|Home | About | Journals | Submit | Contact Us | Français|
The Mec1 and Rad53 protein kinases are essential for budding yeast cell viability and are also required to activate the S-phase checkpoint, which supports DNA replication under stress conditions. Whether these two functions are related to each other remains to be determined, and the nature of the replication stress-dependent lethality of mec1 and rad53 mutants is still unclear. We show here that a decrease in cyclin-dependent kinase 1 (Cdk1) activity alleviates the lethal effects of mec1 and rad53 mutations both in the absence and in the presence of replication stress, indicating that the execution of a certain Cdk1-mediated event(s) is detrimental in the absence of Mec1 and Rad53. This lethality involves Cdk1 functions in both G1 and mitosis. In fact, delaying either the G1/S transition or spindle elongation in mec1 and rad53 mutants allows their survival both after exposure to hydroxyurea and under unperturbed conditions. Altogether, our studies indicate that inappropriate entry into S phase and segregation of incompletely replicated chromosomes contribute to cell death when the S-phase checkpoint is not functional. Moreover, these findings suggest that the essential function of Mec1 and Rad53 is not necessarily separated from the function of these kinases in supporting DNA synthesis under stress conditions.
The integrity of the genome is constantly challenged by DNA damage caused by environmental and intracellular factors. Aberrant DNA replication is a major source of mutations and chromosome rearrangements that can lead to cancer and other diseases in metazoans (reviewed in reference 23). Replication fork progression can be hampered by exogenous or endogenous DNA damage. Furthermore, faithful replication depends on a balanced supply of deoxyribonucleotides (deoxyribonucleoside triphosphates [dNTPs]), whose levels are maintained during S-phase through the action of the ribonucleotide reductase (RNR) activity that converts the ribonucleotides to dNTPs (reviewed in reference 37). Indeed, replication fork pausing can be experimentally induced by genotoxic drugs, such as hydroxyurea (HU), which reduces dNTP pools by inhibiting RNR activity, and the DNA alkylating agent methyl methanesulfonate (MMS) that causes intra-S damage.
Eukaryotic cells respond to replication interference through a complex signal-transduction pathway, known as the S-phase checkpoint, whose key players in the budding yeast Saccharomyces cerevisiae are the Mec1 and Rad53 kinases (reviewed in references 5 and 63). Mec1, together with its interacting protein Ddc2, is recruited to stalled forks, where it activates the effector kinase Rad53. Both kinases act in various ways to respond to replication interference. They are needed to complete DNA replication after exposure to HU or MMS (16, 55) by maintaining the integrity and/or activity of the replication forks (11, 15, 26, 34). Furthermore, they stimulate dNTP production (1, 25, 64, 65) and the transcription of several MCB binding factor (MBF)-regulated genes that are involved in DNA replication (2, 58). Finally, they are required for inhibition of late replication origin firing (45, 49) and for preventing accumulation of aberrant DNA structures, such as reversed forks or excessive single-stranded DNA (ssDNA) (20, 33, 50). Despite their inability to replicate DNA, HU-treated mec1 and rad53 mutant cells proceed to elongate the mitotic spindle and to partition unreplicated or partially replicated DNA (16, 62). This premature chromosome segregation can be the cause of the extensive chromosomal fragmentation that is observed in mammalian cells lacking the Mec1 ortholog ATR (6, 7, 14), indicating that the S-phase checkpoint ensures that DNA replication is complete before cells divide not only in yeast but also in mammals.
Rad53 and Mec1 kinases are essential for cell viability, but cells lacking either Mec1 or Rad53 can be kept alive by overexpression of the RNR genes (16) or by the lack of either the Rnr1 inhibitor Sml1 (64) or the transcriptional repressor of the RNR genes Crt1 (25). Because dNTP pools are limiting even during a normal S phase (40), these findings suggest that the essential function of Mec1 and Rad53 is to provide cells with sufficient dNTP levels to support DNA replication. This checkpoint-mediated regulation of dNTP pools is thought to be distinct from the checkpoint-mediated regulation of S-phase progression under replication stress, because mec1Δ sml1Δ and rad53Δ sml1Δ cells lacking the Sml1-mediated inhibition of RNR activity are still extremely sensitive to agents that cause replication stress, such as HU and MMS.
Given the essential function of these checkpoint kinases in mediating the response to replication stress, a fundamental question to be addressed is which a process(es) regulated by the checkpoint is critical for the maintenance of cell viability. A hypomorphic mec1 mutant (mec1-100) (38), which does not block late origin firing in HU but is much less HU sensitive than mec1Δ sml1Δ cells, argues that regulation of late origin firing plays a relatively minor role in maintaining cell viability after exposure to replication stress (56). Cells lacking Mec1 that are kept viable by SML1 deletion have been shown to accumulate chromosome breakages during HU treatment as a consequence of not fully replicated chromosomes being under persistent tension exerted by the mitotic spindle (19). However, inhibiting spindle formation via nocodazole treatment does not improve viability of mec1Δ sml1Δ and rad53Δ sml1Δ cells during exposure to HU (16, 55), suggesting that precocious chromosome segregation per se is not the reason for the loss of viability of HU-treated rad53 and mec1 mutants. This finding has lead to the proposal that the DNA replication defects arising when mec1 and rad53 mutants experience replication impediments irreversibly commit cells to death during S phase.
To further investigate the role of Mec1 and Rad53 in maintaining cell viability in the presence of replication stress, we searched for extragenic mutations suppressing the hypersensitivity to HU of mec1Δ sml1Δ cells. By characterizing one of the identified suppressor mutations, we provide evidence that decreased activity of the cyclin-dependent kinase (Cdk1 or Cdc28 in yeast) complex suppresses mec1 and rad53 cell lethality not only during exposure to replication stress but also during an unchallenged S-phase. Delaying either the G1/S transition or spindle elongation improves viability of HU-treated mec1 and rad53 mutants and bypasses the Mec1 and Rad53 essential function during an unperturbed S phase. Further investigation of the suppression mechanism suggests that cell death caused by the lack of the S-phase checkpoint may be a consequence of Cdk1 activity forcing unscheduled events, such as the G1/S transition and spindle elongation.
