Checkpoint proteins control morphogenetic events during DNA replication stress in Saccharomyces cerevisiae

Tight coordination of the cellular events associated with every phase of the cell cycle is essential for orderly progression through the cell cycle (Hartwell and Weinert, 1989). Cells arrest the cell cycle when confronted with stresses that may compromise cellular integrity. The mechanisms that halt the cell cycle are referred to as checkpoints (Hartwell and Weinert, 1989). Checkpoints result in key dependency relationships; mutants that fail to complete DNA replication also block nuclear division and cytokinesis, and mutants that fail to complete nuclear division also block cytokinesis (Hartwell and Weinert, 1989). Cells arrest the cell cycle in response to DNA damage. The DNA damage checkpoint was originally defined as the pathway that promotes cell cycle delay in response to DNA damage (Hartwell and Weinert, 1989), although it is now generally accepted that checkpoint responses involve additional processes, such as the activation and recruitment of DNA repair factors (Lisby et al., 2004) and the stabilization of replication forks (Lopes et al., 2001). At least two checkpoints operate during S phase in Saccharomyces cerevisiae; the replication checkpoint, which was originally defined as causing hydroxyurea (HU)-induced cell cycle arrest and inhibition of late-firing origins (Santocanale and Diffley, 1998), and the intra–S phase checkpoint, which reduces the rate of DNA replication and slows cell cycle progression in response to DNA-damaging agents (Paulovich et al., 1997). A series of checkpoints also act to monitor assembly of the mitotic spindle and regulate progression through mitosis (Kops et al., 2005). More recently, additional checkpoints have been identified, including a cell wall morphology checkpoint and a morphogenesis checkpoint, which monitor cell wall synthesis, bud formation, cell size, actin perturbation and, possibly, septin organization (Kellogg, 2003; Lew, 2003; Suzuki et al., 2004).

DNA replication stress and DNA damage induce activation of two phosphoinositide 3-kinase–related kinases, Tel1 and Mec1, which are similar to mammalian ataxia telangiectasia mutated and ataxia telangiectasia mutated and Rad3-related. These function in the activation of downstream protein kinases, including Chk1 and Rad53. Activation of Rad53 (S. cerevisiae Chk2) is mediated by two partially redundant adaptor proteins Mrc1 and Rad9 (Alcasabas et al., 2001). Although in Schizosaccharomyces pombe and higher eukaryotes cell cycle progression is blocked in response to replication stress, mainly by stimulating inhibitory phosphorylation of cyclin-dependent kinases (CDKs), in S. cerevisiae inhibition of Cdk1 (S. cerevisiae CDK) activity does not appear to play a role in S phase checkpoint-induced cell cycle arrest (Amon et al., 1992; Sorger and Murray, 1992). Rather, S. cerevisiae blocks cell cycle progression by directly inhibiting late origin firing and chromosome segregation (Santocanale and Diffley, 1998; Sanchez et al., 1999). Cdk1 in S. cerevisiae appears to have taken on a different function, and it synchronizes bud morphogenesis with the cell cycle (Lew, 2003). In S. cerevisiae, Cdk1 activity is determined by various factors, including Swe1 (S. cerevisiae equivalent of mammalian and S. pombe Wee1), which phosphorylates the conserved Y19 residue on Cdk1, resulting in inhibition of Cdk1. This phosphorylation is removed by the phosphatase Mih1. Swe1 levels are controlled by a pathway often referred to as the morphogenesis checkpoint (Lew, 2003), which responds to the state of the actin cytoskeleton, bud formation, and cell size, resulting in accumulation of Swe1, thereby delaying entry into mitosis (Kellogg, 2003; Lew, 2003). Degradation of Swe1 depends on various factors, including septins. Septins serve as a scaffold to recruit Swe1 and various kinases that phosphorylate Swe1, ultimately targeting it for destruction by an as yet unknown Skp1–Cul1–F-box complex and, possibly, the anaphase-promoting complex (McMillan et al., 2002; Thornton and Toczyski, 2003). A defect in Swe1 degradation results in prolonged inhibition of Cdk1. As a consequence, Cdk1 cannot induce the switch from polar to isotropic bud growth, resulting in the formation of elongated buds (Pruyne and Bretscher, 2000a,b).

