Abstract
The MMS4 gene of Saccharomyces cerevisiae was originally identified due to its sensitivity to MMS in vegetative cells. Subsequent studies have confirmed a role for MMS4 in DNA metabolism of vegetative cells. In addition, mms4 diploids were observed to sporulate poorly. This work demonstrates that the mms4 sporulation defect is due to triggering of the meiotic recombination checkpoint. Genetic, physical, and cytological analyses suggest that MMS4 functions after the single end invasion step of meiotic recombination. In spo13 diploids, red1, but not mek1, is epistatic to mms4 for sporulation and spore viability, suggesting that MMS4 may be required only when homologs are capable of undergoing synapsis. MMS4 and MUS81 are in the same epistasis group for spore viability, consistent with biochemical data that show that the two proteins function in a complex. In contrast, MMS4 functions independently of MSH5 in the production of viable spores. We propose that MMS4 is required for the processing of specific recombination intermediates during meiosis.
THE key step in meiosis is the segregation of homologous chromosomes to opposite poles at the first meiotic division (MI). For proper chromosome disjunction at MI to occur, nonsister chromatids of each pair of homologs must be connected by crossovers (manifested cytologically by chiasmata; Bascom-Slacket al. 1997). Creating crossovers functional for disjunction requires coordination between the recombination machinery and a tripartite proteinaceous structure, called the synaptonemal complex (SC), which is formed between homologous chromosomes.
Many steps in the pathway of recombination predicted by the double-strand-break (DSB) model (Szostaket al. 1983) have now been detected at the molecular level (reviewed in Smith and Nicolas 1998). Furthermore, gene products that are required for many of these steps have been identified. Meiotic recombination is initiated in Saccharomyces cerevisiae by DSBs introduced into the DNA by a topoisomerase-like protein, Spo11p (Klapholzet al. 1985; Bergeratet al. 1997; Keeneyet al. 1997). The Spo11 protein is conserved throughout evolution and phenotypes of spo11 mutants in worms, fruit flies, plants, and mice suggest that its function is conserved as well (Dernburget al. 1998; McKim and Hayashi-Hagihara 1998; Baudatet al. 2000; Romanienko and Camerini-Otero 2000; Grelonet al. 2001; Mahadevaiahet al. 2001). In addition to SPO11, several other genes are required for DSB formation, although their roles in this process remain obscure (Smith and Nicolas 1998). The 5′ ends of the breaks are resected to produce 3′ single-stranded tails, a process that requires MRE11, RAD50, and SAE2/COM1 (Caoet al. 1990; McKee and Kleckner 1997a; Nairz and Klein 1997; Prinzet al. 1997; Tsubouchi and Ogawa 1998). The meiosis-specific RecA ortholog Dmc1p then directs invasion of single strands into duplexes of homologous nonsister chromatids (Bishopet al. 1992; Schwacha and Kleckner 1997). The detection of these single end invasion (SEI) intermediates has recently been published (Hunter and Kleckner 2001). Further processing of these recombination intermediates results in the formation of double Holliday junctions (HJs; Schwacha and Kleckner 1995; Allers and Lichten 2001). For crossovers to occur, the two HJs must be resolved in opposite directions. The bias toward crossing over is promoted by a complex formed between two meiosis-specific MutS orthologs, Msh4p and Msh5p (Ross-Macdonald and Roeder 1994; Hollingsworthet al. 1995; Pochartet al. 1997). Finally, HJs must be cut and the appropriate strands ligated so that nonsister chromatids are no longer covalently attached.
Synapsis between homologous chromosomes in yeast begins when sister chromatids condense upon protein cores called axial elements (AEs), which are composed, at least in part, of three meiosis-specific proteins, Mek1p, Red1p, and Hop1p (reviewed by Roeder 1997). Unstable DNA-DNA interactions between homologous chromosomes allow for connections between nonsister chromatids, which are then stabilized by formation of DSBs and consequent recombination (Kleckner 1996). At these sites of axial association, proteins such as Zip3 and Zip2 are bound and appear to nucleate the insertion of Zip1, a structural component of the central region of the SC (Sym and Roeder 1995; Chua and Roeder 1998; Agarwal and Roeder 2000).
It is clear that in yeast there is an intimate connection between synapsis and recombination. Mutants that fail to initiate recombination also fail to form SCs (Roeder 1997). In addition, only those crossovers that occur in the presence of SCs allow proper disjunction of homologs at MI (Engebrechtet al. 1990). In terms of segregation, the formation of the AE appears to play a more crucial role than the formation of the central region. zip1 mutants form AEs that are aligned and connected by axial associations but lack the central region (Symet al. 1993). Crossing over is reduced only two- to threefold with a twofold reduction in spore viability (Sym and Roeder 1994). In contrast, mutations in components of the AEs (HOP1, RED1, and MEK1) result in large decreases in spore viability due to massive nondisjunction (Hollingsworth and Byers 1989; Rockmill and Roeder 1990, 1991; Leem and Ogawa 1992). Crossing over is reduced, but not eliminated, in these mutants (Hollingsworth and Byers 1989; Rockmill and Roeder 1990, 1991; Mao-Draayeret al. 1996). In red1 and hop1 diploids, those crossovers that do occur do not function in promoting proper MI segregation (Rockmill and Roeder 1990; B. Baumgartner and N. M. Hollingsworth, unpublished results).
In an attempt to connect components of the AEs with proteins involved directly in recombination, a two-hybrid screen was performed using MEK1 as bait. MEK1 encodes a meiosis-specific threonine/serine protein kinase (Rockmill and Roeder 1991; Leem and Ogawa 1992). In addition to spore viability, MEK1 is required for wild-type levels of recombination, sister chromatid cohesion, SC formation, correct partner choice, and the meiotic recombination checkpoint (Rockmill and Roeder 1991; Leem and Ogawa 1992; Xuet al. 1997; Bailis and Roeder 1998; Thompson and Stahl 1999). The meiotic recombination checkpoint blocks cells in prophase if they contain aberrant recombination intermediates (Roeder and Bailis 2000). Kinase activity is required for MEK1 function and Red1p has been shown to be a MEK1-dependent phosphoprotein (Bailis and Roeder 1998; de los Santos and Hollingsworth 1999). The phosphorylation of Red1p by Mek1p is required for the meiotic recombination checkpoint as well as for facilitating the interaction of Red1p with Hop1p (Bailis and Roeder 1998; de los Santos and Hollingsworth 1999).