We searched for spontaneous extragenic mutations suppressing the HU sensitivity of mec1Δ sml1Δ cells. Since 5 mM HU was the minimal HU dose impairing the ability of mec1Δ sml1Δ cells to form colonies, we plated mec1Δ sml1Δ (YLL490) cells on yeast extract-peptone-dextrose (YEPD) plates containing 5 mM HU and searched for clones able to form colonies. This analysis allowed us to identify 20 independent clones able to grow on 5 mM HU. By crossing these clones with a MEC1 sml1Δ strain, we found that the suppressor phenotype for two of them was due to a single-gene recessive mutation. One of these two clones was also temperature sensitive for growth, and this phenotype segregated tightly linked to the suppressor phenotype. We cloned the corresponding gene by transforming the original mutant clone with a yeast genomic DNA library constructed in a LEU2 centromeric plasmid and searching for recombinant plasmids able to inhibit the mutant ability to form colonies on 5 mM HU. Analysis of several positive transformant clones revealed that the minimal complementing region was restricted to a DNA fragment containing the CDC28 gene. Further genetic analysis allowed us to demonstrate that CDC28 was indeed the gene identified by the suppressor mutation.
All yeast strains (see Table S1 in the supplemental material) were derivatives of W303 (ade2-1, trp1-1, leu2-3,112, his3-11,15, ura3, and rad5-535). Gene deletions were generated by one-step gene replacement. The cdc28-as1 mutant, kindly provided by R. Kolodner (San Diego, CA), was backcrossed three times with W303. The cdc28-1N mutant and the strain expressing green fluorescent protein (GFP)-tagged Tub1 were kindly provided by Simonetta Piatti (Montpellier, France). The strain expressing both the nucleoside transporter hENT and the herpes simplex virus thymidine kinase used for bromodeoxyuridine (BrdU) incorporation was kindly provided by J. Diffley (South Mimms, United Kingdom). All of the strains expressing the RAD52-YFP fusion were derivatives of strain W3749/4C, kindly provided by R. Rothstein (New York, NY). Cells were grown in either synthetic minimal medium supplemented with the appropriate nutrients or YEP (1% yeast extract, 2% Bacto peptone, 50 mg of adenine/liter) medium supplemented with 2% glucose (YEPD), 2% raffinose (YEPR), or 2% raffinose and 2% galactose (YEPRG). Benomyl and nocodazole were used at 10 and 5 μg/ml, respectively, in 1% dimethyl sulfoxide. Unless otherwise indicated, the experiments were performed at 25°C.
To visualize the mitotic spindle, cells expressing TUB1-GFP were fixed in 100% ethanol at the time points of interest and then washed in 10 mM Tris (pH 8.0) pending microscopic analysis. The GFP fluorophore was visualized using a band-pass GFP filter. To visualize Rad52-YFP foci, cells expressing RAD52-YFP were grown in synthetic medium supplemented with adenine to minimize autofluorescence. The cells were washed in 0.1 M potassium phosphate buffer at the time points of interest and analyzed immediately at the microscope. Cells were imaged on concanavalin A-coated slides. Microscopy was performed on a Leica TCS resonant STED DMI6000 CS microscope equipped with a multiline argon ion laser. Images of the YFP-stained yeast cells were acquired by collecting between 530 and 600 nm the fluorescence excited by the 27-μW output of the 514-nm line of the argon laser. Both the emission and the transmitted light images have been recorded at 400-Hz scan speed through a 100× HCX PL APO oil objective (numerical aperture = 1.4) after identification of the cellular focal plane by 1-μm step z-scan measurements. Microscopy images were analyzed by using ImageJ.
Nuclear division was scored with a fluorescence microscope on cells stained with propidium iodide. Flow cytometric DNA analysis was determined on a Becton Dickinson FACScan. The pulse-chase BrdU experiment and immunodetection of BrdU-labeled DNA were performed as described previously (48, 61). For spot assays, exponentially growing overnight cultures were counted, and 10-fold serial dilutions of equivalent cell numbers were spotted onto plates containing the indicated media. Experiments involving G1 synchronization were carried out by incubating exponentially growing cells in appropriate media containing 5 μg of α-factor/ml at 25°C for 2 h.
Budding yeast cells lacking Mec1 or Rad53 and kept viable by SML1 deletion (mec1Δ sml1Δ or rad53Δ sml1Δ) die even when exposed to very low HU doses (Fig. 1A and andB).B). To understand the nature of this lethality, we searched for spontaneous mutations that confer increased HU resistance to mec1Δ sml1Δ cells (see Materials and Methods). Given the extremely high HU sensitivity of mec1 mutants, this screening was performed using 5 mM HU, which is the minimal HU dose impairing the ability of mec1Δ sml1Δ cells to form colonies. One of the suppressors turned out to be a mutation in the CDC28 gene (see Materials and Methods), which encodes for the catalytic subunit Cdc28/Cdk1 of cyclin-dependent kinase. This mutation (cdc28-sup) improved viability of mec1Δ sml1Δ cells in the presence of either MMS or low HU doses (Fig. 1A). Suppression was not restricted to the MEC1 deletion, since cdc28-sup also decreased the HU and MMS sensitivity of ddc2Δ sml1Δ cells (Fig. 1A). Cells carrying the cdc28-sup mutation were also temperature sensitive for growth, but they did not show a uniform terminal phenotype when shifted to 37°C (data not shown).
