|Home | About | Journals | Submit | Contact Us | Français|
Rad3, the Schizosaccharomyces pombe ortholog of human ATR and Saccharomyces cerevisiae Mec1, activates the checkpoint kinase Chk1 in response to DNA double-strand breaks (DSBs). Rad3ATR/Mec1 associates with replication protein A (RPA), which binds single-stranded DNA overhangs formed by DSB resection. In humans and both yeasts, DSBs are initially detected and processed by the Mre11-Rad50-Nbs1Xrs2 (MRN) nucleolytic protein complex in association with the Tel1ATM checkpoint kinase and the Ctp1CtIP/Sae2 DNA-end processing factor; however, in budding yeast, neither Mre11 nuclease activity or Sae2 are required for Mec1 signaling at irreparable DSBs. Here, we investigate the relationship between DNA end processing and the DSB checkpoint response in fission yeast, and we report that Mre11 nuclease activity and Ctp1 are critical for efficient Rad3-to-Chk1 signaling. Moreover, deleting Ctp1 reveals a Tel1-to-Chk1 signaling pathway that bypasses Rad3. This pathway requires Mre11 nuclease activity, the Rad9-Hus1-Rad1 (9-1-1) checkpoint clamp complex, and Crb2 checkpoint mediator. Ctp1 negatively regulates this pathway by controlling MRN residency at DSBs. A Tel1-to-Chk1 checkpoint pathway acting at unresected DSBs provides a mechanism for coupling Chk1 activation to the initial detection of DSBs and suggests that ATM may activate Chk1 by both direct and indirect mechanisms in mammalian cells.
DNA double-strand breaks (DSBs), formed by clastogens or from endogenous damage, trigger multiple cellular responses that are critical for maintaining genome integrity. Of particular importance is the cell cycle checkpoint that restrains the onset of mitosis while DSB repair is under way. Chk1 is the critical effector of this checkpoint in the fission yeast Schizosaccharomyces pombe and mammalian cells, whereas the budding yeast Saccharomyces cerevisiae uses both Chk1 and Rad53 (orthologous to human Chk2 and fission yeast Cds1) to delay anaphase entry and mitotic exit. These kinases are regulated by ATM (ataxia-telangiectasia mutated) and ATR (ATM and Rad3-related) checkpoint kinases (5). Curiously, the regulatory connections between ATM/ATR and Chk1/Chk2 orthologs are not strictly conserved between species (Fig. (Fig.1A).1A). In mammals, ATM activates Chk2 while ATR activates Chk1. In S. cerevisiae and S. pombe, ATR orthologs (Mec1 and Rad3, respectively) activate Chk2 orthologs and Chk1, while Tel1 (ATM ortholog) is primarily involved in telomere maintenance (14, 38, 40).
The functions of ATM and ATR orthologs are intimately tied to the detection and nucleolytic processing of DSBs. ATMTel1 localizes at DSBs by interacting with Mre11-Rad50-Nbs1Xrs2 (MRN) protein complex, which directly binds DNA ends (12, 20, 24, 50, 52). The MRN complex is essential for ATMTel1 function in all species. The Mre11 subunit of MRN complex has DNase activities that are critical for radioresistance in S. pombe and mice but not in budding yeast (3, 19, 22, 50). In fission yeast, MRN complex also recruits Ctp1 DNA end-processing factor to DSBs (25, 49). Ctp1 is structurally and functionally related to CtIP in mammals and Sae2 in budding yeast, the latter of which has nuclease activity in vitro (21, 23, 43). Ctp1 and CtIP are essential for survival of ionizing radiation and other clastogens (23, 43, 54), whereas sae2Δ mutants are not radiosensitive except at very high doses of ionizing radiation (IR), although both Ctp1 and Sae2 are required for repair of meiotic DSBs formed by a Spo11/Rec12-dependent mechanism (17, 23, 36). Genetic and biochemical studies indicate that Sae2/Ctp1/CtIP collaborate with MRN complex to initiate the 5′-to-3′ resection of DSBs (7, 23, 28, 43, 53, 55), which leads to the generation of 3′ single-strand overhangs (SSOs) that are critical for DSB repair by homologous recombination (HR). Replication protein A (RPA) binding to SSOs is essential for HR repair of DSBs, but it is also important for recruiting ATRRad3/Mec1, which interacts with RPA through its regulatory subunit ATRIP (Rad26 in fission yeast, Ddc2 in budding yeast) (5, 56). Subsequent phosphorylation of Chk1 by ATR also requires the Rad9-Hus1-Rad1 (9-1-1) checkpoint clamp, which is loaded at the single-strand/double-strand DNA junctions (26, 48, 57), the ATR activating protein TopBP1 (Cut5 in fission yeast), and a checkpoint mediator protein such as Crb2 in fission yeast (34, 41, 48).