Our screen identified MMS4 as a two-hybrid interacting protein with Mek1p. The mms4 mutant has previously been found to be sensitive to alkylating agents such as MMS, weakly sensitive to ultraviolet (UV) light, and resistant to X rays. In addition, mms4 mutants exhibit a weak mutator phenotype and a sporulation defect (Prakash and Prakash 1977; Xiaoet al. 1998; Mullenet al. 2001). The interaction between MMS4 and MEK1 suggested that MMS4 plays a role in recombination during meiosis. Meiotic phenotypes for mms4 diploids were therefore examined to see if the sporulation defect is the result of aberrant recombination triggering the meiotic recombination checkpoint. Our work supports this idea and suggests that MMS4 is required for the processing of recombination intermediates during meiosis.
MATERIALS AND METHODS
Plasmids: Plasmids were constructed by standard procedures (Maniatiset al. 1982) using the Escherichia coli strain BSJ72.
Yeast strains: Strains are listed in Table 1. Liquid and solid media were as described (Vershonet al. 1992; de los Santos and Hollingsworth 1999). Because ade2 strains do not sporulate well under our conditions, all diploids homozygous for ade2 were converted to Ade+ either by transformation with an ADE2-integrating plasmid or by using ADE2-marked alleles for mutagenesis. Deletions were made in the SK1 haploid strains, RKY1145 and S2683 (de los Santos and Hollingsworth 1999), which were then mated to form diploids. MMS4 was deleted using the mms4Δ::hisG-URA3-hisG allele in pWX1604 (provided by W. Xiao, University of Saskatchewan). NH274 was converted to NH274F by recombining out the URA3 gene in each haploid parent using 5-fluoroorotic acid (Boekeet al. 1984), followed by mating to form the diploid. REC104 was deleted using pNH131 (Hollingsworth and Johnson 1993); RED1 was disrupted using pNH119 (Hollingsworth and Johnson 1993); MEK1 was deleted using pTS1 (de los Santos and Hollingsworth 1999); PCH2 was deleted using pSS52 provided by G. S. Roeder (Yale University; San-Segundo and Roeder 1999); and MSH5 was disrupted using pCH2 (Pochartet al. 1997). The MUS81 gene was deleted by PCR using the method of Longtine et al. (1998). PCR was also used to delete MMS4 from the haploid parents of NKY1551, which were subsequently mated to form NH301.
NH298 was generated by a cross between two slow-sporulating strains, BR1373-6DSFαAA, provided by J. Engebrecht (SUNY, Stony Brook), and 5787-21-4 (Hollingsworth and Byers 1989). SPO11 was deleted using pGB324 (Girouxet al. 1989) provided by G. S. Roeder (Yale University); RED1 was deleted using pNH234 (Wolteringet al. 2000); MEK1 was deleted using pNH169 (Hollingsworth and Ponte 1997); and DMC1 was deleted using pMDE379 (Dresseret al. 1997) provided by M. Dresser (University of Oklahoma). NH298::pBB9, NH246::pRS306, and NH331 were converted to Spo13+ by transformation with pNH20-5 (Hollingsworth and Byers 1989).
Time courses: Cells were sporulated as described in de los Santos and Hollingsworth (1999). SK1 strains were sporulated in 2% potassium acetate while non-SK1 strains were sporulated in SPM (Hollingsworth and Johnson 1993). Meiotic progression was monitored by fixing cells with 3.7% formaldehyde and staining them with 4′,6-diamidino-2-phenylindole (DAPI) as described in Woltering et al. (2000). At 33°, varying numbers of immature asci were observed from experiment to experiment. Therefore, only mature asci were scored when assessing sporulation. For analysis of DSBs and crossovers, DNA was digested in plugs as described in Woltering et al. (2000). The gels were prepared for hybridization and probed as described by McKee and Kleckner (1997b). The DSB fragments and crossover bands were quantitated as described in Woltering et al. (2000). Electron microscopic analysis of spread chromosomes was performed as described in Woltering et al. (2000).
S. cerevisiae strains
RESULTS
mms4 diploids arrest in prophase: To determine whether the sporulation defect of mms4 is due to defects in meiosis as opposed to defects in the ability to form spores, meiotic time-course experiments were performed in the isogenic SK1 diploids NH144 (MMS4/MMS4) and NH274F (mms4/mms4). Meiotic divisions were monitored by counting the number of nuclei in each cell. Binucleate cells have completed MI while tetranucleate cells have completed MII. The time course was performed at 33°, a temperature at which the mms4 sporulation phenotype is most severe in the SK1 strain background (MMS4, 75.6% ± 6.0 vs. 1.3 ± 3.3 for mms4, n = 6). The wild-type cells began to enter meiosis I about 4 hr after shift to sporulation medium and ~70% of the cells had completed at least one division by 10 hr. In contrast, <10% of the mms4 cells progressed through MI at 10 hr (Figure 1A). Scoring the number of nuclei in the mms4 cells after 10–12 hr is technically difficult because the pattern of DAPI staining becomes highly fragmented. Mullen et al. (2001) have shown that mms4 diploids arrest at the mononucleate stage in the W303 background.
The mms4 prophase arrest is dependent upon recombination: The mms4 prophase arrest may result from defects in either DNA replication or recombination. If the arrest is dependent upon recombination, then preventing recombination should suppress the meiotic progression defect. REC104 is a meiosis-specific gene required for meiotic recombination (Galbraith and Malone 1992). REC104 was deleted from the mms4 SK1 diploid and the resulting strain (NH349) was assayed for meiotic progression at 33°. Both the rec104 and rec104 mms4 diploids entered meiosis I between 2 and 4 hr after transfer to sporulation medium and produced equivalent numbers of tetranucleate cells (Figure 1A). The diploids containing rec104 progressed through the first meiotic division earlier than the wild-type strain, consistent with previous results (Galbraithet al. 1997).