Mec1 might function in supporting cell viability in the presence of HU because it is required to activate the downstream kinase Rad53. However, mec1Δ sml1Δ cells are considerably more sensitive to HU and other DNA-damaging agents than are rad53Δ sml1Δ cells (16), suggesting that Mec1 and Rad53 might have different roles during DNA replication under stress conditions. Thus, we sought to determine whether the cdc28-sup mutation could suppress also the HU sensitivity of rad53Δ sml1Δ or rad53-K227A mutant cells, the latter expressing a Rad53 mutant variant with reduced kinase activity that supports cell viability even in the presence of Sml1 (53). The ability of rad53Δ sml1Δ cdc28-sup and rad53-K227A cdc28-sup cells to form colonies in the presence of HU was higher than that of rad53Δ sml1Δ and rad53-K227A cells, respectively, indicating that cdc28-sup can suppress also the HU sensitivity caused by Rad53 dysfunction (Fig. 1B). We were unable to determine whether cdc28-sup suppressed the MMS sensitivity of rad53 mutants because the cdc28-sup mutation by itself caused cell lethality in the presence of the amount of MMS that was required to impair viability of rad53 mutants (Fig. 1B).
To assess whether the Cdk1-dependent lethality in HU of mec1 and rad53 cells could be attributed to Cdk1 kinase activity, we constructed mec1 and rad53 mutants expressing the cdc28-as1 allele, which encodes for a kinase with an enlarged ATP-binding pocket, allowing it to bind the nonhydrolyzable ATP analogue 1-NM-PP1. Treatment of cdc28-as1 cells with 1-NM-PP1 results in rapid downregulation of Cdc28 kinase activity, but Cdc28-as1 kinase activity is reduced by ~20% compared to wild type even in the absence of 1-NM-PP1 (4). We found that survival to HU of mec1Δ sml1Δ cdc28-as1 cells was higher than that of mec1Δ sml1Δ cells even in the absence of 1-NM-PP1 (Fig. 1C). Furthermore, the cdc28-as1 mutation diminished the HU sensitivity of both rad53Δ sml1Δ (Fig. 1D) and rad53-K227A cells (Fig. 1E). Therefore, a reduced Cdk1 activity counteracts cell death in mec1 and rad53 mutants exposed to replication impediments. The HU sensitivity of mec1Δ rad53Δ sml1Δ cells was similar to that of mec1Δ sml1Δ cells (Fig. 1C), and cdc28-as1 improved survival in response to HU treatment of mec1Δ sml1Δ and mec1Δ rad53Δ sml1Δ cells to the same extent (Fig. 1C). Thus, suppression in mec1Δ sml1Δ mutant does not appear to require the activity of Rad53.
Interestingly, although mec1Δ sml1Δ cells were more sensitive to MMS than rad53Δ sml1Δ cells (compare Fig. 1C and andD),D), the cdc28-as1 mutation suppressed the sensitivity to MMS of mec1Δ sml1Δ more efficiently to that of both rad53Δ sml1Δ and rad53-K227A cells (Fig. 1C to toE).E). These findings suggest that the causes of death in MMS-treated mec1 and rad53 mutants are different, supporting previous data showing that Mec1 and Rad53 play different roles in the response to MMS treatment (48). Although the MMS sensitivity of mec1Δ rad53Δ sml1Δ cells was similar to that of mec1Δ sml1Δ cells (Fig. 1C), cdc28-as1 improved survival in response to MMS treatment of mec1Δ sml1Δ cells more efficiently than that of mec1Δ rad53Δ sml1Δ cells (Fig. 1C and andD),D), indicating that Rad53 contributes to the cdc28-as1-mediated suppression of the MMS sensitivity caused by the lack of Mec1.
Rad53 and Mec1 are essential for cell viability, prompting us to ask whether a reduced Cdk1 activity could also bypass the essential function of these checkpoint kinases. Diploid strains heterozygous for cdc28-as1 and either mec1Δ or rad53Δ alleles were generated and spore viability was monitored after tetrad dissection. Since the MEC1 and CDC28 genes are linked to each other on chromosome II, mec1Δ and cdc28-as1 alleles were expected to cosegregate in most tetrads from the MEC1 CDC28/mec1Δ cdc28-as1 diploid. As expected, mec1Δ and rad53Δ spores failed to form colonies, whereas all of the mec1Δ cdc28-as1 (Fig. 1F) and rad53Δ cdc28-as1 (Fig. 1G) double-mutant spores formed colonies of almost wild-type size, indicating that a reduced Cdk1 activity rescues the lethality caused by the lack of either Mec1 or Rad53. Collectively, these data indicate that carrying out a certain Cdk1-dependent event(s) causes cell death in the absence of Mec1 or Rad53 independently of exogenous DNA replication stress.
To understand the molecular mechanisms underlying the suppression described above, we analyzed the effects of the cdc28-as1 mutation on the kinetics of DNA replication and cell cycle progression of mec1Δ sml1Δ and rad53Δ sml1Δ cells exposed to a low HU dose. Cell cultures were blocked in G1 with α-factor and released from the G1 arrest either in the absence or in the presence of 20 mM HU (Fig. 2). As expected (16), mec1Δ sml1Δ and rad53Δ sml1Δ cells released in the presence of HU were unable to complete DNA replication even at 360 min after α-factor release (Fig. 2A). However, most cells partitioned incompletely replicated DNA, undergoing nuclear division at 90 min after the release (Fig. 2C). Consistent with the requirement of Cdk1 to perform the G1/S and G2/M transitions, the presence of cdc28-as1 delayed bud formation (Fig. 2B) and nuclear division (Fig. 2C) of wild-type, mec1Δ sml1Δ, and rad53Δ sml1Δ cells released in the presence of HU. Then, HU-treated mec1Δ sml1Δ cdc28-as1 and rad53Δ sml1Δ cdc28-as1 cells exited from mitosis, divided, and initiated a new cell cycle, whereas similarly treated mec1Δ sml1Δ and rad53Δ sml1Δ cells arrested as budded cells with two nuclei, as expected (Fig. 2A to toC).C). Accordingly, the survival rates in HU of mec1Δ sml1Δ cdc28-as1 and rad53Δ sml1Δ cdc28-as1 were higher than those of mec1Δ sml1Δ and rad53Δ sml1Δ cells throughout the experiment (Fig. 2D). Thus, reducing Cdk1 activity in HU-treated mec1Δ sml1Δ and rad53Δ sml1Δ cells restores cell cycle progression and cell viability.
