expression of RusA, although not as well as the single mutants, indicating that Nse5–Nse6 and Mus81 have distinct and overlapping roles. Although nse6Δ behaves like rad51Δ in the HO-induced DSB assay (Figure 1B and C (58)), the hypersensitivity of rad51Δ cells to CPT was not suppressed by RusA expression (Figure 1F).
Due to the similarity of the mitotic phenotypes of nse5Δ, nse6Δ and mus81Δ cells, we tested the role of Nse5–Nse6 in meiosis, an HJ-generating process that is heavily dependent on Mus81–Eme1 resolution activity (8,15,23). Wild-type meiosis produces asci with four haploid spores, with DNA present in each spore (Figure 2A). Asci from nse5Δ and nse6Δ mutant crosses were largely devoid of well-defined spores but infrequently contained one single large spore (Figure 2A), much like mus81Δ asci (15). This result shows that Nse5–Nse6 plays an important role in meiosis.
If Nse5–Nse6, like Mus81–Eme1, promotes JM resolution, then preventing the early stages of meiotic recombination should render Nse5–Nse6 unnecessary for meiotic nuclear division. For these tests, we first eliminated Rec12, which is essential for meiotic DSB formation and recombination (68). Asci from rec12Δ crosses were very heterogeneous: Some asci had four equal-sized spores like wild-type, some lacked spores or had only one like nse6Δ and many had two spores (dyads) like rec12Δ (Figure 2B; (7,15)). As anticipated, spore viability from homozygous nse5Δ or nse6Δ crosses was extremely low: Only ~0.5% of spores germinated and produced colonies, compared to ~80% for wild-type (Figure 2B (15)). Notably, both rec12Δ and rec12Δ nse6Δ crosses yielded similar spore viabilities of ~20%, and, furthermore, their ascus morphologies were indistinguishable (Figure 2B). The same was true for rec12Δ and rec12Δ mus81Δ crosses. The rec12Δ nse6Δ and rec12Δ mus81Δ double mutants had indistinguishable defects from those observed in a rec12Δ deletion, indicating that rec12Δ is epistatic to both nse6Δ and mus81Δ in meiosis.
The morphology of asci produced by crosses of nse6Δ, mus81Δ and mus81Δ nse6Δ double mutants were indistinguishable (Figure 2B). However, the viability of spores from mus81Δ nse6Δ crosses was ~8-fold lower than that of spores from crosses of the single mutants (Figure 2B). Deletion of rec12 rescued the mus81Δ nse6Δ meiotic ascus phenotype, but only partially rescued the low spore viability phenotype, between 1 and 2% spore viability. Nevertheless, the spore viability in the mus81Δ nse6Δ rec12Δ triple mutant was about 1000-fold higher than that in the mus81Δ nse6Δ double mutant (P<0.0001). These results indicate that Nse5–Nse6, like Mus81, acts in meiotic recombination after DSB formation by Rec12 (i.e. in the repair of DSBs).
Next, we exploited the aberrant ascus morphology of nse6Δ cells to determine whether Nse5–Nse6 acts before or after strand invasion promoted by the RecA homologs Rad51 and Dmc1 (4,5). Cells lacking Rad51 fail to repair many meiotic DSBs (73) but can produce asci containing four discrete, although mostly inviable, spores (15), presumably because there are no physical linkages, such as HJs, between the chromosomes to prevent their segregation (Figure 2C). Although not essential for meiotic progression and abundant formation of viable spores, Dmc1 is required for normal levels of meiotic recombination in most genetic intervals (4,9). In contrast to rad51Δ or dmc1Δ single mutants, four-spore asci were not observed in the nse6Δ single or rad51Δ nse6Δ or dmc1Δ nse6Δ double mutants (Figure 2B and C). However, the dmc1Δ rad51Δ double and nse6Δ dmc1Δ rad51Δ triple mutants produced similar asci that sometimes contained four spores, which are not observed for the nse6Δ single mutant (Figure 2B and C). Thus, suppression of the nse6Δ phenotype (to a level equivalent to that of the dmc1Δ rad51Δ double mutant) requires elimination of both strand-exchange proteins. These results indicate that Nse6 acts after strand exchange by Rad51 and Dmc1, perhaps during JM resolution.