The mms4 prophase arrest requires components of the meiotic recombination checkpoint: The dependence of the mms4 prophase arrest on recombination suggests that aberrant or unresolved recombination intermediates are being formed in the absence of MMS4, thereby triggering the meiotic recombination checkpoint (Roeder and Bailis 2000). To test this hypothesis, mms4 was combined with various checkpoint mutants for epistasis analysis in the SK1 background. PCH2 is a meiosis-specific gene required for the prophase arrest/delay exhibited by mutants such as zip1 and dmc1 (San-Segundo and Roeder 1999). At 33°, deletion of PCH2 in our SK1 strains resulted in a delay in the onset of meiosis of ~2 hr. Although by 12 hr only 58% of the pch2 cells had completed at least one division (compared to 90% for NH144; Figure 1B), by 26 hr, 61% of the pch2 cells had formed mature asci, nearly equivalent to the 70% observed for the wild type. The pch2 mutation partially rescued the meiotic progression defect of mms4 (Figure 1B). By 12 hr, only 5% of the mms4 cells had progressed beyond MI, compared to 47% for the mms4 pch2 diploid. The mms4 sporulation defect was only weakly suppressed by pch2, however (0% for mms4 vs. 4% for mms4 pch2).
MEK1 is believed to activate the meiotic recombination checkpoint by phosphorylating Red1p in response to the formation of meiotic DSBs (Bailis and Roeder 2000). Deletion of MEK1 or RED1 has been shown to suppress the prophase arrest or delay conferred by a variety of mutants, including dmc1, zip1, and sae2/com1 (Xuet al. 1997; Wolteringet al. 2000). Meiotic time-course experiments in the SK1 background at 33° demonstrated that both red1 and mek1 completely suppress the meiotic progression defect of mms4 (Figure 1C).
Time-course analyses of various SK1 strains. SK1 diploids were transferred to sporulation medium and shifted to 33°. To monitor meiotic progression (A–C), cells were fixed with formaldehyde, stained with DAPI, and examined by fluorescence microscopy. Binucleate cells were classified as MI, tetranucleate cells as MII. A total of 200 cells were counted for each strain at each timepoint. To monitor viability (D), appropriate dilutions of each diploid were plated on YPAD. Each timepoint is normalized to the viability at the 0 timepoint for that strain. (A) MMS4 (NH144), mms4Δ (NH274F), rec104 (DW11), and rec104 mms4Δ (NH349); (B) MMS4 (NH144), mms4Δ (NH274F), pch2 (NH341), and pch2 mms-4Δ (NH340); (C) MMS4 (NH144), mms4Δ (NH274F), mek1 mms4Δ (NH280Fade2::pRS402), and red1 mms4Δ (NH287); (D) MMS4 (NH-144) and mms4Δ (NH274F).
mms4 SK1 diploids exhibit reduced viability when returned to growth: DSBs induced by meiotic recombination can be repaired by returning cells to growth in vegetative medium. Mutants unable to repair meiotic DSBs exhibit a decrease in cell viability when returned to growth (Arbelet al. 1999). The viability of mms4 in return to growth experiments is greatly reduced as a function of time in sporulation medium (Figure 1D). In contrast to the wild-type strain where viability was relatively unaffected, only 3% of the mms4 cells were viable after 26 hr at 33°.
mms4 diploids have decreased levels of spore viability in the SK1 background: The severity of the mms4 meiotic progression defect is dependent upon temperature in the SK1 background. Although very few mature asci are observed if the cells are sporulated at 33°, higher numbers of mature asci can be obtained by sporulating the mms4 diploid at 30° (MMS4, 82.2% ± 7.8 vs. mms4, 11.5% ± 6.6, n = 10). Dissection of mature asci from cells sporulated at 30° demonstrated a decrease in spore viability in the absence of MMS4. While 96.6% (644 asci) of the spores from NH144 were viable, only 50.9% (892 asci) of the spores from the mms4 diploid formed colonies.
The decrease in spore viability is not due simply to chromosome nondisjunction: A number of mutants that reduce spore viability two- to threefold, similar to mms4 (e.g., zip1, zip2, mer3, msh4, msh5), have previously been identified (Ross-Macdonald and Roeder 1994; Sym and Roeder 1994; Hollingsworthet al. 1995; Chua and Roeder 1998; Nakagawa and Ogawa 1999). The spore inviability of these mutants can be accounted for by MI chromosome nondisjunction. To see whether mms4 is causing meiotic chromosome missegregation, the distribution of viable spores in tetrads was examined.
A hallmark of MI nondisjunction is a decrease in the number of 4+:0− viable-spore tetrads with a corresponding increase in the 2+:2− and 0+:4− classes (for example, see Hollingsworthet al. 1995). This pattern was not observed for mms4 (Figure 2). Although the number of 4+:0− tetrads is greatly reduced, all of the other tetrad classes were increased, including 3+:1− and 1+:3− asci. The mms4 pattern is similar, but not identical, to one predicted by random spore death (χ2, P < 0.001; Figure 2). Another indicator of MI nondisjunction is a bias toward sister spores in two-viable-spore tetrads. Sister spores can be distinguished from nonsister spores using tightly centromere-linked markers such as ARG4. Out of 249 two-viable-spore tetrads produced by the mms4 diploid, only 39% were either +:+ or −:− at the ARG4 locus, indicating a lack of bias for sister spores. Another type of chromosome missegregation that can occur is precocious sister chromatid separation. These events result in three-viable-spore tetrads in which one spore is disomic. Disomy for chromosome III can be detected because the cells are nonmaters. The mms4 diploid produced 211 three-viable-spore tetrads. Of these, only one contained two mating and one nonmating spore. Precocious sister chromatid separation is therefore not occurring at high frequency in the mms4 diploid.
Distribution of viable spores in tetrads of MMS4/MMS4 and mms4Δ/mms4Δ diploids. NH144 and NH274 were sporulated on plates at 30° for 3 days and the resulting tetrads were dissected. A total of 644 and 892 asci were analyzed for NH144 and NH274, respectively. The NH144 data include 337 tetrads originally published in Hollingsworth et al. (1995). Shaded bars indicate the percentage of each tetrad type expected, assuming 50.9% viability and random lethality.