Interestingly, mec1Δ sml1Δ cdc28-as1 and rad53Δ sml1Δ cdc28-as1 cells seemed to have completed the bulk DNA synthesis at the time of nuclear division (Fig. 2A and andC),C), suggesting that lowering the Cdk1 activity suppresses the DNA replication defects of mec1Δ sml1Δ and rad53Δ sml1Δ cells. To follow more directly the DNA replication kinetics, we performed BrdU pulse-chase experiments. Cells were synchronized in G1 with α-factor and released into medium containing 20 mM HU and BrdU for 15 min (Fig. 2E) to label the nascent DNA. The BrdU was then chased by transferring cells to medium containing thymidine at a high concentration and 20 mM HU (Fig. 2E). Labeled nascent DNA replication intermediates, which appeared as a smear in all of the strains after 15 min in HU plus BrdU, rapidly increased in size after the chase in wild-type cells (Fig. 2F). Consistent with a failure of mec1 and rad53 mutants to complete DNA replication, the formation of high-molecular-weight molecules of nascent DNA was delayed in mec1Δ sml1Δ and rad53Δ sml1Δ cells. Strikingly, almost all of the incorporated BrdU in mec1Δ sml1Δ cdc28-as1 and rad53Δ sml1Δ cdc28-as1 cells was present in the high-molecular-weight fraction by 90 to 120 min after α-factor release (Fig. 2F). Altogether, these data suggest that reducing Cdk1 activity increases survival of HU-treated mec1 and rad53 mutants by suppressing their DNA replication defects.
Regulation of Cdk1 activity during the cell cycle depends on the interaction of the Cdk1 kinase with different cyclin subunits. In budding yeast, the partially redundant G1 cyclins Cln1 to Cln3 are required to perform the G1/S transition (43). Interestingly, deletion of both CLN1 and CLN2 was shown to partially bypass the essential requirement of Mec1, whereas overexpression of CLN1 or CLN2 (but not of CLN3) exacerbated the growth defects of mec1 mutants in the absence of genotoxic agents (59). We therefore sought to determine whether the Cdk1-dependent lethality of HU-treated mec1 and rad53 cells could be attributed to a function of Cdk1 in G1. Deletion of CLN2 improved the viability of mec1Δ sml1Δ, rad53Δ sml1Δ, and rad53-K227A mutant cells exposed to moderate HU doses (Fig. 3A), whereas CLN1 deletion did not (data not shown). Similar to what we observed for the cdc28-as1 allele, the lack of Cln2 suppressed the MMS sensitivity of mec1Δ sml1Δ cells, but only a very weak (if any) suppressor effect was detectable in rad53Δ sml1Δ and rad53-K227A cells (Fig. 3A).
We also found that the lack of Cln2 bypasses the essential function of Mec1 and Rad53. In fact, tetrad dissection of diploid strains heterozygous for cln2Δ and mec1Δ alleles showed that all of the mec1Δ cln2Δ double-mutant spores formed colonies, although with a smaller size than wild-type or cln2Δ spores (Fig. 3B). Similar results were obtained by combining the cln2Δ and rad53Δ alleles (data not shown).
The function of Cln1 and Cln2 in G1 is to promote entry into S-phase by activating the S-phase Clb5-6/Cdk1 complexes, which in turn trigger replication origin firing. This Cln/Cdk1 function is accomplished through phosphorylation and degradation of the specific Clb/Cdk1 inhibitor Sic1 (47, 60). At the G1/S transition, the S-phase checkpoint induces transcription of the RNR genes and of several MBF-regulated genes through the inhibition of transcriptional repressors Crt1 (25) and Nrm1 (2, 58), respectively. Whether this checkpoint-mediated transcriptional program helps cell survival in the presence of replication stress is not known. Because CLN2 deletion causes a G1/S transition delay (17), one possibility is that this delay provides mec1 and rad53 mutant cells with a longer time to amend the amount of dNTPs and/or proteins that are required to support DNA replication under stress conditions. If this were the case, the cln2Δ-mediated suppression should be overcome by abrogating the G1/S delay. Indeed, deletion of SIC1, which is known to abolish the delay in S-phase entry caused by the absence of Cln2 (17), counteracted the ability of cln2Δ to suppress the HU and MMS sensitivity of mec1Δ sml1Δ cells. In fact, loss of viability of mec1Δ sml1Δ cln2Δ sic1Δ mutant cells after plating on HU- and MMS-containing media was similar to that of mec1Δ sml1Δ sic1Δ and mec1Δ sml1Δ cells (Fig. 3C). This finding suggests that CLN2 deletion can suppress the sensitivity to replication stress of mec1 and rad53 mutant cells by delaying entry into S phase.
