Double-strand breaks trigger MRX- and Mec1-dependent, but Tel1-independent, checkpoint activation

Authors


  • Editor: Monique Bolotin-Fukuhara

Correspondence: Muriel Grenon, Department of Biochemistry and National Centre for Biomedical Engineering Science, National University of Ireland Galway, Galway, Ireland. Tel.: +353 9149 2060; fax: +353 9149 5504; e-mail: muriel.grenon@nuigalway.ie

Abstract

Together with the Tel1 PI3 kinase, the Mre11/Rad50/Xrs2 (MRX) complex is involved in checkpoint activation in response to double-strand breaks (DSBs), a function also conserved in human cells by Mre11/Rad50/Nbs1 acting with ATM. It has been proposed that the yeast Tel1/MRX pathway is activated in the presence of DSBs that cannot be resected. The Mec1 PI3 kinase, by contrast, would be involved in detecting breaks that can be processed. The significance of a Mec1/MRX DSB-activated DNA damage checkpoint has yet to be reported. To understand whether the MRX complex works specifically with Tel1 or Mec1, we investigated MRX function in checkpoint activation in response to endonuclease-induced DSBs in synchronized cells. We found that the expression of EcoRI activated the G1 and intra-S phase checkpoints in a MRX- and Mec1-dependent, but Tel1-independent manner. The pathways identified here are therefore different from the Tel1/MRX pathway that was previously reported. Thus, our results demonstrate that MRX can function in concert with both Mec1 and Tel1 PI3K-like kinases to trigger checkpoint activation in response to DSBs. Importantly, we also describe a novel MRX-independent checkpoint that is activated in late S-phase when cells replicate their DNA in the presence of DSBs. The existence of this novel mode of checkpoint activation explains why several previous studies had reported that mutations in the MRX complex did not abrogate DSB-induced checkpoint activation in asynchronous cells.

Introduction

DNA lesions that threaten genomic integrity activate signal transduction pathways, known as surveillance mechanisms or DNA damage checkpoints, that are highly conserved from yeast to humans. Once DNA lesions are detected, full activation of these pathways leads to the transcriptional activation of a DNA damage response, temporary cell cycle arrests and activation of DNA repair (Lowndes & Murguia, 2000; Nyberg et al., 2002).

Large protein kinases from the PI3K-like kinase (PIKK) family play key roles in DNA damage response pathways (Shiloh, 2003). They include ATR and ATM in mammals, as well as their respective yeast orthologues, Mec1 and Tel1 in Saccharomyces cerevisiae and Rad3 and Tel1 in Schizosaccharomyces pombe (Nyberg et al., 2002). In human cells, both ATR and ATM have important roles in the DNA damage response, whereas in budding yeast, Mec1 is the major checkpoint kinase while Tel1 is largely redundant with Mec1 and only plays a minor role in most DNA damage checkpoint responses (Greenwell et al., 1995; Morrow et al., 1995; Sanchez et al., 1996; Vialard et al., 1998; Mallory & Petes, 2000).

Together with Mec1 and Tel1, proteins encoded by the S. cerevisiae RAD9 and the RAD24 epistasis group of genes also function in the DNA damage response (de la Torre-Ruiz et al., 1998; Nyberg et al., 2002). These proteins form the Rad24/RFC-like complex, the Rad17/Ddc1/Mec3 PCNA-like complex and the Rad9/Ssa complex (Kondo et al., 1999; Green et al., 2000; Gilbert et al., 2003). They contribute in two additive branches towards full checkpoint activation, and in particular stimulate the phosphorylation and activation of the checkpoint kinases Rad53 (SpCds1 or Chk2) and Chk1 (Vialard et al., 1998; Sanchez et al., 1999; Gilbert et al., 2001; Schwartz et al., 2002; Blankley & Lydall, 2004).

The yeast Mre11/Rad50/Xrs2 (MRX) complex is implicated in several functions related to DNA DSB metabolism (Haber, 1998; D'Amours & Jackson, 2002). These include two pathways of DSB repair, telomere metabolism, meiotic recombination and checkpoint activation in response to DSB-inducing agents. In human cells the existence of a DSB-induced ATM/Mre11/Rad50/Nbs1 (MRN) checkpoint is well established (van den Bosch et al., 2003). NBS1 is phosphorylated in an ATM-dependent manner and this phosphorylation is necessary for a proficient intra-S phase checkpoint (Lim et al., 2000). MRN is also required for ATM localization at the site of DSBs (Uziel et al., 2003; Kitagawa et al., 2004; Falck et al., 2005) and can also directly activate the kinase in vitro (Lee & Paull, 2004, 2005). We and others have shown that the yeast MRX complex is involved in DNA damage responses (D'Amours & Jackson, 2001; Grenon et al., 2001; Usui et al., 2001; Chahwan et al., 2003) and there are several lines of evidence indicating a Tel1/MRX pathway in yeast. First, TEL1 and the genes coding for MRX are in the same epistasis group, distinct and additive to the MEC1 group, for DNA damage sensitivity and telomere maintenance phenotypes (Ritchie & Petes, 2000; D'Amours & Jackson, 2001). Second, Mre11 and Xrs2 are both phosphorylated by Tel1 in response to DNA damage (D'Amours & Jackson, 2001; Usui et al., 2001). Third, both Tel1 and MRX are required for checkpoint arrest in response to S phase damage in the absence of Mec1 and in the presence of mutations that affect DSB processing (Usui et al., 2001). Finally, Tel1 is recruited to DSBs and this recruitment is dependent on MRX (Nakada et al., 2003; Lisby et al., 2004; Shroff et al., 2004). Taken together, these findings suggest that the checkpoint function of ATM and MRN in human cells is conserved in yeast in an equivalent DSB-specific Tel1/MRX-dependent checkpoint (D'Amours & Jackson, 2002; Nyberg et al., 2002; van den Bosch et al., 2003). It has been proposed that this Tel1/MRX pathway would be activated in the presence of DSBs that cannot be resected. The Mec1 pathway, which has not yet been shown to involve the MRX complex, would be activated in response to breaks that have been processed into single-stranded DNA (Usui et al., 2001).