Recombination is initiated at wild-type levels in the mms4 SK1 diploid: Interhomolog recombination was assayed genetically using dissected tetrads from the SK1 diploids, NH144 (MMS4) and NH274 (mms4). In four-viable-spore tetrads, the mms4 diploid exhibited higher levels of gene conversion at all three loci analyzed, although only the value at HIS4 was statistically significant (Table 2A). Recently it has been discovered that recombination frequencies can be altered depending upon whether a strain is auxotrophic for various nutrients (Abdullah and Borts 2001). Although in this experiment, the mms4 strain is Ura+ while the wild-type strain is Ura−, the observed differences in gene conversion are likely to be real. Identical results were obtained when MUS81 (the gene encoding the protein partner of MMS4; see below) was deleted in the isogenic strain background using the kanamycin-resistance gene instead of URA3. In this experiment, both strains were Ura− (B. Larkin and N. M. Hollingsworth, unpublished data).
In contrast to the increase in gene conversion, a 1.3-fold reduction in crossing over in the MAT-HIS4 interval was observed in the mms4 diploid when four-viable-spore asci were analyzed (Table 2A). Crossing over was also examined in three-viable-spore tetrads by inferring the genotype of the dead spores on the basis of the assumption that HIS4 and MAT segregated 2+:2−. One explanation for the mms4 distribution of viable spores in tetrads is that crossovers are initiated normally, but “half-crossover” events result in the repair of only one chromatid while the unrepaired broken chromosome creates a dead spore (Chamberset al. 1996). The dead spore in a three-viable-spore tetrad would therefore contain one broken recombinant chromosome while asci with two, three, or four dead spores would contain two, three, or four different broken chromosomes that segregated to different spores. This model predicts that crossover recombinants should be enriched in three-viable-spore asci. Comparison of the MAT-HIS4 map distances in the three- and four-viable-spore tetrads from the MMS4 diploid revealed no difference (Table 2A). In contrast, the three-viable-spore tetrads from the mms4 diploid exhibited 37.6 cM between MAT and HIS4 compared to just 30.6 cM for the four-viable-spore tetrads from the same strain (Table 2A). These data further support the idea that a wild-type number of recombination events are initiated in the absence of MMS4 but that processing of the recombinants is faulty. For those cells that are able to escape the checkpoint, reduced spore viability may occur as a result of failure to repair one side of the DSB break (see discussion).
The tetrad dissection data may be biased because only a small fraction of the mms4 cells are able to form asci. To eliminate this bias, heteroallelic recombination was monitored in the wild-type and mms4 diploids using return-to-growth experiments. The same leu2 heteroalleles assayed by tetrad dissection were analyzed. The mms4 mutant reproducibly exhibits a slight delay in the onset of gene conversion but the kinetics and level of recombinants mirrors the MMS4 diploid until 8 hr after the induction of meiosis (Figure 3). After 8 hr, the level of recombinants in the wild type remains constant while the mms4 level drops (Figure 3). This decrease coincides with the onset of the first meiotic division in the mms4 strain (data not shown). We conclude that a wild-type number of recombination events are initiated in mms4 cells, but that in many cells the chromosomes are not properly repaired, thereby resulting in cell death.
Double-strand breaks are made and processed in mms4 mutants: Meiotic recombination is initiated by the introduction of DSBs, which are then resected to create 3′ single-stranded tails. DSB formation was assayed in isogenic MMS4 (NKY1551) and mms4 (NH301) SK1 diploids using the well-characterized HIS4/LEU2 recombination hotspot (for example, see McKee and Kleckner 1997a). DNA was isolated from the wild-type and mms4 diploids at different times after transfer to sporulation medium. The two predicted DSB fragments of 3.9 and 6.8 kb were observed in the wild-type and mms4 diploids. The bands were diffuse, indicating that the ends of the fragments are being processed to create single-stranded tails of heterogeneous lengths (Figure 4A). The time at which the maximum number of DSBs was observed was delayed ~2 hr in the mms4 mutant. By 10 hr, the majority of DSB fragments had disappeared in wild type, while for mms4Δ the DSB fragments did not disappear until 12 hr (Figure 4B).
Recombination in mms4 diploids
Heteroallelic recombination measured by return-to-growth in MMS4 and mms4 SK1 diploids. NH144 (MMS4/MMS4) and NH274F (mms4Δ/mms4Δ) were transferred to sporulation medium at 33° and appropriate dilutions were plated onto SD-leu and YPAD at different timepoints. The averages of two or three single colonies are plotted for NH144 and NH274F, respectively.
DSB formation was also examined at the naturally occurring YCR048w hotspot on chromosome III in the SK1 strains NH144 (MMS4/MMS4) and NH274F (mms4Δ/mms4Δ) sporulated at 33° (Wu and Lichten 1994). At this locus, the kinetics of DSB formation and disappearance were equivalent for mms4Δ and wild type. However, a twofold increase in the number of DSBs was observed in mms4 compared to wild type (data not shown). DSBs were also analyzed in the isogenic diploids, NH280F::pRS402 (mek1 mms4Δ) and NH287 (red1 mms4Δ). In both strains DSBs were observed (data not shown), indicating that red1 and mek1 do not suppress the meiotic recombination checkpoint triggered by mms4Δ simply by preventing DSBs, consistent with previous observations in the literature (Xuet al. 1997).
Crossovers are reduced approximately twofold in mms4 SKI diploids: Because the genetic data measuring crossovers could be biased by the small number of mms4 cells able to form asci, a physical analysis was used as an alternative way of monitoring crossing over. Crossing over was assayed at the HIS4/LEU2 hotspot that was used for analyzing DSB formation (Storlazziet al. 1995). DNA was isolated from NKY1551 (MMS4/MMS4) and NH301 (mms4Δ/mms4Δ) at 2-hr intervals after the induction of meiosis. DNA was digested with XhoI, which produces recombinant bands of 18.5 and 13.8 kb (R1 and R2, respectively, in Figure 5; Storlazziet al. 1995). Crossovers were detected in both strains starting at 6 hr. The wild-type cells entered MI beginning at 6 hr in this time course (data not shown). Quantitation of the two mms4 recombinant bands, R1 and R2, at the 10-hr time point revealed that they contained 4.9 and 2.1% of the total DNA, respectively (Figure 5B). This represents a twofold reduction in crossing over compared to wild type.
