|Home | About | Journals | Submit | Contact Us | Français|
DNA double-strand breaks (DSBs) can arise at unpredictable locations after DNA damage or in a programmed manner during meiosis. DNA damage checkpoint response to accidental DSBs during mitosis requires the Rad53 effector kinase, whereas the meiosis-specific Mek1 kinase, together with Red1 and Hop1, mediates the recombination checkpoint in response to programmed meiotic DSBs. Here we provide evidence that exogenous DSBs lead to Rad53 phosphorylation during the meiotic cell cycle, whereas programmed meiotic DSBs do not. However, the latter can trigger phosphorylation of a protein fusion between Rad53 and the Mec1-interacting protein Ddc2, suggesting that the inability of Rad53 to transduce the meiosis-specific DSB signals might be due to its failure to access the meiotic recombination sites. Rad53 phosphorylation/activation is elicited when unrepaired meiosis-specific DSBs escape the recombination checkpoint. This activation requires homologous chromosome segregation and delays the second meiotic division. Altogether, these data indicate that Rad53 prevents sister chromatid segregation in the presence of unrepaired programmed meiotic DSBs, thus providing a salvage mechanism ensuring genetic integrity in the gametes even in the absence of the recombination checkpoint.
Chromosomal breaks can occur at unpredictable locations in the genome of eukaryotic cells as a result of ionizing radiation, radiomimetic chemicals, or DNA replication across nicked DNA. Moreover, they are introduced in a programmed manner to initiate meiotic recombination during meiosis, the specialized differentiation process in which two rounds of chromosome segregation follow one round of DNA replication (reviewed in references 21 and 26). In the first meiotic division (meiosis I) homologous chromosomes pair and separate, whereas sister chromatids segregate from each other in the second division (meiosis II). Both unprogrammed and programmed double-strand breaks (DSBs) can be repaired by homologous recombination. The primary function of homologous recombination in mitotic cells is to repair DSBs, whereas, during meiosis, it is essential to establish a physical connection between homologous chromosomes, thus ensuring their correct pairing and subsequent segregation at the first meiotic division. In any case, for crossovers to be functional in promoting meiosis I execution in Saccharomyces cerevisiae, they must occur in the context of a proteinaceous tripartite structure, the synaptonemal complex, which connects the axes of homologs along their entire lengths via a close-packed array of transverse filaments (reviewed in reference 36).
Programmed meiotic DSB formation requires the product of the meiosis-specific gene SPO11, which, together with several other factors, breaks both strands of a DNA molecule, creating a DSB with covalent linkages between the newly created 5′ DNA ends and a Spo11 catalytic tyrosine residue (18, 19). In S. cerevisiae, the highly conserved Sae2 protein and the MRX (Mre11-Rad50-Xrs2) complex catalyze the endonucleolytic cleavage of Spo11 from the 5′ DSB ends (11, 27, 28, 29, 32, 38). After Spo11 removal, one or more nucleases resect the break to generate 3′-ended single-stranded DNA (ssDNA) overhangs. The RecA-like strand exchange proteins Rad51 and Dmc1 bind such tails to form presynaptic nucleoprotein filaments, which engage in the search for homologous templates (reviewed in reference 31).
Both accidental and programmed DSB repair are coupled to cell cycle progression by surveillance mechanisms, named DNA damage checkpoint and recombination checkpoint, which delay mitotic and meiotic cell cycle progression, respectively, until DSB repair is achieved (reviewed in references 14, 21, and 22). Mechanistically, the two checkpoints are related to each other. In fact, DSB detection is accomplished in both cases by highly conserved protein kinases, among which mammalian ataxia telangiectasia mutated (ATM) and ataxia telangiectasia and RAD3-related (ATR), as well as their S. cerevisiae orthologues Tel1 and Mec1. During the mitotic cell cycle, Tel1/ATM appears to bind unprocessed DSBs via the MRX/MRN complex, and its signaling activity is disrupted when DSB termini are resected (25, 30). By contrast, Mec1/ATR is thought to recognize ssDNA regions that arise after DSB processing (55).
