A Tel1/MRX-Dependent Checkpoint Inhibits the Metaphase-to-Anaphase Transition after UV Irradiation in the Absence of Mec1

S. cerevisiae strains and media.

The relevant genotypes of all S. cerevisiae strains used in this study are listed in Table 1. All strains were constructed during this study, and all were derivatives of W303 (K699) (MATa or MATα ade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3 rad5-535). The rad5-535 mutation confers very slight sensitivity to methyl methanesulfonate (MMS) (23).

Standard yeast genetic techniques and media (80) were used. Cells were grown in YEP medium (1% yeast extract, 2% Bacto Peptone, 50 mg of adenine per liter) supplemented with 2% glucose (YEPD) or 2% raffinose (YEP+Raf) or 2% raffinose and 1% galactose (YEP+Raf+Gal).

Deletions of the MEC1, RAD53, CHK1, SML1, TEL1, and RAD14 genes have been obtained as previously described (50, 51, 69, 74, 75). Deletion of MRE11 was generated from strain K699, where 2,111 bp of the MRE11 coding region was replaced by the HIS3 gene by one-step PCR disruption method (107). The rfa1-t11 mutant was kindly provided by R. D. Kolodner (University of California, La Jolla).

Strains carrying fully functional MRE11-HA3, CHK1-HA3, and PDS1-HA3 alleles at their corresponding chromosomal loci were constructed as previously described (6, 73, 75). W303-derived strains carrying either deletions of the PDS1 and MAD2 genes, or the GAL-CLB2 allele integrated in one copy at the TRP1 locus or in seven copies at the CLB2 locus (7XGAL-CLB2) (96), or a galactose-induced GAL-ubiCDC6 fusion at the CDC6 locus, were kindly provided by S. Piatti (University of Milano-Bicocca, Milan, Italy), while a W303-derived strain expressing the Scc1 version cleavable by the tobacco etch virus (TEV) protease (scc1Δ::HIS3 SCC1-TEV GAL-TEV) was kindly provided by F. Uhlmann (Cancer Research UK, London, United Kingdom). All these strains were used to disrupt MEC1 or TEL1 or both, as reported above. Strain YLL1437, carrying three copies of the GAL-SIC1ΔNT-MYC-HIS fusion integrated at the URA3 locus, was obtained by transforming strain K699 with ApaI-digested plasmid pLD1, kindly provided by J. Diffley (Clare Hall Laboratories, South Mimms, United Kingdom). The accuracy of integration was verified by Southern blot analysis.

The mec1Δ tel1Δ sml1Δ triple mutants were constructed by crossing mec1Δ sml1Δ and tel1Δ sml1Δ strains. The resulting diploid strain mec1Δ/MEC1 tel1Δ/TEL1 sml1Δ/sml1Δ was allowed to sporulate, and tetrads were dissected on YEPD plates. Forty-eight hours after tetrad dissection, segregant clones were streaked on YEPD plates and were used 24 h later for the experiments described in the text and concomitantly analyzed for the presence of the mec1Δ and tel1Δ markers.

Other techniques.

Synchronization experiments were performed as described previously (74). Mechloretamine (nitrogen mustard or HN2) and mononitrogen mustard (2-dimethylaminoethylchloride hydrochloride or HN1) were purchased from Sigma-Aldrich. Flow cytometric DNA analysis was performed on a Becton-Dickinson FACScan. Nuclear division on cells stained with propidium iodide was scored with a fluorescence microscope. tet operators integrated at the URA3 locus of chromosome V (35 kb from the centromere) using the GFP-tetR fusion were visualized as described previously (60). For Western blot analysis, protein extracts were prepared by trichloroacetic acid precipitation as previously described (51). Rad53 was detected using anti-Rad53 polyclonal antibodies kindly provided by J. Diffley (Clare Hall Laboratories), and Rad9 was detected using anti-Rad9 polyclonal antibodies kindly provided by N. Lowndes (University of Ireland, Galway). Secondary antibodies were purchased from Amersham, and proteins were visualized by an enhanced chemiluminescence system according to the manufacturer.

Cellular responses to different genotoxic agents in the absence of Mec1.

In order to gain new knowledge of the interrelationships between Mec1 and Tel1 in the DNA damage checkpoint response, we analyzed the abilities of mec1Δ cells to progress through the cell cycle and to activate Rad53 after treatment with different genotoxic agents. Rad53 activation was examined by monitoring its phosphorylation, which is known to be required for its activation as a kinase, and can be detected as an electrophoretic mobility shift (31, 84, 85, 95). It is worth pointing out that all strains carrying the mec1Δ allele also carried the sml1Δ allele (Table 1), which suppresses the lethality of MEC1 deletion, but not its effect on the DNA damage response (113).

In the experiments shown in Fig. 1A, B, and C, G₁-arrested wild-type and mec1Δ cells were either released into the cell cycle under unperturbed conditions or UV irradiated before release or released in the presence of the alkylating agent MMS or the radiomimetic drug bleomycin. When 95% of cells had budded after release, α-factor was added back in order to prevent cells from entering a second cell cycle. As expected, due to activation of the DNA damage checkpoint by the genotoxic treatments, damaged wild-type cells slowed down cell cycle progression and started cell division much later than the untreated cell cultures (Fig. 1A). Accordingly, phosphorylated Rad53 appeared in wild-type cells immediately after UV irradiation or concomitantly with S-phase entry in the presence of MMS and bleomycin (Fig. 1B). Interestingly, although both untreated and UV-, MMS-, and bleomycin-treated mec1Δ cells completed DNA replication with similar kinetics, the damaged mec1Δ cells consistently went through the subsequent cell cycle phases slower than the untreated cells did. Moreover, they accumulated phosphorylated Rad53, which was detected concomitantly with completion of damaged DNA replication about 60 min after genotoxic treatment (Fig. 1A and B).

As shown in Fig. 1C, not only Rad53 but also Mre11 was phosphorylated in mec1Δ cells about 30 min after UV irradiation or bleomycin or MMS addition, whereas its phosphorylation in wild-type cells was detected only in the presence of bleomycin. Since Mre11 phosphorylation is known to be Tel1 dependent (17, 34) (see Fig. 3B), these data indicate that Tel1, which can substitute for the absence of Mec1 under specific circumstances, is activated in response to DNA damage in the absence of Mec1 and therefore might be responsible for the observed Mec1-independent phosphorylation events.