As Mec1 is mostly involved in DNA damage-induced checkpoints, we wanted to determine if MRX could also be involved in checkpoint activation induced by DSBs together with Mec1. Our previous study strongly suggested that MRX is involved in checkpoint activation in G1 and G2 phase specifically in response to ionizing radiation and that the checkpoint defect of MRX mutants is easier to assess in the G1 phase of the cell cycle (Grenon et al., 2001). However, one single HO DSB does not induce a checkpoint in G1 due to the poor DNA end resection capacity of G1 cells that have low cyclin-dependent kinase (CDK) activity (Pellicioli et al., 2001; Ira et al., 2004). We thought that it was formally possible that the presence of multiple DSBs may be sufficient to trigger checkpoint activation in G1 cells despite their low CDK activity. In this work, we investigated MRX involvement in checkpoint activation induced by endonuclease generated DSBs in G1 synchronized cells. We chose endonuclease-expression to induce DSBs to avoid the complication that ionizing radiation can cause many base lesions in addition to DSBs. To generate multiple breaks in a controlled manner, we used yeast strains expressing the gene for Escherichia coli EcoRI endonuclease under the control of a galactose-inducible promoter (Barnes & Rine, 1985; Lewis et al., 1998).

We found that induction of DSBs by EcoRI activates the G1 and the intra-S phase DNA damage checkpoints. Thus, in agreement with its involvement in checkpoint activation in response to DSB-inducing agents, the MRX complex is also required for the G1 and intra-S checkpoints activated by EcoRI DSBs. Interestingly, these MRX-dependent responses specifically require Mec1, but not Tel1. Our work therefore clearly establishes MRX as a co-factor for both the Mec1 and Tel1 DNA damage-dependent checkpoint kinases. Interestingly, we have also identified a novel MRX-independent checkpoint pathway that responds to DSBs and results in Rad53 phosphorylation in late S phase. In asynchronous cells, this pathway masks the checkpoint defect due to the absence of MRX and explains why the involvement of MRX in DSB-specific checkpoint activation had not previously been observed in these cells (Lee et al., 1998; Lewis et al., 1999; Grenon et al., 2001, and this work).

Materials and methods

With the exception of the experiment shown in Fig. 5b, all other experiments have been repeated at least three times.

Figure 5.

 An Mre11-independent checkpoint is activated in the presence of EcoRI DSBs in late S phase. (a) Schematic of experiments presented in (b) and (c). (b) Western blot analysis of Rad53 phosphorylation and Sic1 accumulation in indicated strains released into S phase having expressed (+GAL-EcoRI) or not (−GAL-EcoRI) the EcoRI endonuclease for 6 h in G1 (t=0′). (c) Western blot analysis of Rad53 phosphory-lation in indicated strains released in S phase or maintained in G1 in the presence of EcoRI DSBs induced for 6 h in G1 (t=0′).

Strains

The strains used in this study are available from M.G. and S.P.J. All the strains are in the MATa T334 background, derived from MATa T334 (MATaura3-52 leu2-3,112 Δtrp1 :: hisG reg1-501 gal1 pep4-3 prb1-1122 [E. Perkins, obtained from K. Lewis (Lewis et al., 1998)]). This strain contains the mutant alleles reg1-501, alleviating glucose repression of the GAL1 and GAL10 promoters, and gal1, blocking the metabolism of galactose. The addition of galactose (2% final) to these cells cultured in glucose medium induces the GAL1 promoter while permitting the cells to continue growth on glucose.