Next, we tested the impact of absence of Nse6 on meiotic recombination in three genomic intervals. In the large ade7–leu1 interval, frequency of crossover among viable spores from a wild-type cross was 41.5%, not significantly different from the 40.8% in nse6Δ mutants (Figure 3A). In two shorter intervals, lys4–his4 and ura1–prh1 (which contains the mbs1 DSB hotspot used in DNA analyses below (61)), frequencies of crossover were reduced by a factor of about 2 in nse6Δ mutants, relative to wild-type (Figure 3A), a statistically significant reduction (P<0.0001). However, in contrast to the reductions by factors of 20–100 in mus81Δ cells (8,26), the meiotic crossover defects of nse6Δ cells are mild.
Through genetic methods, we can measure crossovers only in the ~0.5% of spores that are viable after nse6Δ meiosis; thus, our observed frequencies may overrepresent the frequencies in most of the cells. To circumvent this limitation, we physically measured the crossovers between two markers closely flanking the DSB hotspot mbs1 in the entire meiotic population, as previously analyzed in mus81Δ mutants (8). The crossover-specific product R2 (Figure 3B) was quantified in wild-type, mus81Δ, nse6Δ and nse6Δ mus81Δ strains at 7 h after induction of synchronous meiosis. Consistent with previous data (8), there was a severe defect, a reduction by a factor of 7, in crossover formation in mus81Δ and nse6Δ mus81Δ cells (Figure 3B). Importantly, in nse6Δ cells, crossover formation was reduced by a factor of about 2, as in the genetic assay for ura1–mbs1–prh1 recombinants (Figure 3A), suggesting that Nse5–Nse6 promotes but is not absolutely essential for meiotic crossover formation (Figure 3B).
The similarities of the vegetative and meiotic phenotypes of mus81Δ and nse6Δ led us to investigate whether DNA JMs accumulate during meiosis in nse6Δ, as they do in mus81Δ (8). Synchronous meiosis was induced in pat1-114 nse6Δ and nse6Δ mus81Δ diploids, and the formation of branched DNA structures, indicative of both replication and recombination intermediates, was analyzed via 2D gel electrophoresis as previously performed for mus81Δ (8). Branched DNA structures produced during replication were visible at 2.5 h and decreased toward 3 h (Figure 4A and B), the time at which DNA content shifts from 2n to 4n as measured by flow cytometry ((8); our unpublished data). After replication was complete and DSB formation had begun (Figure 4C), the amount of X-shaped DNA increased again at 4 h, persisted at 4.5 h and disappeared after 5 h in the wild-type strain (Figure 4A and B), as expected for the formation and resolution of HJs. In the nse6Δ strain, the recombination intermediates were formed with the same timing as in wild-type, but they did not disappear; the same was true for mus81Δ and nse6Δ mus81Δ mutants, although the double mutant showed a slight delay in JM formation (Figure 4A and B; (8)). As expected, early but not late appearing JMs were detected in an nse6Δ rec12Δ haploid double mutant (haploid and diploid analyses are indistinguishable (74) and our unpublished data). This result indicates that, as in mus81Δ mutants (8), the early arising JMs are replication derived, and the late arising JMs are recombination intermediates and are consistent with the meiotic phenotype of nse6Δ depending on DSB formation by Rec12 (Figure 2B). Thus, both nse6Δ and mus81Δ mutants accumulate recombination intermediates with a similar timing during meiosis.
Next, we determined whether the JMs that accumulate in the nse6Δ mutant specifically form between sister chromatids or between homologs, and whether they are single HJs, as they are in a mus81Δ mutant (8), and not double HJs, a plausible alternative. A strain with heterozygous restriction site mutations that flank the DSB hotspot mbs1 was used to assay the relative interhomolog (IH) and intersister (IS) HJs formed during DSB repair. As was the case in the mus81Δ strain (8,9), IS HJs outnumbered IH HJs by ~3.5:1 (Figure 5A). To distinguish single versus double HJs, the DNA in the agarose gel after the first dimension of electrophoresis was heated to 65°C to promote branch migration. The unequal length arms of homologous DNA in single HJs, which have two recombinant DNA strands, prevent the junction from migrating past the ends of the arms; the absence of recombinant DNA strands in double HJs and the equal length homologous arms of IS HJs allow their junctions to readily migrate off the end (8). When the DNA was heated, the IS HJs readily branch migrated and disassociated into the linear forms, whereas the IH HJs were resistant and remained intact (Figure 5B). This stability indicates that the IH-branched DNA structures are single HJs. This is the same behavior observed previously for branched DNA that accumulates in the mus81Δ mutant (8). These physical analyses further support a role for Nse6 in the resolution of DNA JMs during meiosis.