In this mechanism of DNA damage checkpoint signaling, DNA end resection is critical for ATR (Rad3/Mec1) activation, and therefore resection defective mutants should be unable to mount a fully active checkpoint response (44). However, Rad53 activation is not diminished in budding yeast sae2Δ mutants that suffer an irreparable DSB by expressing HO endonuclease. In fact, there is a defect in turning off the checkpoint signal (6). A similar effect is observed in S. cerevisiae strains expressing the mre11-H125N nuclease-defective form of Mre11. Moreover, overexpression of SAE2 strongly inhibits Rad53 activation (6). The reasons for these phenotypes are unknown, since neither Sae2 nor Mre11 nuclease activity are required for DSB resection or radioresistance. However, deleting Sae2 delays resection while at the same time enhancing a cryptic Tel1-to-Rad53 checkpoint pathway (6, 47). These effects correlate with delayed disassembly of Mre11 foci at DSBs in sae2Δ cells, suggesting that Sae2 may negatively regulate checkpoint signaling by modulating Mre11 association at damaged DNA (1, 6, 24). Enhancement of a Tel1-to-Rad53 checkpoint pathway by eliminating Sae2 suggests that the signaling pathways between ATM/ATR and Chk1/Chk2 checkpoint kinases are not hard wired but are adaptable to changes in DNA end processing (47). However, as yet there is no evidence that ATMTel1 can activate Chk1 in any organism.
Since SAE2 deletion or overexpression has unexpected effects on Rad53 activation in budding yeast, we decided to explore the relationship between Ctp1 and Chk1 activation in fission yeast. Here, we show that Chk1 activation is substantially diminished in ctp1Δ cells exposed to ionizing radiation. These data are consistent with studies showing that CtIP is required for efficient Chk1 activation in mammalian cells treated with camptothecin (CPT), a topoisomerase I poison that causes replication fork collapse (43, 53). We also investigate the role of Mre11 nuclease activity and find that while ablating Mre11 nuclease activity enhances Rad53 activation in budding yeast, the equivalent Mre11 mutation in fission yeast severely impairs Chk1 activation by ionizing radiation. Furthermore, we find that deleting Ctp1 reveals a previously unknown Tel1-to-Chk1 signaling pathway in S. pombe, a finding analogous to the enhancement of a Tel1-to-Rad53 checkpoint pathway by eliminating Sae2 in S. cerevisiae (47). This Tel1-to-Chk1 pathway also requires Mre11 nuclease activity. These data establish that Tel1ATM can activate Chk1 independently of Rad3ATR, which has implications for studies linking ATM to Chk1 activation in mammalian cells (16, 31). Characterization of this pathway allows us to propose a more detailed model of how Chk1 is activated in response to DSBs.
Strains used in these studies were made by using standard techniques (13) and are listed in Table Table1.1. Cell survival assays were performed by plating cells in triplicate and exposing to indicated agents. Colonies were counted and expressed as a percentage relative to untreated samples. For DNA damage sensitivity assays, cells were grown in YES medium (yeast extract, glucose, and supplements) to log phase, plated as 5-fold serial dilutions, and treated as indicated. For checkpoint analysis, cells were grown to log phase and subjected to centrifugal elutriation to isolate a population of cells synchronized in early G2 phase. These cells were treated with IR and scored for progression through the cell cycle. The rad3Δ allele used in study deletes the kinase domain. A recent investigation found that this allele, but not a full deletion of the rad3 open reading frame, mediates Tel1 binding to telomeres by a pathway that does not require the C terminus of Nbs1 (46). The use of a full-deletion allele of rad3 did not alter the outcome of Chk1 phosphorylation assays in a ctp1Δ background (data not shown), suggesting that the Tel1 recruitment to DSBs is not dependent on the N terminus of Rad3, as is the case with telomeres.