Although UV irradiation can cause replication fork stalling, like MMS and bleomycin (reviewed in reference 57), the highest level of Rad53 phosphorylation and the most pronounced delay in cell cycle progression in mec1Δ cells were induced by UV treatment, while MMS and bleomycin were more efficient than UV in causing these effects in wild-type cells (Fig. 1A and B). These results suggest that UV-induced DNA alterations might be more suitable than those caused by MMS and bleomycin of eliciting a Mec1-independent checkpoint response.

Besides inducing formation of cyclobutane pyrimidine dimers and pyrimidine-pyrimidone (6-4) photoproducts in DNA, UV irradiation can also generate strand breaks and cross-links between duplex DNA molecules (24), which constitute some of the most serious damage to DNA. Since previous studies have shown that DNA interstrand cross-links (ICLs) are capable of eliciting a checkpoint-dependent G₂/M arrest in both S. pombe and humans (2, 43), we verified the ability of ICLs to induce Rad53 phosphorylation both in the presence and absence of Mec1. To this end, exponentially growing wild-type and mec1Δ cells were arrested in G₁ with α-factor and released in the presence of the bifunctional alkylating agent nitrogen mustard HN2, which forms both monoadducts and cross-links (43, 83).

As shown in Fig. 1D (left), partial Rad53 phosphorylation took place in mec1Δ cells 60 min after release from the G₁ block in the presence of HN2, concomitantly with completion of S phase (data not shown), while its phosphorylation occurred in similarly treated wild-type cells 30 min after release as soon as cells initiated DNA replication (data not shown). Conversely, Rad53 phosphorylation was not observed when the same cell cultures were released in the presence of a monofunctional HN2 analogue, HN1, which causes only monoadduct formation (43, 83) (Fig. 1D, left). Although we could not rule out the possibility that HN2 formed monoadducts more efficiently than HN1 under our experimental conditions, these results suggest that HN2-dependent Rad53 phosphorylation might be induced by ICLs and not by monoadducts. Accordingly, HN2 treatment induced a delay in the G₂/M transition in both wild-type and mec1Δ cells, while HN1 did not affect cell cycle progression at all (data not shown). Moreover, ICLs seem to trigger Tel1 activation in the absence of Mec1, since Mre11 phosphorylation was strongly induced by HN2 treatment in mec1Δ cells, while its level, if any, was very low in wild-type cells under the same conditions (Fig. 1D, right).

UV irradiation in the absence of Mec1 leads to Scc1-dependent metaphase arrest.

We analyzed the in vivo consequences of UV-induced DNA damage in the absence of Mec1 in more detail by monitoring the kinetics of progression through the cell cycle, separation of sister chromatids, and nuclear division of mec1Δ cells released from α-factor arrest either undamaged or after UV irradiation. All strains carried the tetR-GFP/tetO system that allows detection of the separation of sister chromatids on one arm of chromosome V (60). The kinetics of cell cycle progression, sister chromatid separation, and nuclear division after α-factor release were similar in untreated wild-type and mec1Δ cells (Fig. 2A, top, and data not shown). After UV irradiation, wild-type cells progressed through the S phase slower than the untreated cultures did, due to G₁/S checkpoint activation, completing DNA replication within 150 min (Fig. 2A, bottom). As expected, when the checkpoint is turned off, the cells completed the cell cycle and started to undergo sister chromatid separation (Fig. 2B) and nuclear division (Fig. 2C) 150 and 210 min after UV irradiation, respectively, and arresting with 1C DNA content after cell division, due to the addition of α-factor. Conversely, UV-treated mec1Δ cells, which progressed through S phase much faster than similarly treated wild-type cells and completed DNA replication within 75 min (Fig. 2A, bottom), failed to undergo anaphase and arrested mostly with duplicated but unseparated chromosomes (Fig. 2B and C).

Sister chromatids are held together by cohesin, a multisubunit complex (reviewed in reference 100), and proteolytic cleavage of the Scc1 cohesin subunit at the metaphase-to-anaphase transition is essential for the separation of sister chromatids (102). If the metaphase arrest in UV-irradiated mec1Δ cells were due to some physical impediment, such as incompletely replicated DNA molecules not allowing sister chromatid separation, cleavage of Scc1 should not be sufficient to trigger anaphase in these cells. We took advantage of the fact that one Scc1 cleavage site can be exchanged with the recognition sequence for the TEV protease (101), and we generated isogenic wild-type and mec1Δ strains carrying both the SCC1-TEV allele at the SCC1 chromosomal locus and a galactose-induced GAL-TEV fusion at the TRP1 locus. Cell cultures were then arrested in G₁ with α-factor and released from the α-factor block, either unirradiated or after UV irradiation, in medium containing galactose to induce TEV protease production (Fig. 2). As expected, both mec1Δ and mec1Δ SCC1-TEV cells completed DNA replication about 75 min after UV irradiation, faster than wild-type and SCC1-TEV cells under the same conditions (Fig. 2A, bottom). UV-irradiated mec1Δ cells then remained arrested with unseparated sister chromatids and single nuclei until the end of the experiment (Fig. 2B and C), whereas in mec1Δ SSC1-TEV cells, sister chromatids started to separate and nuclei started to divide 105 and 180 min after UV irradiation, respectively. It is worth pointing out that cell division in galactose-induced SCC1-TEV and mec1Δ SSC1-TEV cells was delayed compared to cell division in wild-type cells, as indicated by the accumulation of cells with 2C DNA content (Fig. 2A) and separated sister chromatids (Fig. 2B and data not shown), and that this delay was independent of UV treatment. This cell cycle delay might be due to activation of the spindle assembly checkpoint triggered by the lack of tension at kinetochores caused by Scc1 cleavage, since Scc1 cleavage has been shown to delay cell division in a Mad2-dependent manner (89, 92). Since Rad53 phosphorylation persisted in galactose-induced, UV-treated SCC1-TEV cells longer than in wild-type cells under the same conditions (Fig. 2D), it is also possible that Rad53 contributes to the cell division delay observed in these cells.

