Here we examine the roles of budding-yeast checkpoint proteins in regulating degradation of dsDNA to ssDNA at unprotected telomeres (in Cdc13 telomere-binding protein defective strains). We find that Rad17, Mec3, as well as Rad24, members of the putative checkpoint clamp loader (Rad24) and sliding clamp (Rad17, Mec3) complexes, are important for promoting degradation of dsDNA in and near telomere repeats. We find that Mec1, Rad53, as well as Rad9, have the opposite role: they inhibit degradation. Downstream checkpoint kinases Chk1 and Dun1 play no detectable role in either promoting degradation or inhibiting it. These data suggest, first, that the checkpoint sliding clamp regulates and/or recruits some nucleases for degradation, and, second, that Mec1 activates Rad9 to activate Rad53 to inhibit degradation. Further analysis shows that Rad9 inhibits ssDNA generation by both Mec1/Rad53-dependent and -independent pathways. Exo1 appears to be targeted by the Mec1/Rad53-dependent pathway. Finally, analysis of double mutants suggests a minor role for Mec1 in promoting Rad24-dependent degradation of dsDNA. Thus, checkpoint proteins orchestrate carefully ssDNA production at unprotected telomeres.
TO understand the biochemical mechanisms underlying the DNA-damage response, it is essential to understand how checkpoint proteins interact with damaged DNA and affect repair pathways. Although cell cycle arrest and DNA repair were initially thought to be independent cellular responses to damaged DNA (Weinert and Hartwell 1988), this view has evolved and the current consensus is that checkpoint and repair pathways are partially interdependent (Zhou and Elledge 2000; Carr 2002; D’Amours and Jackson 2002; Nyberget al. 2002; Rouse and Jackson 2002a). It is now thought that an effective DNA-damage response requires that DNA repair and checkpoint pathways interact and affect each other to improve the efficiency of repair.
Understanding the myriad roles of checkpoint proteins in regulating cell cycle arrest, transcription, DNA repair, and apoptosis, including roles in DNA metabolism discussed in this report, benefits from understanding the composition and proposed roles of different checkpoint protein complexes. Checkpoint signal transduction cascades comprise DNA-damage sensors that directly interact with damaged DNA, signal transduction proteins that transmit the checkpoint signal, mediators that act among signalers, and targets that affect cell cycle progression (Nyberget al. 2002). In budding yeast, several checkpoint sensor complexes bind damaged DNA independently of each other. The cardinal complex comprises the Mec1 PI3 kinase type kinase, which is orthologous to mammalian ATR, which exists in a heterodimer with Ddc2 and binds damaged DNA independently of other checkpoint complexes (Kondoet al. 2001; Meloet al. 2001; Rouse and Jackson 2002b; Zou and Elledge 2003). Mec1/Ddc2 is involved in virtually all known DNA-damage checkpoint pathways in budding yeast. However, its role in cell signaling is dependent on other checkpoint protein complexes.
Other checkpoint complexes include a proliferating cell nuclear antigen (PCNA)-type heterotrimer consisting of Rad17, Mec3, and Ddc1 in budding yeast. This complex also binds damaged DNA (Kondoet al. 2001; Meloet al. 2001) and is called either the checkpoint sliding clamp or the 9-1-1 complex (after the human and Schizosaccharomyces pombe orthologs comprising Rad9, Rad1, and Hus1). This Rad17 checkpoint sliding clamp complex is loaded onto damaged DNA by a third complex consisting of Rad24 and Rfc2-5 subunits called a “checkpoint clamp loader” (Greenet al. 2000; Bermudezet al. 2003; Majka and Burgers 2003).
A fourth protein important in checkpoint signaling is Rad9, which has been classified as a checkpoint mediator or adaptor protein (Melo and Toczyski 2002; Osbornet al. 2002). In vertebrates and yeast cells, this class of proteins includes claspin (Kumagai and Dunphy 2000), BRCA1 (Nyberget al. 2002), 53BP1 (Wanget al. 2002), MDC1 (Lou et al. 2003a,b; Stewartet al. 2003), and MRC1 (Alcasabaset al. 2001; Tanaka and Russell 2001). Mediator proteins are thought to facilitate interactions between upstream checkpoint proteins such as ATR and ATM and downstream checkpoint kinases such as Chk1 and Chk2. In budding yeast, Rad9 appears to mediate interactions between the upstream kinase Mec1/Ddc2 and two downstream kinases, Rad53 (a budding-yeast Chk2 ortholog) and Chk1 (Gardneret al. 1999; Sanchezet al. 1999; Gilbertet al. 2001; Schwartzet al. 2002; Blankley and Lydall 2004).