During the meiotic cell cycle, unrepaired programmed DSBs are stably generated in S. cerevisiae sae2Δ or rad50s mutants, where Spo11 remains covalently attached to the DSB ends that therefore cannot be resected (27, 38). These meiotic aberrant intermediates activate a Tel1-dependent recombination checkpoint, which slows down meiosis I (47, 52). Unrepaired meiosis-specific DSBs with unusually long single-stranded tails are instead generated in S. cerevisiae cells lacking the strand exchange protein Dmc1 (5). These cells are competent to remove Spo11 from the DSB ends but are defective in strand invasion. Similarly to the DNA damage checkpoint, where generation of 3′-ended ssDNA results in Mec1 recruitment and Mec1-dependent checkpoint activation (55), activation of the recombination checkpoint in dmc1Δ mutants is dependent on Mec1 and its regulators Rad24 and Rad17 (23). Despite the persistence of unrepaired meiotic DSBs, sae2Δ and rad50s cells display only a transient Tel1-dependent delay of meiosis I, whereas dmc1Δ cells exhibit a permanent Mec1-dependent meiosis I block (47, 52), suggesting that Tel1 can sense and signal meiotic DSBs less efficiently than Mec1. Consistent with this hypothesis, Mec1 responds to a single DSB in mitosis, whereas Tel1 signaling activity becomes apparent only when multiple DSBs are generated in the absence of Mec1 (25). Interestingly, Sae2 undergoes Mec1- and Tel1-dependent phosphorylation during meiosis, with a peak at the time of DSB generation (8). Mutations altering the Sae2 [S/T]Q motifs preferred for phosphorylation by ATM/ATR-like kinases lead to the accumulation of unprocessed DSBs, as does the simultaneous absence of Mec1 and Tel1 (8), suggesting that the latter may allow DSB resection by phosphorylating Sae2.
Propagation of the checkpoint signals to the downstream targets occurs in two different ways, depending on whether the checkpoint response is elicited by accidental DSBs or by programmed meiotic DSBs. In fact, DNA damage checkpoint activation requires the effector kinase Rad53 and its adaptor Rad9 (12, 44). Rad9 first promotes Mec1-Rad53 interaction and Mec1-mediated Rad53 phosphorylation/activation (44) and then acts as a scaffold to facilitate in trans Rad53 autophosphorylation (12). Despite their essential role in activating the DNA damage checkpoint in response to mitotic DSBs, Rad9 and Rad53 do not appear to be involved in controlling meiosis I progression in response to meiotic programmed DSBs (23). This control instead requires the meiosis-specific proteins Mek1, Red1, and Hop1. In particular, meiotic DSB formation leads to Mec1- and Tel1-dependent Hop1 phosphorylation, which is required for Mek1 activation (6, 33, 34, 49). However, inactivation of HOP1, RED1, or MEK1 in dmc1Δ cells leads to efficient repair of the breaks via intersister recombination, indicating that meiotic progression in these cells is a consequence of inappropriate repair rather than an arrest relief (33, 34, 52).
In this study, we investigated the role of Rad53 in responding to DSBs during the meiotic cell cycle. We show that Rad53 is not phosphorylated and activated as soon as programmed meiosis-specific DSBs occur, suggesting that such DSBs are hidden from the canonical Rad53-dependent DNA damage checkpoint machinery. However, Rad53 phosphorylation is triggered when unrepaired meiotic DSBs escape the recombination checkpoint-mediated prophase I arrest. This Rad53 phosphorylation and activation result in the slowing down of meiosis II.
Yeast strains used for this work are listed in Table Table1.1. All the strains were SK1 derivatives that were isogenic with the NKY3000 (MATa/MATα HO/HO lys2/lys2 ura3::hisG/ura3::hisG leu2::hisG/leu2::hisG) strain, kindly provided by N. Kleckner (Harvard University, Cambridge, MA) and R. Cha (Medical Research Council, London, United Kingdom). Heterozygous diploid strains carrying deletions of the MEC1, TEL1, SML1, DMC1, SAE2, MEK1, RAD9, NDT80, RAD54, or SPO11 gene were obtained by one-step PCR disruption. Diploid strains homozygous for the above deletions were obtained after tetrad dissection of the corresponding heterozygous strains and self-diploidization of the spore carrying the desired deletion. Heterozygous diploid strains carrying the MEK1-HA3 allele at the MEK1 locus were generated by PCR one-step tagging, as previously described (8). Homozygous MEK1-HA3 diploid strains were obtained by self-diploidization of spores with the desired tagged allele derived from the corresponding heterozygous strain. The MEK1-HA3 allele was shown to be fully functional, since all the strains carrying it were undistinguishable from the isogenic untagged strains with respect to meiotic cell cycle progression and meiotic DSB repair. pRS316 DDC2-RAD53-3FLAG (DDC2-RAD53) and pRS316 DDC2-rad53K227A D339A-3FLAG (DDC2-rad53kd) plasmids, used to transform wild-type (NKY3000) and dmc1Δ strains, were kindly provided by D. Stern (University of California, San Francisco) (20). The accuracy of all gene replacements and integrations was verified by Southern blot analysis or PCR.