Exponential cells grown in YES or minimal medium (supplemented with uracil, adenine, and histidine) were treated with 90 Gy of IR using a 137Cs source. Cells were harvested after removal from the irradiator. Samples within each blot were treated and harvested simultaneously. For Chk1, whole-cell extracts were prepared from exponentially growing cells in standard lysis buffer. Protein amounting to ~150 mg was resolved by SDS-PAGE using 15-by-18-cm 8% gels with an acrylamide/bisacrylamide ratio of 99:1. Proteins were transferred to nitrocellulose membranes, blocked with 5% milk in TBST (0.05% Tween), and probed with anti-HA (12CA5) antibody (Roche catalog no. 11583816007). Ponceau-stained blots and a polyclonal antibody against Cdc2 raised in rabbits was used for a loading control. For H2A, cells were lysed in 0.2 M sulfuric acid and proteins precipitated by addition of 20% trichloroacetic acid as previously described (37). Pellets were washed with ethanol, resuspended in loading buffer, and resolved by SDS-PAGE using 4 to 20% gradient Tris-glycine gels (Invitrogen). Proteins were transferred to nitrocellulose membranes, blocked with 5% bovine serum albumin in TBST, and probed with a phospho-specific H2A (γH2A) antibody made in rabbit (39) or total H2A (Upstate catalog no. 07-146). The data shown are representative of at least two independent experiments. Immunoblots were quantified by using ImageJ software (http://rsb.info.nih.gov/ij/), with error bars representing standard error of three independent experiments.
Chromatin immunoprecipitation (ChIP) experiments were performed as previously described (23, 49, 50) and are representative of three independent experiments. HO endonuclease expression was induced by growth in thiamine-free minimal medium at the indicated time points. Microscopic analysis confirmed that cells arrested division as a result of HO endonuclease expression, indicating efficient cutting at the HO site.
Sae2 negatively regulates Mec1-to-Rad53 signaling in budding yeast cells suffering unrepairable DSBs formed by HO endonuclease, whereas CtIP is required for efficient ATR-to-Chk1 signaling in mammalian cells exposed to the topoisomerase I poison CPT that collapses replication forks (6, 43, 53). Since RPA localization at DSBs is diminished in ctp1Δ cells of fission yeast (23), we decided to investigate the role of Ctp1 in activating Chk1 in response to repairable DSBs formed by IR. We performed immunoblot analysis to detect the phosphorylation of Chk1 that occurs when the DNA damage checkpoint is activated. These experiments were performed with asynchronous populations of cells that are predominantly (~80%) in G2 phase prior to IR treatment. Chk1 phosphorylation was monitored for 4 h after exposure to 90 Gy of IR. Chk1 phosphorylation peaked in both wild-type and ctp1Δ cells ~30 min after IR exposure, but the Chk1 phosphorylation in ctp1Δ cells was substantially lower than in the wild type (Fig. (Fig.1B).1B). It is worth noting that in untreated cells, we detected a low level of Chk1 phosphorylation in ctp1Δ cells but not in the wild type. This low level of constitutive Chk1 phosphorylation in ctp1Δ cells likely reflects the important role that Ctp1 plays in repair of endogenous DNA damage, which when left unrepaired leads to persistent activation of the DNA damage checkpoint. We quantified the ratio of phosphorylated Chk1 versus unmodified Chk1, subtracting the ratio observed in untreated ctp1Δ cells to reflect only the Chk1 phosphorylation resulting from IR treatment and confirmed that ctp1Δ cells had an ~2-fold decrease in the peak of Chk1 phosphorylation relative to the wild type (Fig. (Fig.1C).1C). Thus, the diminished RPA localization at DSBs in fission yeast ctp1Δ cells correlates with decreased Chk1 activation, as predicted by current checkpoint models.
Rad3 is essential for Chk1 activation in S. pombe (27, 48). As controls for the Chk1 activation assays, we monitored Chk1 phosphorylation in a rad3Δ mutant and in ctp1Δ rad3Δ cells. As expected, Chk1 phosphorylation was abolished in rad3Δ cells. Surprisingly, we found that IR triggered Chk1 phosphorylation in ctp1Δ rad3Δ cells (Fig. (Fig.1D).1D). We quantified these results and found that the relative amount of Chk1 that was phosphorylated in ctp1Δ rad3Δ cells was reduced in comparison to the wild type or ctp1Δ mutant, but it was clearly evident, unlike in rad3Δ cells (Fig. (Fig.1E).1E). These findings were reminiscent of a study that found that deleting Sae2 enhances Rad53 phosphorylation in budding yeast mec1 mutants treated with methyl methanesulfonate, a DNA alkylating agent that blocks replication forks (47).
We measured cell cycle progression to independently assess the checkpoint response in ctp1Δ rad3Δ cells. We observed that IR caused a substantial checkpoint arrest in ctp1Δ rad3Δ cells, whereas the checkpoint delay was abolished in rad3Δ cells (Fig. (Fig.1F).1F). Thus, restoration of Chk1 phosphorylation in ctp1Δ rad3Δ cells correlates with reestablishment of a checkpoint arrest.