The effects of mms4Δ on meiosis-specific double-strand breaks at the HIS4/LEU2 hotspot. (A) DNA was isolated from NKY1551 (MMS4/MMS4) or NH301 (mms4Δ/mms4Δ) taken at different times after transfer to sporulation medium at 33°. DNA was digested with PstI and fractionated on a 0.6% agarose gel. The gel was probed with a 1.8-kb PstI/BglII fragment from pNKY291, which detects two meiosis-specific DSB fragments of 3.9 kb (DSB I) and 6.2 kb (DSB II). (B) Quantitation of the amount of the 3.9-kb DSB fragment as a function of time.
mms4 SK1 diploids exhibit aberrant synapsis and a delay in pachytene: To determine the effects of deletion of MMS4 on chromosome synapsis, nuclei from NH144 (MMS4/MMS4) and NH274F (mms4Δmms4Δ) were spread from cells taken at different times after transfer to sporulation medium at 33° and the chromosomes were examined by electron microscopy. The extent of synapsis observed between mms4 nuclei was highly variable, ranging from short fragments of SC to reasonably complete SC. Even those mms4 nuclei classified as having complete SCs were not equivalent to wild type, however. The mms4 SCs tend to be clumped and the spreads contain polycomplexes, a sign that synapsis is not proceeding normally (Figure 6, a–c). Specifically, many SCs appeared locally split with axial elements often thickened at these sites. Also, entangled SCs were seen, which may be due to failure to resolve interlocking or to involvement of split axes in nonhomologous associations. The mms4 cells exhibited a long delay in pachytene. The peak number of cells containing complete SCs occurred at 4 hr in wild type with only 5% of the cells containing complete SCs by 5.5 hr. In contrast, cells with complete SCs in the mms4 diploid peaked at 5.5 hr and SCs persisted until 10 hr (Figure 6e).
The mms4 sporulation and spore viability defects are not rescued by spo13: The spo13 mutation results in a single meiotic division, thereby producing two diploid spores (Klapholz and Esposito 1980). Because there is only one division, the requirement for crossing over to promote segregation at MI is removed. A number of meiotic mutants that decrease spore viability in wild-type meioses generate viable spores in the presence of spo13 (Peteset al. 1991). Mutants that are rescued by spo13 tend to be defective in the initiation of recombination (e.g., spo11, mei4, rad50) or in synapsis (hop1, red1). In contrast, mutants that are blocked after the initiation of recombination and that trigger the meiotic recombination checkpoint (e.g., rad50S) are not rescued by spo13 (Alaniet al. 1990).The mms4Δ spo13 diploid (NH299) was created in a slow-sporulating strain background, and sporulation and spore viability were compared with the isogenic spo13 strain, NH298. A 15-fold decrease in the frequency of mature dyads was observed in NH299 compared to NH298. Spore viability was reduced 2.8-fold by mms4 (Table 3).
To ensure that mms4 defects in sporulation and spore viability were due to recombination in this strain background, a mutation in SPO11 was introduced. The spo11 spo13 diploid produced wild-type levels of dyads and 92.9% of the spores were viable (Table 3). The spo11 mms4 spo13 strain was phenotypically identical to the spo11 spo13 strain (Table 3), providing a second demonstration that preventing recombination suppresses the mms4 meiotic mutant phenotypes.
mms4 has variable effects on interhomolog recombination in non-SK1 strains: Crossing over was examined in four intervals in non-SK1 spo13 diploids. Because spo13 does not rescue the mms4 spore inviability, these data represent a selected set of dyads, those capable of forming mature asci containing two viable spores. Variable effects of mms4 on crossing over were seen. In the MAT-CENIII and ARG4-THR1 intervals on chromosomes III and VIII, respectively, crossing over was reduced 2.5-fold. In contrast, no difference was seen in the CENIII-LEU2 interval, while the map distance between LEU2 and HIS4 was increased 1.8-fold in the mms4 diploid (Table 2B).
The effect of mms4Δ on meiotic crossing over at the HIS4/LEU2 hotspot. (A) DNA was isolated from NKY1551 (MMS4/MMS4) and NH301 (mms4Δ/mms4Δ) at different times after transfer to sporulation medium at 33°. DNA was digested with XhoI and fractionated on a 0.6% gel. The gel was probed with pNKY155, which detects two parental bands (P1, 19.9 kb; P2, 12.4 kb) and two recombinant bands (R1, 18.5 kb; R2, 13.8 kb). (B) Quantitation as a function of time of R1 or R2 as a percentage of the total DNA.
red1 and mek1 differentially affect the ability to rescue the sporulation and spore viability defects of mms4 spo13 diploids in slow-sporulating strains: Although abrogation of the meiotic recombination checkpoint allows mms4 cells to proceed through the meiotic divisions, the question remains as to whether mms4 recombination intermediates are repaired before the chromosomes segregate or not. Analysis of the spore viability of the SK1 pch2 mms4 asci indicates that the chromosomes have not been repaired when the meiotic recombination checkpoint is eliminated by mutation of PCH2. Dissection of asci sporulated at 30° showed that while the pch2 strain, NH341, produced 97.4% viable spores (48 asci), the pch2 mms4 diploid (NH340) produced the same number of viable spores (30.2%, 122 asci) as mms4 alone (33.3%, 48 asci for NH274F). Therefore, although the mms4 cell cycle progression defect is ameliorated by pch2, the DNA damage occurring due to a lack of mms4 is still present.
