Regulation of meiotic progression by the meiosis-specific checkpoint kinase Mek1 in fission yeast

Eukaryotic cell division consists of a highly regulated sequence of events that must occur in the appropriate order. Checkpoints are control mechanisms that prevent initiation of late events until earlier events have been successfully completed, thus ensuring faithful transmission of genetic information to the progeny (Hartwell and Weinert, 1989). In response to defective cellular processes and/or alterations of genome integrity, these surveillance mechanisms arrest or delay cell cycle progression. Checkpoints pathways are composed of sensors that detect the cellular defect or DNA lesions, generating a signal that is transmitted through a transduction pathway usually formed by protein kinases. Ultimately, checkpoint effectors act on cellular targets, triggering various responses, including cell cycle arrest or delay and DNA repair. In mammals, defects in checkpoint responses cause genomic instability, leading to tumor development (reviewed by Weinert, 1998; Lowndes and Murguia, 2000; Melo and Toczyski, 2002).

Meiosis is a specialized type of cell division that generates haploid gametes from diploid parental cells because a single round of DNA replication is followed by two consecutive nuclear divisions. During meiotic prophase, a complex series of interactions between homologous chromosomes (or homologs) occur. First, chromosomes search for and associate with the homologous partners (pairing). In most (but not all) organisms, these associations are stabilized by synapsis, which is the formation of an elaborate proteinaceous structure (the synaptonemal complex; SC) that holds homologs close together along their entire length. Concomitantly, DNA recombination between homologous chromosomes takes place. In addition to the exchange of genetic information, the result of these interactions is the formation of physical connections between homologs, called chiasmata, which promote correct chromosome segregation during the first meiotic division (reviewed by Roeder, 1997; Smith and Nicolas, 1998; Zickler and Kleckner, 1999; Lee and Amon, 2001).

Meiotic cells possess a surveillance mechanism referred to as the `pachytene checkpoint' or the `meiotic recombination checkpoint' that monitors these critical meiosis-specific events. Meiotic recombination is initiated by DNA double-strand breaks (DSBs), which are repaired using nonsister chromatids as templates. In response to defects in recombination that lead to accumulation of unrepaired DSBs and/or other recombination intermediates, the pachytene checkpoint triggers meiotic cell cycle arrest or delay to prevent meiotic chromosome missegregation (Roeder and Bailis, 2000).

A number of studies in the budding yeast Saccharomyces cerevisiae have identified several components of the pachytene checkpoint (Roeder and Bailis, 2000). DNA damage checkpoint proteins that respond to DSBs in vegetative cells also monitor these lesions during meiosis; however, there are differences between the mitotic DNA damage checkpoint and the meiotic recombination checkpoint. First, some DNA damage checkpoint proteins (e.g., Chk1, and Rad9) are not required for the pachytene checkpoint (Lydall et al., 1996) (P.A.S.-S. and G. S. Roeder, unpublished). Second, it has been proposed that meiotic DSBs are monitored in a meiosis-specific chromosomal context and are not recognized as `general' damage (Xu et al., 1997). Third, some crucial pachytene checkpoint proteins, such as the nucleolar silencing factor Pch2 or the Mek1 kinase, are produced only during meiosis (Rockmill and Roeder, 1991; San-Segundo and Roeder, 1999). Fourth, the DNA damage checkpoint and the pachytene checkpoint act on different targets of the cell cycle machinery to block cell cycle progression (Leu and Roeder, 1999).

The pachytene checkpoint has been extensively studied only in S. cerevisiae, but its operation in worms, flies and mammals has been also reported (Edelmann et al., 1996; Pittman et al., 1998; Yoshida et al., 1998; Ghabrial and Schupbach, 1999; Gartner et al., 2000; MacQueen and Villeneuve, 2001; Abdu et al., 2002). In fact, most (if not all) yeast pachytene checkpoint proteins have homologs in other organisms (Roeder and Bailis, 2000). However, although the fission yeast Schizosaccharomyces pombe is a model organism widely used in checkpoint studies during the mitotic cell cycle (Murakami and Nurse, 2000), little is known about surveillance mechanisms of meiosis-specific processes, in particular meiotic recombination.