As shown in Fig. 2D, Scc1 cleavage did not seem to affect Rad53 phosphorylation or activation, since both the level and kinetics of Rad53 phosphorylation were similar in mec1Δ SCC1-TEV and mec1Δ cells after UV irradiation and release in medium containing galactose.

A Tel1/MRX-dependent checkpoint inhibits the metaphase-to-anaphase transition in UV-irradiated mec1Δ cells.

Since the inability to undergo anaphase of UV-irradiated mec1Δ cells requires the cohesins to hold sister chromatids together, this cell cycle arrest likely involves the activation of a checkpoint. Therefore, we asked whether Tel1 and the MRX complex, which were very good candidates for mediating such a checkpoint response, were required for mec1Δ cells to arrest in metaphase after UV irradiation in G₁. Exponentially growing cells of mec1Δ tel1Δ and mec1Δ mre11Δ double mutant strains, together with the appropriate control strains, were arrested in G₁, UV irradiated, and allowed to proceed through the cell cycle; α-factor was added back to the cultures when 95% of the cells had budded (Fig. 3). Unlike UV-treated mec1Δ cells, which did not undergo sister chromatid separation and nuclear division throughout the experiment, mec1Δ tel1Δ and mec1Δ mre11Δ double mutant cells started to exhibit sister chromatid separation and nuclear division about 105 and 150 min after UV irradiation, respectively (Fig. 3A); cell division followed (data not shown).

Both Tel1 and Mre11 were required in mec1Δ cells for UV-induced phosphorylation not only of Rad53 but also of the Rad9 adaptor and of the G₂/M checkpoint kinase Chk1. In fact, phosphorylated forms of Rad53, Rad9, and Chk1, which were detected in mec1Δ cells 60 to 75 min after UV irradiation (Fig. 3B) concomitantly with completion of DNA replication (data not shown), were absent throughout the experiment in mec1Δ tel1Δ and mec1Δ mre11Δ double mutant cells under the same conditions. Thus, both these phosphorylation events and the inhibition of the metaphase-to-anaphase transition in UV-irradiated mec1Δ cells involve the activation of a Tel1/MRX-dependent checkpoint. Tel1 was also responsible for the phosphorylation of Mre11, which was detected in mec1Δ cells 30 min after UV irradiation (Fig. 3B), indicating that Tel1 is activated in the absence of Mec1 after UV-induced damage.

If Chk1 and Rad53, which act independently from one another to inhibit the G₂/M transition after DNA damage in G₂ (85), were responsible for the Tel1-dependent metaphase arrest in UV-irradiated mec1Δ cells, a rad53Δ chk1Δ double mutant should be unable to carry on the Tel1-dependent response and should therefore undergo anaphase even in the absence of Mec1 after UV irradiation in G₁ and release into the cell cycle. As shown in Fig. 4A, both mec1Δ rad53Δ chk1Δ and rad53Δ chk1Δ cell cultures completed DNA replication about 90 min after UV irradiation in G₁ and release into the cell cycle, similar to mec1Δ cells under the same conditions. They then started to exhibit nuclear separation about 120 min after UV irradiation (Fig. 4B), followed by cell division and arrest with 1C DNA content, due to α-factor readdition, while UV-treated mec1Δ cells remained arrested with undivided nuclei throughout the experiment (Fig. 4A and B). It is interesting that the kinetics of cell cycle progression and nuclear division in mec1Δ rad53Δ chk1Δ and rad53Δ chk1Δ cell cultures were similar to those of mec1Δ tel1Δ cells (Fig. 4A and B), indicating that Tel1 inhibits the metaphase-to-anaphase transition acting exclusively through Rad53 and Chk1.

Both Rad53 and Chk1 contributed to the metaphase arrest of UV-irradiated mec1Δ cells, since mec1Δ rad53Δ and mec1Δ chk1Δ cell cultures underwent nuclear division with similar kinetics and less efficiently than mec1Δ rad53Δ chk1Δ cells after UV irradiation (Fig. 4A and B). Thus, Chk1 and Rad53 are required to inhibit the metaphase-to-anaphase transition in UV-irradiated mec1Δ cells.

The spindle assembly checkpoint protein Mad2 contributes to the G₂/M arrest of UV-irradiated mec1Δ cells.

Some delay in nuclear division was still apparent in cells carrying the mec1Δ tel1Δ, mec1Δ mre11Δ, mec1Δ rad53Δ chk1Δ, and rad53Δ chk1Δ deletions, when they were released into the cell cycle after UV irradiation in G₁ (Fig. 3 and 4). These observations suggested that although the Tel1-dependent checkpoint seemed to have a major role in the UV-induced G₂/M block in the absence of Mec1, other mechanisms might contribute to this arrest. We reasoned that replication of damaged DNA in the absence of Mec1 could lead to changes in kinetochores, which are assembled at centromeres and mediate spindle microtubule attachment. These changes might in turn activate the Mad2-dependent spindle assembly checkpoint that prevents the onset of anaphase in response to changes in the structure or tension of kinetochores (reviewed in reference 64).

As shown in Fig. 4, Mad2 does not play a role in slowing down progression through the cell cycle in response to UV treatment when the DNA damage checkpoint is fully functional, since wild-type and mad2Δ cells exhibited delays in cell cycle progression and phosphorylated Rad53 with very similar kinetics after UV irradiation in G₁ and release into the cell cycle. However, the absence of Mad2 advanced nuclear division of rad53Δ chk1Δ double mutants and partially relieved the metaphase arrest of mec1Δ cells after UV irradiation (Fig. 4A and B). In fact, UV-irradiated rad53Δ chk1Δ mad2Δ cells released from G₁ arrest underwent nuclear division earlier than UV-treated rad53Δ chk1Δ cells did (Fig. 4B), while both cell cultures completed DNA replication at the same time (Fig. 4A). Moreover, unlike mec1Δ cells, which arrested with undivided nuclei after completion of UV-damaged DNA replication, mad2Δ mec1Δ cells underwent nuclear division, although much later than the untreated cultures, followed by mitotic exit and cell division (Fig. 4A and B). Rad53 phosphorylation was not affected by the absence of Mad2, since Rad53 was phosphorylated in mec1Δ mad2Δ cells after UV irradiation and release into the cell cycle (Fig. 4C).

Pds1/securin inactivation bypasses the metaphase arrest of UV-irradiated mec1Δ mutant.