We extended these analyses to genotoxin survival assays that included nbs1Δ strains. Mutants lacking Nbs1 or Ctp1 are equally sensitive to IR, and combining the mutations had no additive effects (23). However, we found that nbs1Δ and ctp1Δ mutations have quite different effects in a rad3Δ background: nbs1Δ rad3Δ cells grew poorly relative to either single mutant, whereas ctp1Δ rad3Δ cells grew as well as either single mutant (Fig. (Fig.2A,2A, left panel). These differences can be explained by the critical role of Nbs1 in recruiting Tel1 to telomeres; nbs1Δ rad3Δ cells lose telomeres because Rad3 and Tel1 share a function essential for telomere maintenance (4, 33). In contrast, ctp1Δ rad3Δ cells maintain functional telomeres, showing that Ctp1 is not essential for Tel1 function at telomeres (23).
Having constructed nbs1Δ rad3Δ and ctp1Δ rad3Δ strains, we compared them and the corresponding single mutant strains in DNA damage survival assays using IR, the topoisomerase I poison CPT, or UV irradiation. As expected from the telomere defects, nbs1Δ rad3Δ cells were acutely sensitive to these DNA-damaging agents. In contrast, ctp1Δ rad3Δ cells were more resistant to DNA damage than the more sensitive single mutants (Fig. (Fig.2A).2A). The exact relationship varied depending on the genotoxin. In response to UV, rad3Δ cells were more sensitive than ctp1Δ cells, and the ctp1Δ mutation partially rescued the rad3Δ UV sensitivity. On the other hand, when treated with IR or CPT, ctp1Δ cells were more sensitive than rad3Δ cells, and the rad3Δ mutation partially rescued ctp1Δ cells (Fig. (Fig.2A2A).
To specifically test whether the positive genetic interaction between ctp1Δ and rad3Δ mutations arises from restoration of Chk1 activation, we introduced the chk1Δ mutation into the ctp1Δ rad3Δ background. We also included the cds1Δ allele in these assays, because Cds1 checkpoint kinase (orthologous to mammalian Chk2 and budding yeast Rad53) is activated by Rad3 as part of the checkpoint response to replication fork arrest (38). Elimination of Cds1 modestly impaired IR and UV resistance, but it strongly decreased CPT resistance (Fig. 2B and C). However, Chk1 was critical for survival of all of the tested genotoxins in the ctp1Δ rad3Δ background (Fig. 2B and C).
These data established that deleting Ctp1 allows another kinase to step in and substitute for Rad3 in phosphorylating Chk1. Tel1 was a candidate because it is a structurally similar enzyme that is recruited to DSBs where it and Rad3 phosphorylate histone H2A (32, 52), and in budding yeast Tel1 is responsible for the Mec1-independent phosphorylation of Rad53 that is enhanced by eliminating Sae2 (47). We tested this hypothesis by creating a ctp1Δ rad3Δ tel1Δ strain. Immunoblot analysis showed that IR-induced Chk1 phosphorylation was lost in this strain (Fig. (Fig.2D).2D). To confirm these results, we transformed ctp1Δ rad3Δ tel1Δ cells with a plasmid expressing tel1+ and found that tel1+ expression restored Chk1 phosphorylation in response to IR (Fig. (Fig.2E).2E). These data showed that Tel1 is required for Chk1 phosphorylation triggered by DSBs in ctp1Δ rad3Δ cells.
We next investigated the mechanism by which eliminating Ctp1 allows Tel1 to activate Chk1. Since Ctp1 localizes at DSBs by binding Nbs1 (49), we hypothesized that Ctp1 activity might determine the duration of MRN complex binding to DSBs. Sustained binding of MRN complex in ctp1Δ cells may allow persistent signaling of Tel1 to Chk1. Consistent with this model, immunofluorescence microscopy studies showed that Mre11 foci persist in S. cerevisiae sae2Δ mutants treated with IR or HO endonuclease (6, 24). To test our model, we used chromatin immunoprecipitation (ChIP) to measure Mre11 binding to a defined DSB in ctp1+ or ctp1Δ backgrounds. This experiment was performed in a strain that possesses a unique HO endonuclease cleavage site in chromosome I and expresses HO nuclease from the thiamine-repressible nmt41 promoter. In ctp1+ cells, ChIP analysis detected transient binding of Mre11 ~0.2 kb away from the DSB. In contrast, in a ctp1Δ background, enrichment of Mre11 at the DSB is massively increased and persists for the duration of the experiment (Fig. (Fig.3).3). From this, we propose that deletion of Ctp1 results in persistent binding of MRN complex at DSBs, extending what might be normally a transient Tel1-to-Chk1 signal.