Checkpoint sensor proteins probably perform several functions once they bind to DNA damage, including regulation of DNA-damage metabolism. A role for checkpoint proteins in DNA-damage metabolism was first shown by examining budding-yeast mutant cells with a defect in the telomere-binding protein Cdc13 (Weinert and Hartwell 1993; Garviket al. 1995; Lydall and Weinert 1995). Telomeres are not protected by Cdc13 in temperature-sensitive cdc13-1 cells grown at restrictive temperatures (Garviket al. 1995; Mitton-Fryet al. 2002). This absence of Cdc13-dependent protection at telomeres leads to binding of checkpoint sensor proteins (Meloet al. 2001; Rouse and Jackson 2002b), cell cycle arrest (Weinert and Hartwell 1993; Lydall 2003), accumulation of single-stranded DNA (ssDNA) on the 3′-strand of telomeres (Garviket al. 1995), and recombination (Grandinet al. 2001).
Studies of the generation of ssDNA in cells with a defect in the Cdc13 telomere-binding protein revealed that checkpoint proteins in fact regulate ssDNA production themselves (Lydall and Weinert 1995). We found that although both Rad9 and Rad24 are required for cell cycle arrest, remarkably, each protein has a profound and opposite effect on ssDNA production (Lydall and Weinert 1995; Boothet al. 2001): Rad24 is required for maximal levels of ssDNA, while Rad9 inhibits ssDNA production.
Since it is generally thought that ssDNA is an important and universal component for recruiting checkpoint complexes and for activating cell cycle arrest (Lydall and Weinert 1995; Sogoet al. 2002; Vazeet al. 2002; Carr 2003; Zou and Elledge 2003), and in the light of the current classifications of different checkpoint protein complexes, here we extend our analyses of differing checkpoint proteins in regulating ssDNA production in cells with exposed telomeres. By examining ssDNA production in cells defective in the Mec1/Ddc2 protein kinase, the Rad53 and Chk1 effector kinases, and the Rad17-checkpoint sliding clamp, we have developed a model of how checkpoint proteins regulate ssDNA production. We suggest that the Rad24-Rfc2-5 checkpoint clamp loader loads the Rad17-checkpoint sliding clamp onto defective, unprotected telomeres and that the Rad17-checkpoint sliding clamp then recruits or tethers nucleases that convert double-stranded DNA (dsDNA) to ssDNA (Boothet al. 2001; Majka and Burgers 2003). To counteract or limit the ssDNA production, Mec1 activates Rad9 so that it inhibits degradation through both Rad53-dependent and Rad53-independent mechanisms. Mec1 has another minor role in activation of ssDNA production that acts through the Rad24 or Rad17 complexes. Finally, cell death associated with ssDNA may arise by mechanisms that require Mec1, Rad53, and Exo1.
MATERIALS AND METHODS
Yeast strains: Unless otherwise indicated, the strains used in this study are in the W303 background and were RAD5, rather than rad5-535 (Fanet al. 1996; see online supplemental Table 1 available at http://www.genetics.org/supplemental/). To construct strains, standard genetic procedures of transformation and tetrad analysis were followed (Adamset al. 1997). Since W303 strains contain an ade2-1 mutation, yeast extract, peptone, and dextrose (YEPD) medium was supplemented with adenine at 50 μg/liter.
Serial dilution and growth on plates: Colony-purified yeast strains were inoculated into 2 ml YEPD (ade) grown overnight with aeration at 23°. Fivefold dilution series were set up in 96-well plates, and small, 3- to 4-μl aliquots of the dilution series were transferred to duplicate YPD (ade) plates using metal prongs. Plates that had been inoculated with samples of the fivefold dilution series were incubated until colonies formed and then were photographed.