The strains of interest were patched on YEPD (1% yeast extract, 2% Bacto peptone, 50 mg/liter adenine, 2% glucose) plates from a −80°C glycerol stock and incubated overnight at 30°C. To obtain synchronous G1/G0 cell population, overnight liquid YEPD cell cultures from these patches were diluted to a final concentration of 1 × 107 cells/ml in 200 ml YPA (1% yeast extract, 2% Bacto peptone, 1% potassium acetate) in a 2-liter flask and grown with vigorous shaking for 13 h at 30°C. Cells were then washed and transferred into the same volume of SPM (0.3% potassium acetate, 0.02% raffinose) to induce meiosis. Flow cytometric DNA analysis was determined on a Becton-Dickinson FACScan sorter. Nuclear division was scored with a fluorescence microscope in propidium iodide-stained cells.
Genomic DNA was purified from cells collected from synchronized meiotic cultures, digested with EcoRI, and separated on native agarose gels. DSBs at the THR4 hot spot were detected with a 32P-labeled 1.6-kb fragment spanning the 5′ region of THR4 as described in reference 16. This probe was obtained by PCR using oligonucleotides PRP686 (5′-GGG GTA CCC CCA AGG TAA AAT TTC ACC GCG-3′) and PRP687 (5′-GGG GTA CCC CGG CGT GCA ATA ATT GCA GAA-3′) as primers and genomic DNA as the template.
DSB end resection at the THR4 hot spot was analyzed on alkaline agarose gels. The single-stranded probe used to detect DSB resection was obtained by in vitro transcription using Promega Riboprobe System-T7 and plasmid pML601 as the template. The latter was constructed by inserting in the pGEM-7Zf EcoRI site a 700-bp fragment containing part of the THR4 locus (coordinates 212503 to 213199 on chromosome III), obtained by PCR by using yeast genomic DNA as template and PRP924 (5′-CGG AAT TCC ATG GAT GTT CTT GGG CTG GAT-3′) and PRP925 (5′-CGG AAT TCT GCA TGA AGA ACT GTG CCG TGA-3′) as primers.
The in situ autophosphorylation assay (ISA) was performed as described in reference 37. For Western blot analysis, protein extracts were prepared by trichloroacetic acid precipitation as previously described (35). Hemagglutinin-tagged proteins were detected with the monoclonal antibody 12CA5. FLAG-tagged proteins were detected with monoclonal anti-FLAG antibodies purchased from Sigma. Rad53 was detected using anti-Rad53 polyclonal antibodies kindly provided by J. Diffley (Clare Hall Laboratories, South Mimms, United Kingdom). Rad9 was detected using anti-Rad9 polyclonal antibodies kindly provided by N. Lowndes (National University of Ireland, Ireland).
Mec1- and Tel1-dependent phosphorylation of Mek1 and Rad53 is required for their activation as protein kinases and can be detected as changes in their electrophoretic mobility (6, 12, 33, 39, 43, 44). It is known that Mek1 phosphorylation is induced by programmed DSBs that occur during meiosis (3, 6), whereas exogenous DNA lesions trigger Rad53 phosphorylation during mitosis (39). Thus, we asked whether treatment with DSB-inducing agents of cells undergoing meiosis in the absence of programmed DSBs, due to the lack of Spo11, could trigger Rad53 and Mek1 phosphorylation. Because meiotic programmed DSBs occur after completion of premeiotic DNA synthesis (time = 210 min), spo11Δ cells were treated for 30 min with the radiomimetic drug phleomycin 210 min after meiosis induction (Fig. (Fig.1A).1A). As expected, due to their inability to generate meiotic DSBs, spo11Δ cells failed to phosphorylate Mek1 after meiosis induction in the absence of phleomycin (Fig. (Fig.1B,1B, top). In contrast, both Rad53 and Mek1 were phosphorylated after phleomycin addition in spo11Δ cells (Fig. (Fig.1B),1B), indicating that exogenous DNA damage during meiosis can trigger both Rad53 and Mek1 phosphorylation. Phleomycin-treated spo11Δ cells slowed down meiosis I compared to the untreated cells, and this delay was dependent on the recombination checkpoint (Fig. (Fig.1C).1C). In fact, phleomycin-treated spo11Δ mek1Δ cells underwent meiosis I with kinetics similar to those of untreated spo11Δ mek1Δ cells and faster than those of phleomycin-treated spo11Δ cells (Fig. (Fig.1C).1C). In contrast, when Rad53 phosphorylation was prevented by eliminating its regulator Rad9 (Fig. (Fig.1B,1B, bottom), phleomycin-treated spo11Δ rad9Δ cells still slowed down meiosis I (Fig. (Fig.1C),1C), indicating that Rad53 activation was not responsible for this delay. Thus, although Rad53 can be phosphorylated and activated in response to chemically induced DSBs, it does not induce arrest of meiosis I. Interestingly, although phleomycin-treated spo11Δ mek1Δ cells performed meiosis I with kinetics similar to those of the isogenic untreated cells, they still suffered a 60-min delay of meiosis II (Fig. (Fig.1C),1C), suggesting that the phleomycin-induced meiosis II delay was not simply the consequence of the meiosis I delay.