To study the cellular effect of EcoRI production in G1 phase, we created a MATa T334 strain by switching the mating type of MATα T334, which cannot be arrested in G1 by alpha factor. The GAL1 :: EcoRI TRP1, or GAL-EcoRI, cassette was obtained by PCR on plasmid pLKL31 (K. Lewis) and was integrated at the LYS2 locus in MATa T334, essentially as described previously (Lewis et al., 1998). The correct integration of the EcoRI expression cassette was confirmed by PCR analysis. The level of EcoRI expression was verified by monitoring the phosphorylation of Rad53 in response to EcoRI induction in asynchronously growing cells. MATa T334 with or without the GAL-EcoRI cassette were termed wild-type –GAL-EcoRI (MATa T334) and wild-type +GAL-EcoRI (MATa T334 Δlys2 :: GAL1 :: EcoRI TRP1), respectively. The following strains were obtained by transformation of PCR-amplified deletion cassettes, adapted from Longtine et al. (1998), of both wild-type strains, – or +GAL-EcoRI. Strains were checked by PCR as well as for DNA damage sensitivity and short-telomeres phenotypes as appropriate: sml1Δ (sml1 :: URA3), sml1Δmec1Δ (sml1 :: URA3 mec1 :: trp1 :: LEU2), mre11Δ (mre11 :: URA3), rad9Δ (rad9 :: URA3), rad24Δ (rad24 :: KANMX) rad9Δrad24Δ (rad9 :: URA3 rad24 :: KANMX), tel1Δ (tel1 :: URA3). tel1Δ100G was obtained by streaking tel1Δ four times to allow the short-telomeres phenotype due to the lack of Tel1 to be fully expressed (Lustig & Petes, 1986).

Strains deprived from mitochondrial DNA (rho0 strains) were obtained by plating MATaT334, sml1Δmec1Δ, mre11Δ and tel1Δ, with or without the GAL-EcoRI cassette, on yeast extract peptone dextrose (YPD) containing ethidium bromide.

Cell cycle analyses

Alpha factor G1 arrest and release, as well as cell cycle progression, were normal in MATa T334 cells, which behaved similarly to W303a. After removal of the alpha factor, cells re-entered cell cycle normally and progressed through S phase with a replication rate similar to that of W303a as judged by budding index and fluorescence activated cell sorter (FACS) profile. Furthermore, the periodic accumulation and degradation of Sic1, an inhibitor of Cdc28/Clb complexes, was similar to that observed with W303a (data not shown) and to previously published studies (Schwob et al., 1994; Drury et al., 2000).

Because checkpoint induction in response to EcoRI expression takes several hours ((Lewis et al., 1998, 1999), Fig. 1a), we verified that the MATa T334 strain could be arrested in G1 phase for long enough for the checkpoint to be induced. Firstly, we used the MATa T334 treated with the standard dose of 5 μg of purified alpha factor/mL of culture. From time 0′, when the alpha factor was added, cells were centrifuged and the medium was changed every 3 h to remove the Bar1 protease that is synthesized and secreted by MATa cells and specifically degrades the alpha factor. Similar results were obtained with a bar1Δ strain, which was arrested using alpha factor at 0.5 μg mL−1 and maintained in this medium (data not shown). Both strains were treated for 6–10 h with alpha factor in the conditions specified above, and then released in medium without alpha factor. Both strains arrested within 2 h and were able to maintain the arrest for a further 6 or 8 h as shown by their budding index remaining <5% (data not shown). Despite the length of the alpha factor treatment, both strains were able to re-enter the cell cycle normally as demonstrated by the budding index and the Sic1 degradation (data not shown). Cell cycle progression by FACS analysis during the experiment described above was studied in rho0 strains to avoid accumulation of mitochondrial DNA during the experiment.

Figure 1.

EcoRI DSBs induce an Mre11-indepen-dent checkpoint response in asynchronous cells. (a) Analysis of cell cycle arrest in asynchronous wild-type (MATa T334) and mre11Δ cells arrest in response to induced EcoRI DSBs (□- GAL-EcoRI, ▪+GAL-EcoRI). (b) Western blot analysis of Rad9 and Rad53 hyperphosphorylation in exponentially growing wild-type and C/mre11Δ cells in the presence (upper panels) or in the absence (lower panels) of EcoRI DSBs induction. *indicates a non-specific signal.

Checkpoint experiments

For the asynchronous experiment, cells were grown to exponential phase (5 × 106 cells mL−1) in YPD and galactose was added at 2% final concentration to induce expression of EcoRI DSBs (time 0′). Samples for determining the budding index by microscopy and trichloroacetic acid (TCA) extracts were taken at the time indicated.

For the G1 experiment, cells were grown to exponential phase (5 × 106 cells mL−1) in YPD and arrested by addition of alpha factor (5 μg mL−1 final concentration). Once arrested, galactose was added at 2% final concentration to induce expression of EcoRI and induction of DSBs (time 0′). Medium containing alpha factor was replaced every 3 h during the experiment to maintain the G1 arrest. Samples for the budding index and TCA extracts were taken at the time indicated.

For the cell cycle experiment, rho0 cells were grown to exponential phase (5 × 106 cells mL−1) in YPD and arrested with alpha factor (5 μg mL−1 final concentration). EcoRI DSBs were induced for 6 h by galactose treatment. Cells were then washed and released in rich medium without alpha factor and containing galactose and 5 μg mL−1 of nocodazole (time 0′). Samples for budding index, FACS analysis and TCA extracts were then taken at indicated time. In the experiment described in Fig. 5c, cultures were split in two at time 0′. One half was maintained in medium containing alpha factor and galactose while the other half was released from the arrest as described above.