Liquid culture, medial nuclear division, viability assays, and ssDNA protection: Cells were cultured and collected for monitoring nuclear divisions and viability assays as described before (Lydall and Weinert 1997). Single-stranded DNA was measured as before (Boothet al. 2001). In a recent series of experiments, we found that the effect of RAD9 on ssDNA production in cdc13-1 mutants is most readily observed at PDA1, rather than at YER186C (compare Figure 2, H and I; M. Zubko, S. Guillard and D. Lydall, unpublished results), whereas previously we also observed large differences between RAD9 and rad9Δ strains at YER186C (Lydall and Weinert 1995; Boothet al. 2001). This may be because cdc13-1 and cdc15-2 mutations in most cases have been introduced by transformation rather than by backcrossing. Irrespective of such differences, it is clear that in all experiments RAD9 inhibits while RAD24 activates ssDNA production at telomeres of cdc13-1 mutants.
Quantitative amplification of ssDNA: Quantitative amplification of ssDNA (QAOS) has been described in detail by Booth et al. (2001). In brief, high-quality DNA isolated from yeast nuclei was prepared using a QIAGEN column-based protocol. For all real-time PCR measurements, samples and standards were measured in triplicate in 96-well plates in an ABI7700 real-time PCR machine, using oligonucleotides and conditions described previously (Boothet al. 2001). DNA was quantified by real-time PCR, standards containing 4, 2, and 0.4 ng/μl yeast genomic DNA, and primers directed to the PAC2 locus, close to the centromere of chromosome V. DNA concentrations were then adjusted to 2 ng/μl and ssDNA was measured by QAOS in comparison with standards that contained 51.2, 3.2, and 0.8 (or 0.2)% boiled (single-stranded) DNA.
Rad17 and Mec3 are required to generate ssDNA at telomeres in Cdc13-defective cells: The RAD24 genetic epistasis group is composed of RAD24, RAD17, MEC3, and DDC1. Mutations in any or all of these genes result in a similar set of phenotypes (Lydall and Weinert 1995; Paciottiet al. 1998). RAD24 is important for generating ssDNA near the telomeres of cdc13-1 and cdc13-1 rad9Δ mutants at the restrictive temperature, but other members of the RAD24 group have not been tested (Lydall and Weinert 1995; Boothet al. 2001). Therefore we combined rad17Δ and mec3Δ mutations with cdc13-1 (bar1 and cdc15-2 mutations) and measured cell cycle arrest and ssDNA production at telomeres. BAR1 encodes a secreted protease and a bar1 mutation makes cells more sensitive to the peptide, mating pheromone, and α-factor. CDC15 is involved in mitotic exit and a cdc15-2 mutation ensures that checkpoint-defective cdc13-1 mutants do not undergo additional rounds of DNA replication at restrictive temperatures (Lydall and Weinert 1997).
Cell cycle arrest was determined from nuclear morphology of fixed cells (Lydall and Weinert 1997). ssDNA levels were determined using a quantitative real-time PCR-based protocol that detects ssDNA but not dsDNA (Boothet al. 2001). The principles of QAOS are shown in Figure 1. By using real-time PCR to measure the accumulation of ssDNA-dependent PCR product and by measuring each DNA sample in triplicate, it is possible to accurately determine the amount of ssDNA at single-copy loci in the yeast genome in the range of 0.2-50% (Figure 1, D and E; Boothet al. 2001).
We measured cell cycle arrest and ssDNA production in cells as they progressed through one cell cycle with a defective telomere-binding protein, Cdc13. We found that rad17Δ and mec3Δ mutants fail to arrest at the G2/M checkpoint in the absence of Cdc13, as previously reported (Figure 2, B and D). The slight delay shown by the rad17Δ strain in this experiment was not observed in repeated experiments.