The finding that Rad53 can be phosphorylated in response to chemically induced DSBs during the meiotic cell cycle prompted us to ask whether it can be phosphorylated/activated in response to meiosis-specific DSBs. Because the inability to repair meiotic DSBs is known to activate the recombination checkpoint, we analyzed the pattern of Rad53 and Mek1 phosphorylation not only during an unperturbed meiosis but also when DSB repair was prevented by the lack of Dmc1 (Fig. (Fig.2A).2A). We also monitored in the same samples the kinetics of DSB formation at the THR4 hot spot (Fig. (Fig.2B)2B) by Southern blot analysis of EcoRI-digested genomic DNA run on a native agarose gel. As shown in Fig. Fig.2,2, Mek1 was phosphorylated in both wild-type and dmc1Δ cells after transfer to sporulation medium, concomitantly with DSB formation (time = 210 min). Then, phosphorylated Mek1 decreased when DSBs were repaired in wild-type cells (time = 300 min), whereas it persisted until the end of the experiment in dmc1Δ cells (Fig. (Fig.2A,2A, top), which remained arrested with undivided nuclei (data not shown) and accumulated unrepaired hyperresected DSBs (Fig. (Fig.2B).2B). In contrast, Rad53 was not phosphorylated in wild-type cells and only slightly in dmc1Δ diploid cells after transfer to SPM medium (Fig. (Fig.2A,2A, bottom). The inability to phosphorylate Rad53 was not due to competition with Mek1 for Mec1- and Tel1-mediated phosphorylation. In fact, Rad53 was only slightly phosphorylated even in meiotic mek1Δ cells (Fig. (Fig.2A,2A, bottom). Thus, meiosis-specific programmed DSBs fail to trigger Rad53 phosphorylation both in the presence and in the absence of Mek1.
Because Mec1 and Tel1 are required to phosphorylate and activate Mek1 in response to meiotic programmed DSBs, whereas chemically induced DSBs trigger Rad53 phosphorylation by the same kinases, the choice of using Mek1 instead of Rad53 in the recombination checkpoint cannot likely be exerted at the levels of Mec1- and/or Tel1-mediated DSB recognition. One possibility is that the DSB signals can be easily transduced to Mek1, Red1, and Hop1, because they are structural components of the meiosis-specific chromosome structure, whereas Rad53 activation might be prevented by its inability to detect Mec1/Tel1 signaling at meiotic DSBs. If this were the case, artificial targeting of Rad53 to Mec1 by fusing it with the Mec1 regulatory subunit Ddc2 (35) should result in Rad53 phosphorylation/activation in response to meiotic DSB formation. In order to analyze this possibility, wild-type and dmc1Δ cells were transformed with a plasmid carrying a DDC2-RAD53-FLAG in-frame fusion, whose expression was driven by the RAD53 upstream regulatory sequences (20). The encoded Ddc2-Rad53 fusion protein was shown to be activated in a Mec1- and Tel1-dependent manner upon DNA damage and to circumvent the Ddc2, Rad24, Rad17, Mec3, and Rad9 requirement for Rad53 activation (20). It is known that Rad53 activation requires Mec1- and Tel1-dependent phosphorylation, which allows Rad53 molecules to undergo in trans autophosphorylation, thus completing the Rad53 activation process (12, 37, 44). We therefore monitored the activation of the Ddc2-Rad53 chimera by evaluating its electrophoretic mobility by Western blot analysis, as well as its autophosphorylation activity by an ISA (37). As a control for Rad53 activation, we analyzed also isogenic cells expressing a Ddc2-Rad53kd fusion, where Ddc2 was fused to the kinase-defective Rad53K227A D339A variant (20). As shown in Fig. Fig.3A,3A, wild-type and dmc1Δ cells expressing either the Ddc2-Rad53 or the Ddc2-Rad53kd chimeras initiated and completed premeiotic S phase with similar kinetics. Strikingly, slowly migrating forms of the Ddc2-Rad53 fusion (indicated as Ddc2-Rad53), presumably due to phosphorylation events, appeared in both wild-type and dmc1Δ cells (Fig. (Fig.3B)3B) concomitantly with meiotic DSB formation (time = 210 min; Fig. Fig.3C).3C). Then, they decreased in wild-type cells when DSBs were repaired (time = 360 min), whereas they were detectable as the predominant forms until the end of the experiment in dmc1Δ cells (Fig. (Fig.3B),3B), where DSBs could not be repaired (data not shown). Consistent with the finding that Rad53 phosphorylation was not induced by meiotic DSB formation (Fig. (Fig.2A),2A), the endogenous Rad53 protein (indicated as Rad53) did not show significant changes in electrophoretic mobility in either wild-type or dmc1Δ cells (Fig. (Fig.3B3B).