Western blotting

We performed TCA extracts and western blot analysis as described (Vialard et al., 1998). Serum NLO16 was used to detect Rad53 (10% poly-acrylamide gel). NLO16 preferentially recognizes the non-phosphorylated forms of Rad53, therefore loss of signal and slower mobility shift in sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) are both diagnostic of Rad53 hyperphosphorylation (Grenon et al., 2001). Serum NLO5 was used to detect Rad9 (de la Torre-Ruiz et al., 1998; Vialard et al., 1998). Serum JDI47, gift from J. Diffley, was used to detect Sic1 (Drury et al., 2000).

Results

EcoRI expression induces checkpoint activation in asynchronous cells and this does not require the MRX complex

Checkpoint activation in response to galactose-induced EcoRI expression in asynchronous Saccharomyces cerevisiae cells elicits the accumulation of large budded cells in response to the formation of EcoRI DSBs (Fig. 1a; Lewis et al., 1998). We further investigated this checkpoint activation at a molecular level by examining Rad53 and Rad9 hyperphosphorylation, which depend on Mec1 in response to most forms of DNA damage (Sanchez et al., 1996; Sun et al., 1996; Emili, 1998; Vialard et al., 1998) and on the MRX complex specifically in response to DSBs (Grenon et al., 2001; Usui et al., 2001). Both Rad53 and Rad9 phosphorylation became detectable after approximately 4 h of galactose induction, and were only observed in the strain containing the GAL-EcoRI cassette (+GAL-EcoRI; Fig. 1b). At this time, low levels of genomic DNA cleavage could already be observed on an agarose gel or by Southern blotting analysis of these cells (Barnes & Rine, 1985; Lewis et al., 1998, and data not shown). Thus, DSBs induced by EcoRI result in a cell cycle arrest at the metaphase/anaphase transition with features highly similar to the arrest observed in response to other DNA damaging treatments.

Next we examined checkpoint activation after induction of EcoRI in a strain deleted for the gene encoding Mre11, which results in loss of the active MRX complex (Usui et al., 1998) and in defective Rad53 activation in response to γ-irradiation in G1 and G2 (Grenon et al., 2001). Unlike cells harboring mutations in other checkpoint genes, asynchronously growing mre11Δ cells arrested as large budded cells in response to EcoRI induction (Fig. 1a, right panel; Lewis et al., 1999). In agreement with the G2/M arrest, we observed that both Rad53 and Rad9 were hyperphosphorylated after EcoRI induction in asynchronously growing mre11Δ cells with similar kinetics as for these events in MRE11+ cells (Fig. 1b and c). Thus, the absence of Mre11 does not preclude checkpoint activation in exponentially growing cells after EcoRI induction, as after γ-irradiation (Grenon et al., 2001).

EcoRI activates the G1 checkpoint and this requires MRX and Mec1 but not Tel1

We have shown that the involvement of the MRX complex in checkpoint activation is easily detected in synchronized cells (Grenon et al., 2001). Therefore, we analyzed EcoRI-dependent activation of the DNA damage checkpoint in G1 arrested cells. To do this, we needed to develop a methodology that allowed normal G1 arrest to be maintained efficiently for at least 8 h (see Experimental Procedures). Thus, wild-type strains, either with or without the GAL-EcoRI cassette, were treated with alpha factor. Once the cells were G1 arrested, 2% galactose was added and the arrest was maintained until the end of the experiment, as demonstrated by a budding index of <5% and the presence of Sic1, a G1 phase-specific protein (data not shown). We found that both Rad53 and Rad9 were phosphorylated from 4 to 8 h of DSB induction by EcoRI, with kinetics similar to those observed in exponentially growing cells expressing EcoRI (compare Figs 2a and 1a). Thus, unlike a single, irreparable DSB induced by the HO endonuclease (Pellicioli et al., 2001; Ira et al., 2004), multiple DSBs induced by the EcoRI endonuclease are able to induce checkpoint activation in the G1 phase of the cell cycle (Fig. 2a).

Figure 2.

 Genetic dependency of Rad53 phosphorylation induced by EcoRI DSBs in G1 arrested cells. (a) Western blot analysis of Rad9 and Rad53 hyper-phosphorylation in the presence (+GAL-EcoRI) or in the absence (−GAL-EcoRI) of EcoRI DSBs in alpha factor G1 arrested wild-type cells. (b) Western blot analysis of Rad53 phosphorylation in indicated checkpoint mutant strains having expressed (+) or not (−) the EcoRI endonuclease for 8 h in G1. tel1Δ 100G are tel1Δ cells that have been through 100 generations. *indicate non-specific signals. (c) Western blot analysis of Rad53 phosphorylation kinetics induced by EcoRI DSBs in G1 arrested wild-type, sml1Δmec1Δ and mre11Δ cells. Western blot analysis of Sic1 protein accumulation in G1 arrested cells is shown in the lower panel.