We then tested whether Rad17 and Mec3 regulate ssDNA production in cdc13-1 mutants. We found that rad17Δ and mec3Δ mutants, like rad24Δ mutants, fail to generate large amounts of ssDNA, measured 14.5 kb from the right telomere of chromosome V (YER186C, Figure 2I) (Lydall and Weinert 1995; Boothet al. 2001). rad17Δ and mec3Δ mutants failed to generate ssDNA in either RAD9+ cells or rad9Δ mutant cells (Figure 2G and data not shown). Therefore, we conclude that Rad17 and Mec3, and thus the checkpoint sliding clamp complex, are required for robust ssDNA production that accumulates near telomeres of cdc13-1 mutants. These experiments are consistent with the idea that Rad24 regulates ssDNA accumulation in Cdc13-defective cells by recruiting the Rad17-checkpoint sliding clamp complex, which then may tether an exonuclease that directly degrades DNA near telomeres.
Rad53 inhibits ssDNA production at telomeres in Cdc13-defective cells: Rad9 is required to activate the Rad53-checkpoint kinase, which is important for signaling and cell cycle arrest (Gilbertet al. 2001). We confirmed the partial arrest phenotype of rad53Δ mutants in response to cdc13-1-induced damage (Figure 3, B and C; Gardneret al. 1999). We have shown previously that Rad9 inhibits production of ssDNA at telomeres in cells with defective Cdc13 telomere-binding proteins (Lydall and Weinert 1995; Boothet al. 2001). Because Rad9 acts through Rad53 to cause cell cycle arrest, we wondered whether Rad9 also acts through Rad53 to inhibit ssDNA production near telomeres. Therefore we examined the effect of Rad53 on ssDNA production at several telomeric loci, data from two of which are shown in Figure 3, D and E; one locus is YER186C, 15 kb from the telomere, and a second locus is PDA1, 30 kb from the telomere.
We found that Rad53, like Rad9, inhibits ssDNA production, although its inhibitory role is not as profound as that of Rad9. In cdc13 rad53 mutants, ssDNA accumulates at the locus 30 kb from the telomere within 200 min (Figure 3D), clearly more rapidly than cdc13-1 RAD+ cells, which accumulate very little ssDNA at this locus but less rapidly than cdc13-1 rad9Δ cells, which accumulate ssDNA by 120 min (Figure 3G). Checkpoint double-mutant cdc13-1 rad9Δ rad53 strains accumulated ssDNA 30 kb from the telomere at a similar rate to cdc13-1 rad9Δ mutants (Figure 3D); thus the rad9Δ mutation is epistatic to the rad53 mutation. ssDNA production in rad53 mutants appears to occur by the same mechanism as it does in rad9Δ mutants, because ssDNA production in both rad53Δ and rad9Δ mutants requires an intact RAD24 gene; that is, ssDNA production at the 15- and 30-kb loci is eliminated in rad53Δ rad24Δ double mutants (Figure 3, D and E).
In summary we conclude that Rad53, like Rad9, inhibits Rad24- and Rad17-dependent ssDNA production at the telomeres in Cdc13-defective cells. The slower rate of ssDNA production in cdc13-1 rad53Δ compared to cdc13-1 rad9Δ mutants suggests that Rad9 inhibits ssDNA production in part by activating the downstream kinase Rad53 and in part by a second mechanism independent of Rad53. We favor a model in which the second mechanism involves Rad9 directly inhibiting nuclease activity (see discussion and data in Figure 7). However, we cannot exclude the possibility that the slower rate of ssDNA production in cdc13-1 rad53Δ compared to cdc13-1 rad9Δ mutants may be due to an indirect effect of slow cell cycle progression kinetics; the slower cell cycle progression in rad53 mutants may somehow slow degradation itself. We think this indirect mechanism of cell cycle progression on degradation is unlikely because wild-type and chk1Δ mutants have similar rates of ssDNA accumulation, although they have very different rates of cell cycle progression (see below).