The data presented above imply that the role of Mec1 and Rad53 in inducing transcription is important to maintain cell survival in the presence of replication stress. To further assess this possibility, we sought to determine whether the lack of the transcriptional repressor Crt1 or Nrm1 improves the survival in response to HU and/or MMS of mec1Δ sml1Δ and rad53Δ sml1Δ cells. Indeed, the lack of Crt1, which was previously shown to bypass the essential function of Mec1 (25), decreased the HU and MMS sensitivity of mec1Δ sml1Δ cells and the HU sensitivity of rad53Δ sml1Δ cells (Fig. 3D), whereas the lack of Nrm1 did not (Fig. 3E). Thus, checkpoint-mediated inhibition of Crt1 and subsequent transcriptional activation of the RNR genes help cells to survive in the presence of replication stress.
In S. cerevisiae, the two B-type cyclins—Clb5 and Clb6—stimulate the function of Cdk1 in promoting replication origin firing, with Clb5 playing the major role (18, 28, 46). Decreasing Clb5/Clb6-Cdk1 activity might improve the viability of HU- and/or MMS-treated mec1 and rad53 mutants by lowering the total number of replication forks and therefore the chance to block them. This does not seem to be the case, since neither CLB5 deletion nor CLB6 deletion suppressed the sensitivity to HU or MMS of mec1 and rad53 cells (Fig. 4A). On the contrary, mec1Δ sml1Δ clb5Δ and mec1Δ sml1Δ clb6Δ cells displayed enhanced sensitivity to HU and MMS compared to each single mutant (Fig. 4A). Moreover, consistent with previous data (21), mec1Δ sml1Δ clb5Δ cells grew less efficiently than mec1Δ sml1Δ cells even in the absence of HU or MMS (Fig. 4A), and the deletion of CLB5 was lethal in both rad53Δ sml1Δ and rad53-K227A mutant cells (Fig. 4B).
Loss of viability of rad53 clb5Δ and mec1 clb5Δ double-mutant cells even in the absence of the Sml1-mediated inhibition of RNR activity might be due to a diminished number of fired replication origins and therefore of active replication forks that can complete DNA replication before cells divide. Since origin firing throughout S-phase requires also the Cdc7-Dbf4 (DDK) complex (reviewed in reference 29), we analyzed the consequences of introducing the cdc7-4 allele in mec1Δ sml1Δ and rad53Δ sml1Δ mutants. No viable rad53Δ sml1Δ cdc7-4 and mec1Δ sml1Δ cdc7-4 spores could be found after tetrad dissection of homozygous sml1Δ diploid strains, which were heterozygous for cdc7-4 and either mec1Δ or rad53Δ, whereas each kind of single mutant spores was obtained at normal frequency (Fig. 4C).
Thus, a decreased S-phase function of Cdk1 cannot account for the suppressor effect on the HU sensitivity of mec1 and rad53 mutants observed by reducing Cdk1 activity. Instead, disabling S-phase Cdk1 complexes impairs viability of these mutants even in the absence of exogenous replication impediments, possibly because it diminishes the number of active replication forks and therefore the chance to complete DNA replication.
To investigate whether the Cdk1-dependent lethality of mec1 and rad53 cells could implicate also some mitotic Cdk1 functions, we sought to determine whether the cdc28-1N allele, which is specifically defective in the interaction with the mitotic cyclins (54), could suppress the HU and MMS sensitivity of mec1 and rad53 mutants. Furthermore, because Cdk1 association with the mitotic cyclins Clb1 to Clb4 drives spindle assembly and progression to metaphase (41), we also analyzed the consequences of deleting the mitotic cyclin gene CLB2. The cdc28-1N allele increased survival to HU of mec1Δ sml1Δ and rad53Δ sml1Δ cells (Fig. 5A). Furthermore, viability on HU of mec1Δ sml1Δ clb2Δ and rad53Δ sml1Δ clb2Δ cells was higher than that of mec1Δ sml1Δ and rad53Δ sml1Δ cells, respectively (Fig. 5B). Thus, a mitotic function of Cdk1 contributes to impair viability of HU-treated mec1 and rad53 mutants.
Although mec1Δ sml1Δ and rad53Δ sml1Δ cells are unable to complete DNA replication in the presence of HU, they prematurely elongate the spindles and partition unreplicated or partially replicated chromosomes (16, 62). Since Cdk1 activity is required to assemble a mitotic spindle (41), decreased Cdk1 activity might suppress the sensitivity of mec1 and rad53 mutants to replication stress by providing the cells with a longer time to finish replicating the bulk of their chromosomes before they segregate. If this were the case, cell death in these mutants should be suppressed by disrupting kinetochore-microtubule attachment via nocodazole or benomyl treatment. Indeed, benomyl addition to HU-containing medium improved the viability of mec1Δ sml1Δ and rad53Δ sml1Δ cells (Fig. 5C). Furthermore, the survival rates of G1-arrested mec1Δ sml1Δ and rad53Δ sml1Δ cells after release into YEPD medium containing 20 mM HU and nocodazole were higher than those of the same cells released into YEPD containing only 20 mM HU (Fig. 5D). Interestingly, nocodazole addition suppressed the HU sensitivity of mec1Δ sml1Δ cells more efficiently than that of rad53Δ sml1Δ cells (Fig. 5D), suggesting that the DNA replication defects in HU-treated mec1 and rad53 mutants do not completely overlap.