At the onset of anaphase, a caspase-related protease (separase) destroys the link between sister chromatids by cleaving the cohesin subunit Scc1 (reviewed in reference 100). During most of the cell cycle, separase is kept inactive by binding to an inhibitory protein called Pds1/securin, which is destroyed by ubiquitin-mediated proteolysis shortly before the metaphase-to-anaphase transition (14, 25, 114). One of the mechanisms by which both the DNA damage and spindle assembly checkpoints prevent the metaphase-to-anaphase transition is the inhibition of Pds1 ubiquitination by the ubiquitin ligase anaphase-promoting complex (APC), thus stabilizing the securin. In particular, the spindle assembly checkpoint proteins Bub3, Mad2, and Mad3 physically interact with the APC regulatory subunit Cdc20, required for Pds1 ubiquitination, likely leading to Cdc20 inhibition (reviewed in reference 64). Conversely, Rad53 activation seems to affect the Pds1-Cdc20 interaction, whereas Chk1 is responsible for Pds1 phosphorylation, which in turn appears to inhibit the ubiquitination reaction itself (1, 13).

Therefore, we asked whether Pds1 was involved in the metaphase arrest of UV-irradiated mec1Δ cells. When cell cultures were UV irradiated in G₁ and allowed to proceed through the cell cycle, UV-treated mec1Δ pds1Δ cells, which completed DNA replication faster than similarly treated wild-type and pds1Δ cells (data not shown), started to exhibit sister chromatid separation (Fig. 5A, top) and nuclear division (Fig. 5A, middle) about 90 and 150 min after release, respectively, while mec1Δ cells under the same conditions were still arrested with undivided sister chromatids and nuclei at the end of the experiment. Once the nuclei divided, mec1Δ pds1Δ cell cultures underwent cell division as indicated by fluorescence-activated cell sorting (FACS) analysis (data not shown) and by the decrease in the percentage of binucleate cells about 300 min after UV irradiation (Fig. 5A, middle). As expected, the absence of Pds1 did not prevent the Tel1-dependent Rad53 phosphorylation, which took place with similar kinetics in mec1Δ and mec1Δ pds1Δ UV-irradiated cells (Fig. 5A, bottom).

We then monitored the phosphorylation and level of Pds1 of G₁-arrested, UV-irradiated wild-type, tel1Δ, mec1Δ, and mec1Δ tel1Δ cells carrying a fully functional PDS1-HA tagged allele after the cells were allowed to proceed through the cell cycle. As previously reported (1), Pds1 appeared as a triplet in exponentially growing undamaged cells, with the fastest and constitutive migrating band exhibiting greater intensity than the two slower bands, one of which was weakly detected (Fig. 5B, top). After UV irradiation, the intensities of the two slower bands were expected to increase due to Mec1-dependent phosphorylation (1). As expected, Pds1 was absent in all α-factor-arrested cell cultures and increased in level when cells progressed through S phase. As shown in Fig. 5B, there was no change in the intensities of the two slower migrating bands of Pds1 in UV-irradiated mec1Δ tel1Δ double mutant cells. Conversely, the intensities of the two slower bands increased in similarly treated wild-type, tel1Δ, and mec1Δ cells (Fig. 5B, top), indicating that Tel1 contributed to the DNA damage-induced Pds1 phosphorylation. Moreover, UV-irradiated mec1Δ cells, which arrested with 2C DNA content (Fig. 5B, bottom), maintained constant Pds1 levels until the end of the experiment, while the securin was degraded after completion of DNA replication in wild-type, tel1Δ, and mec1Δ tel1Δ cells (Fig. 5B, top). Altogether, these data indicate that Pds1 is necessary for the Tel1-dependent metaphase arrest of UV-irradiated mec1Δ cells, where its phosphorylation and stabilization depend on Tel1, suggesting that its persistence may be responsible for the inability of these cells to undergo anaphase.

Signals triggering Tel1/MRX-dependent phosphorylation events in response to UV irradiation.

The pathway primarily responsible for removing UV-induced DNA lesions is the nucleotide excision repair (NER) pathway, which is also necessary for checkpoint activation after UV irradiation in both the G₁ and G₂ phases of the cell cycle (30, 69). In the absence of a functional NER pathway, not only is the removal of UV-induced lesions impaired (79), but replication fork stalling and accumulation of recombination intermediates are also observed (69). Therefore, NER inactivation by deletion of the RAD14 gene, acting at early stages of the pathway, might influence the generation of checkpoint signals leading to Tel1 activation in mec1Δ cells. Indeed, when G₁-arrested cells were UV irradiated with a very low UV dose (5 J/m²) and then allowed to proceed through the cell cycle, Rad53 phosphorylation was detected in mec1Δ rad14Δ cells about 75 min after UV irradiation (Fig. 6A, bottom), concomitantly with completion of S phase in the presence of irreparable DNA damage (Fig. 6A, middle). Conversely, since small amounts of UV-induced DNA lesions are rapidly repaired when the NER pathway is functional, the extent of Rad53 phosphorylation under the same conditions was very low in wild-type cells and even lower in mec1Δ cells (Fig. 6A, bottom). As expected, this low UV dose led rad14Δ cells to arrest in S phase with high levels of phosphorylated Rad53 (Fig. 6A, middle and bottom) due to the activation of a Mec1-dependent checkpoint (69). Wild-type, mec1Δ, and rad14Δ mec1Δ cells under the same conditions completed DNA replication with similar kinetics (Fig. 6A, middle). Strikingly, rad14Δ mec1Δ cells then arrested with 2C DNA content (Fig. 6A, middle), persistent Rad53 phosphorylation (Fig. 6A, bottom), and undivided nuclei (data not shown), while wild-type and mec1Δ cells underwent nuclear and cell division, as indicated by the reappearance of cells with 1C DNA content (Fig. 6A, middle). Both Rad53 phosphorylation and the metaphase arrest of rad14Δ mec1Δ cells were dependent on Tel1, since similarly treated rad14Δ mec1Δ tel1Δ triple mutant cells did not phosphorylate Rad53 and underwent cell division (Fig. 6A). Thus, RAD14 deletion appears to amplify the UV-induced signals leading to Tel1 activation in the absence of Mec1. Since the absence of Rad14 causes the complete loss of NER proteins at the damaged site, possibly by affecting the structure of the complexes involved in lesion recognition (79), these data also indicate that NER processing of UV-induced lesions is not required to activate the Tel1-dependent checkpoint.