Roles of other checkpoint protein kinases: Chk1 and Dun1 have no effect on ssDNA production near telomeres: One explanation for the more rapid ssDNA production in rad9Δ mutants than in rad53Δ mutants is that Rad9 inhibits ssDNA production through both Rad53 and Chk1, two protein kinases known to act in parallel downstream of Rad9 in cell cycle arrest (see Introduction). And, as previously observed, we find that chk1Δ mutants, like rad53Δ mutants, are only partially defective at the metaphase/anaphase checkpoint in response to cdc13-1-induced damage (Figure 4, B and C; Sanchezet al. 1999), and the delay in chk1Δ mutants is entirely dependent on RAD9 and RAD24 (Figure 4, B and C). Therefore, we tested whether Chk1 regulates ssDNA production near telomeres in Cdc13-defective cells. We found that Chk1 neither contributes to ssDNA production nor inhibits it; chk1Δ mutants and RAD+ CHK+ cells behave nearly identically in ssDNA production at both the 15-kb and 30-kb loci tested (compare Figure 4, D and E, with Figure 4F). Note particularly that the time of onset of ssDNA production at the 15-kb locus, 120 min, is nearly identical in chk1Δ and CHK1 strains. (It is important to note that we do not consider the twofold higher levels of ssDNA in chk1Δ rad9Δ compared to rad9Δ mutants to be significant because of interexperimental variability.) Further evidence that Chk1 plays little role in ssDNA production and that the Rad53-independent role of Rad9 is not via Chk1 is given by the finding that cdc13-1 chk1Δ rad53Δ strains accumulate ssDNA at similar rates to cdc13-1 rad53Δ strains (compare Figure 3, D and E, with Figure 4, D and E).
We also tested whether the Dun1 checkpoint protein kinase that appears to act downstream or interdependently with Rad53 in cell cycle arrest might also have a role in ssDNA production. We confirmed that dun1Δ mutants have a partial arrest defect similar to that of rad53Δ mutants (Gardneret al. 1999; Figure 5, B and C). We found that dun1Δ mutants generate ssDNA virtually identically to chk1Δ and to wild-type cells (Figure 4, D-F).
On this basis we conclude that protein kinase Chk1, which acts downstream of Rad9, and Dun1, which most likely acts downstream of Rad53 in cell cycle arrest, have no role in regulation of ssDNA production in Cdc13-defective cells. This suggests that Rad9 inhibits ssDNA production via both Rad53 and a second pathway that is yet to be defined.
Mec1 inhibits ssDNA production at telomeres: Mec1 is a central checkpoint protein in budding yeast required for virtually all known DNA-damage checkpoint pathways. Mec1 is required for phosphorylation and activation of Rad9 and Rad53, both of which inhibit ssDNA production in Cdc13-defective cells (Emili 1998; Vialardet al. 1998; Pellicioliet al. 1999). Mec1 is also required to phosphorylate Ddc1 (Paciottiet al. 1998), which, with Rad17 and Mec3, is part of the hetero-trimeric checkpoint sliding clamp that contributes to the generation of ssDNA (Figure 2I). Therefore it is possible that Mec1 contributed both to the generation of ssDNA through regulation of the Rad17-Mec3-Ddc1 complex and to the inhibition of ssDNA production through Rad9 and Rad53.
To test for roles of Mec1, we analyzed cell cycle arrest and ssDNA as performed with other checkpoint mutants in this study. We confirm that a mec1Δ mutation like rad9Δ, rad17Δ, rad24Δ, and mec3Δ mutations causes a complete failure of the DNA-damage responsive metaphase/anaphase checkpoint (Figure 6, B and C). On analyzing ssDNA production in mec1Δ mutants, we found greater production of ssDNA in mec1Δ mutants than in wild type at the locus 30 kb from the telomere and found that the accumulation of ssDNA apparently occurs by the same mechanism because it requires an intact RAD24 gene (Figure 6, D and E). This indicates that Mec1, like Rad9 and Rad53, has a prominent role in inhibiting ssDNA production.
In an attempt to then determine if Mec1 acts entirely through Rad9 to inhibit ssDNA production, we made an interesting observation that suggests dual roles for Mec1. We found that cdc13-1 mec1Δ mutants and cdc13-1 mec1Δ rad9Δ mutants accumulate ssDNA at the 30-kb locus at the same time, ∼200 min after release from G1 (Figure 6D). Recall that rad9Δ mutants accumulate ssDNA ∼120 min after release from G1 (Figure 6G), a full 80 min earlier than mec1Δ or mec1Δ rad9Δ mutants do. This shows that the mec1 mutation, in contrast to the rad53Δ mutation, is dominant to the rad9Δ mutation in regulating ssDNA production 30 kb from the telomere. We suggest that this result may be explained if Mec1, but not Rad53, plays some minor role in activating ssDNA production (see discussion and Figure 9).