Rad3-catalyzed activation of Chk1 requires recruitment of additional checkpoint factors to DSBs. These factors include the PCNA-like 9-1-1 protein complex, a checkpoint clamp comprised of Rad9, Hus1, and Rad1, which is loaded on single-strand/double-strand DNA junctions at DNA damage sites by a RFC-like protein complex in which the large subunit is replaced by Rad17 (26, 48, 57). To determine the role of the 9-1-1 protein complex in the Tel1-mediated Chk1 signaling pathway, we deleted the Hus1 subunit of the complex and found that Chk1 phosphorylation was greatly diminished in the ctp1Δ rad3Δ background (Fig. (Fig.4A).4A). Resistance to DNA-damaging agents was also lost (Fig. (Fig.4B4B).
Rad3-mediated phosphorylation of Chk1 also requires Crb2, a checkpoint mediator protein related to mammalian 53BP1 and budding yeast Rad9 (41). Crb2 associates with both Rad3 and Chk1 and thereby mediates phosphorylation of Chk1 by Rad3 (29). We found that deletion of Crb2 completely abolished Chk1 phosphorylation in ctp1Δ rad3Δ cells (Fig. 4C and D). Furthermore, deletion of Crb2 abolished the ctp1Δ rad3Δ rescue (Fig. 4E to G). These data indicate that the functional interactions between Rad3 and Crb2 that are required for Chk1 activation also likely occur between Tel1 and Crb2.
There are two pathways that lead to Crb2 localization at sites of DNA damage (10). One pathway requires two independent histone modifications: (i) phosphorylation of the C-terminal tail of histone H2A (γH2A) and (ii) methylation of histone H4 on lysine 20 (H4K20me) (32, 42). The C-terminal half of Crb2 has tandem pairs of Tudor and BRCT domains that directly bind H4K20me and γH2A, respectively (2, 18). The other pathway for Crb2 localization at DSBs requires an interaction with Cut5, the homolog of mammalian TopBP1, via a cyclin-dependent kinase (CDK) phosphorylation of threonine-215 in Crb2 (10). Whereas deleting Crb2 abolished Chk1 phosphorylation in ctp1Δ rad3Δ cells, eliminating either Crb2 threonine-215 phosphorylation (crb2-T215A) or histone H2A phosphorylation (H2A-AQ) reduced but did not completely abrogate Chk1 phosphorylation (Fig. 4C and D). This relationship was also reflected in genotoxin survival assays, where elimination of either one of the pathways reduced viability in the ctp1Δ rad3Δ background (Fig. 4E to G). In fact, the crb2-T215A mutation in the ctp1Δ rad3Δ background caused a phenotype very similar to a full deletion of Crb2, suggesting that the remaining Chk1 phosphorylation in this strain is unable to cause a significant resistance to DNA damage. Thus, as with activation of Chk1 by Rad3, either of the pathways for recruiting Crb2 to DSBs is sufficient for some signaling of Tel1 to Chk1, but the pathway of Crb2 recruitment dependent on its interaction with Cut5 may be dominant.
Tel1 phosphorylation of histone H2A at DSBs depends on its interaction with the Nbs1 subunit of MRN complex (52). To determine whether activation of Chk1 by Tel1 in ctp1Δ rad3Δ cells also requires MRN complex, we deleted Mre11 in this strain. Chk1 phosphorylation was abolished in this strain, indicating that this Tel1-mediated signaling requires the canonical pathway of MRN complex localizing to DSBs and recruiting Tel1 (Fig. (Fig.5A5A).
Mre11 possesses endonuclease and 3′ to 5′ exonuclease activities. Site-directed mutational studies of S. pombe Mre11 guided by a crystal structure of Pyrococcus furiosus Mre11 dimers bound to duplex DNA identified alleles that ablate only exonuclease activity (mre11-H68S) or both nuclease activities (mre11-H134S). The mre11-H68S mutation has little effect in fission yeast, whereas mre11-H134S causes severe radiosensitivity (50). The mre11-H129N mutation in murine Mre11, which is equivalent to fission yeast mre11-H134S, is homozygous lethal in mice and causes acute radiosensitivity and genome instability in assays of mre11-H129N/Δ cells (3). On the other hand, equivalent mutations in S. cerevisiae (e.g., mre11-H125N) cause only mild radiosensitivity (19, 22). Rad53 undergoes prolonged activation in budding yeast mre11-H125N cells expressing HO endonuclease (6), whereas Chk1 phosphorylation triggered by UV irradiation is impaired in mouse mre11-H129N/Δ cells (3).