Because mek1 and red1 cause a more severe spore viability phenotype than mms4 in a two-division meiosis, one cannot assess whether the DNA damage arising from a lack of MMS4 is present in the spores from strains where MEK1 or RED1 has been deleted. Since red1 and mek1 give high spore viability in the spo13 background, we assayed sporulation and spore viability in red1 mms4 spo13 and mek1 mms4 spo13 strains. In our slow-sporulating spo13 strain background, mek1 mutants sporulated as well as spo13 alone and produced 90.4% viable spores (Table 3). Deletion of MEK1 in the mms4 spo13 strain created a strain that resembled mms4 spo13 rather than mek1 spo13, producing few mature asci and only 18.9% viable spores (Table 3). A different result was obtained in epistasis experiments using red1. Like mek1, deletion of RED1 has no effect on sporulation and produces high levels of viable spores in the presence of spo13 (Table 3). A partial rescue of the mms4 spo13 sporulation and spore viability phenotypes was observed when RED1 was absent. Sporulation and spore viability in the red1 mms4 spo13 diploid were increased 7.6- and 2.8-fold, respectively, compared to mms4 spo13 (Table 3). The absence of MMS4 is therefore more deleterious if recombination is occurring in the presence of RED1.
red1 and mek1 similarly rescue the sporulation defects of dmc1 spo13 diploids: Deletion of DMC1 in spo13 SK1 strains causes sporulation and spore viability defects that can be suppressed by mutation of RED1 (Bishopet al. 1999). Similar results were obtained for dmc1 in our spo13 non-SK1 strain background. While dmc1 alone produced only 6.4% dyads, the red1 dmc1 strain sporulated at levels equivalent to wild type (Table 3). The dmc1 spo13 diploid produced 72.1% viable spores, consistent with previous results for dmc1 in a slow-sporulating strain background (Rockmill and Roeder 1994). The dmc1 spo13 spore viability was also improved by addition of red1 (Table 3; χ2, P < 0.001). The effects of mek1 on sporulation and spore viability of dmc1 spo13 are indistinguishable from those of red1 (Table 3). The fact that only red1 can suppress mms4 while both red1 and mek1 can suppress dmc1 suggests that the mechanism of red1 suppression differs for the two mutants.
Time course of synapsis in SK1 wild-type and mms4 diploids. NH144 (MMS4/MMS4; d) and NH274F (mms4Δ/mms4Δ; a–c) were sporulated at 33° for 5.5 hr and nuclei were spread and examined by electron microscopy. Arrowheads indicate polycomplexes and arrows indicate aberrant SCs. (e) Quantitation of cells exhibiting complete SCs. NH144 and NH274F were sporulated at 33° and cells were taken at various timepoints and spread for electron microscopy. The number of nuclei containing complete SCs was scored for at least 100 cells for each strain.
mms4 and msh5 are in different epistasis groups with regard to spore viability: Deletion of MSH5 in SK1 results in 37.5% viable spores (Hollingsworthet al. 1995). This frequency of viable spores is similar to that observed in mms4 and could arise by three possible mechanisms:
msh5 and mms4 act on the same pathway for spore viability. This model predicts that the msh5 mms4 diploid will produce the same frequency of viable spores as msh5 or mms4 alone.
msh5 and mms4 act on parallel pathways to promote crossing over. In this case, viable spores should be eliminated in the double mutant. These two possibilities seem unlikely given that spore inviability of msh5 can be attributed to meiosis I nondisjunction, which does not appear to be increased by mms4.
The msh5 and mms4 effects on spore viability are independent of each other.
A msh5 mms4 SK1 diploid (NH370) was created and sporulated and 365 tetrads were dissected. The spore viability of mms4 msh5 is 21.8%, similar to the 19.1% predicted if the two genes act independently. The percentage of four-viable-spore tetrads observed for msh5 is 19.5 (Hollingsworthet al. 1995) compared to 14.0 for mms4. The mms4 msh5 diploid produced 3.3% tetrads in which all four spores were viable, comparable to the 2.7% predicted if the genes act independently. MSH5 and MMS4 therefore play different roles during meiosis in the production of viable spores.
MMS4 is in the same epistasis group as MUS81 for spore viability: Recently mms4 was discovered to function in the same epistasis group as mus81 for the vegetative phenotypes of UV and MMS sensitivity (Mullenet al. 2001). Like mms4, mus81 diploids exhibit reduced sporulation and produce only 40.0% viable spores in the SK1 strain background (Interthal and Heyer 2000). To determine whether mus81 and mms4 function in the same epistasis group for spore viability, mus81 was deleted in our SK1 strains. Deletion of MUS81 produced 40.5% viable spores (1357 asci). Dissection of the mus81 mms4 diploid, NH372, produced 45.4% viable spores (66 asci). This value is similar to both single mutants, indicating that mms4 and mus81 act in the same pathway during meiosis to produce viable spores.
DISCUSSION
MMS4 was originally identified as a mutant specifically sensitive to MMS in vegetative cells (Prakash and Prakash 1977). Xiao et al. (1998) later showed that the mms4 sensitivity extends to a variety of alkylating agents and also that mms4 has a weak spontaneous mutator phenotype. Recently MMS4 was identified as a mutant that causes a synthetic lethal phenotype in combination with a deletion of SGS1, the RecQ helicase ortholog in S. cerevisiae required for genomic stability (Mullenet al. 2001). All of these results indicate that MMS4 has a function in DNA metabolism in vegetative cells. The work presented here demonstrates an important role for MMS4 in meiotic DNA metabolism as well.
Sporulation and spore viabilities in non-SK1 spo13 diploids
A number of experiments indicate that MMS4 acts sometime downstream of the initiation of recombination. First, wild-type levels of meiosis-specific DSBs are observed in mms4 diploids. Second, the mms4 sporulation and spore inviability phenotypes are not rescued by the single division meiosis conferred by spo13. Third, in dissected tetrads and return-to-growth experiments, the levels of interhomolog gene conversion are equivalent or elevated relative to wild type, indicating that wild-type levels of heteroduplex DNA have been formed. Finally, consistent with a post-initiation defect, the mms4 sporulation defect is due to activation of the meiotic recombination checkpoint. Preventing the formation of DSBs is sufficient to suppress the meiotic progression and sporulation defects of mms4. Furthermore, mutation of several meiosis-specific components of the meiotic recombination checkpoint allows mms4 cells to progress through the meiotic divisions. The extent of the mms4 prophase arrest is dependent upon temperature in the SK1 background, a phenomenon previously observed for other mutants that trigger this checkpoint (Nakagawa and Ogawa 1999; V. Boerner and N. Kleckner, personal communication).