When the same extracts were tested by ISA, a phosphorylated form corresponding to the Ddc2-Rad53 fusion protein was detected in both wild-type and dmc1Δ cells (time = 210 to 240 min) (Fig. (Fig.3D),3D), and the corresponding phosphorylation reaction was dependent on Rad53 kinase activity. In fact, no phosphorylated forms were detectable by ISA in either wild-type or dmc1Δ cells expressing the Ddc2-Rad53kd fusion protein (Fig. (Fig.3D).3D). Moreover, the simple proximity of Rad53 to Ddc2 in the Ddc2-Rad53 fusion was not sufficient to activate Rad53 in the absence of meiotic DSBs. In fact, similarly to the Ddc2-Rad53 mobility shift, Rad53 autophosphorylation activity became detectable in wild-type cells at the time of DSB formation (time = 210 to 240 min) and it decreased when DSBs were repaired (time = 360 min) (Fig. 3C and D), while it persisted until the end of the experiment in dmc1Δ cells that failed to repair meiotic DSBs (Fig. (Fig.3D)3D) and arrested with undivided nuclei due to the presence of Mek1 (data not shown). Thus, targeting Rad53 to Mec1 through its fusion with Ddc2 triggers Rad53 activation in response to meiotic programmed DSBs. Consistent with our previous observation that Rad53 activation by exogenous DSBs did not result in meiosis I delay, wild-type cells expressing the Ddc2-Rad53 or the Ddc2-Rad53kd fusion protein progressed through meiosis with similar kinetics (data not shown).
If the chromosome structure specifically formed during meiosis I inhibits Rad53 access to the meiotic DSB signals, unrepaired meiotic DSBs might be capable of inducing Rad53 phosphorylation once homologous chromosomes have separated from each other and cells enter meiosis II. Because the lack of Sae2 allows meiotic cells to perform meiosis I in the presence of unprocessed DSBs (Fig. 4C and D) (47, 52), we monitored Rad53 phosphorylation in sae2Δ cells after meiosis induction (Fig. 4A and B). Rad53 phosphorylation was detectable in sae2Δ cells about 300 min after meiosis induction, and it was DSB dependent, because it was prevented in spo11Δ sae2Δ cells (Fig. (Fig.4B,4B, top). Rad53 phosphorylation in sae2Δ cells occurred concomitantly with homologous chromosome segregation (time = 300 min) (Fig. (Fig.4B,4B, top, and C), well after DSB formation (Fig. (Fig.4D),4D), whereas Mek1 phosphorylation became detectable in the same sae2Δ cells at the time of meiotic DSB formation (time = 210 min) (Fig. (Fig.4B,4B, bottom, and D). This suggests that unrepaired meiotic DSBs become capable of activating Rad53 after homologous chromosome segregation, whose inhibition might therefore prevent Rad53 phosphorylation in meiotic sae2Δ cells. To address this point, we monitored Rad53 phosphorylation in sae2Δ cells lacking the meiosis-specific transcription factor Ndt80, which is required to activate transcription of middle meiosis genes (9) and whose lack causes meiotic cells to arrest at the pachytene stage of meiosis I (51). We found that Rad53 was not phosphorylated in sae2Δ ndt80Δ cells (Fig. (Fig.4B,4B, top), which, as expected, failed to divide nuclei (Fig. (Fig.4C).4C). The inability of sae2Δ ndt80Δ cells to phosphorylate Rad53 was not due to the failure to generate meiotic DSBs, because both these cells and sae2Δ cells phosphorylated Mek1 and accumulated unrepaired meiotic DSBs with similar kinetics (Fig. (Fig.4B,4B, bottom, and D). Thus, unrepaired meiotic DSBs seem to induce Rad53 phosphorylation only after homologous chromosome segregation.