To summarize, Mec1 and Rad53 have similar roles in inhibiting ssDNA because both single mutants have increased ssDNA production compared to wild-type cells. However, we infer that Mec1 has an additional role in promoting ssDNA production because MEC1+rad9Δ mutants generate ssDNA more quickly than do mec1Δ rad9Δ mutants. Rad53 does not have an additional role in promoting ssDNA production because rad53Δ rad9Δ and RAD53+ rad9Δ mutants have identical rates of ssDNA production. Therefore it appears that Mec1 contributes both to the generation of ssDNA through regulation of the Rad17, Mec3, Ddc1 complex and to the inhibition of ssDNA production through regulation of Rad9 and Rad53.
Mec1 and Rad53, in contrast to Rad9, specifically inhibit Exo1-dependent ssDNA: We asked if there is a relationship between the two Rad9 inhibitory mechanisms, one Rad53 dependent and one Rad53 independent, and Exo1 activity. This is because recently we showed that Rad9 inhibits both Exo1 (a 5′-3′ exonuclease) and an unidentified nuclease (or nucleases) termed “ExoX” (M. Zubko, S. Guillard and D. Lydall, unpublished results). Both Exo1 and ExoX act at telomeres of cdc13 mutants (Maringele and Lydall 2002; M. Zubko, S. Guillard and D. Lydall, unpublished results). We infer the existence of ExoX because cdc13 rad9Δ exo1Δ strains still possess nuclease activity and generate high levels of ssDNA at telomeres (M. Zubko, S. Guillard and D. Lydall, unpublished results). As previously observed, we found that ssDNA levels are high in cdc13 rad9Δ exo1Δ strains, low in cdc13 exo1Δ strains (M. Zubko, S. Guillard and D. Lydall, unpublished results), and low in cdc13 rad53Δ exo1Δ and cdc13 mec1Δ exo1Δ strains (Figure 7, A and B). One explanation for these data is that Rad9 inhibits an Exo1-independent nuclease activity (ExoX) by Mec1- and Rad53-independent mechanisms. That is, in mec1 and rad53 mutants, Rad9 is still present and able to inhibit ExoX. We further infer that normally Mec1, Rad9, and Rad53 work together to regulate Exo1 activity.
ssDNA accumulation and loss of viability: We previously reported that a rad24 mutation rescues the poor viability of rad9Δ mutant cells with defects in Cdc13 (Lydall and Weinert 1995). In that study we found a correlation between amount of ssDNA and cell death in several single and double mutants tested, and we concluded that ssDNA was leading to cell death. We recently found a mutation that separates the presence of ssDNA from cell death; an exo1Δ mutation suppressed cell death in cdc13 rad9Δ cells but ssDNA still accumulated (M. Zubko, S. Guillard and D. Lydall, unpublished results). We suggest that the Exo1 protein is somehow involved in converting ssDNA into some lethal event. Given this role of Exo1, we asked if other checkpoint proteins might also be involved in converting ssDNA to lethal lesions. We measured cell viability in cdc13 checkpoint single mutants and double mutants (e.g., cdc13 mec1Δ and cdc13 mec1Δ rad9Δ) and found three classes of response. First, we found that mec3Δ and rad17Δ mutations, like the rad24Δ mutation reported previously, both suppressed the loss of viability and decreased the amount of ssDNA that accumulates in cdc13-1 rad9Δ mutants. We believe the high viability in cdc13 rad17Δ rad9Δ strains, for example, is because ssDNA does not accumulate (data not shown). Second, we found that dun1Δ and chk1Δ mutations do not affect cell viability of cdc13 rad9Δ mutants, which is not surprising since they do not affect accumulation of ssDNA. Third, we find that rad53Δ and mec1Δ mutations act like an exo1Δ mutation; rad53Δ and mec1Δ mutations increase cell viability of cdc13 rad9Δ mutants yet they do not dramatically decrease the amount of ssDNA that accumulates in cdc13-1 rad9Δ cells, as rad24Δ, mec3Δ, and rad17Δ mutations do (presumably because Exo1 is still active). One simple explanation for our data is that ssDNA is not lethal per se, but is converted to some lethal event by the action of Mec1, Rad53, and Exo1. Recombination does not seem to be the lethal event because deletion of Rad52, which is required for all types of recombination events in budding yeast, does not rescue the loss in viability of cdc13-1 rad9Δ cells (Figure 8G).