Here we show that the meiotic recombination checkpoint does indeed operate in S. pombe, and we describe a role for a meiosis-specific kinase, Mek1, in this control mechanism. Mek1 contains a forkhead-associated (FHA) domain. FHA motifs are usually implicated in protein-protein interactions regulated by phosphorylation (Durocher et al., 1999). We present evidence indicating that fission yeast Mek1 prevents entry into the first meiotic division (MI) until recombination is completed. The Mek1-dependent negative regulation of MI entry is achieved by maintaining phosphorylation of Cdc2 at Tyr15, at least in part, through inhibition of the Cdc25 phosphatase.

Fission yeast strains used in this work are listed in Table 1. To delete the mek1⁺ and meu13⁺ genes, a PCR-based strategy was used (Bahler et al., 1998). The whole ORF was replaced for the kanMX6 or the ura4⁺ marker. The same approach was used for tagging mek1⁺ with three copies of the HA epitope immediately before the stop codon at its genomic locus. Strains expressing mek1⁺ tagged with GFP integrated in the genome under control of the thiamine-repressible nmt1 promoter were constructed by transformation with plasmid pLV1 (see below) cut with NruI to target the construct to the leu1-32 locus. Strain S1299 carrying the cdc25-9A allele at the genomic locus was constructed as follows. An NdeI-SmaI fragment containing the cdc25-9A gene was obtained from plasmid pGEX2TN(HpaI)-Cdc25(9A) (Zeng and Piwnica-Worms, 1999) and used to transform a cdc25-22 temperature-sensitive strain (PN35). Transformants were initially selected on YES plates at 36°C; then, several colonies were patched and replica-plated to YES-phloxin plates containing 0.05% and 0.005% methylmethane sulphonate (MMS). Transformants showing a defective response to MMS were selected and the presence of the nine mutated phosphorylation sites in cdc25-9A was verified by DNA sequencing. Growth conditions and strain manipulations were described previously (Moreno et al., 1991). Diploid strains homozygous for the mating type (h^-/h^-) were generated by protoplast fusion (Sipiczki and Ferenczy, 1977). Synchronous meiosis in h^-/h^- pat1-114/pat1-114 thermosensitive diploid strains was carried out essentially as described previously (Blanco et al., 2001).

To clone the mek1⁺ gene, the mek1⁺ cDNA was amplified by PCR using cDNA obtained from a 3-hour pat1 meiotic culture (see Fig. 1) with primers mek1-N 5′-TTTTCTCGAGCATATGGACTTTTTATCACATGCCATG-3′ (XhoI site, underlined; NdeI site, italicized) and mek1-C 5′-TTTTCCCGGGCTAGCGGCCGCCTAGCCGGGAATGTTTAAGAGG-3′ (SmaI and NotI sites, underlined; an added stop codon, italicized). The PCR product was digested with XhoI-SmaI and cloned into the same sites of the pREP3X vector (Forsburg, 1993), producing plasmid pSS123, which contains mek1⁺ cDNA under the nmt1(3×) promoter. The NdeI-SmaI fragment from pSS123 containing mek1⁺ was cloned into the same sites of pREP41-EGFP-N (Craven et al., 1998), generating plasmid pSS124, which expresses mek1⁺, N-terminally tagged with GFP, from the nmt1(41×) promoter. Plasmid pLV1 was constructed by cloning a PstI-EcoRI fragment from pSS124 containing nmt1-GFP-mek1⁺ into the same sites of the integrative vector pJK148 (Keeney and Boeke, 1994).

Spore viability was assayed by tetrad dissection. The frequency of meiotic intergenic recombination was determined by random spore analysis. Crosses were performed on MEA plates and, after 2 days, spores were isolated, grown on YES plates and replica-plated to minimal medium. The number of recombinant spores was counted and normalized to the total number of viable spores.

RNA preparation and northern blot analysis were performed as described previously (Blanco et al., 2001) using a mek1⁺ PCR fragment amplified with oligomers mek1-N and mek1-C as a probe. Total protein extracts were prepared as described elsewhere (Blanco et al., 2000). For western blot analysis, ∼60 μg of total extracts were run on 12% SDS-PAGE gels, transferred to nitrocellulose and probed with the following antibodies: mouse monoclonal anti-HA (12CA5; 0.15 mg/ml), rabbit polyclonal anti-Cdc2 (C2; 1:200 dilution), rabbit polyclonal anti-phospho-Cdc2(Tyr15) (Cell Signaling Technology; 1:1000 dilution) and mouse monoclonal anti-tubulin (TAT1; 1:1000 dilution). Goat anti-rabbit or goat anti-mouse antibodies conjugated to horseradish peroxidase (Amersham) were used as secondary antibodies (1:3500 and 1:2000 dilution, respectively). Immunoblots were developed using the Luminol Reagent (Santa Cruz Biotechnology) or the SuperSignal kit (Pierce).