In order to further investigate this possibility, we asked whether Rad53 underwent phosphorylation in dmc1Δ cells that were allowed to segregate their homologous chromosomes due to the lack of Mek1. Because MEK1 deletion allows repair of meiotic DSBs in dmc1Δ cells by using sister chromatids (34), we also deleted the RAD54 gene that is required for this repair process in dmc1Δ mek1Δ cells (34). As shown in Fig. Fig.5,5, Rad53 phosphorylation became detectable in rad54Δ dmc1Δ mek1Δ cells at the time of homologous chromosome segregation (time = 300 min). In contrast, it was under the detection level in dmc1Δ cells (Fig. (Fig.5A),5A), which failed to complete meiosis I (Fig. (Fig.5B),5B), and in dmc1Δ mek1Δ cells (Fig. (Fig.5A),5A), where meiotic DSBs were repaired (34; data not shown). Therefore, unrepaired meiotic DSBs induce Rad53 phosphorylation after execution of meiosis I.
Both sae2Δ cells and dmc1Δ mek1Δ rad54Δ cells exhibit a delay in segregating sister chromatids during meiosis II (Fig. (Fig.4C4C and and5B).5B). Since meiosis II is functionally equivalent to mitosis, we asked whether Rad53 activation was responsible for the sae2Δ-induced meiosis II delay. We were unable to synchronize rad53Δ sml1Δ meiotic cell cultures, because the lack of Rad53 impairs cell viability and mitotic cell cycle progression even in the absence of Sml1. We therefore inactivated Rad53 by deleting the RAD9 gene, whose product is required for Rad53 activation in response to DNA damage during the mitotic cell cycle (12, 44), while it is not involved in the recombination checkpoint (23). Rad9 is known to undergo phosphorylation by cyclin-dependent kinases in the mitotic cycle (46) as well as Mec1- and Tel1-dependent hyperphosphorylation in response to DNA damage (48). We found that Rad9 was hyperphosphorylated about 240 min after meiosis induction in sae2Δ cells (asterisk in Fig. Fig.6A),6A), similarly to Rad53 in the same cells (Fig. (Fig.6C).6C). Furthermore, meiotic sae2Δ rad9Δ cells did not phosphorylate Rad53 (Fig. 6B and C), confirming that Rad53 activation after segregation of unrepaired homologous chromosomes depends on Rad9. Consistent with the knowledge that the Mek1-dependent checkpoint is responsible for the meiosis I delay of sae2Δ cells (52), both sae2Δ and sae2Δ rad9Δ cells showed similar delays in meiosis I execution compared to wild-type and rad9Δ cells (Fig. (Fig.6D).6D). In contrast, meiosis II started earlier in sae2Δ rad9Δ cells than in sae2Δ cells (Fig. (Fig.6D),6D), although it was still delayed compared to wild-type cells, due to the Mek1-dependent slowing down of meiosis I. Thus, Rad9-dependent activation of the Rad53 kinase is likely responsible for the meiosis II delay in sae2Δ cells.
The sae2Δ mutation prevents Spo11 removal from meiotic DSB ends, leading to the accumulation of unresected DSBs, which in turn triggers a delay of both meiosis I and meiosis II (1, 8). This delay is dependent on both Mec1 and Tel1, with Tel1 playing the major role (47). In contrast, Tel1 is dispensable for the meiosis I arrest of dmc1Δ cells, which are known to accumulate unrepaired DSBs with unusually long single-stranded tails that are monitored by Mec1 (5, 23).
To investigate whether phosphorylation of Mek1 and Rad53 in meiotic sae2Δ cells had the same genetic requirements, we monitored such phosphorylation events after meiosis induction in sae2Δ tel1Δ and sae2Δ mec1Δ cells, the latter being kept viable by SML1 deletion (54). The absence of Mec1 or Tel1 did not affect the kinetics of either premeiotic DNA replication (data not shown) or DSB accumulation (Fig. (Fig.7B).7B). Although inactivation of either Mec1 or Tel1 affected both Mek1 and Rad53 phosphorylation in sae2Δ cells, their effects were quantitatively different. In fact, Mek1 phosphorylation was dramatically reduced in sae2Δ tel1Δ cells compared to sae2Δ cells, whereas it was only slightly affected in sae2Δ mec1Δ cells under the same conditions (Fig. (Fig.7A,7A, left). In contrast, Rad53 phosphorylation was undetectable in sae2Δ mec1Δ cells, whereas its amount was reduced in sae2Δ tel1Δ cells compared to sae2Δ cells (Fig. (Fig.7A,7A, right). Thus, Tel1 has a major role in triggering Mek1 phosphorylation in sae2Δ cells, while Rad53 phosphorylation in the same cells is primarily dependent on Mec1, suggesting that the signals eliciting Mek1 and Rad53 phosphorylation in sae2Δ cells undergoing meiosis I and meiosis II, respectively, are different.