Although it has been shown that RPA-coated ssDNA represents one of the signals activating the Mec1 kinase in response to DSBs induced by HO endonuclease (27, 45, 54, 115), it is still unclear whether UV-induced lesions in G₁ need to be processed to ssDNA in order to be sensed by the checkpoint. To address this question, we used the rfa1-t11 mutant, because its alteration in the 70-kDa subunit of the RPA complex renders cells sensitive to UV radiation and impairs loading of checkpoint proteins onto damaged DNA, thus delaying checkpoint activation (103, 115). As shown in Fig. 6B, when G₁-arrested cells were kept in G₁ by α-factor addition after UV irradiation, rfa1-t11 cells exhibited defective phosphorylation of Rad53, which was normally phosphorylated in similarly treated wild-type cells, suggesting that ssDNA generation is necessary for the checkpoint machinery to detect UV-induced DNA lesions in G₁. Moreover, Rad53 phosphorylation under these conditions was completely abolished in mec1Δ cells, while it still took place in tel1Δ cells (Fig. 6B), indicating that Mec1 is strictly required to respond to UV-induced lesions during the G₁ phase.

In order to determine whether RPA-coated ssDNA is responsible for the Tel1-dependent Rad53 phosphorylation observed in UV-irradiated mec1Δ cells after the cells were allowed to proceed through the cell cycle, we released UV-irradiated mec1Δ rfa1-t11 double mutant cells and the appropriate control strains from G₁ arrest. As shown in Fig. 6C, phosphorylated Rad53 appeared immediately in wild-type cells and 30 min after α-factor release in rfa1-t11 cells, while it was detected only 75 min after release in both rfa1-t11 mec1Δ and mec1Δ cells, concomitantly with S-phase completion (data not shown). Importantly, the level of phosphorylated Rad53 was significantly lower in the rfa1-t11 mec1Δ double mutant cells than in mec1Δ cells (Fig. 6C), suggesting that RPA-coated ssDNA may be part of the signals activating the Tel1-dependent G₂/M checkpoint response.

Tel1-dependent checkpoint activation after UV irradiation requires DNA replication and occurs concomitantly with completion of the S phase.

As shown in Fig. 7A, Rad53 phosphorylation could also be detected concomitantly with S-phase completion when mec1Δ cells were allowed to proceed through the cell cycle after UV irradiation in G₂. Under these conditions, phosphorylated Rad53 appeared in mec1Δ cells about 105 min after release from the G₂ block, when cells completed the subsequent DNA replication (Fig. 7A). This response was dependent on Tel1, since similarly treated mec1Δ tel1Δ cells did not undergo Rad53 phosphorylation at any time after release (data not shown). Moreover, these Tel1-dependent phosphorylation events required entry into the cell cycle. In fact, no phosphorylated Rad53 was detected when G₂-arrested mec1Δ cells were held in G₂ by the addition of nocodazole after UV irradiation (Fig. 7B), while wild-type cells were able to phosphorylate Rad53 immediately after UV irradiation in G₂, independently of progression through the cell cycle (Fig. 7A and B).

The observation that Rad53 phosphorylation in UV-treated mec1Δ cells occurred concomitantly with the completion of damaged DNA replication, independently of whether the damage was generated in G₁ or G₂, suggested that S-phase entry and/or progression might be required for Tel1-dependent checkpoint activation. To further explore this possibility, we blocked initiation of DNA replication by preventing prereplicative complex (pre-RC) assembly through depletion of the Cdc6 protein (12, 77). To this end, isogenic wild-type and mec1Δ strains, both carrying an integrated copy of a galactose-induced GAL-CDC6 fusion (78) as the sole Cdc6 source, were blocked in G₂ with nocodazole in medium containing galactose and transferred to medium containing nocodazole and glucose for 1 h to completely repress GAL-CDC6. Cell cultures were then released from the nocodazole arrest by transfer into fresh medium containing glucose in the presence of α-factor, followed by release from α-factor under glucose-repressed conditions of the G₁-arrested cells, either unirradiated or after UV irradiation. As expected, GAL-CDC6 and mec1Δ GAL-CDC6 cells did not undergo DNA replication under these conditions independently of UV irradiation (Fig. 7C, top and middle), and the appearance of cells with less than 1C DNA content about 105 to 120 min after α-factor release indicated that they eventually underwent mitosis in the absence of DNA replication as previously described (77). Rad53 phosphorylation, which took place immediately after UV irradiation in wild-type and GAL-CDC6 cells and concomitantly with S-phase completion in UV-treated mec1Δ cells, was not detected in UV-irradiated mec1Δ GAL-CDC6 cells throughout the experiment (Fig. 7C, bottom).

Thus, UV-induced DNA damage requires entry into S phase to trigger Tel1-mediated Rad53 phosphorylation, suggesting that UV-damaged DNA needs to be replicated in order to be able to activate the Tel1-dependent checkpoint.

Links between cyclin-CDK complexes and Tel1-dependent Rad53 phosphorylation.

In both budding yeast (S. cerevisiae) and fission yeast (S. pombe), entry and progression into S and M phases require activation of the CDK1/Cdc28/Cdc2 cyclin-dependent kinase bound to S-phase- and M-phase-specific B-type cyclins (reviewed in reference 63). In budding yeast, the S-phase Clb5-CDK1 and Clb6-CDK1 complexes trigger the firing of origins that had previously formed pre-RCs, while activation of M-phase Clb1-, Clb2-, Clb3-, and Clb4-CDK1 complexes promote the formation of a mitotic spindle and APC activation. The observation that Tel1-dependent checkpoint activation in UV-irradiated mec1Δ cells occurs concomitantly with completion of S phase suggests that it may require Clb-CDK1 complexes specifically active at the end of S phase. If this were the case, ectopic Clb-CDK1 activation in G₁ by overexpression of the major mitotic cyclin gene CLB2 might be expected to trigger premature Tel1-dependent Rad53 phosphorylation after UV irradiation.