Flow cytometric analysis was performed on a Becton-Dickinson FACSscan using propidium iodide staining of cells (Sazer and Sherwood, 1990).

For the analysis of meiotic progression, cells were fixed in 70% ethanol and processed for DAPI staining of nuclei as described previously (Moreno et al., 1991). To study the subcellular localization of Mek1-HA, immunofluorescence analysis was performed essentially as described previously (Santos and Snyder, 1997), except that PEMBAL buffer (100 mM PIPES pH 6.9, 1 mM EGTA, 1 mM MgSO4, 1% BSA, 0.1% azide, 0.1 M L-lysine HCl) was used instead of PBS. Mouse monoclonal anti-HA antibody (HA.11, Convance) was used at 1:150 dilution. Goat anti-mouse antibody conjugated to CY3 (Jackson ImmunoResearch Labs) was used as the secondary antibody (1:200 dilution). Cells were visualized using a Zeiss Axioplan2 fluorescence microscope equipped with a Hamamatsu CCD camera.

Bioinformatic analysis of the fission yeast genome (http://www.sanger.ac.uk/Projects/S_pombe/) revealed that S. pombe possesses a putative open reading frame (SPAC14C4.03), which encodes a 445 amino-acid protein with a high degree of similarity to the S. cerevisiae meiosis-specific kinase Mek1 (Fig. 1A; 34% identitity, 54% similarity, BLAST e-value: 4×10^-58). In addition to the kinase domain (amino acids 160-421), S. pombe Mek1 contains a FHA domain (amino acids 62-116), which is characteristic of a conserved family of kinases involved in different checkpoint responses in eukaryotic cells (Fig. 1B) (Durocher et al., 1999).

The S. pombe mek1⁺ gene was cloned by polymerase chain reaction (PCR) using specific primers and genomic DNA or cDNA as the template (see Materials and Methods; Fig. 1C). No PCR product corresponding to mek1⁺ was amplified when cDNA obtained from vegetative cells was used (Fig. 1C). However, a PCR fragment of the expected size was obtained when cDNA from S. pombe cells at the early stages of meiosis was used (Fig. 1C), indicating that the mek1⁺ gene is only expressed in meiotic cells. The difference in size of the mek1⁺ fragment amplified from genomic DNA or from cDNA is consistent with the presence of the two predicted introns in the gene.

Expression of the mek1⁺ gene was monitored by northern blot analysis during a pat1-driven synchronous meiosis (Fig. 2A-C). Consistent with the results shown above, no expression of mek1⁺ is detected in vegetatively growing cells; its expression is induced at the same time as the onset of premeiotic S phase and reaches the maximum level during the period corresponding to meiotic prophase; then, mek1⁺ mRNA levels decrease as cells enter into the first meiotic division (Fig. 2A-C). The production of the Mek1 protein was analyzed by western blot using anti-HA antibodies in a meiotic time course of pat1-114 diploid cells expressing a functional version of mek1⁺ tagged with three copies of the HA epitope. The kinetics of Mek1-HA production is similar to the one described above for mek1⁺ mRNA (Fig. 2D).

To determine the subcellular location of Mek1, immunofluorescence analysis of pat1-114 mek1-HA diploid cells was carried out using anti-HA antibodies. The Mek1 protein localizes to the nucleus of meiotic cells during the horse-tail movement period (Fig. 2E). No staining is detected in control cells lacking the HA epitope (data not shown).

To study Mek1 function during meiosis in fission yeast, the mek1⁺ gene was deleted. The mek1 mutant completes meiotic divisions and sporulation, generating morphologically normal four-spore asci. However, tetrad dissection revealed that spore viability is reduced in the mek1 mutant compared to wildtype (64% versus 89%, respectively). Although the overall decrease in spore viability of mek1 is not dramatic, the fraction of tetrads containing four viable spores is significantly reduced compared to wildtype (∼16% versus ∼78%; Fig. 3A). No excess of tetrads containing 4, 2 and 0 viable spores is observed, suggesting that spore death in the mek1 mutant is not due to chromosome nondisjunction during meiosis I. Intergenic recombination was examined in the leu1-his5 interval on chromosome II. The mek1 mutant displays a ∼2.4-fold reduction in meiotic recombination in this region (Fig. 3B).