Because Mec1 is known to detect and signal ssDNA (55), we wondered whether sae2Δ unresected DSB ends might undergo some processing after homologous chromosome segregation, thus allowing their detection by Mec1 and subsequent Rad53 phosphorylation. To test this hypothesis, we monitored ssDNA formation at two meiosis-specific DSBs within the THR4 locus (Fig. (Fig.7C).7C). As a control for DSB end resection, we also analyzed dmc1Δ cells, which are known to accumulate hyperresected DSBs. As shown in Fig. Fig.7D,7D, 3′-ended ssDNA resection products could be detected in meiotic sae2Δ cells at both DSBs (rDSB I and rDSB II), although they appeared later and in lower amounts than in dmc1Δ cells under the same conditions. The ssDNA regions appeared in sae2Δ cells about 270 to 300 min after meiosis induction, at the time of Rad53 phosphorylation, when most cells had completed meiosis I (data not shown). Therefore, segregation of the homologous chromosomes containing unrepaired DSBs in sae2Δ cells may allow some DSB processing by unknown mechanisms, thus generating ssDNA regions that can be detected by Mec1 and induce Rad53 phosphorylation.
Accumulation of unrepaired meiosis-specific DSBs is known to activate the recombination checkpoint that arrests the meiotic cell cycle prior to meiosis I (Fig. (Fig.8B).8B). On the other hand, the DNA damage checkpoint senses and signals DSBs that arise at unpredictable locations as a consequence of DNA damage, thus delaying the mitotic G2/M transition (reviewed in reference 21). Although these two checkpoint mechanisms share the sensor kinases Mec1/ATR and Tel1/ATM, the meiosis-specific Mek1 kinase is the effector of the recombination checkpoint, while the Rad53 kinase is known to be essential for transducing the DNA damage checkpoint signals during mitosis.
Here, we provide evidence that Rad53 can be phosphorylated during the meiotic cell cycle after generation of chemically induced DSBs, indicating that Rad53 can be activated in response to DNA damage also during meiosis. However, Rad53 activation does not result in meiosis I delay, suggesting that the regulators of meiosis I progression cannot be targeted by Rad53. Interestingly, neither Rad53 nor its activator Rad9 is phosphorylated and activated when programmed meiosis-specific DSBs arise during meiosis I, even when their repair is prevented by the lack of Dmc1. However, targeting Rad53 to Mec1 by a Ddc2-Rad53 chimera allows its phosphorylation and activation in response to meiotic programmed DSBs. Thus, the reason why Rad53 activation by meiotic programmed DSBs is prevented may be that Rad53 and/or Rad9 is not reachable by Mec1 signaling from the meiotic recombination sites. On the other hand, a Rad53-dependent checkpoint response, which causes a delay of the second meiotic division, is elicited when unrepaired meiosis-specific DSBs escape the recombination checkpoint-mediated prophase I arrest (Fig. 8C and D). In fact, Rad53 is phosphorylated in dmc1Δ mek1Δ rad54Δ cells, which fail to repair meiotic DSBs due to the absence of Rad54, but are allowed to segregate homologous chromosomes containing hyperresected DSBs due to the absence of Mek1. Moreover, Rad53 is phosphorylated in sae2Δ cells, which are known to perform meiosis I in the presence of unprocessed DSBs. This phosphorylation requires DSB formation and causes a delay of the second meiotic division. Unlike Mek1, whose phosphorylation is detectable concomitantly with meiotic DSB generation, Rad53 phosphorylation in both sae2Δ and dmc1Δ mek1Δ rad54Δ cells occurs at the time of meiosis I completion. Moreover, Rad53 phosphorylation is prevented in sae2Δ cells by eliminating Ndt80, which causes a meiosis I arrest at late prophase. This suggests that unrepaired meiosis-specific DSBs can elicit a Rad53-dependent checkpoint only when homologous chromosomes segregate from each other, and this checkpoint causes a delay in sister chromatid separation at meiosis II (Fig. 8C and D). The Rad53-dependent delay of meiosis II is transient even if meiotic DSBs are not repaired. Because meiotic cells progress through meiosis and form spores even if a DSB remains unrepaired (24), one possibility is that the checkpoint mechanism is less responsive to DNA damage during meiosis than during mitosis.