To explore this possibility, G₁-arrested wild-type and mec1Δ cells, either lacking or carrying one (GAL-CLB2) or seven (7XGAL-CLB2) integrated copies of a galactose-induced GAL-CLB2 fusion (96), were UV irradiated and allowed to proceed through cell cycle in the presence of galactose to induce CLB2 expression from the GAL promoter. In all untreated cell cultures, CLB2 overexpression did not induce Rad53 phosphorylation (Fig. 8A, bottom). Phosphorylated Rad53 appeared immediately after UV irradiation in all strains expressing functional Mec1 (Fig. 8A, bottom), independent of the different kinetics of S-phase entry and progression, which were delayed by GAL-CLB2 expression (Fig. 8A, top). Conversely, all UV-irradiated mec1Δ strains completed DNA replication about 75 to 90 min after release, faster than the MEC1 strains (Fig. 8A, top). Strikingly, phosphorylated Rad53 was detected about 60 and 30 min after release in mec1Δ GAL-CLB2 and mec1Δ 7XGAL-CLB2 cells, respectively, while it was detected in similarly treated mec1Δ cells only after 90 min (Fig. 8A, bottom), when cells had completed DNA replication and reached the G₂ phase (Fig. 8A, top). The advanced Rad53 phosphorylation induced by high levels of Clb2 was still dependent on Tel1, since both tel1Δ mec1Δ GAL-CLB2 and tel1Δ mec1Δ 7XGAL-CLB2 cells did not phosphorylate Rad53 in response to UV irradiation after induction of CLB2 overexpression (data not shown). Thus, ectopic CLB2 overexpression triggers UV-induced Rad53 phosphorylation in the absence of Mec1 even if cells have not yet completed replication of damaged DNA, indicating that these phosphorylation events are likely stimulated by a mitotic CDK.

If unscheduled Tel1-dependent checkpoint activation caused by Clb2 overproduction required Clb/CDK1 kinase activity, Clb/CDK1 inhibition should influence UV-induced Rad53 and/or Mre11 phosphorylation. Previous studies have shown that overexpression of a truncated form of the Clb/CDK1 inhibitor Sic1 (Sic1ΔNT), lacking 50 N-terminal amino acids that target Sic1 for degradation, efficiently inhibits Clb/CDK1 activity (18, 71). Therefore, we analyzed the effects caused by overexpressing the galactose-induced GAL-SIC1ΔNT allele on UV-induced Rad53 and Mre11 phosphorylation in wild-type and mec1Δ cells (Fig. 8B). To this end, cell cultures exponentially growing in raffinose were arrested in G₁ with α-factor, UV irradiated, and first released from G₁ in medium containing raffinose in the presence of nocodazole, to allow DNA replication and Clb/CDK1 activation, followed by G₂ arrest. After DNA replication was completed and cells were arrested in G₂, SIC1ΔNT expression from the GAL promoter was induced in half of each cell culture by the addition of galactose. As shown in Fig. 8B, Rad53 was phosphorylated immediately after UV irradiation in wild-type cells, and 120 min later, concomitantly with S-phase completion, in mec1Δ cells. Phosphorylated Rad53 was still apparent after 300 min in both cell types in the absence of galactose (Fig. 8B, top), while it started to decrease with similar kinetics in both strains about 15 min after GAL-SIC1ΔNT induction (Fig. 8B, bottom). Thus, Clb/CDK1 activity is required to maintain Rad53 phosphorylation and activation both in the presence and absence of Mec1; therefore, some components of the Mec1- and Tel1-dependent checkpoint networks are likely activated by these CDKs. Interestingly, Clb/CDK1 inhibition did not affect Mre11 phosphorylation, which, as expected, was detected in UV-treated mec1Δ cells 30 min after release, concomitantly with S-phase entry, and persisted throughout the experiment independent of GAL-SIC1ΔNT induction (Fig. 8B).

Checkpoint-dependent metaphase arrest after UV irradiation in the absence of Mec1.

We found that UV irradiation in the absence of Mec1 leads to metaphase arrest and concomitant phosphorylation of the adaptor protein Rad9 and of the downstream checkpoint kinases Rad53 and Chk1. All these events require both Tel1 and Mre11, indicating that a Tel1/MRX-dependent checkpoint, whose activation results in the inhibition of the metaphase-to-anaphase transition, can be activated in response to UV irradiation in the absence of Mec1.

Although Tel1 and MRX can be activated in response to UV irradiation in the absence of Mec1, tel1Δ cells show neither sensitivity to DNA-damaging agents nor defects in Rad53 phosphorylation after UV irradiation in any cell cycle stage, suggesting that activation of the Tel1 kinase becomes apparent only in the absence of Mec1. Although Tel1 associates with DSBs in an Xrs2-dependent manner in the presence of Mec1 (68), it may be excluded from the sites of damage by Mec1 when processing of UV-induced lesions generates ssDNA, which is known to be recognized by the Mec1-Ddc2 complex (115). Consistent with this hypothesis, Tel1 and Mec1 are localized to telomeric ends in a reciprocal manner, with Mec1 preferentially bound to chromosomal ends when ssDNA was generated (97). In this view, Tel1 might be recruited to sites of UV-induced lesions only in the absence of Mec1, thus leading to Mre11, Rad9, Rad53, and Chk1 phosphorylation and checkpoint activation. Since Tel1-dependent Rad53 phosphorylation also occurs in mec1kd mutants with defective kinase (our unpublished observation), the absence of Mec1 kinase activity appears to be sufficient to allow Tel1/MRX activation, suggesting that Mec1 may regulate Tel1 association with damaged DNA through phosphorylation events.

However, since Mec1 is required to maintain the integrity of the replication forks under stress conditions, as well as for processive DNA replication in the presence of hydroxyurea and MMS (52, 91, 98), we cannot exclude the possibility that replication of damaged templates in the absence of functional Mec1 results in the production of potentially lethal events that trigger a Tel1-dependent checkpoint response. Indeed, the checkpoint pathway plays a role during DNA replication even during unperturbed conditions. In fact, defects in the checkpoint genes acting during S phase in the presence of DNA damage increased the rates of gross chromosome rearrangements, suggesting that one normal role of S-phase checkpoint functions is to facilitate nonmutagenic repair of the DNA damage that normally occurs during DNA replication (40, 65, 66).