To investigate the role of Mre11 nuclease activity in checkpoint signaling in fission yeast, we monitored IR-induced Chk1 phosphorylation in mre11-H134S strains. Importantly, previous studies had shown that this allele does not diminish Mre11 abundance, nor does it impair Mre11, Nbs1, or Ctp1 binding to a DSB, indicating that the mre11-H134S mutations specifically ablates Mre11 nuclease activity without affecting Mre11 stability or it interactions with Rad50, Nbs1, or Ctp1 (49, 50). We found that Chk1 phosphorylation was strongly diminished in mre11-H134S cells (Fig. (Fig.5B).5B). Chk1 phosphorylation was completely eliminated in mre11-H134S rad3Δ cells, showing that Rad3 catalyzed the weak activation of Chk1 in mre11-H134S cells. Surprisingly, deleting Ctp1 in mre11-H134S rad3Δ cells did not restore Chk1 phosphorylation (Fig. (Fig.5B)5B) and abolished the suppressive genetic interactions between ctp1Δ and rad3Δ (Fig. (Fig.5C).5C). From these results, we conclude the Mre11 nuclease activity is critical for both Rad3 and Tel1 signaling to Chk1.
These data suggested that Mre11 nuclease activity might be essential for Tel1 activity at DSBs. To test this hypothesis, we examined C-terminal phosphorylation of histone H2A. Confirming previous studies (32), we found that Rad3 and Tel1 have redundant activities in creating phospho-H2A (γH2A) in response to IR (Fig. (Fig.5D).5D). This activity was abolished in mre11Δ rad3Δ cells, which is consistent with studies showing that MRN complex recruits Tel1 to DSBs. However, unlike mre11Δ rad3Δ cells, the mre11-H134S rad3Δ strain was fully proficient in phosphorylating histone H2A (Fig. (Fig.5D).5D). Since phosphorylation of histone H2A by Tel1 requires MRN to recruit Tel1 to DSBs, these results reinforce the evidence that Mre11-H134S protein is fully capable of forming a stable MRN complex and recruiting Tel1 to DSBs.
We repeated Chk1 and histone H2A phosphorylation assays with another nuclease dead allele of Mre11, mre11-D65N (35, 51), and obtained identical results with that of mre11-H134S (Fig. 5E and F). Taken together, these data show that Mre11 nuclease activity is not required to activate Tel1 at DSBs, but it is critical for a downstream step required for Chk1 phosphorylation by Tel1. The likely nature of this step is discussed below.
This study was initiated with the goal of defining the relationships between nucleolytic processing of DNA ends and checkpoint signaling in fission yeast. We found that IR-induced phosphorylation of Chk1 is strongly decreased in cells lacking Mre11 nuclease activity or Ctp1. These findings suggest a checkpoint mechanism in which Ctp1 and MRN complex initiate DSB resection, leading to RPA binding ssDNA, recruitment of Rad3 and other checkpoint proteins, and finally to phosphorylation of Chk1 by Rad3. These data are consistent with evidence that CtIP-dependent resection is rate limiting for ATR activation of Chk1 in mammalian cells treated with CPT (43, 44, 53). In contrast, elimination of Sae2 does not impair Mec1-dependent activation of Rad53 in budding yeast cells suffering DSBs; in fact, Rad53 activation is enhanced in sae2Δ cells and suppressed in cells that overexpress SAE2. Moreover, HO-induced DSBs caused robust and sustained activation of Rad53 in a strain lacking Mre11 nuclease activity (6).
Interestingly, we also found that deleting Ctp1 uncovers a second Chk1 activation pathway that uses Tel1 instead of Rad3. This pathway requires Mre11 nuclease activity, the 9-1-1 checkpoint clamp, and Crb2. In fact, the pathway of Crb2 recruitment mediated by its interaction with Cut5TOPBP1 via threonine-215 seems to be the dominant one, which suggests that this pathway is important in the checkpoint response prior to extended resection. Parallel studies in S. cerevisiae have shown that deleting Sae2 enhances Tel1-mediated activation of Rad53, a Cds1/Chk2 ortholog. As for the Tel1-to-Chk1 pathway in fission yeast, this “TM” pathway requires Mre11 and the Crb2-related protein Rad9. However, the 9-1-1 complex seems to be dispensable, and the role of Mre11 nuclease activity is unknown (47).