The DNA-damage response relies on coordinated interactions between DNA repair and cell cycle arrest responses. DNA repair, like replication, is a highly choreographed yet potentially disruptive series of biochemical events that lead to restoration of an undamaged DNA structure. Checkpoint-dependent cell cycle arrest ensures that repair is completed before critical cell cycle transitions occur. It is thought that DNA repair events regulate the cell cycle response and vice versa, but few details are known about how these two aspects of the DNA-damage response affect each other (Carr 2002; Rouse and Jackson 2002a). Defects in the Cdc13 telomere-binding protein provide an opportunity to examine the interaction between arrest and repair by a single set of checkpoint proteins. Although the biological role for the metabolism of ssDNA by checkpoint proteins acting near unprotected telomeres under study here is still unclear, we think the role of checkpoint proteins in activating nucleolytic degradation, yet limiting production of extensive amounts of ssDNA by inhibition of degradation, probably occurs in other types of damaged DNA as well.
In this article we have examined how a large set of DNA-damage checkpoint protein complexes and pathways interact to regulate the accumulation of ssDNA. Our experiments support a model in which specific checkpoint proteins have one of three roles: promoting ssDNA production at unprotected telomeres of cdc13-1 mutants, activating a cascade to inhibit ssDNA production, or having no effect on ssDNA production (Figure 9). One protein complex, Mec1-Ddc2, appears to be able to both promote and inhibit degradation.
Checkpoint protein complexes required to activate ssDNA production: We have found that components of both the Rad24-Rfc2-5-checkpoint clamp loading complex and the Rad17-checkpont sliding clamp complex are required for production of ssDNA (Figure 2, G-I). Others have shown that Rad24p, in combination with the four small Rfc subunits, loads the Rad17, Mec3, and Ddc1 hetero-trimeric PCNA-type ring onto damaged telomeres (Greenet al. 2000; Meloet al. 2001; Majka and Burgers 2003). The Rad24-Rfc2-5 complex may recruit the Rad17 complex to sites of damage, which in turn may recruit nucleases to the sites of damage. The putative RAD24-dependent nuclease(s) that we term ExoX remains unknown (Figure 9). The nuclease activity of ExoX is either an intrinsic property of checkpoint sliding clamp proteins or a separate nuclease recruited to damaged telomeres by checkpoint sliding clamp proteins. We have shown previously that Exo1 plays a role in generating ssDNA at telomeres of cdc13-1 mutants (Maringele and Lydall 2002) but, unlike Rad24, Exo1 is not required to generate high levels of ssDNA in cdc13-1 rad9Δ mutants (M. Zubko, S. Guillard and D. Lydall, unpublished results). Therefore, ExoX is not Exo1.
The Mec1 protein kinase appears also to contribute to the activity of ExoX because cdc13-1 rad9Δ mec1Δ mutants do not generate ssDNA as rapidly as cdc13-1 rad9Δ MEC1+ mutants (Figure 1, G and H, and Figure 5, D and E). Since Mec1 is required for phosphorylation of Ddc1 after DNA is damaged (Paciottiet al. 1998), phosphorylation of Ddc1 may contribute to the stability of the Rad17/Mec3/Ddc1 complex or to the activity of ExoX (Figure 9).