We show that the Rad53 and Chk1 kinases are required for the Tel1-dependent metaphase arrest of UV-irradiated mec1Δ cells, since the kinetics of nuclear division of UV-treated mec1Δ tel1Δ, rad53Δ chk1Δ, and mec1Δ rad53Δ chk1Δ cells were similar. Nonetheless, UV-irradiated mec1Δ tel1Δ, rad53Δ chk1Δ, and mec1Δ rad53Δ chk1Δ cells still show some anaphase delay compared to the untreated cultures, suggesting that other factors may be responsible for this delay. Indeed, we found that the absence of the spindle assembly checkpoint Mad2 protein, which prevents the onset of anaphase in the presence of changes in the structure or tension of the kinetochore (reviewed in references 46 and 64), partially allows the bypass of inhibition of the metaphase-to-anaphase transition in UV-irradiated mec1Δ cells and advances nuclear division in rad53Δ chk1Δ cells under the same conditions. This suggests that replication of UV-damaged chromosomes in the absence of Mec1 may affect kinetochore assembly or structure, thus triggering the activation of the Mad2-dependent mitotic checkpoint. Alternatively, since Mad2 is known to prevent sister chromatid separation by inhibiting Cdc20/APC and the Mad2-Cdc20 complex is detected even during an unperturbed cell cycle (37), the absence of Mad2 may allow a portion of Cdc20 to be activated. This may in turn be sufficient to trigger the onset of anaphase when the DNA damage checkpoint is not fully functional. Whatever the mechanism, Mad2 appears to contribute to UV-induced metaphase arrest independently of the DNA damage checkpoint. In fact, the nuclei in rad53Δ chk1Δ mad2Δ cells divided earlier than in mec1Δ tel1Δ, rad53Δ chk1Δ, and mec1Δ rad53Δ chk1Δ cells after UV irradiation. Accordingly, Rad53 is still phosphorylated in UV-treated mad2Δ mec1Δ cells.

Cross talk between spindle assembly and DNA damage checkpoints in arresting the cell cycle has been reported in other studies. In fact, MAD2 deletion was shown to partially relieve the cell cycle arrest caused either by the yku70Δ allele (56) or by mutations affecting DNA replication (26). Moreover, cdc13-induced DNA damage triggers Rad53-dependent phosphorylation of the spindle orientation checkpoint protein Bfa1, which primarily acts together with Bub2 by inhibiting the Tem1 GTPase at the top of the mitotic exit network (36). Unlike the Mad2-dependent pathway, this branch of the mitotic checkpoint does not seem to be involved in the G₂ arrest of UV-irradiated mec1Δ cells that persists even in the absence of Bub2 (our unpublished observation).

Like the MAD2-dependent checkpoint, the Tel1/MRX-dependent checkpoint appears to block the metaphase-to-anaphase transition in mec1Δ cells by preventing degradation of the anaphase inhibitor Pds1/securin, which does not allow cleavage of the cohesin Scc1 by keeping the caspase-related protease Esp1 inactive (reviewed in reference 100). In fact, we observed Tel1-dependent Pds1 phosphorylation and stabilization in UV-irradiated mec1Δ cells, which are allowed to undergo sister chromatid separation and nuclear division both by artificially cleaving the Scc1 cohesin and by eliminating Pds1.

Signals in the Mec1- and Tel1/MRX-dependent response to UV irradiation.

It has been proposed that ssDNA is a key structure for activation of the Mec1-Ddc2 complex in response to HO-induced DSB and cdc13-induced cell cycle arrest (27, 45, 54). This hypothesis is supported by the finding that RPA-coated ssDNA enables ATR/ATRIP association with DNA and stimulates target phosphorylation (115). Moreover, the Rfa1-t11 variant of the largest subunit of the RPA complex (103) has been shown to suppress the HO-induced cell cycle arrest and to cause defective recruitment of Mec1/Ddc2 to damaged DNA sites (45, 103, 115).

Our finding that UV-induced Rad53 phosphorylation or activation in G₁ is Mec1 dependent but Tel1 independent indicates that only the Mec1 kinase can sense and transduce checkpoint signals in response to UV damage during the G₁ phase. Since this Mec1-dependent Rad53 phosphorylation is dramatically reduced in rfa1-t11 cells, which were shown to exhibit defective recruitment of Mec1/Ddc2 to HO-induced DSBs, RPA-coated ssDNA appears to be responsible for the activation of the Mec1-dependent checkpoint in response to UV damage. This implies that in order to generate enough checkpoint signals, UV lesions in G₁ need to be processed, for example, by DNA repair proteins. Consistent with this hypothesis, activation of the Mec1-dependent checkpoint in response to DNA damage outside of S phase depends entirely upon the NER pathway (30, 69).

While UV-induced lesions can trigger Mec1-dependent Rad53 phosphorylation and activation independent of S-phase entry and progression, UV-damaged DNA needs to be replicated in order to activate the Tel1/MRX-dependent checkpoint. In fact, Tel1-dependent Rad53 phosphorylation does not take place when initiation of DNA replication is prevented after UV irradiation in G₁ by inhibiting pre-RC assembly, a condition that, on the other hand, allows cells with unreplicated DNA to accumulate normal mitotic cyclin levels and to undergo haploid mitosis (77, 89). Moreover, UV irradiation in G₂ triggers Rad53 phosphorylation in mec1Δ cells only when they complete the subsequent S phase. These observations imply that the Tel1 kinase is not able to respond to unreplicated UV-induced DNA damage, even in the absence of Mec1. Indeed, UV-induced DNA lesions may undergo structural transformations and metabolic processing during DNA replication, thus generating secondary lesions that may amplify the signals able to trigger a checkpoint response. If UV-induced lesions are not repaired prior to initiation of DNA replication, a replication fork encountering a bulge in the template may stall and resume replication by repriming DNA synthesis downstream of the lesion (reviewed in reference 16). This mechanism would generate highly recombinogenic daughter strand gaps, thus converting primary lesions into intrinsically unstable ssDNA regions, which are promptly channeled into homologous recombination pathways (reviewed in references 39 and 57). Moreover, ssDNA accumulated during replication of UV-damaged template can be converted to DSBs under certain conditions (42, 61). Stalled replication forks, DNA strand breaks, ssDNA regions, and/or DNA recombination intermediates, which may result from replicating UV-damaged DNA, may activate Tel1. The requirement of DNA replication for signaling to the DNA damage checkpoint is also critical in other eukaryotes. In fact, the inhibition of DNA initiation in Xenopus egg extracts also prevents ATR-dependent checkpoint activation in response to UV irradiation (53, 93).