The requirement for Mre11 nuclease activity shows unequivocally that this activity can function independently of Ctp1 in vivo and suggests that DNA end processing by Mre11 is required for Chk1 activation by Tel1. This activity is not required for Tel1 activity per se, since we found that Tel1 efficiently phosphorylates histone H2A in both mre11-H134S and mre11-D65N backgrounds. Instead, Mre11 nuclease activity likely generates a DNA structure that is essential for Chk1 activation, probably a short single-stranded overhang (SSO) that is used for loading the Rad9-Hus1-Rad1 checkpoint clamp. Such a nuclease activity has been described for Mre11-Rad50 complex from Pyrococcus furiosus (15).
How does deleting Ctp1 uncover a Tel1-to-Chk1 pathway? Our data show that Mre11 residence at a DSB is fleeting in wild type, but sustained and enhanced in ctp1Δ cells (Fig. (Fig.3B).3B). Since Tel1 binds the Nbs1 subunit of MRN complex (52), we propose that sustained binding of MRN complex allows Tel1 to remain at DSBs in ctp1Δ cells, revealing a Tel1-to-Chk1 checkpoint pathway. It is uncertain whether this pathway is transiently activated in wild-type cells. Active Tel1 associates with DSBs in rad3Δ cells; however, Chk1 phosphorylation has not been detected in rad3Δ cells, nor is a checkpoint delay evident in rad3Δ strains. On the other hand, weak IR-induced Chk1 phosphorylation was observed in cells arrested in G1 phase (27), a phase of the cell cycle when Ctp1 is not expressed (23), leaving open the possibility that resection-independent signaling through Tel1 is a cell cycle-regulated event. It is important to note that in mammalian cells, Mre11 complex also associates with DSBs through chromatin-based interactions with Mdc1, which binds phospho-histone H2AX surrounding DSBs (45), allowing for the possibility of sustained recruitment of ATM through its interaction with Mdc1.
Our data show that MRN complex persistently binds to DSBs in the absence of Ctp1. Ctp1 associates with DSBs via an interaction with the FHA domain at the N terminus of Nbs1, but Ctp1 does not a stably bind the MRN complex (23). Thus, the duration of MRN binding at DSBs is potentially determined by the affinity of Ctp1 for the MRN complex. As Tel1 associates with DSBs by binding to the C terminus of Nbs1, the affinity of Ctp1 for Nbs1 may regulate signaling by Tel1. These relationships may explain why Ctp1 is not a stable subunit of MRN complex, because if it were, there would be little opportunity for Tel1 to associate with DSBs. Consistent with these observations, recent studies with Xenopus egg extracts and mammalian cells show that the appearance of Nbs1 and ATM at DSBs precedes that of CtIP (53).
From these data we can propose a biphasic or handoff mechanism of checkpoint signaling at DSBs that can operate whenever Ctp1/CtIP recruitment is significantly delayed relative to Tel1/ATM (Fig. (Fig.6).6). As illustrated for fission yeast, the first step is MRN complex binding to DSBs, followed by Tel1 recruitment mediated by Nbs1. Tel1 phosphorylates histone H2A in chromatin flanking the DSB, activating one of two pathways that can recruit Crb2. Mre11 endonuclease creates a short 3′ overhang required for loading the checkpoint clamp and Cut5. With these factors in place, Crb2 mediates phosphorylation of Chk1 by Tel1.
The second phase of the checkpoint response commences with Ctp1 binding the N-terminal FHA domain of Nbs1 (25, 49). Ctp1-dependent resection disengages the MRN complex and presumably Ctp1 itself from DNA ends. It is unknown how this happens, but it is noteworthy that MRN disengagement appears to also require Mre11 endonuclease activity (49). One possibility is that Mre11 complex has poor affinity for resected DNA ends. Alternatively, formation of an RPA-ssDNA filament may displace MRN complex from DNA ends. In vitro studies support the latter possibility, since MRN complex in HeLa cell extracts preferentially binds blunt-ended duplex DNA substrates versus DNA substrates with single-strand overhangs, whereas purified MRN complex displayed no preference (44). In either case, Ctp1-dependent resection of DNA terminates MRN complex association with DNA ends, and with it the DSB association of Tel1 and Ctp1 itself. Thus, both MRN and Ctp1 self-limit their association DNA ends, promoting a handoff to downstream elements of the HR repair pathway. Displacement of MRN, Ctp1, and Tel1 from DNA ends by an RPA-ssDNA filament would provide for a seamless transition to the second phase of the checkpoint response in which RPA recruits Rad3-Rad26 complex. The handoff between Tel1 and Rad3 can ensure that checkpoint responses are both fast and yet can be maintained while DSBs are processed for HR repair.