Using pat1 strains to induce synchronous meiosis, kinetics of meiotic progression were examined in the mek1 mutant in comparison with the wildtype (Fig. 4A). In the mek1 mutant, the first meiotic division occurs reproducibly ∼30 minutes faster than in the otherwise isogenic wildtype (Fig. 4A). Since mek1⁺ expression is induced at the time of premeiotic S phase (although its peak of expression is reached at prophase; Fig. 2), DNA replication was carefully monitored by FACS during synchronous meiosis in mek1 and wild-type cells at 15 minute time points (Fig. 4B). This analysis revealed that premeiotic DNA replication takes place with the same kinetics in both wild-type and mek1 cells. Therefore, since entry into meiosis I occurs earlier in mek1, this implies that meiotic prophase is shorter in the absence of Mek1.

To further characterize this observation, the effect of expressing high levels of Mek1 was also studied. A green fluorescent protein (GFP)-tagged version of mek1⁺ was placed under control of the thiamine-regulated nmt1 promoter and integrated at the leu1 locus (see Materials and Methods). To induce overexpression of mek1⁺, thiamine was removed 14 hours prior to transferring the cells to medium lacking nitrogen (Fig. 4C). The production of Mek1-GFP was followed by microscopic examination of the cells throughout the experiment. A control culture, in which thiamine was always present and therefore mek1⁺ expression was repressed, was also examined. FACS analysis revealed that both cultures were blocked in G1 to the same extent and underwent premeiotic S phase with similar kinetics (Fig. 4D); however, nmt1-driven expression of mek1⁺ resulted in a significant delay (∼1 hour) of the first meiotic division (Fig. 4E). Interestingly, the Mek1-GFP signal disappeared as cells entered meiosis I, and binucleate cells containing GFP signal were rarely observed (data not show). Thus, these results suggest that the Mek1 kinase negatively regulates entry into meiosis I.

In order to understand how Mek1 regulates cell cycle progression, high levels of the protein were produced in vegetative cells using the nmt1 promoter. Interestingly, ectopic overproduction of Mek1 results in inhibition of growth (Fig. 5A). Microscopic examination revealed that Mek1-overproducing cells are highly elongated and contain a single undivided nucleus (Fig. 5B), a phenotype that resembles the G2/M arrest induced by activation of the DNA integrity checkpoints or by overproduction of the Cds1 or Chk1 checkpoint kinases (Furnari et al., 1997; Boddy et al., 1998).

In principle, the G2/M arrest triggered by mek1⁺ ectopic overexpression may be caused by inhibition of the Cdc25 phosphatase or activation of the Wee1 or Mik1 kinases. To identify which cell cycle regulator(s) is the target of Mek1, the protein was overproduced in mutants defective in either Cdc25 or Wee1 function. Since cdc25⁺ is an essential gene, to analyze whether Cdc25 is involved in the Mek1-dependent arrest, a cdc2-3w strain, which bypasses the requirement for Cdc25 was used (Russell and Nurse, 1986). Like in the wild-type cells, overexpression of mek1⁺ in cdc2-3w results in cell elongation, but this phenotype is largely suppressed in a cdc2-3w cdc25Δ strain (Fig. 5C). By contrast, wee1-50 cells overexpressing mek1⁺ at the restrictive temperature still manifest the elongation phenotype (Fig. 5C). These observations suggest that Cdc25, but not Wee1, is a target of Mek1.

Our results are consistent with the possibility that S. pombe Mek1 may be the meiosis-specific counterpart of the FHA family Cds1 checkpoint kinase (Fig. 1). When DNA integrity checkpoints are activated, Cds1 and Chk1 phosphorylate and inhibit Cdc25; several residues of Cdc25 phosphorylated by Cds1 have been identified (Zeng et al., 1998; Furnari et al., 1999; Zeng and Piwnica-Worms, 1999). In order to investigate whether the Mek1-induced G2/M arrest in vegetative cells is also mediated by phosphorylation of Cdc25, the effect of Mek1 overproduction was examined in the cdc25-9A mutant, which contains nine Cds1 phosphorylation sites changed to alanine and is impaired in the checkpoint response to DNA damage and replication blocks (Zeng and Piwnica-Worms, 1999). Importantly, the cell cycle arrest phenotype caused by Mek1 overproduction is significantly less severe in cdc25-9A cells compared to wildtype (Fig. 5D). Thus, our results suggest that Mek1 phosphorylates Cdc25 on at least some of the same residues as Cds1, leading to inactivation of Cdc25 function and resulting in cell cycle arrest. Nevertheless, since nmt1-mek1⁺cdc2-3w cdc25Δ and nmt1-mek1⁺cdc25-9A cells, in the absence of thiamine, still show a partial arrest (Fig. 5; data not shown), targets of Mek1 other than Cdc25 may exist.