Whatever the structures resulting from replication of UV-damaged DNA are, the finding that the rfa1-t11 allele decreases Tel1-dependent Rad53 phosphorylation and activation suggests that ssDNA generated during replication of UV-damaged DNA is part of the signals that can be sensed by Tel1. In this view, the amount of ssDNA generated by NER-dependent processing of UV-induced lesions in G₁ may not be sufficient to activate the Tel1-dependent response, while replication of the same lesions might allow reaching the ssDNA threshold required to activate Tel1. Since human ATM seems to respond primarily to DSBs (reviewed in reference 90) and Tel1 associates with DSBs through an Xrs2-dependent mechanism (68), Tel1 might also be activated by DSBs that can arise after replication of UV-damaged DNA and/or by ssDNA that can be generated by MRX-dependent DSB processing.

The requirement of a damage threshold for signaling to the Tel1/MRX-dependent checkpoint is also suggested by the finding that the absence of the NER protein Rad14 amplifies the signals for Rad53 activation in response to UV irradiation in mec1Δ cells. Since UV-treated NER-deficient cells accumulate Y-DNA intermediates, likely due to replication fork stalling at the site of damage and Rad52-dependent Holliday junction formation (69), all these structures may increase the amount of signals able to activate the Tel1-dependent checkpoint.

Cyclin-CDK1 complexes in DNA damage checkpoints.

Tel1/MRX-dependent Rad53 phosphorylation requires replication of UV-damaged DNA and this phosphorylation is detected after damaged DNA replication is completed, suggesting that it likely requires factors that are expressed or activated specifically at that cell cycle stage. Cell cycle-dependent checkpoint activation was also observed while studying checkpoints in response to a single DNA break in S. cerevisiae. In fact, there is no evident checkpoint kinase response to a single DSB in G₁-arrested cells, whereas the G₂/M-arrested cells respond to this kind of lesion, suggesting that DSBs may need to be processed by unidentified G₂-specific factors in order to be detected by the checkpoint machinery (76).

Since the cell cycle transitions are controlled by the sequential action of cyclin-CDK complexes, possible candidates for a role in triggering Tel1-dependent Rad53 phosphorylation may be the mitotic CDKs Clb1-CDK1 to Clb4-CDK1 in budding yeast, which are active at the G₂/M transition (reviewed in reference 63). Indeed, ectopic overexpression of the major mitotic cyclin gene, CLB2, leads to premature Rad53 phosphorylation in UV-irradiated mec1Δ cells that are still replicating their DNA, suggesting that Clb2-CDK1 activity may be required to transduce the checkpoint signals to the downstream kinases. This hypothesis is also supported by the finding that expressing high levels of the Clb-CDK1 inhibitor, Sic1, after UV-induced checkpoint activation leads to Rad53 dephosphorylation in both wild-type and mec1Δ cells, indicating that Clb-CDK1 activity is required at least to keep both the Mec1- and Tel1-dependent checkpoints active. Conversely, high levels of Sic1 do not affect the maintenance of Tel1-dependent Mre11 phosphorylation after UV irradiation. This observation, together with the fact that phosphorylated Mre11 is detected as soon as these cells initiate DNA replication and therefore earlier than Rad53 phosphorylation in the same cells, suggests that Clb-CDK1 complexes may influence only subsets of Tel1 functions.

A role for cyclins in the response to DNA damage is also observed in other systems. In fact, loss of the S-phase cyclins Clb5 and Clb6 renders yeast cells sensitive to MMS, UV, and IR (59). Moreover, the S. pombe cyclin B-CDK1 complex influences recombinational repair of radiation-induced DSBs during the G₂ phase (9). In fact, defects in CDK1 kinase activity affect the formation of Rhp51 (Rad51 ortholog) foci in response to IR, and cyclin B-CDK1 activity is required, together with the checkpoint protein Crb2, to maintain topoisomerase III function (9). Finally, budding yeast telomerase-negative cells require Clb2-CDK1 to generate Rad50-dependent postsenescence survivors, suggesting that this kinase may control the Rad50-dependent homologous recombination pathway (32). Thus, cyclin B-CDK1 activity could regulate the expression of key repair or recombination enzymes and/or could directly phosphorylate proteins involved in the response to DNA damage. In fact, CDK-dependent phosphorylation has been reported for several proteins involved in DNA damage response, including human RPA (19, 70), the S. cerevisiae Srs2 helicase (47), and S. pombe Crb2 (22), as well as its S. cerevisiae ortholog Rad9 (99, 106). Since we found that Clb-CDK1 activity is required to keep the UV-induced checkpoint response active, it is possible that Clb-CDK1 complexes may influence the activity of proteins required to generate or maintain the checkpoint signals and/or to transduce the signals to the downstream kinases. Although no functions for RPA phosphorylation have been uncovered so far, a possible candidate for such regulation may be the RPA complex, which has been proposed to compete with Rad51 in generating RPA-ssDNA nucleofilaments that are in turn detected as checkpoint signals (38, 94). Moreover, since Rad9 is involved in the formation of ssDNA near telomeres in cdc13 mutants (27, 54) and is required to trigger Rad53 phosphorylation and activation (31, 104), Clb-CDK1 activity itself may exert a regulatory role on Rad9.

However, the finding that Rad53 cannot be phosphorylated when G₁ or G₂ UV-irradiated mec1Δ cells are prevented from entering the subsequent S phase without affecting cyclin/CDK activity indicates that active Clb-CDK1 complexes by themselves are not sufficient to trigger Tel1-dependent Rad53 phosphorylation, which still requires replication of UV-damaged DNA.

Altogether our data provide evidence that a Tel1/MRX-dependent checkpoint can be activated after UV irradiation and that its activation specifically triggers a metaphase arrest. Moreover, the mechanisms that couple DNA damage recognition to enhanced phosphorylation of substrates by Mec1 and Tel1 appear to be different for the two kinases, thus indicating that ATR/Mec1- and ATM/Tel1-dependent pathways are not redundant and highlighting the functional distinction between these pathways.