We then checked the dependency of this response on certain checkpoint genes by analyzing the phosphorylation of Rad53 8 h after galactose induction in different checkpoint mutants. We found that rad9Δ, rad24Δ, and rad9Δrad24Δ were all defective for Rad53 phosphorylation in response to EcoRI induction in G1 arrested cells (Fig. 2b). Similarly, this phosphorylation was defective in the sml1Δmec1Δ double mutant, although Rad53 phosphorylation was normal in the sml1Δ control strain (Fig. 2b). Rad53 was also normally phosphorylated in the absence of Tel1. To exclude the possibility that tel1 mutants only exhibit checkpoint defects when they have short telomeres, we looked at Rad53 phosphorylation in tel1Δ cells that have been through 100 generations (tel1Δ 100G), at which time they fully exhibit the short-telomere phenotype (Lustig & Petes, 1986). Rad53 is still normally phosphorylated in these cells (Fig. 2b).

Next, we asked if the MRX complex is involved in the G1 checkpoint induced by EcoRI-DSBs. We analyzed the kinetics of Rad53 phosphorylation in response to EcoRI-induced DSBs in G1 arrested mre11Δ cells and in a checkpoint-defective sml1Δmec1Δ strain. In the wild-type strain, Rad53 was phosphorylated from 4 to 8 h of EcoRI induction (Fig. 2c). G1 arrest throughout the experiment was confirmed by the detection of elevated levels of Sic1 (Fig. 2c, lower panels) as well as the lack of budded cells (data not shown) in all three strains. Most importantly, as for the mec1 mutant strain, the Mre11-defective strain was defective in phosphorylation of Rad53 in response to EcoRI-induced DSBs (Fig. 2c). Thus, the Mre11 complex is needed for checkpoint activation in the G1 phase of the cell cycle in response to DSBs generated by the EcoRI-endonuclease. This is in agreement with its specific role in G1 checkpoint activation in response to ionizing radiation (IR), but not UV treatment (Grenon et al., 2001). Finally, activation of the DNA damage checkpoint pathway of G1 cells by endonuclease-induced DSBs requires the Rad9, Rad24, Mec1 checkpoint proteins as well as MRX, but does not require Tel1.

DSB formation by EcoRI induces Mre11- and Mec1-dependent but Tel1-independent cell cycle delays

Phosphorylation and activation of Rad53 correlate with transient cell cycle delays after DNA damage. We determined whether the induction of EcoRI in G1 arrested cells for 6 h was able to induce such delays. In this experiment, during which EcoRI was induced continuously, cells were released into medium without alpha factor, but containing nocodazole to arrest in G2/M all cells that had successfully completed S phase. Cell cycle progression was monitored by both budding index, to measure exit from G1 therefore allowing quantification of the G1 checkpoint activation, and FACS analysis of DNA content, to measure S phase progression to quantify the activation of the intra-S phase checkpoint in the presence of EcoRI-induced DSBs.

As shown in Fig. 3a, cell cycle re-entry was delayed by 10–15 min in wild-type cells expressing EcoRI, relative to cells that were not expressing the endonuclease. As this transient delay was abolished in a mec1 mutant, it is clearly checkpoint-dependent and this checkpoint dependency was further supported by the absence of activated Rad53 in these cells (Fig. 2b and c). As for Mec1, Mre11 was required for this transient, EcoRI-dependent, G1 delay, as it was not observed in mre11Δ cells (Fig. 3a). This is also in agreement with the results obtained for Rad53 phosphorylation in G1 arrested cells in response to EcoRI induction (Fig. 2c). Hence, the cellular response to the multiple endonuclease-induced DSBs caused by EcoRI expression is similar to the cellular response to γ-irradiation and unlike the response to a single irreparable DSB break induced by the HO endonuclease, which does not activate the G1 checkpoint (J. Diffley, personal communication). Similar responses are observed in G1 cells after UV-irradiation, but in this case the response is Mec1-dependent but Mre11-independent (Vialard et al., 1998; Grenon et al., 2001).

Figure 3.

EcoRI DSBs induce Mec1- and Mre11-dependent G1 and S phase cell cycle delays. (a) Analysis of the G1 cell cycle checkpoint in response to 6 h of EcoRI DSBs production in the indicated strains (□- GAL-EcoRI, ▪+GAL-EcoRI). Note that the 40′ time point is atypical and there is no residual delay in the mre11 mutant cells (see Fig. 4a). (b) Analysis of the intra-S phase checkpoint response in the indicated strains expressing (bottom panels) or not (upper panels) EcoRI DSBs. In this experiment, cells were washed away from alpha factor and released in medium containing galactose and nocodazole.