Thus far, studies on the S-phase checkpoints have mainly focused on regulation of DNA replication, replication fork stabilization, and chromosome segregation. We show a novel role for components of the replication stress checkpoints in control of morphogenesis during replication stress. Our results are consistent with a model in which checkpoint proteins promote timely degradation of Swe1, thereby restricting bud growth during replication stress.

We observed that checkpoint mutants frequently have morphological aberrations. For instance, mec1 tel1 double mutant cells are commonly misshapen, have somewhat elongated buds, and often fail to complete cytokinesis (Fig. 1 A). rad9 and mrc1Δ single mutants did not have morphological aberrations (unpublished data), whereas mutants lacking both Mrc1 and Rad9, which function upstream of Rad53 (Alcasabas et al., 2001), as well as rad53Δ mutants had a phenotype similar to that of mec1 tel1 mutants (Fig. 1 A). Deletion of DUN1, which acts downstream of Rad53, resulted in a milder phenotype, whereas mec1 and chk1 mutants did not have a morphology defect (unpublished data). In addition, mec1 tel1 and mrc1Δ rad9 double mutants, and rad53Δ single mutants, frequently deposited abnormally large amounts of chitin (Fig. 1 A), not only at the bud neck but often at other apparently random sites of the cell wall (unpublished data), suggesting a possible defect in orchestrating cell wall architecture. Because cells with defective cell walls lyse in the presence of SDS or Calcofluor white (Lussier et al., 1997; Bickle et al., 1998), we tested the sensitivity of various mutants to these chemicals. mec1, tel1, and chk1 single mutants were no more sensitive to Calcofluor white and SDS than wild-type cells, whereas rad53Δ mutants were one to two orders of magnitude more sensitive (Fig. 1 B), which is indicative of a defective cell wall architecture. mec1 tel1 double mutants showed severely reduced viability on yeast extract/ peptone/dextrose (YPD), probably because these cells suffer major endogenous DNA damage from lack of DNA repair and checkpoint functions (Myung et al., 2001), and these mutants were sensitive to Calcofluor white (Fig. S1 A). However, because of the dramatic growth defects of mec1 tel1 double mutants, we decided to focus on rad53Δ and mrc1Δ rad9 mutants instead. mrc1Δ rad9 mutants were as sensitive to Calcofluor white as rad53Δ mutants (Fig. S1 A), whereas dun1 and chk1 mutants were only weakly sensitive to Calcofluor white (Fig. S1 A). Therefore, Rad53 seems to be a critical mediator of resistance to Calcofluor white, whereas there appear to be redundancies between Mec1 and Tel1, between Rad9 and Mrc1, and, potentially, between Chk1 and Dun1 (not tested), paralleling the roles of these proteins in checkpoints (Cobb et al., 2004). Cells with cell wall defects are usually more sensitive to zymolase (Ovalle et al., 1998). Untreated wild-type and rad53Δ mutants did not lyse when incubated with a hypotonic buffer containing zymolase (Fig. 1 C). However, pretreatment with HU increased sensitivity of rad53Δ mutants to zymolase, whereas wild-type cells remained unaffected. Cell lysis typically occurred at the bud tips (unpublished data), indicating that the replication checkpoint promotes formation of a healthy bud during replication stress (see below). Cell wall stress activates the Pkc1–Slt2 pathway (Cid et al., 1995), and, consistent with our findings that checkpoint mutants have cell wall defects, Slt2 is hyperphosphorylated in mec1 tel1 and rad53Δ mutants (Fig. 1 D). These results suggest that mec1 tel1 and mrc1Δ rad9 double mutants, and rad53Δ single mutants, have considerable defects in control of cell morphology and cell wall structure. Finally, we noticed that in rad53Δ mutants bud scars are frequently positioned at distal poles or at random positions on the cell (Fig. 1 E), indicating that rad53Δ mutants may have a bud site–selection defect because wild-type haploid cells have an axial budding pattern, forming buds at just one pole (Fig. 1 E; Pruyne and Bretscher, 2000a). As shown in Fig. 1 F, the majority of wild-type cells had an axial budding pattern. However, ∼45% of log phase rad53Δ mutants had bud scars that deviated from that pole. Cells expressing a kinase-dead mutant of RAD53 (rad53-KD; K227A, D319A, and D339A; see Smolka et al. on p. 743 of this issue), which results in a checkpoint deficiency similar to that of rad53Δ mutants, had a bud site selection defect similar to that of a rad53Δ mutant (Fig. 1 F), indicating that the kinase function of Rad53 is essential for correct bud site selection. Furthermore, whereas chk1, mec1, and tel1 mutants had very mild or no bud site selection defects, mec1 tel1 double mutants had a bud site selection defect similar to that of rad53Δ mutants, whereas rad9 and dun1 mutants had a more intermediate phenotype. bud1 mutants were examined as a control and were found to have a 100% random distribution of bud scars (unpublished data). In conclusion, we found that various checkpoint proteins contribute to cell wall architecture and maintenance of cell polarity.