The meiotic recombination checkpoint is activated by mutations defective at a variety of different stages of recombination. The earliest block occurs in rad50S, mre11S, and sae2/com1 mutants, which produce DSBs that fail to get processed (Xuet al. 1997; Usuiet al. 2001). As a result, DSB fragments appear as discrete bands and accumulate with time. The DSB fragments in mms4 are heterogeneous, an indication that they are getting processed and do not accumulate. MMS4 therefore functions downstream of DSB resection.
A second stage of recombination that triggers the checkpoint is revealed by dmc1 mutants. DMC1 encodes a meiosis-specific ortholog of the bacterial RecA strand transfer protein. In dmc1 mutants, DSBs are hyperresected and persist because they are unable to efficiently invade the homologous chromosome without DMC1 (Bishopet al. 1992; Hunter and Kleckner 2001). The sporulation and spore inviability of dmc1 can be suppressed if a second RecA ortholog, Rad51p, is activated by overexpression of RAD54 (Bishopet al. 1999). Overexpression of RAD54 has no effect on the mms4 sporulation defect (T. de los Santos and N. M. Hollingsworth, unpublished data), suggesting that MMS4 is acting after strand invasion. An alternative route for the repair of dmc1 DSBs utilizes sister chromatid recombination. In a typical diploid meiosis, constraints that promote recombination between homologs as opposed to sister chromatids are imposed. When these constraints are released, either by return to growth or by mutation of the axial element component, RED1, the DSBs present in dmc1 are repaired (Schwacha and Kleckner 1997; Arbelet al. 1999; Bishopet al. 1999). An increase in the ratio of intersister joint molecules compared to interhomolog joint molecules has been directly observed in red1 mutants (Schwacha and Kleckner 1997). We assume that the suppression of the dmc1 arrest by mek1 (Xuet al. 1997; this work) is due to a similar change in chromosome bias because mek1 and red1 exhibit similar phenotypes with respect to reduced levels of sister chromatid cohesion as well as a change in partner choice toward sister chromatids (Bailis and Roeder 1998; Thompson and Stahl 1999).
MMS4 differs from DMC1 in that it appears to function downstream of the decision for partner choice (and therefore strand invasion). In haploid or diploid dmc1 strains induced in meiosis and then returned to growth, viability remains high throughout the time course. This result is due to repair of the dmc1 DSBs using RAD54-dependent recombination between sister chromatids (Arbelet al. 1999). In rad54 strains, there is no intersister recombination and so rad54 haploids lose viability in meiotic time courses but rad54 diploids do not (repair can occur using the homologous chromosome). A dmc1 rad54 diploid shows low viability because neither homologs nor sister chromatids can be used to repair the DSBs (Arbelet al. 1999). The viability of mms4 diploids is similar to that observed for the dmc1 rad54 diploid. This finding implies either that MMS4 is required for repair of DSBs using both sisters and homologs or, alternatively, that the block to recombination in mms4 occurs after the decision of which chromatid to use in recombination has been made.
Unlike dmc1 spo13, the sporulation and spore viability defects of mms4 spo13 can be partially rescued only by red1, and not by mek1. This result further argues against the idea that sister chromatid recombination is repairing the mms4 DSBs since mek1 and red1 mutants appear to similarly relieve the constraints against sister chromatid exchange. A major phenotypic difference between red1 and mek1 in slow-sporulating strain backgrounds is that red1 mutants make no axial elements or SC while substantial amounts of SC are present in the absence of MEK1 (Rockmill and Roeder 1990, 1991). In addition, mek1 mutants generate more viable spores, presumably because crossovers are being generated in the context of the SC and are therefore more effective for disjunction. In contrast, in the SK1 background where mek1 produces <1% viable spores, the synapsis defect of mek1 is much more severe (J. Loidl, unpublished results). When our slow-sporulating spo13 strains were converted to Spo13+, mek1 produced 23% viable spores compared to just 3% for red1 (data not shown). These results suggest that SC formation is occurring in our slow-sporulating mek1 strains. It is possible, therefore, that recombination occurring in the presence of the SC creates recombination intermediates that require MMS4 for completion while those occurring in the absence of the SC can utilize alternative pathways such as synthesis-dependent strand annealing (Paques and Haber 1999).
An alternative explanation for the difference between the ability of mek1 and red1 to rescue the mms4 spore inviability in the spo13 background would be if MMS4 required MEK1 for activation, perhaps by a phosphorylation event. This idea seems unlikely given that the mek1 spo13 diploid produces highly viable spores. If Mek1p was necessary to phosphorylate Mms4, then the mek1 mutant should contain unprocessed recombination intermediates and therefore broken chromosomes, which is clearly not the case.
What, then, is the function of MMS4 during meiotic recombination? An important clue lies in the biochemical activity of the protein, which has recently been elucidated by Brill and colleagues. Previously Mullen et al. (2001) showed that Mms4p physically interacts with a second protein, Mus81p. Genetic evidence supports the notion that MMS4 and MUS81 function together because mus81 mms4 double mutants are phenotypically identical to either single mutant (Mullenet al. 2001; this work). Mus81p shares homology with Rad1p, a protein involved in nucleotide excision repair (Interthal and Heyer 2000; Mullenet al. 2001). Like Mus81, Rad1 acts in a complex with a second protein, Rad10. The Rad1/10 complex is a duplex 3′ single-stranded junction-specific endonuclease (Bardwellet al. 1994). A preferred substrate for Rad1/10 is a simple Y form. In contrast, a preferred substrate for Mms4/Mus81 is a branched structure in which only the 3′ end is single stranded (Figure 7A; Kaliramanet al. 2001).
Model for MMS4 function in meiotic recombination. (A) Structure of a preferred substrate for Mms4/Mus81 in vitro (Kaliramanet al. 2001). (B) Schematic diagram of the strand-displacement and annealing model adapted from Allers and Lichten (2001). Asterisks indicate the junctions at which Mms4/Mus81 is proposed to cleave. Newly synthesized DNA is indicated by a dotted line. Arrowheads indicate 3′ ends.