Next, we looked at S phase progression in response to EcoRI-induced DSBs. In the absence of EcoRI-induced DSBs, wild-type cells re-entered the cell cycle quickly and, after release from a 6h alpha factor arrest, completed S phase in 40 min. As shown by the FACS analysis profile in Fig. 3b, wild-type cells that produced EcoRI endonuclease during the G1 arrest progressed very slowly through S phase. More than 2 h after being released, most of the cell population had not yet reached a 2C DNA content, behaving in a very similar way to cells treated with MMS or Bleocin/Bleomycin (Paulovich & Hartwell, 1995; D'Amours & Jackson, 2001; Grenon et al., 2001). Therefore, EcoRI-induced DSBs dramatically slow down S phase progression. This response was due to the intra-S phase checkpoint activation and not to physical impairment of DNA replication since the delay in DNA replication observed in the wild-type strain was abolished in the absence of Mec1. Indeed, the sml1Δmec1Δ double mutant behaved very similarly, irrespective of whether EcoRI was produced or not (Fig. 3b). The MRX complex is also required for the EcoRI-induced intra-S phase checkpoint as mre11Δ cells, similar to sml1Δmec1Δ cells, were able to progress through S phase despite the fact that EcoRI had been induced (Fig. 3b). Thus, both Mec1 and the Mre11 protein are required for the intra-S phase checkpoint induced by EcoRI-generated DSBs. Interestingly, multiple endonuclease-generated DSBs behave more like Bleocin/Bleomycin-induced DSBs (this work; D'Amours & Jackson, 2001; Grenon et al., 2001) than a single irreparable break induced by the HO endonuclease, which results in neither G1 nor S phase delays (Lee et al., 1998; Pellicioli et al., 2001).

The Mre11 complex has been reported to act together with Tel1 in checkpoint activation during S phase in response to DNA damage (D'Amours & Jackson, 2001; Usui et al., 2001). Therefore, we next analyzed the putative involvement of Tel1 in the G1 and intra-S phase delays induced by EcoRI DSBs. In contrast to our observations with mre11Δ cells, we found that EcoRI production in tel1Δ cells induced a 15–20-min delay in cell cycle re-entry similar to the one observed in wild-type cells (Fig. 4a). This result supports our observation that Rad53 phosphorylation in response to EcoRI-induced DSBs is not dependent on the presence of Tel1 (Fig. 2b and data not shown). We also investigated the ability of tel1Δ cells to slow down DNA replication in response to EcoRI DSBs (Fig. 4b). The tel1Δ cells behaved similarly to wild-type cells after induction of EcoRI by progressing extremely slowly through S phase. Similarly to previous observations with bleomycin (D'Amours & Jackson, 2001; Usui et al., 2001), they reproducibly showed slightly more progression through S phase. However, the subtle defect in S phase progression in response to DSBs was minor and in marked contrast with the behavior of cells lacking the MRX complex. These results show that, unlike Mec1 and MRX, Tel1 does not play a critical role in either the G1 checkpoint or in delaying S phase progression in response to DSBs induced by EcoRI.

Figure 4.

 The G1 and intra-S phase checkpoint-induced EcoRI DSBs does not require the Tel1 kinase. (a) Analysis of the G1 cell cycle checkpoint in response to 6 h of EcoRI DSBs production in the indicated strains (□- GAL-EcoRI, ▪+GAL-EcoRI). (b) Analysis of the intra-S phase checkpoint response in the indicated strains expressing (bottom panels) or not (upper panels) EcoRI DSBs. In this experiment, cells were washed away from alpha factor and released in medium containing galactose and nocodazole.

Activation of a late S phase checkpoint that is independent of the MRX complex

For a systematic study of intra-S checkpoint activation in response to the induction of DSBs by EcoRI, we analyzed Rad53 phosphorylation during S phase progression in the arrest and release experiments described above or in cells where the arrest was maintained (Fig. 5a). As expected, in wild-type cells, the level of Rad53 phosphorylation observed during the 6-h period of EcoRI expression in G1 was maintained during the release throughout the time course of the experiment (Fig. 5b). In sml1Δmec1Δ cells, consistent with the absence of any intra-S phase checkpoint (Fig. 3b), there was a complete absence of detectable Rad53 phosphorylation both during the alpha factor arrest and throughout the release (Figs 2b, 2c and 5b). As in sml1Δmec1Δ cells, EcoRI expression did not result in Rad53 phosphorylation during the alpha factor arrest of mre11Δ cells (Fig. 2c and time 0′ in Fig. 5b). Indeed in this strain, Rad53 remained unphosphorylated for up to 40 min after release from alpha factor. However, in mre11 mutant cells (Fig. 5b), between 40 and 80 min post-alpha factor release, Rad53 became phosphorylated and was still phosphorylated at time 120′ post release. At these times, most of the mre11Δ cells showed a close to 2C DNA content, indicating that they had completed a significant amount of DNA replication (data not shown). Since, at this time point, the cells' DNA content had already changed from 1C to 2C, the effect of Rad53 activation on S phase progression was impossible to assess by FACS analysis. Importantly, it should be noted that although spontaneous DSBs do occur during a normal S phase and their frequency increases in the absence of the Xenopus MRN complex (Costanzo et al., 2001), the phosphorylation of Rad53 that we have observed (Fig. 5b) cannot simply be due to an increased accumulation of spontaneous replication-dependent DSBs in mre11Δ cells, as this Rad53 phosphorylation did not occur when EcoRI was not produced in mre11 mutant cells (Fig. 5b, -GAL-EcoRI).

Finally, we verified that Rad53 phosphorylation after EcoRI induction in mre11Δ cells was specific for cell cycle release: mre11Δ cells maintained in G1 did not activate Rad53 unlike cells released from G1 into S phase (Fig. 5c). Taken together, these observations suggest that a checkpoint is activated in mid to late S phase in response to EcoRI-induced DSBs in the absence of the MRX complex. This MRX-independent checkpoint is most certainly impeding the detection of the requirement of the MRX complex on checkpoint activation in asynchronously growing cells in response to EcoRI DSBs (Fig. 1a and c; Lewis et al., 1998) or in response to other DSB-inducing agents (Lee, et al., 1998; Grenon et al., 2001).