The X-shaped DNA structures seen in the 2D gel assays could be either HJs or a related four-stranded DNA structure, a hemicatenane, with two single DNA strands, or two pairs of strands, interwound. We first tested the sensitivity of these structures to the E. coli HJ resolvase RuvC (75). The JMs of the nse6Δ and mus81Δ strains were similarly sensitive to RuvC (Figure 6A and B), indicating that at least some of the X-shaped structures formed in both mus81Δ and nse6Δ strains are the same and are HJs. However, the JMs of the nse6Δ mus81Δ strain were largely resistant, suggesting that these X-shaped structures are not HJs but might be related structures such as hemicatenanes.
Although structurally similar to HJs, hemicatenanes have exposed ss DNA that would be susceptible to S1, a nuclease more active on ss DNA than on double-stranded DNA (76). To test this possibility, meiotic DNA prepared at 4.5 or 5h after induction of mus81Δ or nse6Δ strains was treated with S1 before 2D gel electrophoresis. The accumulated branched DNA structures from the mus81Δ strain were insensitive to S1 treatment, but those from the nse6Δ and nse6Δ mus81Δ strains were partially sensitive (Figure 6A and B). We propose that both HJs and an S1-sensitive form of DNA, such as hemicatenanes, accumulate in nse6Δ mutants but only HJs accumulate in mus81Δ mutants during meiosis (Figure 6A and B).
Our biochemical analyses provide evidence that nse6Δ mutants accumulate meiotic HJs, as observed in mus81Δ mutants (8,15). For in vivo validation of this conclusion, we used the RusA resolvase to probe the nature of the meiotic impediment in nse5Δ and nse6Δ cells. Our laboratories and others have shown that heterologous expression of E. coli RusA reduces the level of HJs that accumulate either on treatment with various DNA damaging agents or during meiosis in various DNA repair mutants (15,26,30–32). We expressed RusA and its catalytically inactive mutant, D70N, in nse6Δ cells and analyzed both spore formation and viability. Strikingly, RusA expression significantly rescued the meiotic defects of nse6Δ cells, yielding asci that often contained a wild-type complement of four spores (Figure 6C). This rescue required RusA endonuclease activity, as the RusA-D70N catalytically inactive mutant provided no benefits (Figure 6C). The spore viability was increased more than 20-fold when RusA was expressed in nse6Δ cells compared to an empty vector control (Figure 6D). Spore viability was also increased in nse5Δ cells (Figure 6D). Likewise, although more dramatic due to their initial lower viability, mus81Δ spore viability was increased more than 800-fold by RusA expression. Interestingly, spore viability was around 1.5% in nse5Δ and 3% in nse6Δ mutants expressing RusA, which was significantly lower (P<0.005) than in mus81Δ-expressing RusA (7%; Figure 6D). This result is consistent with the accumulation of both HJs and an S1-sensitive form of DNA, such as hemicatenanes, in nse5Δ and nse6Δ meiosis, but only HJs in mus81Δ meiosis (Figure 6A and B). RusA expression did not so strongly improve the ascus morphology or spore viability in the nse6Δ mus81Δ double mutant (Figure 6C and D), which is consistent with the RuvC insensitivity of DNA isolated from these cells (Figure 6A and B). Overall, the remarkably similar suppression of both the ascus morphology and spore viability defects in nse5Δ or nse6Δ and mus81Δ by RusA endonuclease suggests that Nse5–Nse6 plays a role in facilitating the HJ resolvase activity of Mus81–Eme1.
Failure of the HR pathways underlies many human diseases including cancer and can cause birth defects through aberrant meiotic chromosome segregation. Here, we have identified a novel function for Nse5–Nse6 of the Smc5–Smc6 complex in processing mitotic and meiotic HR intermediates. Genetic and physical analyses indicate that nse6Δ mutants accumulate JMs resembling the HJs that accumulate in mus81Δ during meiosis. Thus, to our knowledge, Nse5–Nse6 is only the second factor in fission yeast required for the endonucleolytic processing of HJs, besides Mus81–Eme1, which has the catalytically active site for HJ resolution.