The results reported above revealed that the consequences of high levels of Mek1 in vegetative cells to a certain extent mimic the DNA integrity checkpoint responses. However, mek1⁺ is normally expressed only during meiotic prophase; therefore, we investigated whether Mek1 carries out a checkpoint function during meiosis in fission yeast. The alterations observed in meiotic progression when mek1⁺ is deleted or when mek1⁺ is overexpressed (Fig. 4) are also consistent with such a regulatory role for Mek1.

In several organisms, including budding yeast, C. elegans and mouse, defects at intermediate steps in the meiotic recombination process trigger the so-called `pachytene checkpoint' or `meiotic recombination checkpoint', which blocks meiotic cell cycle progression, thus preventing the formation of defective gametes. To study whether a similar response also occurs in fission yeast, meiotic progression in the S. pombe meu13 mutant, which is defective in chromosome pairing and meiotic recombination (Nabeshima et al., 2001), was carefully examined using pat1-driven synchronous meiosis. Meu13 is the homolog of the S. cerevisiae Hop2 protein; the hop2 mutant triggers the pachytene checkpoint in budding yeast (Leu et al., 1998). The meu13 mutant completes meiosis and sporulation, as described previously (Nabeshima et al., 2001), but displays a ∼30 minute delay in entering meiosis I compared to wildtype (Fig. 6A). For example, at the 4 hour time point, only ∼15% of meu13 cells had undergone meiosis I, compared with ∼50% in the wild-type strain (see arrows in Fig. 6A; a representative time course is presented, but the meu13 delay has been observed in three independent experiments). Introduction of a rec12 mutation, which abolishes initiation of meiotic recombination suppresses this delay (data not shown), suggesting that it is due to the presence of recombination intermediates that trigger the meiotic checkpoint. Interestingly, mutation of mek1⁺ also alleviates the meiotic delay of meu13 (Fig. 6A); the meu13 mek1 double mutant proceeds to the first meiotic division with similar kinetics to that of the mek1 single mutant (i.e., even faster than wildtype; Fig. 4). Moreover, whereas spore viability in meu13 only shows a slight reduction in comparison with wildtype (62% versus 77%, respectively), it is significantly reduced in the meu13 mek1 double mutant (40%). Thus, during meiosis in fission yeast, Mek1 participates in a surveillance mechanism that delays cell cycle progression in response to defective recombination, which is important to promote viability of the meiotic progeny.

The above results (Fig. 5) suggest that Cdc25 is a target of Mek1. Since Cdc25 dephosphorylates Tyr15 of Cdc2 promoting G2/M transition, the status of Tyr15 phosphorylation was examined during synchronized meiosis of wild-type, mek1, meu13 and meu13 mek1 strains (Fig. 6B). In agreement with the checkpoint-dependent meiotic delay of the meu13 mutant, phosphorylation of Tyr15 persists longer than in wildtype (compare the 4 hour time point in Fig. 6B). By contrast, in both mek1 and meu13 mek1 strains, dephosphorylation of the Cdc2 tyrosine 15 occurs earlier, consistent with higher levels of Cdc25 phosphatase activity in the absence of Mek1 and correlating with a faster meiotic progression in these strains.

Our overexpression studies in vegetative cells suggest that the Mek1-dependent regulation of Cdc2 tyrosine 15 phosphorylation is exerted through inhibition of Cdc25. To directly demonstrate that Cdc25 is required for the fission yeast meiotic recombination checkpoint, meiotic progression was examined in strains carrying the phosphorylation-deficient cdc25-9A allele (Fig. 7). Like deletion of mek1⁺, the cdc25-9A mutant alleviates the meiotic delay of meu13; for example, at the 4 hour time point, ∼20% of the meu13 cells had undergone meiosis I, compared with ∼60% in the cdc25-9A meu13 double mutant (see arrows in Fig. 7).