Both bud site selection and cell wall synthesis are controlled by factors that regulate the actin cytoskeleton (Pruyne and Bretscher, 2000a). To test whether S-phase checkpoint proteins function in regulation of the actin cytoskeleton, we treated log-phase wild-type cells with HU for 4 h and visualized F-actin using rhodamine-phalloidin. Untreated wild-type, chk1, rad53Δ, and mrc1Δ rad9 mutants had a polarized actin cytoskeleton, with actin cables extending from the mother cell into the bud (Fig. 2, A and B). Treatment of wild-type cells with HU arrested the cells with large buds and a depolarized actin cytoskeleton (Fig. 2, A and B). In contrast, the actin cytoskeleton of both rad53Δ and mrc1Δ rad9 mutant cells remained polarized upon HU treatment (Fig. 2, A and B). Incubation of rad53Δ mutants at 42°C for 5 min resulted in complete actin depolarization, showing that rad53Δ mutants did not have a general defect in stress responses (unpublished data). HU also failed to induce actin depolarization in rad53-KD mutants. dun1 mutants, as well as sgs1 mutants (a DNA helicase that is thought to play roles in both the replication and intra–S phase checkpoints; Frei and Gasser, 2000), also failed to fully depolarize the actin cytoskeleton. tel1 and chk1 mutants were similar to wild-type cells, and in mec1 mutants, HU treatment only partially depolarized the actin cytoskeleton (Fig. 2 B, Fig. S1 B, and not depicted). Together, these results suggest that checkpoint proteins like Sgs1, Dun1, Rad53, and a combination of Mrc1 and Rad9 affect the polarity state of the actin cytoskeleton upon induction of replication stress.

Cytoskeletal polarity is guided by the Cdc24–Cdc42 pathway (Pruyne and Bretscher, 2000a). We found that in wild-type cells, treatment with HU resulted in the disappearance of Cdc24 from the bud tip membrane (Fig. 2 C), as might be expected from cells that have arrested with large buds. However, in rad53Δ and dun1 single mutants, and in mrc1Δ rad9 double mutants, Cdc24 remained at the membrane (Fig. 2 D), whereas chk1 mutants were similar to wild-type cells. Similar results were obtained with cells expressing Sec4-GFP, a Rab GTPase that is an essential component of the secretory machinery found primarily at sites of polarized growth (Guo et al., 1999; Fig. S1 C). Therefore, these results indicate that checkpoint proteins, including Rad53, Dun1, and Mrc1, in combination with Rad9, promote removal of the bud growth machinery from the bud tip upon treatment with HU.