Tel1 is thought to recognize unprocessed DSBs, whereas Mec1 senses and signals ssDNA regions that arise after DSB processing and are covered by the replication protein A complex (25, 55). Consistent with previous data showing that the lack of Sae2 impairs DSB processing (10), Tel1 has the major role in triggering Mek1 phosphorylation in meiotic sae2Δ cells. In contrast, Mec1 has a more critical role than Tel1 in triggering Rad53 phosphorylation in the same cells. These results suggest that a subset of meiotic DSBs are processed after homologous chromosome segregation in sae2Δ cells, thus generating ssDNA regions that are detected by Mec1.
How can Rad53 activation in response to programmed meiotic DSBs be prevented? Inhibition of Rad53 phosphorylation during meiosis I recalls the ability of meiotic cells to generate the so-called “barrier to sister chromatid repair” (33). In fact, one of the differences between mitosis and meiosis is that meiotic DSBs are repaired using an intact homologous nonsister chromatid, whereas mitotic recombination occurs preferentially between sister chromatids (15, 40). The Mek1, Hop1, and Red1 proteins, which are structural components of the meiosis-specific chromosome structures that favor the association between homologous chromosomes, are essential to establish the correct meiotic recombination partner choice (6, 33, 34, 41, 49). Thus, one possibility is that this meiosis-specific structure (or some specific components) may not only suppress intersister DSB repair but also hide programmed meiosis-specific DSBs from being signaled as DNA damage to the Rad53 kinase, thus preventing activation of the Rad53-dependent DNA damage checkpoint during meiosis I. When homologous chromosome segregation takes place and interhomolog bias is abolished, meiotic DSBs that are not yet repaired could then be monitored as DNA damage by the canonical Rad53-dependent DNA damage checkpoint machinery. Exogenous DSBs during meiosis I may in turn cause a local disruption of the meiosis-specific chromosome structure, thus allowing Rad53 to be phosphorylated and activated.
Given that DSB-induced Mek1 activation is required to ensure the formation of interhomolog crossovers (34), the meiosis-specific propagation of the checkpoint signals through Mek1, Red1, and Hop1 instead of Rad53 is likely critical for the formation of genetically balanced gametes. In fact, reduced Mek1 phosphorylation would allow meiosis to proceed without the correct repair partner choice and formation of chiasmata, which are critical for proper meiotic chromosome segregation.
The meiosis-specific large-scale structure does not prevent sensing and signaling of meiotic programmed DSBs by Mek1, Red1, Hop1, Mec1, and Tel1, possibly because they are part of the normal recombination machinery. In fact, Mek1, Red1, and Hop1 proteins are structural components of the meiotic chromosome axes (3, 50, 52). Moreover, Mec1 loss of function leads to a number of meiotic defects, including aberrant chromosome synapsis, reduced recombination frequency and spore viability, and loss of interhomolog bias and of crossover control (reviewed in reference 7). In higher eukaryotes, both ATR and ATM associate with different sites along meiotically pairing chromosomes (17), and ATM-deficient mice show aberrant synapsis with unpaired axial cores and fragmented synaptonemal complexes (4, 53). Finally, mutations in the RAD17, RAD24, or MEC3 gene, encoding regulators of Mec1 activity, reduce meiotic interhomolog recombination frequency, while increasing the frequency of ectopic recombination events and of illegitimate repair from the sister chromatids (2, 13, 42, 45).
In conclusion, whereas accidental DSBs induce a Rad53-dependent DNA damage response during both mitosis and meiosis, meiotic DSB repair is monitored by a meiosis-specific checkpoint mechanism involving integral components of the chromosomal structures specifically formed during meiosis. On the other hand, when meiosis I takes place despite unrepaired meiotic DSBs, the latter can trigger a Rad53-dependent DNA damage checkpoint slowing down the second meiotic division, which is functionally equivalent to mitosis. The possibility of activating this checkpoint might provide a salvage mechanism preventing chromosome rearrangements and/or loss in the gametes even in the absence of the recombination checkpoint, thereby further protecting the offspring from birth defects and cancer predisposition.
We thank R. Cha, J. Diffley, N. Kleckner, N. Lowndes, and D. Stern for yeast strains and plasmids; M. Foiani for critical reading of the manuscript; and all the members of our laboratory for useful discussions and criticisms.
This work was supported by grants from Associazione Italiana per la Ricerca sul Cancro and Cofinanziamento MIUR/Università di Milano-Bicocca to M.P.L. H. Cartagena-Lirola was supported by an EC Research Training Network Grant (HPRN-CT-2002-00238).
Published ahead of print on 27 May 2008.