The key data supporting a function for Nse5–Nse6 in the resolution of JMs are as follows: (i) like mus81Δ mutants (15), nse6Δ mutants form few viable spores (Figure 2), due to the failure of chromosome segregation; (ii) like mus81Δ (15), rec12Δ is epistatic to nse6Δ in meiosis, demonstrating that Nse5–Nse6 acts after DSB formation (Figure 2); (iii) the double nse6Δ mus81Δ mutant also forms few viable spores and is suppressed, at least partially, by rec12Δ, indicating that Mus81 and Nse6 act in the same or closely related steps of meiosis (Figure 2); (iv) the rad51Δ dmc1Δ combination is epistatic to nse6Δ, demonstrating that Nse5–Nse6 acts after the formation of JMs (Figure 2); (v) JMs accumulate in meiotic nse6Δ and nse6Δ mus81Δ cells, and these JMs are temporally, genetically and electrophoretically indistinguishable from the HJs that accumulate in mus81Δ cells (Figures 4 and and5,5, (8)); (vi) at least some of these JMs are sensitive to E. coli RuvC HJ resolvase (Figure 6) and to RusA HJ resolvase, because, as for mus81Δ cells (15), expression of RusA partially rescues the meiotic defects of nse6Δ cells (Figure 6C and D) and (vii) crossovers are modestly reduced in nse6Δ cells, indicating that Nse5–Nse6 is important but not absolutely essential for HJ resolution, which requires Mus81–Eme1 (Figure 3).
In addition to HJ resolution, Nse5–Nse6 appears to have a second role because (i) the double nse6Δ mus81Δ mutant grows more slowly and is more CPT sensitive than either single mutant (Figure 1D); (ii) the double mutant is not quite as well suppressed to CPT resistance by expression of RusA as the single mutants (Figure 1D); (iii) unlike mus81Δ, nse6Δ is defective for mitotic DSB repair, as is rad51Δ (Figure 1); (iv) crossovers are formed, although at reduced level compared to wild-type, in nse6Δ but not in mus81Δ or nse6Δ mus81Δ mutants (Figure 3) and (v) the JMs that accumulate in nse6Δ and nse6Δ mus81Δ meiotic cells are at least partially sensitive to S1 nuclease, whereas those that accumulate in mus81Δ meiotic cells are not (Figure 6A and B), suggesting that Mus81–Eme1 and Nse6 have distinct, and overlapping, roles.
To reconcile these observations, we propose that in meiotic cells Nse5–Nse6 stimulates the resolution of HJs by Mus81–Eme1 and the resolution of other structures, such as hemicatenanes, which may arise during mitotic growth or during meiosis in the absence of Mus81–Eme1. These structures may arise during mitotic replication, as suggested by others (45,77), thereby partially accounting for the mitotic phenotypes of nse6Δ mutants. The structure of the JMs that accumulate in nse6Δ and nse6Δ mus81Δ mutants is not entirely clear, although those that accumulate in mus81Δ mutants are clearly single HJs (8). The JMs that accumulate in nse6Δ mutants are sensitive to both S1 nuclease and RuvC HJ resolvase, whereas those that accumulate in mus81Δ mutants are sensitive to RuvC and those in nse6Δ mus81Δ mutants are sensitive to S1 nuclease (Figure 6A and B). The stimulation of Mus81 HJ resolvase activity by the Nse5–Nse6 complex may be direct or indirect.
We propose that in nse6Δ mutants, Mus81–Eme1 slowly resolves HJs while some are converted into another structure and that in nse6Δ mus81Δ mutants, most or all of the HJs are converted into this structure. This proposal is consistent with the slight reduction in meiotic crossover frequency, among the few viable spores that arise, in nse6Δ mutants but strong reduction in mus81Δ mutants (Figure 3A (8,23,26)); with the slight reduction in total crossover DNA in nse6Δ mutants but strong reduction in mus81Δ and nse6Δ mus81Δ mutants (Figure 3B (8)); and with the suppression of nse6Δ, but not nse6Δ mus81Δ, by expression of the RusA HJ resolvase (Figure 6C and D).