Mutations in genes involved in regulation of the actin cytoskeleton often render cells sensitive to pharmacological actin inhibitors like latrunculin A (Ayscough et al., 1997). Therefore, we speculated that rad53Δ mutants might also be sensitive to latrunculin A. As shown in Fig. 2 E, rad53Δ mutants were more sensitive to latrunculin A than wild-type cells, whereas mec1, tel1, chk1, and dun1 single mutants were no more sensitive to latrunculin A than wild-type cells (Fig. 2 E and not depicted). Interestingly, a swe1 deletion suppressed the sensitivity of rad53Δ mutants to latrunculin A (Fig. 2 E), and we found that Rad53 may control Swe1 (see the following section).

In addition to failure to depolarize the actin cytoskeleton, treatment of rad53Δ, but not wild-type, cells with HU for extended periods of time (16–20 h, although visible after 6 h [see Smolka et al. on p. 743 of this issue], after which >80% of the cells are still alive, as indicated by staining with vital dyes; Fig. S2 A) resulted in formation of elongated buds (Fig. 3, A and C). Analysis of W303 RAD5 and W303 RAD5 rad53Δ sml1 strains revealed a similar effect of a rad53Δ mutation (unpublished data). Formation of elongated buds is a common feature of mutants that fail to properly control levels of Swe1 (Pruyne and Bretscher, 2000b). This, and the finding that deletion of SWE1 suppressed the sensitivity of rad53Δ mutants to latrunculin A, raised the possibility that the elongated bud phenotype of HU-treated rad53Δ mutants is caused by failure to down-regulate Swe1. Indeed, deletion of SWE1 completely rescued the HU-induced elongated bud phenotype of the rad53Δ mutant (Fig. 3, B and C). Similar results were obtained with cells harboring the rad53-KD allele (Fig. 3 C and Fig. S2 B). Deletion of DUN1 also caused cells to form elongated buds after treatment with HU (Fig. 3 C), and deletion of DUN1 in a rad53Δ background resulted in an augmented phenotype (Fig. 3 C).

We next tested the effect of a range of mutations in genes encoding checkpoint functions on bud morphogenesis during replication stress. rfc5-1, tel1, mec1, mrc1Δ, and rad9 single mutations all caused no, or a very small, increase in HU-induced elongated bud growth. In contrast, rad9 mrc1Δ and rad9 mrc1-AQ double mutants, the latter of which express an allele of Mrc1 that is proficient in DNA replication, but unable to activate the replication checkpoint (Osborn and Elledge, 2003), showed elongated bud growth upon HU treatment. Furthermore, mec1 tel1 double mutants also showed an increase in elongated bud growth; however, we noticed that cells that attempted to elongate their buds frequently lysed, possibly because of their severe cell growth and cell wall defects, and this may obscure the phenotype. The dpb11-1, tof1, csm3, chk1, rad17, and rad24 mutations alone did not increase HU-induced elongated bud growth (Fig. 3 C). In contrast, cells harboring the rfa1-t11 allele displayed considerable HU-induced bud elongation, which is consistent with the function of the replication protein A complex in the DNA replication stress response. sgs1 mutants also showed HU-induced elongated bud growth, which was SWE1-dependent, and the extent of elongation of these buds was often severe (Fig. S2 B). The sgs1 rad24 double mutant had a more severe HU-induced elongated bud phenotype compared with the respective single mutants, similar to a previous report on the effects of sgs1 and rad24 mutations on the sensitivity to HU (Frei and Gasser, 2000). Other combinations of mutations did not reveal significant genetic interactions (Fig. 3 C). Finally, mutations in MRE11, RAD50, and XRS2, which are the three components of the MRX complex, resulted in HU-induced elongated bud growth. These results identify Sgs1, the MRX complex, replication protein A, Mec1 in combination with Tel1, Mrc1 in combination with Rad9, Rad53, and Dun1 as important mediators in controlling bud morphology during replication stress.