Discussion

Checkpoint activation in response to EcoRI DSBs

By using galactose-dependent expression of the gene encoding the EcoRI endonuclease, we have obtained new insights into checkpoint activation in response to DNA DSBs in Saccharomyces cerevisiae. The EcoRI endonuclease cleaves DNA once it binds its recognition sequence (Terry et al., 1983) and it is unlikely to have effects unrelated to its endonuclease activity in yeast. Expression of EcoRI in yeast has previously been shown to generate DSBs in genomic DNA as well as at specific EcoRI sites (Barnes & Rine, 1985; Lewis et al., 1998, 1999; Mills et al., 1999). These breaks induce a cell cycle arrest at the metaphase/anaphase transition, which is dependent on the checkpoint protein Rad9 (Barnes & Rine, 1985; Lewis et al., 1998). However, despite well-documented reports on survival in different DSB repair mutants, which suggested that EcoRI lesions formed are repaired by the nonhomologous-end-joining DSB repair pathway (Lewis et al., 1998, 1999), little information was available to address the ability of EcoRI to induce cell cycle checkpoints other than the G2/M checkpoint in asynchronous cells, or on the genetic dependency of the cellular responses it induces. We have shown that, similar to data generated with other DNA-damaging agents, the G2/M arrest induced by the production of this endonuclease in exponentially growing cells is associated with hyper-phosphorylation of Rad9 and Rad53. Thus, EcoRI-induced DSBs are recognized as DNA damage by the checkpoint machinery and induce checkpoint activation and cell cycle arrest similarly to those induced by other forms of DNA-damaging agents, such as γ- and UV-irradiations, and various genotoxic drugs.

As shown by cell cycle delays and Mec1-dependent activation of Rad53, EcoRI induced DSBs activate the G1 and the intra-S checkpoints in synchronized cells. The activation of these two checkpoints is dependent on the MRX complex. Therefore, our work demonstrates for the first time that ‘clean’ cohesive DNA ends produced by the EcoRI endonuclease are able to induce cell cycle checkpoint in G1 and S phase, as is the case for ‘complex’ DNA ends generated by γ-irradiation or radiomimetic drugs (this work; D'Amours & Jackson, 2001; Grenon et al., 2001).

However, we have noted one difference with the inducible EcoRI system used in this study relative to other damaging treatments, which might explain the low decrease in survival in response to EcoRI expression: the kinetics of checkpoint protein phosphorylation are much slower in the case of EcoRI. This might at least partially be due to the low-level expression of EcoRI in yeast cells as well as a low cutting efficiency of the 4368 EcoRI sites in the S. cerevisiae genome (Patchmatch results from Saccharomyces Genome Database; http://www.yeastgenome.org/). Analysis of genomic DNA in agarose gels at early times of induction (2, 4 and 8 h) showed that relatively low levels of DSBs were generated by this enzyme, although the exact number of breaks formed was impossible to assess ((Barnes & Rine, 1985; Lewis et al., 1998) and data not shown). In addition to the likely slow accumulation of the enzyme in the nucleus of yeast cells, DNA cleavage will be affected by the chromatin organization surrounding each EcoRI recognition sequence. Thus, cutting at most sites will be inefficient, with perhaps only a few sites being recognized relatively efficiently. Even though such a comparison is difficult because of the distinct nature of DSBs induced by IR and EcoRI, it is interesting to compare survival after these treatments: 4–8 h of EcoRI expression results in similar effects on survival as treatment with 50–100 Grays of γ-irradiation (i.e. decrease of 10–30% in survival; Lewis et al., 1998, 1999, and data not shown). Estimates of the amount of DSBs generated by IR suggest that 100 Grays produce about 10 DSBs per cell (Redon et al., 2003).

Activation of an Mre11-independent checkpoint pathway in late S phase in response to EcoRI DSBs

Our results have also revealed the existence of a checkpoint that can be activated in the absence of Mre11 in response to EcoRI breaks. Strikingly, analysis of Rad53 phosphorylation during a synchronous release of mre11Δ cells from G1 into S phase in the presence of EcoRI-induced DSBs indicated that this checkpoint response was activated under these conditions. It is important to note that Rad53 is activated only in the presence of EcoRI DSBs, excluding the possibility that the checkpoint is activated because of spontaneous DSBs persisting in the absence of MRX (Costanzo et al., 2001). This result suggests the existence of an MRX-independent intra-S phase checkpoint pathway capable of sensing DSBs in late S phase revealed by the absence of the MRX complex. Alternatively, in the absence of the MRX complex and in the presence of DSBs in S phase, a different type of structure might be generated that can form a secondary lesion distinct from DSBs that would therefore accumulate and be detected independently of the MRX complex. Additional experiments will be required to fully characterize this checkpoint. In any case, our finding of this MRX-independent checkpoint pathway explains why it has not been possible to show a clear checkpoint defect in asynchronously growing MRX mutant cells, as these cells are still able to activate a checkpoint pathway when they progress in S phase in the presence of DSBs (Fig. 1; Lee et al., 1998; Lewis et al., 1998; Grenon et al., 2001). In agreement with our findings, it has recently been reported that a single irreparable HO DSB formed in G2 synchronized cells induces Rad53 phosphorylation in an Mre11-dependent manner (Ira et al., 2004) when, in contrast, mre11Δ cycling cells are able to arrest at the G2/M transition in response to this type of damage (Lee et al., 1998; Lewis et al., 1998; Grenon et al., 2001). The involvement of the MRX complex in checkpoint activation can therefore only be properly studied in synchronized cells. Under these conditions, the crucial role of the complex can clearly be demonstrated in the G1, S and G2 phases of the cell cycle (this work; Grenon et al., 2001; Ira et al., 2004), whereas it is not detectable if asynchronous cultures are used (Fig. 1; Lee et al., 1998; Lewis et al., 1998; Grenon et al., 2001).