Premature spindle elongation and DNA partitioning in HU-treated mec1 and rad53 mutants is due to upregulation of the major kinesin motor protein Cin8 (27), which is known to contribute to the formation, stability and extension of the spindles (24, 44, 52). Since Cdk1 activity drives the assembly of mitotic spindles by restraining Cin8 proteolysis (13), delaying spindle extension by deleting CIN8 might be expected to improve the survival of mec1 and rad53 mutants to HU and/or MMS treatment. Indeed, the viability of mec1Δ sml1Δ cin8Δ, rad53Δ sml1Δ cin8Δ, and rad53-K227A cin8Δ cells in the presence of HU was considerably higher than that of mec1Δ sml1Δ, rad53Δ sml1Δ, and rad53-K227A cells, respectively (Fig. 6A), suggesting that premature spindle elongation contributes to kill HU-treated mec1 and rad53 mutants. Similar to cdc28-as1 and cln2Δ, cin8Δ efficiently suppressed the sensitivity to MMS of mec1Δ sml1Δ cells but suppressed very poorly that of rad53Δ sml1Δ cells (Fig. 6A). Moreover, inactivation of both Cin8 and Cln2 suppressed the sensitivity to HU and MMS of mec1Δ sml1Δ cells more efficiently than did the single Cin8 or Cln2 inactivation (Fig. 6B), in agreement with the notion that both the G1/S and the G2/M functions of Cdk1 are involved in determining this sensitivity.
To characterize the cin8Δ-mediated suppression, cells were arrested in G1 with α-factor and released in YEPD in the presence or absence of 20 mM HU. As expected, although HU-treated mec1Δ sml1Δ and rad53Δ sml1Δ mutants failed to complete DNA replication (Fig. 7A), a significant proportion of them elongated the spindles (Fig. 7C), divided the nuclei (Fig. 7D), and then arrested as binucleate cells with elongated spindles. Deletion of CIN8 did not affect bud emergence (Fig. 7B) but slowed down spindle elongation and nuclear division in both mec1Δ sml1Δ and rad53Δ sml1Δ cells (Fig. 7C and andD).D). Moreover, the survival rates of HU-treated mec1Δ sml1Δ cin8Δ and rad53Δ sml1Δ cin8Δ were higher than those of mec1Δ sml1Δ and rad53Δ sml1Δ cells throughout the experiment (Fig. 7E). Consistent with the finding that CIN8 deletion (as well as nocodazole addition) suppresses the HU sensitivity of mec1Δ sml1Δ more efficiently than that of rad53Δ sml1Δ cells (Fig. 7E), most HU-treated mec1Δ sml1Δ cin8Δ cells exited from mitosis and resumed cell cycle progression within 210 min after release from G1 arrest, whereas most rad53Δ sml1Δ cin8Δ cells were still arrested after 360 min (Fig. 7A to toDD).
Interestingly, while spindle elongation and nuclear division in HU-treated mec1Δ sml1Δ and rad53Δ sml1Δ cells took place when DNA was not fully replicated, the bulk of DNA synthesis seemed to be completed at the time of spindle elongation/nuclear division in mec1Δ sml1Δ cin8Δ and rad53Δ sml1Δ cin8Δ cells under the same conditions (Fig. 7A to toD).D). This indication was confirmed for mec1Δ sml1Δ cin8Δ cells by BrdU pulse-chase experiments during HU exposure. The cells were synchronized in G1 with α-factor and released into medium containing 20 mM HU and BrdU for 15 min (Fig. 7F) to label the nascent DNA. The BrdU was then chased by transferring cells to medium containing thymidine at a high concentration and 20 mM HU (Fig. 7F). As expected, the formation of high-molecular-weight molecules of nascent DNA was delayed in mec1Δ sml1Δ cells compared to wild-type cells (Fig. 7G), whereas mec1Δ sml1Δ cin8Δ cells contained almost all of the incorporated BrdU into the high-molecular-weight fraction by 90 to 120 min after α-factor release (Fig. 7G). Collectively, these data indicate that delaying spindle elongation suppresses the DNA replication defects of mec1 and rad53 mutants exposed to a mild HU dose, possibly by providing additional time for completing DNA synthesis before spindle elongation and chromosome segregation take place.
Cells lacking Mec1 or Rad53, but carrying wild-type SML1, die even in the absence of exogenous replication stress because they replicate their DNA with suboptimal dNTP levels (64). As HU depletes the dNTP pools, the cause of death in mec1Δ SML1 and rad53Δ SML1 cells might mimic that of mec1Δ sml1Δ and rad53Δ sml1Δ cells replicating their DNA in the presence of HU. Thus, we sought to determine whether delaying nuclear division by CLB2 or CIN8 deletion could bypass the essential function of Mec1 and Rad53. Tetrad dissection of diploid strains heterozygous for cin8Δ and either mec1Δ or rad53Δ alleles showed that most mec1Δ cin8Δ and rad53Δ cin8Δ double-mutant spores formed colonies, although with a smaller size than wild-type or cin8Δ spores (Fig. 8A). Similar results were obtained by combining the mec1Δ or rad53Δ mutation with the clb2Δ mutation (Fig. 8B). Thus, delaying spindle elongation not only suppresses the sensitivity of mec1 and rad53 to replication stress but also bypasses their essential function, suggesting that loss of viability in mec1Δ SML1 and rad53Δ SML1 cells is at least partially due to segregation of incompletely replicated chromosomes.