In summary, our results indicate that the meiotic recombination checkpoint in fission yeast inhibits entry into meiosis I by Mek1-dependent inhibitory phosphorylation of Cdc25, which contributes, at least in part, to maintaining Cdc2 phosphorylated on tyrosine 15 (Fig. 8).

Here we report the functional characterization of the meiosis-specific mek1⁺ gene product in fission yeast. S. pombe Mek1 belongs to the family of protein kinases containing FHA domains. One group of this family, including S. cerevisiae Rad53, S. pombe Cds1 or mammalian Chk2, are key regulators of the checkpoint responses to DNA damage and/or replication blocks in mitotically dividing cells. Upon checkpoint activation, phosphorylation of different substrates by Rad53/Cds1/Ckh2 brings about different cellular responses, such as cell cycle arrest/delay and DNA repair (Rhind and Russell, 2000). By contrast, other members of the FHA family of kinases, including S. cerevisiae Mek1, C. elegans Chk2 and the S. pombe Mek1 protein described in this paper, function during meiosis. Consistent with their meiotic roles, these genes display a meiosis-specific pattern of expression (this work) (Rockmill and Roeder, 1991; Higashitani et al., 2000; MacQueen and Villeneuve, 2001; Oishi et al., 2001).

The mek1 mutant of S. cerevisiae is proficient in pairing, but displays reduced (although not abolished) meiotic recombination, reduced spore viability, makes only short stretches of SC and is defective in meiotic sister chromatid cohesion (Rockmill and Roeder, 1991; Bailis and Roeder, 1998). By contrast, inactivation of the chk-2 gene in C. elegans results in a strong pairing defect and lack of crossover recombination, but SC formation and chromosome morphogenesis are apparently normal (Higashitani et al., 2000; MacQueen and Villeneuve, 2001; Oishi et al., 2001). Our results indicate that, like the budding yeast homolog, S. pombe Mek1 is also required for normal levels of meiotic interhomolog recombination and spore viability. Because fission yeast lacks SC, no role for S. pombe Mek1 in SC development can be proposed, but the characteristic horse-tail morphology adopted by the prophase nucleus in S. pombe (Chikashige et al., 1994) appears to be normal in the mek1 mutant (L. P.-H., S.M. and P.A.S.-S., unpublished).

Despite the different meiotic phenotypes resulting from inactivation of these meiotic FHA kinases during an unperturbed meiosis in these model organisms, there is a common role for them; both budding and fission yeast Mek1, as well as Chk2 in worms, are essential components of the meiotic cell cycle control mechanism called pachytene checkpoint or meiotic recombination checkpoint, which arrests or delays meiotic cell cycle progression when recombination is incomplete (this work) (Bailis and Roeder, 2000; MacQueen and Villeneuve, 2001). It is possible that checkpoint function, like that of the mitotic counterparts of this protein family, could be the evolutionarily conserved role for these proteins, and the diverse meiotic phenotypes observed may reflect the peculiarities of meiosis in the different organisms as revealed, for example, by the different relationship between synapsis and initiation of recombination in S. cerevisiae and C. elegans or the absence of SC in S. pombe (Villeneuve and Hillers, 2001).

We have shown that the mek1 mutant in S. pombe shows reduced interhomolog meiotic recombination, decreased spore viability and a shorter meiotic prophase. In principle, the defect in meiotic recombination can be explained by a reduction in the number of initiating events (i.e., DSBs) and/or by increasing the number of DSBs repaired using a sister (instead of a nonsister) chromatid. In S. cerevisiae, Mek1 appears to be required both for the generation of wild-type levels of DSBs (Xu et al., 1997) as well as for the proper choice of recombination partner (Thompson and Stahl, 1999). Alternatively, the accelerated progression through prophase in the fission yeast mek1 mutant may cause entry into meiosis I with unrepaired/unresolved recombination intermediates in a fraction of cells, resulting in reduced recombination frequency and spore inviability. The observed random pattern of spore death in mek1 is consistent with this possibility and suggests that nondisjunction of homologs at meiosis I, as a consequence of the failure to recombine, is not the only cause of spore death in mek1 as this would result in an excess of asci with two or zero viable spores (Molnar et al., 1995).