If one assumes that the preferred substrate of Mms4/Mus81 in vitro is similar to the structure cleaved in vivo, the question remains as to how this structure could be generated during meiotic recombination. One possibility would be if there was heterology on one of the 3′ ends of the break so that it was unable to anneal properly, thereby creating a 3′ flap (Kaliramanet al. 2001). This model is inconsistent with the finding that mms4 still exhibits a sporulation-defective phenotype in HO-derived diploids where the number of mismatches is very low (N. M. Hollingsworth, unpublished data). We propose instead that MMS4 is required for an alternative pathway of meiotic DSB repair recently postulated by Allers and Lichten (2001). This pathway of “strand displacement and annealing” initiates with a SEI (Figure 7B). DNA synthesis extends the length of the invading strand, which then dissociates from the nonsister chromatid and anneals to the resected 3′ end on the other side of the break. If extension of the invading strand creates a single-stranded tail that is longer than the resected end on the other side of the break, a 3′ “flap” will result after the two strands anneal (Figure 7B). The resulting structure looks identical to the preferred substrate for Mms4/Mus81 in vitro. We propose that Mms4/Mus81 is required to cleave this flap so that one side of the break can be repaired by ligation. Failure to remove this flap would result in one broken chromatid, while resolution of the double Holliday junctions on the other side of the break would create an intact chromatid.
Allers and Lichten (2001) proposed the strand displacement and annealing model to account for a class of joint molecules, which they observed during meiotic recombination, that cannot be explained by the canonical DSB repair model. This class of joint molecules (JM2) contains double Holliday junctions downstream of heteroduplex DNA instead of flanking it. A testable prediction of our model is that the number of JM2 molecules should be reduced in an mms4 mutant.
In addition to the biochemistry, genetic and cytological data support our model for MMS4 function. We propose that the majority of cells are arrested in prophase because unprocessed 3′ flaps trigger the meiotic recombination checkpoint. For those cells that escape the checkpoint and form asci, spore viability is decreased due to the formation of half crossovers. Tetrads containing more than one dead spore could occur if more than one pair of homologs sustain a half crossover and the broken chromatids segregate into different spores. As predicted by the half-crossover model, recombinants are enriched in mms4 asci with only three viable spores (Chamberset al. 1996). A similar result is seen in the three-viable-spore tetrads from mus81 diploids, arguing against this observation being a statistical fluke (B. Larkin and N. M. Hollingsworth, unpublished data). Also consistent with the half-crossover model is the fact that crossovers are reduced twofold by physical analyses. A similar reasoning has been used to explain the presence of recombinants in rad52 diploids (Haber and Hearn 1985; Bortset al., 1986) where it is known, in meiosis, that rad52 mutants are unable to progress from SEI invasion intermediates to double HJs (N. Hunter and N. Kleckner, personal communication). The mms4 interval-specific effects on crossing over could be explained if the ratio of crossovers arising from canonical DSB repair vs. the strand displacement mechanism varies with different regions of the chromosomes. Cytologically, the synapsis defect of mms4 appears as though it is occurring in late zygotene/early pachytene, the time at which SEI intermediates are being processed (Hunter and Kleckner 2001).
Another appealing aspect of our model is that it can explain the red1 suppression of the mms4 sporulation/spore viability phenotypes. In the absence of RED1, the ratio of resection on one side of the break to the new synthesis occurring on the invading strand at the other side may be altered such that the resected end is always longer than the end of the invading strand. Alternatively, the strand displacement pathway of recombination may not occur in the absence of synapsis, in which case there would be no requirement for MMS4. Our model is also compatible with the observation that MMS4 and MSH5 act independently of each other since the spore inviability of mms4 comes from broken chromosomes resulting from failure to repair one side of the DSB, whereas the msh5 spore inviability is due to aneuploidy resulting from MI nondisjunction.
Acknowledgments
We thank JoAnne Engebrecht, Neil Hunter, and Aaron Neiman for comments on the manuscript. Rhona Borts, Steve Brill, JoAnne Engebrecht, Jim Haber, Neil Hunter, Michael Lichten, Aaron Neiman, and Frank Stahl provided helpful discussions. Steve Brill, Jim Haber, and Neil Hunter generously communicated results prior to publication. In particular we acknowledge Rhona Borts for the half-crossover idea and Neil Hunter for the suggestion that overreplication is responsible for creating the 3′ tail cleaved by Mms4p. Doug Bishop, Mike Dresser, JoAnne Engebrecht, Bob Malone, Shirleen Roeder, Rolf Sternglanz, and Wei Xiao provided strains and/or plasmids. We thank Lihong Wan for her help with tetrad dissection and Bridget Baumgartner, Cindy Lee, Lisa Ponte, and Dana Woltering for excellent technical support. This work was supported in part by a grant from the National Institutes of Health (GM-50717) and by Research grant no. 1-FY00-193 from the March of Dimes Birth Defects Foundation to N.M. H.J.L. was supported by the Austrian Science Fund (grant S 8202).
Footnotes
-
Note added in proof: Boddy et al. (M. N. Boddy, P.-H. L. Gaillard, W. H. McDonald, P. Shanahan, J. R. Yates 3rd and P. Russell, 2001, Mus81-Eme1 are essential components of a Holliday junction resolvase. Cell 107: 537–548) have recently described the biochemical activities of the Mus81/Eme1 complex from Schizosaccharomyces pombe. Eme1 is thought to be a homolog of Mms4 (Boddy et al. 2001; S. Brill, personal communication). The authors suggest that Mus81/Eme1 acts as a Holliday junction resolvase in the repair of stalled replication forks and during meiotic recombination. A similar model for human Mus81 activity has also been proposed (Chen, X-B., R. Melchionna, C.-M. Denis, P.-H. L. Gaillard, A. Blasina, I. Van de Weyer, M. N. Boddy, P. Russell, J. Vialard and C. H. McGowan, 2001, Human Mus81-associated endonuclease cleaves Holliday junctions in vitro. Mol. Cell 8: 1117–1127). Our experiments show that there is only a twofold reduction in meiotic crossovers in the absence of MMS4. This indicates that, at least in Saccharomyces cerevisiae, the Mus81/Mms4 complex does not function as the unique Holliday junction resolvase during meiotic recombination.
-
Communicating editor: M. Lichten
- Received May 30, 2001.
- Accepted September 24, 2001.
- Copyright © 2001 by the Genetics Society of America