Segregation of partially replicated chromosomes may contribute to cell death because it causes lethal DNA breaks. Since DNA breaks are subjected to the action of the recombination protein Rad52 (30), we investigated whether conditional inactivation of Mec1 in the presence of wild-type SML1 leads to the formation of Rad52 foci. We failed to efficiently deplete Mec1 by using repressible promoters or inducible degradation systems, and therefore we conditionally inactivated Mec1 by using the temperature-sensitive mec1-14 allele that we identified previously (32). Consistent with our finding that the cdc28-as1 and clb2Δ alleles bypass the essential function of Mec1, both the cdc28-as1 and clb2Δ mutations improved the viability of mec1-14 cells at 34°C (Fig. 8C). When wild-type and mec1-14 cell cultures were arrested in G1 with α-factor at 25°C and released into the cell cycle at 37°C, both cell types showed similar kinetics of bud emergence (Fig. 8D). However, mec1-14 cells delayed completion of DNA replication compared to wild-type cells, indicating a DNA replication defect (Fig. 8E). Moreover, mec1-14 cells dramatically accumulated Rad52 foci about 90 min after α-factor release at 37°C, concomitantly with nuclear division (Fig. 8F and andG),G), suggesting that lethal chromosome breaks occur when mec1-14 cells enter mitosis. These Rad52 foci could arise as a consequence of the action of endonucleases, which process the replication intermediates that persist until mitosis in mec1-14 cells. Alternatively, segregation of incompletely replicated chromosomes can be the cause of Rad52 focus formation in Mec1-deficient cells. Since the endonucleases Mms4 and Yen1, which are known to resolve recombination intermediates, are activated at the G2/M transition (35), we analyzed their contribution in the generation of Rad52 foci in mec1-14 cells. We found that the lack of Mms4, Yen1, or both did not impair Rad52 focus formation in mec1-14 cells (Fig. 8H). Rather, the lack of both Mms4 and Yen1 caused an increase of Rad52 foci in mec1-14 cells even at 25°C (time zero in Fig. 8H), possibly because these two nucleases are involved in repairing the double-strand breaks that arise in mec1-14 cells. Interestingly, when mec1-14 cells were released from a G1 block at 37°C in the presence of nocodazole, Rad52 focus formation was greatly reduced (Fig. 8G), supporting the hypothesis that premature chromosome segregation can be the cause of DNA break formation in Mec1-deficient cells.
Why cells deficient for the S-phase checkpoint die in the presence of replication stress is a long-standing question. Here we show that reducing Cdk1 activity suppresses the loss of viability and the DNA replication defects of mec1Δ sml1Δ and rad53Δ sml1Δ mutant cells that are treated with low HU doses. Furthermore, reducing the Cdk1 activity bypasses the essential function of Mec1 and Rad53 kinases during an unchallenged S phase. Thus, the execution of some Cdk1-dependent events is detrimental when the S-phase checkpoint is not functional independently of the presence of exogenous replication impediments, suggesting that the essential function of Mec1 and Rad53 is not necessarily separated from the function they exert under replicative stress conditions. Although our work has been carried out in budding yeast, lowering Cdk1 activity appears to counteract the deleterious effects caused by a deficient DNA damage checkpoint also in mammalian cells. In fact, downregulation of the Cdk activator CDC25A rescues the replicative stress occurring after inhibition of the checkpoint proteins Chk1 and ATR (3, 51, 57).
The Cdk1-dependent lethality of mec1 and rad53 mutants involves a function of Cdk1 in G1. In fact, the lack of the G1 cyclin Cln2, which causes a G1/S delay by impairing proteolysis of the Clb-Cdk1 specific inhibitor Sic1 (17), suppresses the sensitivity of mec1Δ sml1Δ and rad53Δ sml1Δ cells to low HU doses. Accordingly, the deletion of SIC1, which abrogates the G1/S delay caused by the cln2Δ mutation (17), counteracts the ability of cln2Δ to suppress the HU sensitivity of mec1Δ sml1Δ and rad53Δ sml1Δ cells. Thus, a delay in S-phase entry appears to support cell survival to replication stress in the absence of Mec1 or Rad53. This finding is complementary to earlier observations in mammalian systems, where ATR and Chk1 inhibitors were shown to be particularly toxic for p53-deficient cells, which lack the G1 checkpoint and therefore experience a less restrictive S-phase entry in the presence of DNA damage (42, 57).
It has been shown that Mec1 and Rad53 stimulate production of dNTPs by inducing the expression of the RNR genes at the G1/S transition via inactivation of the transcriptional repressor Crt1 (25). Indeed, the lack of Crt1 not only bypasses the essential function of Mec1 and Rad53 (25) but also improves survival in response to HU treatment of mec1Δ sml1Δ and rad53Δ sml1Δ mutant cells (Fig. 3D). Therefore, the role of Mec1 and Rad53 in inducing transcription of the RNR genes contributes to maintain cell viability also under replication stress. In this scenario, the S-phase delay in cln2Δ cells could provide mec1 and rad53 mutants with a longer time to reach high enough dNTP levels to support DNA replication. Consistent with previous data showing that increased dNTP pools promote replication of MMS-damaged DNA (9, 40), the lack of Cln2 or Crt1 improves survival of mec1 mutants also to MMS treatment.
The Cdk1-dependent lethality of mec1 and rad53 mutants cannot be entirely ascribed to the G1 function of Cdk1. In fact, the lack of the mitotic cyclin Clb2 also improves the viability of HU-treated mec1Δ sml1Δ and rad53Δ sml1Δ mutants. Although mec1 and rad53 mutants are unable to complete DNA replication in the presence of HU, they prematurely undergo spindle elongation due to upregulation of the kinesin motor protein Cin8 (27). Because Cdk1 activity is required to assemble a mitotic spindle by restraining Cin8 proteolysis (13), a reduced Cdk1 activity might suppress the sensitivity of mec1 and rad53 mutants to replication stress by delaying spindle elongation. Indeed, we found that both restraining spindle extension by CIN8 deletion and disrupting kinetochore-microtubule attachments by nocodazole improve the survival in response to HU of mec1 and rad53 mutants. Interestingly, the lack of Cin8 allows mec1Δ sml1Δ and rad53Δ sml1Δ cells to complete DNA synthesis in the presence of HU, suggesting that the DNA replication defects in these mutants can be overcome by providing cells with additional time to complete DNA replication before chromosome segregation takes place. This hypothesis implies that the segregation of incompletely replicated DNA contributes to kill HU-treated mec1 and rad53 mutants. Because HU-treated mec1Δ sml1Δ cells can duplicate their centromeres (19) and therefore are proficient for bipolar attachment that generates tension within the spindle, this precocious spindle elongation can lead to lethal chromosome breaks due to disjunction of incompletely replicated chromosomes (19). Although the force exerted by a bipolar spindle might be insufficient to generate chromosome breakage, the presence of ssDNA in HU-treated mec1 and rad53 mutants (20, 50) may cause spindle-induced breakage.