The resistance to branch migration of the IH JMs in nse6Δ mutants suggests that these JMs, between heterozygous restriction sites, are single HJs with recombinant length strands (Figure 5B). This is because either double HJs, which have parental length strands, or hemicatenanes with either recombinant or parental length strands should dissociate into separate duplexes on heating (e.g. (8,76,78)). IS JMs would dissociate in any case, as observed. IH and IS JMs may differ in structure, but their low level has precluded our determining their sensitivity to S1 nuclease and RuvC. Although the S1 nuclease sensitivity is consistent with some JMs being hemicatenanes, to our knowledge, this structure has not been clearly demonstrated to arise in cells, for example by electron microscopy or comparison with synthetic DNA molecules. Further investigation is required to establish the structure of the population of non-HJ containing JMs in the absence of Nse6.
As Nse5–Nse6 acts as part of the Smc5–Smc6 complex, which has multiple roles in chromosome metabolism (44), it is no surprise that Nse5–Nse6 has multiple roles. One tempting hypothesis is that during meiosis, the primary role of Nse5–Nse6 is to stimulate the Mus81–Eme1 HJ resolvase and that during mitotic growth it plays both this and a second role. During both stages of the life cycle, expression of RusA suppresses at least partially the nse6Δ phenotype (Figures 1D and and6C6C and D), indicating that HJ resolution is stimulated by Nse5–Nse6 in both stages. The second role of Nse5–Nse6 might be to regulate any of the multiple functions of the Smc5–Smc6 complex (44). Further investigation is required to elucidate this function, but the requirement for Rad51 and Nse6 but not Mus81–Eme1 in mitotic DSB repair (Figure 1) suggests that Nse5–Nse6 is also important for the synapsis phase of mitotic DSB repair.
Although the Nse1 and Nse2 E3 ligase activities, like Nse5–Nse6, facilitate mitotic DNA repair (79–81), neither of these E3 ligases is required for meiotic nuclear division and recombination [(82) our unpublished data]. Thus, although the mitotic functions of Nse1, Nse2 and Nse5–Nse6 have not been dissected, in meiosis, Nse5–Nse6 acts independently of posttranslational modifications catalyzed by Nse1 and Nse2. Nevertheless, on the basis of our previous analyses showing that hypomorphic mutants of the essential Smc5–Smc6 subunits exhibit catastrophic meioses (83), we propose that Nse5–Nse6 acts in conjunction with the Smc5–Smc6 holocomplex to execute its meiotic HR role.
Budding yeast smc5–smc6 mutations also disrupt meiotic nuclear division (84), but the underlying defects appear strikingly different from those of nse6Δ fission yeast. Key differences are that a spo11 (rec12 homolog) mutation is not epistatic to an smc5–smc6 mutation, and crossovers are not affected in smc6 temperature-sensitive (Ts) mutant cells (84). Thus, the authors concluded that the crucial role of Smc5–Smc6 is executed during premeiotic S phase in budding yeast and not after the initiation of meiotic recombination. It is unclear whether this reflects real differences in the functions of Smc5–Smc6 between species or if the unavoidable disruption of both the essential and repair roles in smc6 (Ts) budding yeast masks functions analogous to those of fission yeast Nse5–Nse6.
HJ resolution must be carefully controlled, both during mitotic growth and in meiosis. When a DNA strand lesion blocks mitotic replication, the fork can regress and form a structure whose center is identical to that of an HJ. Were it resolved by Mus81–Eme1, a DSB would be formed and require further processing to allow completion of replication. Alternatively, the regressed fork can migrate back to the original position and allow immediate continuation of replication. Thus, Mus81–Eme1 may be kept inactive during this time. If strand exchange between sisters or homologs forms an HJ that is present at the time of mitosis, HJ resolution would appear to be the most expedient means to allow chromosome segregation, and Mus81–Eme1 may be activated at this time [e.g. (19)]. Similarly, during meiosis, many dozens of HJs must arise to account for the ~45 crossovers in a fission yeast cell (85), and Mus81–Eme1 is clearly highly active in meiotic cells. We surmise that Nse5–Nse6, likely as part of the Smc5–Smc6 complex, regulates directly or indirectly the activity of Mus81–Eme1 allowing it to function at the proper time and place to maintain chromosome integrity and cell viability.
Scholar Award from the Leukemia and Lymphoma Society (to M.N.B.); National Institutes of Health (NIH) of the United States of America [GM068608 and GM081840 to M.N.B. and GM032194 to G.R.S.]. Funding for open access charge: NIH, United States.
Conflict of interest statement. None declared.
The authors thank Ken Marians for a generous supply of RuvC protein, Tim Humphrey for his kind gift of the minichromosome assay strain, Sue Amundsen for helpful comments on the article and Jay Patel for media preparation.