In contrast to the original checkpoint definition (Hartwell and Weinert, 1989), the direct participation of checkpoint proteins in the monitored cell cycle event appears to be the rule rather than the exception. For example, in addition to the mek1 phenotypes mentioned above, other pachytene checkpoint mutants in budding yeast, such as rad24, rad17 and mec1-1, show decreased crossing over, increased ectopic recombination, increased unequal sister-chromatid exchange, defective chromosome synapsis and reduced spore viability (Lydall and Weinert, 1995; Lydall et al., 1996; Grushcow et al., 1999; Thompson and Stahl, 1999). The same observation applies for DNA damage and replication checkpoint proteins such as Rad53 (another FHA Kinase member; Fig. 1) and Mec1, which perform essential functions during DNA replication in the mitotic cell cycle (Desany et al., 1998).

We have observed that kinetics of meiotic progression in fission yeast depends on Mek1 dosage. Lack of Mek1 results in a more rapid entry into meiosis I, whereas high levels of Mek1 lead to delayed meiotic progression. Premeiotic S phase is not affected, suggesting that Mek1 function negatively regulates the prophase to meiosis I transition. Supporting this notion, western blot analysis of Mek1 throughout meiotic time courses revealed that the protein rapidly disappears as cells enter the first meiotic division.

Since little is known about the molecular mechanisms controlling the meiotic cell cycle in fission yeast in comparison with the regulation of the mitotic cell cycle, we used ectopic overexpression of mek1⁺ in vegetative cells as a tool for identifying potential cell cycle targets of Mek1. We found that high levels of Mek1 in vegetative cells lead to G2/M arrest, the same effect observed when the homologous Cds1 checkpoint kinase is overproduced (Boddy et al., 1998). In S. pombe, the G2/M transition depends on the phosphorylation status of Cdc2 on Tyr15 (reviewed by Moser and Russell, 2000). The Wee1 and Mik1 kinases inhibit Cdc2 activity by phosphorylation of Tyr15, whereas the Cdc25 phosphatase activates Cdc2 by removing the phosphate of Tyr15, thus promoting the G2/M transition. Our results suggest that Mek1-induced cell cycle arrest partly results from inhibition of Cdc25 rather than activation of Wee1, because the effect of Mek1 overproduction, as manifested by cell elongation, is diminished in cdc25Δ and cdc25-9A strains, but is not significantly altered in the absence of Wee1 function. The cdc25-9A mutant lacks the relevant Cds1 phosphorylation sites and shows a much weaker cell cycle arrest response to high Mek1 levels. Given the sequence similarity between Mek1 and Cds1 (Fig. 1), these results strongly suggest that Mek1 directly phosphorylates Cdc25, promoting its inhibition. Nevertheless, our results indicate that Cdc25 is not the only target of Mek1; the possible involvement of Mik1 in the Mek1-dependent regulation of cell cycle remains to be investigated.

Our results show that defects at intermediate steps in the recombination pathway (induced by a meu13 mutation) trigger a meiotic cell cycle delay in fission yeast mediated by inhibitory phosphorylation of the cyclin-dependent kinase Cdc2 on Tyr15. Like its budding yeast homolog, S. pombe Mek1 is an important component of this meiotic recombination checkpoint; mutation of mek1⁺ alleviates the meu13 delay. Although the checkpoint-induced delay is not very prolonged (∼30 minutes), it appears to be important for the viability of the meiotic products. The extent of the meiotic delay in meu13 roughly correlates with a delay in DSBs repair (Shimada et al., 2002); therefore it is formally possible that S. pombe mutants with more profound defects in recombinational repair of DSBs may exhibit stronger meiotic delays. However, although detailed analysis of meiotic progression in such mutants has not been reported, there is no evidence for a robust meiotic block in fission yeast. For example, the rhp51 mutant is able to complete meiosis and sporulation despite the presence of unrepaired DSBs (Zenvirth and Simchen, 2000; Boddy et al., 2001).

The pachytene checkpoint pathway has been extensively studied in S. cerevisiae, and several components have been identified (see introduction) (reviewed by Roeder and Bailis, 2000). It has been proposed that Mek1-dependent phosphorylation of Red1 is required for checkpoint-induced arrest in response to unrepaired recombination intermediates. Once recombination has been completed, dephosphorylation of Red1 by Glc7 allows pachytene exit and entry into meiosis I (Bailis and Roeder, 1998; Bailis and Roeder, 2000). Pachytene arrest is achieved, at least in part, by inhibition of Cdc28/Clb1 activity, both by Swe1-mediated inhibitory phosphorylation of Cdc28 on Tyr19 (the equivalent of Tyr15 in S. pombe Cdc2) and by limiting Ndt80-dependent transcription of CLB1 (Chu and Herskowitz, 1998; Hepworth et al., 1998; Leu and Roeder, 1999; Tung et al., 2000). An additional branch of the checkpoint, which targets the Sum1 transcriptional repressor has been recently reported (Lindgren et al., 2000). When the pachytene checkpoint is activated, the Swe1 kinase accumulates in a hyperphosphorylated (and presumably activated) form (Leu and Roeder, 1999); however, the molecular mechanisms regulating Swe1 stability and phosphorylation are unknown.