MRX/Mec1-dependent but Tel1-independent checkpoint pathways

Importantly, our study identified the MRX complex as a crucial player in the Mec1 (ATR orthologue)-dependent checkpoint response. Surprisingly, the G1 and intra-S checkpoints activated by EcoRI endonuclease breaks did not require the Tel1 kinase, the counterpart of human ATM. The pathways identified here are therefore different from the Tel1/MRX pathway identified previously (Usui et al., 2001; Giannattasio et al., 2002; Nakada et al., 2003), which is believed to be activated in the presence of DSB that cannot be processed (Usui et al., 2001). This model for DSB-dependent checkpoint activation also proposed that a Mec1 pathway would take care of lesions that can be processed into single-strand DNA (ssDNA). An involvement of MRX in this Mec1 pathway has never been established previously. Recent work from M. Clerici and collaborators (Clerici et al., 2004) gives clear evidence for the existence of a Tel1-specific pathway similar to the one proposed (Usui et al., 2001). Their work demonstrates the necessity of the presence of replication forks for Tel1 to respond to UV lesions, which otherwise cannot be detected in the absence of Mec1. The interaction of replication forks with UV lesions may generate a lesion, or lesions, specific to the Tel1 pathway that can be sensed independently of DSB processing. Here, we show that EcoRI DSBs activate the G1 and intra-S checkpoints independently of Tel1. These pathways require the Mec1 kinase, suggesting that EcoRI DSBs are likely to be processed. Perhaps surprisingly, our work establishes that MRX collaborates with Mec1 and not Tel1 in activating these cellular responses to EcoRI DSBs.

Together with work from other groups, our analyses have shown that MRX, whether DSBs formed can be processed or not, has the ability to function with both Mec1 and Tel1. The ability of yeast MRX to act with both PIKKs in DSB-induced checkpoints suggests that the DSB specificity of the pathways induced is due primarily to the PIKKs themselves and not to the MRX complex. In yeast cells, MRX could be a ‘privileged’ partner of the Mec1 kinase. Indeed, it has been shown recently that MRX also cooperates with the nuclease Exo1 to promote Mec1-dependent signaling induced by hydroxyurea, phleomycin, a HO-induced DSB and UV damage in G2 (Nakada et al., 2004). The contribution of Tel1 in checkpoint activation in these conditions has not been assessed. A role of MRN, the MRX homologue in higher eukaryotes, in ATR-dependent phosphorylation events induced by replication block (hydroxyurea) and single-strand gap (UV irradiation) is also conserved in human cells (Stiff et al., 2005). However, MRN is known to function with ATM in response to DSBs and its possible role with ATR in DSB signaling remains to be determined. Although the prevalent function of the Mec1/MRX pathway in responding to DSBs in yeast might be unusual to this organism, it is tempting to speculate that the MRN complex is involved in ATR-, as well as ATM-dependent signaling of DSBs in the mammalian system. It will thus be interesting to test whether ATR/MRN-dependent signaling requires processing of DSBs, whereas ATM/MRN-dependent signaling might not. The likely greater emphasis placed on ATM/MRN signaling in human cells perhaps indicates a greater requirement for these cells to respond to DSBs without the necessity for prior DSBs processing. Such processing of DSBs into ssDNA is consistent with repair mechanisms based on homologous recombination rather than non-homologous end joining and it is notable that human cells, unlike yeast cells, are not proficient at the former repair mechanism, particularly in the G1 phase of the cell cycle.

Acknowledgements

We thank Kevin Lewis and Michael Resnick for strain MATa T334 and plasmid pLKL31, Serge Gravel for bar1Δ wild-type strains and John Diffley and Lucy Drury for Sic1 antibodies. We thank John Diffley for permission to cite his unpublished data. We thank Serge Gravel, Ali Jazayeri, Paul Modrich, John Rouse and Thomas Wenner for useful discussion, comments and support. We also thank Ronan Bree, Serge Gravel, Ciaran Morrison, Aisling O'Shaughnessy and Alain Verreault for critical reading of the manuscript. The work in the SPJ laboratory was funded by Cancer Research UK.

Ancillary