Interestingly, the lack of Cin8 suppresses the sensitivity to HU of mec1 and rad53 cells more efficiently than nocodazole addition. Indeed, nocodazole disrupts the kinetochore-microtubule attachment, thus causing the release of the kinetochores from the spindle pole bodies to which they were first attached. It has been shown that premature spindle elongation caused by Cin8 deregulation in nocodazole-treated cells leads to mis-segregation of sister chromatids because it impairs the recapture and biorientation of microtubules after nocodazole removal (10, 31). Thus, one possibility is that the unrestrained Cin8-dependent spindle elongation in HU-treated mec1 and rad53 mutants limits the ability of nocodazole to rescue mec1 and rad53 cell lethality in HU.
Noteworthy, although mec1Δ sml1Δ cells are more sensitive to HU and MMS than rad53Δ sml1Δ cells, both nocodazole addition and CIN8 deletion improve survival to HU of mec1 cells more efficiently than that of rad53 cells. Furthermore, delaying spindle elongation suppresses the MMS sensitivity of mec1Δ sml1Δ cells, but it only slightly reduces that of rad53Δ sml1Δ cells. Although HU slows down DNA synthesis by depleting dNTPs, MMS-induced lesions block the progression of DNA replication forks because replicative polymerases cannot accommodate 3-methyl-adenine in their catalytic sites (22, 39). Interestingly, deletion of the EXO1 nuclease gene has no effect on the sensitivity of mec1 mutants to MMS, whereas it suppresses cell lethality and the fork progression defects of MMS-treated rad53Δ sml1Δ cells (48). It is therefore tempting to propose that, while the Mec1 requirement in supporting replication of MMS-damaged DNA can be bypassed by providing additional time to complete DNA replication, the lack of Rad53 results in defects at the replication forks that become substrates for irreversible Exo1-dependent replication fork breakdown or resection events. In line with this view, it has been shown that Exo1 can process aberrant DNA intermediates in rad53-deficient cells (12) and that Rad53-dependent phosphorylation of Exo1 may act to limit ssDNA accumulation (36).
In any case, delaying spindle elongation can rescue the HU-induced lethality of mec1 and rad53 mutants only during exposure to low HU doses. This finding is consistent with previous data showing that the addition of nocodazole does not improve viability of rad53 and mec1 cells following transient exposure to a high HU dose (200 mM) (16, 55). Thus, high HU levels seem to irrecoverably commit checkpoint mutants to death, possibly because they induce the stalling of the replication forks that are defective in restarting replication (15) and/or can undergo irreversible pathological transitions, such as the replisome dissociation from the nascent DNA chains (11, 26, 34).
Is the essential function of the S-phase checkpoint related to its function in supporting DNA replication under mild replication stress? Mec1 and Rad53 are essential for cell viability, and their essential function can be bypassed by increasing dNTP levels through SML1 deletion (64), suggesting that the lethality of mec1- and rad53-null mutants is due to DNA replication occurring in the absence of adequate dNTP accumulation. We found that, like the lack of Cln2 (59) (Fig. 3B), the lack of Clb2 or Cin8 not only suppresses the sensitivity of mec1Δ sml1Δ and rad53Δ sml1Δ cells to replication stress but also bypasses the essential function of Mec1 and Rad53. Therefore, cells lacking MEC1 or RAD53, but carrying wild-type SML1, appear to die through a mechanism similar to that killing mec1Δ and rad53Δ cells exposed to low HU doses in the absence of the Sml1-mediated inhibition of RNR activity. In both cases, cells experience defective DNA replication caused by a condition of nucleotide depletion, and cell death can be overcome by providing cells with additional time to accumulate dNTPs and accomplish DNA replication before chromosome segregation takes place. As proposed for HU-treated mec1Δ sml1Δ and rad53Δ sml1Δ cells (19), partition of incompletely replicated DNA in mec1Δ SML1 and rad53Δ SML1 cells could give rise to lethal chromosome breaks. Consistent with this hypothesis, conditional inactivation of Mec1 by temperature-sensitive mec1 alleles gives rise to the accumulation of chromosome breakage (8) and of Rad52 foci that can be reversed by nocodazole addition (Fig. 8G).
In summary, we demonstrate that Cdk1-driven events impair cell viability when the S-phase checkpoint is not functional. Given that some cancer cells (such as those expressing oncogenes or lacking tumor suppressors) undergo high levels of replicative stress and that Cdk inhibitors are considered relevant candidates as anticancer drugs, this link between the S-phase checkpoint and Cdk1 activity may be important in developing new targeted strategies improving the efficiency of cancer treatments.
We thank J. Diffley, R. Kolodner, S. Piatti, and R. Rothstein for providing yeast strains. We are grateful to M. Bazzi for preliminary data, L. D'Alfonso and M. Collini for valuable advice at the microscope, M. Lopes and M. Foiani for helpful discussions, and M. Clerici and S. Piatti for helpful suggestions and critical reading of the manuscript.
This study was supported by grants from the Associazione Italiana per la Ricerca sul Cancro (grant IG11407), Cofinanziamento 2008 MIUR/Università di Milano-Bicocca to M.P.L., and Cofinanziamento 2009 MIUR/Università di Milano-Bicocca to G.L. N.M. was supported by a fellowship from Fondazione Italiana per la Ricerca sul Cancro.
Published ahead of print 8 October 2012
Supplemental material for this article may be found at http://mcb.asm.org/.