Checkpoint protein complexes required to inhibit ssDNA production: We have shown previously that Rad9 inhibits ssDNA production near telomeres. We now show that Mec1 and Rad53, as well as Rad9, are all involved in inhibiting ssDNA production at telomeres in Cdc13-defective cells. Others have shown that Mec1-dependent phosphorylation of Rad9 occurs after DNA is damaged (Emili 1998; Vialardet al. 1998) and that phosphorylated Rad9 can bind and activate the Rad53 kinase (Gilbertet al. 2001; Schwartzet al. 2002). Therefore our data suggest that Mec1 may also activate Rad9, which then activates Rad53, which then inhibits ssDNA production (Figure 9A, pathway a). Our experiments further suggest that this Mec1-Rad9-Rad53 pathway specifically inhibits Exo1-dependent accumulation of ssDNA. Rad9 is also able to inhibit ExoX by a Mec1- and Rad53-independent route (Figure 9A, pathway b). The evidence for pathways a and b is that ExoX-dependent ssDNA accumulates in cdc13 exo1Δ rad9Δ mutants (disabled in pathways a and b) but not in cdc13 exo1Δ rad53Δ or cdc13 exo1Δ mec1Δ mutants (disabled in pathway a, but proficient in pathway b). Therefore, inhibitory pathway b is still able to inhibit ExoX in mec1 or rad53 mutants. The existence of Rad9-dependent, but Mec1-independent, responses to DNA damage (Figure 9A, pathway b) is supported by the observation that rad9 mec1 double mutants are more sensitive than corresponding single mutants to DNA damage (Paulovichet al. 1997).
There are several plausible mechanisms by which Rad9 and Rad53 may counteract nuclease activity at telomeres. Rad9 or Rad53 may directly bind to nucleases to inhibit their activity. Rad53 may phosphorylate nucleases or nuclease inhibitors. Rad9 or Rad53 may affect chromatin structure and thereby inhibit nuclease movement through chromatin. For example, Rad53 interacts with the chromatin-remodeling protein Asf1 (Emiliet al. 2001; Huet al. 2001). Finally, Rad9 or Rad53 may activate repair pathways (e.g., repair polymerases) that counteract the activity of nucleases. Future experiments will be necessary to distinguish between these numerous potential mechanisms of nuclease inhibition.
ssDNA and cell death: We previously observed a good correlation between the rapid accumulation of ssDNA at telomeres and rapid loss of viability strains carrying combinations of cdc13-1, rad9Δ, and rad24Δ mutations (Lydall and Weinert 1995). However, in this article and elsewhere we have found that mec1Δ, rad53Δ, and exo1Δ mutations each suppress the rapid loss in viability of cdc13-1 rad9Δ mutants but ssDNA still accumulates (this work and M. Zubko, S. Guillard and D. Lydall, unpublished results). One plausible explanation for this is that ssDNA is cytotoxic only if Mec1, Rad53, and Exo1 convert it into a lethal lesion, for example, to a lesion that is not easily repaired (Figure 9B). Consistent with this idea are numerous examples of DNA repair pathways converting nonlethal lesions to lethal lesions in cells. For example, homologous recombination pathways convert ultraviolet (UV)-induced DNA damage to lethal lesions in helicase-defective srs2 mutants or spontaneous lesions to lethal lesions in srs2 sgs1 double mutants (Gangloffet al. 2000). Similarly, Sae2 converts alkylating agent methyl methanesulfonate (MMS)-induced lesions to lethal lesions in mec1 mutants (Usuiet al. 2001). Unlike UV, MMS, or spontaneous DNA damage, cdc13-induced damage is located at telomeres and is highly specific. Therefore, in comparison with these other types of damage it may be comparatively simple to understand the mechanisms responsible for the death of cdc13 rad9Δ cells.
In summary we have shown that orchestrated interactions between checkpoint pathways and damaged telomeres regulate cell cycle arrest, ssDNA production, and cell viability of cdc13 mutants. Further understanding of these interactions will be critical for understanding the DNA-damage response. Future studies should define the mechanisms by which Rad9 and Rad53 inhibit Exo1 activity and how other nuclease(s) are regulated by the checkpoint sliding clamp. It seems likely that similar mechanisms will regulate the DNA-damage response at other types of DNA damage and in other eukaryotic cells.
We thank all members of our labs for input and comments on the manuscript. We thank M. P. Longhese for providing strains. X.J. was partially supported by the award of an Overseas Research Scholarship. D.L. is a Wellcome Senior Research Fellow in Basic Biomedical Science.
Communicating editor: A. Nicolas
- Received July 31, 2003.
- Accepted November 7, 2003.
- Copyright © 2004 by the Genetics Society of America