As mentioned above, in S. cerevisiae, phosphorylation of Red1 (a component of the lateral elements of the SC) by Mek1 is important for transducing the pachytene checkpoint signal. In S. pombe, however, there is no SC and, although SC-like structures called linear elements have been described (Bahler et al., 1993), BLAST searches reveal that no obvious RED1 homolog exist in the fission yeast genome (http://www.sanger.ac.uk/Projects/S_pombe/). Therefore, targets of Mek1 other than Red1 must exist in fission yeast. Our experiments expressing Mek1 in different S. pombe mutant backgrounds during vegetative growth suggest that Mek1 phosphorylates Cdc25. Moreover, we have shown that the phosphorylation-deficient cdc25-9A mutant is impaired in the meiotic checkpoint response to the presence of recombination intermediates. We thus provide the first evidence of a direct connection between an effector kinase of the meiotic recombination checkpoint (Mek1) and a key component of the cell cycle machinery (Cdc25). Consistent with our findings, an essential role for Cdc25 in promoting meiosis I has been reported (Iino et al., 1995).

As in budding yeast, here we show that the meiotic recombination checkpoint in S. pombe regulates the phosphorylation of Cdc2 on Tyr15; however, we propose that the checkpoint-induced meiotic delay would, in part, be mediated by the Mek1-dependent inhibitory phosphorylation of Cdc25 and not by the activation of Wee1. This contrasts with the situation in S. cerevisiae, in which Mih1 (the Cdc25 homolog) seems to be dispensable for pachytene checkpoint function (Leu and Roeder, 1999). Thus, although the meiotic checkpoint arrests or delays meiosis by maintaining inhibitory phosphorylation of the cyclin-dependent kinase in both budding and fission yeast, different cell cycle regulators are targeted in each organism. It has been recently described that the Mrc1 protein is required for activation of Rad53 and Cds1 during the replication checkpoint in both budding and fission yeast (Alcasabas et al., 2001; Tanaka and Russell, 2001). Whether activation of Mek1 by the pachytene checkpoint also requires Mrc1 or other adaptor proteins remains to be tested.

In agreement with our observations, during the preparation of this paper, the existence of a meiotic recombination checkpoint in S. pombe has also been reported (Shimada et al., 2002). These authors show that the mitotic DNA integrity checkpoint Rad proteins also respond to unrepaired DSBs during meiosis in fission yeast, as they do in S. cerevisiae (Lydall et al., 1996). It has been recently shown that Mec1 and the Rad24 group of budding yeast checkpoint proteins do indeed localize to the sites of DSBs, acting as sensors of damage (Kondo et al., 2001; Melo et al., 2001; Hong and Roeder, 2002). Combining our results and those of Shimada et al., we propose the following model for the action of the meiotic recombination checkpoint in fission yeast (Shimada et al., 2002) (Fig. 8). The presence of ongoing recombination (presumably unrepaired DSBs) is sensed by the group of Rad checkpoint proteins, generating a signal that results in activation of the Mek1 kinase. Mek1, in turn, phosphorylates Cdc25, and possibly other as yet unknown substrate(s), which contribute to the maintenance of Tyr15 phosphorylation of Cdc2, thus inhibiting meiosis I entry. When recombination has been completed, the inhibitory signal disappears and dephosphorylation of Cdc2 on Tyr15 promotes the first meiotic division.

We thank Miguel Blanco, Cristina Martín, Rafael Daga and Beatriz Santos for comments on the manuscript. We are grateful to Miguel Blanco and Rafael Daga for invaluable advice and discussion throughout this work. We also thank Helen Piwnica-Worms and Juan Carlos Ribas for providing plasmids and strains. L.P.-H. is a recipient of a predoctoral fellowship from CSIC, Spain. P.A.S.-S. is a `Ramón y Cajal' investigator of the Ministry of Science and Technology of Spain. This work was supported by grants from CICYT and the European Union to S.M.