In support of this model, a recent study using DNA templates added to HeLa cell nuclear extracts provided strong evidence for an ATM-to-ATR switch coinciding with formation of single-stranded overhangs (SSOs) (44). Double-stranded ends and single-stranded/double-stranded DNA junctions with 5- to 25-nucleotide SSOs activated ATM, but longer SSOs attenuated ATM activation and stimulated ATR. Using Chk2 (Cds1) and Chk1 as ATM- and ATR-specific substrates, respectively, we found that ATM and ATR are consecutively activated in HeLa cells treated with IR. A period of overlapping Chk2 and Chk1 phosphorylation occurring ca. 10 to 30 min after IR treatment was hypothesized to arise from asynchronous processing of DSBs, but our studies of fission yeast suggest an alternative explanation, namely, that ATM may contribute directly to the early period of Chk1 phosphorylation.
In mammalian cells, both ATR and ATM are critical for Chk1 activation in response to DSBs (16, 31). ATM was recently shown to be required for efficient recruitment of CtIP to DSBs, suggesting that a resection deficiency may underlie the IR-induced Chk1 activation defect in ATM-null cells (43, 53). However, regulating CtIP may not fully explain the role of ATM in Chk1 activation, since Chk1 is significantly activated in CtIP knockdown cells treated with CPT, which creates DSBs in S-phase (43, 53). By extrapolating from fission yeast, our data raise the possibility that ATM may be directly responsible for a portion of Chk1 activation observed in the absence of CtIP. This may explain the severe IR sensitivity, genetic instability and cancer predisposition of A-T patients. From the data currently available, the only crucial difference between fission yeast and mammalian cells with respect to resection and checkpoint signaling is that Tel1 is not required for efficient DNA end processing, hence is not required for Chk1 activation by Rad3 in fission yeast.
In the present study we found that the UV sensitivity of rad3Δ cells is partially suppressed by ctp1Δ, even though the ctp1Δ mutation itself causes moderate UV sensitivity. Suppression of rad3Δ by ctp1Δ requires Chk1, indicating that it occurs through the partial reestablishment of a Chk1-mediated checkpoint response that delays the onset of mitosis. However, we also found that rad3Δ partially suppresses the IR and CPT sensitivity of ctp1Δ cells, even though the rad3Δ mutation causes substantial IR and CPT sensitivity. This suppression also requires Chk1. These genetic interactions are more challenging to explain. One possibility is that as both Rad3 and Tel1 activate Chk1 in ctp1Δ cells, deleting Rad3 abbreviates but does not abolish the checkpoint response, which might confer some survival advantage to a subset of the ctp1Δ cells that are unable to complete DNA repair. Indeed, cells lacking Dis2, a putative Chk1 phosphatase, were shown to accumulate phosphorylated Chk1, have a prolonged checkpoint arrest, and acute sensitivity to DNA damaging agents such as the alkylating agent, methyl methanesulfonate (8). Another possibility is that Rad3 (but not Tel1) might negatively regulate an activity that can compensate for the absence of Ctp1. Elimination of the Ku protein complex required for nonhomologous end joining partially rescues ctp1Δ (23). This suppression requires Exo1, a 5′-to-3′ exonuclease, suggesting that Ku prevents Exo1 from substituting for Ctp1 in catalyzing the resection of DNA ends required for homologous recombination repair of DSBs. Studies in both human cells and budding yeast have indicated that phosphorylation of Exo1 negatively regulates its activity. Exo1 phosphorylation appears to be regulated by the Mec1/ATR pathways in budding yeast and human cells (11, 30). Extrapolating to fission yeast, it is plausible that eliminating Rad3 enhances Exo1 activity and thereby allows Exo1 to substitute for Ctp1. Future experiments will address this possibility.
We thank Charly Chahwan, Yoshiki Yamada, Li-Lin Du, Michael N. Boddy, and members of the Russell Lab for critical discussions, support, and advice. We also thank Devon Nieto for technical support and Christophe Redon for the generous gift of γH2A antibody.
This study was funded by National Institutes of Health grants awarded to P.R. (CA77325 and GM59447) and N.R. (GM069957).
Published ahead of print on 22 November 2010.