The aforementioned results indicate that a pathway requiring Rad53 may control Cdk1 activity during S phase, when most bud growth takes place. Because different checkpoint mutants undergo HU-induced, Swe1-dependent elongated bud growth, we analyzed lysates of HU-treated cells by Western blotting using Swe1 antibodies (Fig. 4 A). Treatment of wild-type cells with HU for 1 or 2 h resulted in accumulation of moderately phosphorylated Swe1, which then became hyperphosphorylated after 3 h, ultimately resulting in its destruction (4 h), which is consistent with the results of Fig. 1 A of Liu and Wang (Liu and Wang, 2006). In contrast, in HU-treated rad53Δ mutants, only moderately phosphorylated species of Swe1 accumulated, and Swe1 largely failed to get degraded (Fig. 4 A); this was not caused by cell cycle effects, as the budding index of rad53Δ mutants was similar to that of wild-type cells and cells did not reenter the cell cycle (Fig. S3 A). Importantly, this resulted in hyperphosphorylation of Y19 of Cdk1 (Fig. 4 B), which, as expected, was Swe1-dependent (unpublished data). A similar effect of a rad53Δ mutation was seen with W303 RAD5 and W303 RAD5 rad53Δ sml1 strains (unpublished data). When analyzed as a control, Y19 of Cdk1 also became hyperphosphorylated in hsl1 and elm1 mutants, which are known to be defective in degradation of Swe1 (Fig. S3 B). Furthermore, high levels of Swe1 accumulated in rad9 mrc1Δ mutants (Fig. 4 A), which is consistent with the fact that Rad9 and Mrc1 are upstream regulators of Rad53. Swe1 levels were also elevated in cla4 mutants (Fig. 4 A), which is consistent with the known role of Cla4 in phosphorylating Swe1 to target it for destruction.

For Swe1 to be hyperphosphorylated it needs to localize to the bud neck (Lee et al., 2005), which is dependent on various factors, including septins. We could not detect endogenous GFP-tagged Swe1 (unpublished data), but, when overexpressed, we could detect bud neck–localized Swe1-GFP in 50–60% of the cells (Fig. 4 C), in accordance with a previous study (Asano et al., 2005). Importantly, we found that Swe1 localization in rad53Δ mutants was similar to that of wild-type cells, even after treatment with HU (unpublished data), showing that the defect in Swe1 degradation does not result from failure to localize Swe1 to the bud neck. Altogether, these data show that a Rad53-dependent pathway restricts bud growth when cells are confronted with DNA replication stress by controlling Swe1-Cdk1.

Various pathways are known to control the actin cytoskeleton and bud morphogenesis (Pruyne and Bretscher, 2000a,b). Because we found that checkpoint proteins control bud morphology during replication stress, one would predict that defects in known pathways that regulate bud morphology might also result in HU-induced elongated bud growth. Based on knowledge from the literature (Pruyne and Bretscher, 2000a,b), we tested several candidates in an attempt to identify the pathways that may be used by the replication stress checkpoints to ensure proper bud growth during replication stress. The results of this analysis, described in this section, show that regulation of bud morphology is an essential part of the response to DNA replication stress, and that cells use a network of diverse pathways to ensure proper bud morphology when confronted with replication stress.

The p21-activated kinase–like kinase Cla4 is a key factor in regulation of the actin cytoskeleton, and is a central component of the Cdc42 pathway. We not only found that the Cdc24–Cdc42 pathway may not be properly down-regulated in several checkpoint mutants after HU treatment (Fig. 2 D) but also that cla4 mutants accumulate Swe1 during replication stress (Fig. 4 A). Therefore, we tested whether cla4 mutants might be defective in HU-induced actin depolarization. Indeed, cla4 mutants failed to depolarize the actin cytoskeleton and dramatically elongated their buds upon HU treatment (Fig. 5 A; quantified in Fig. 5, B and C), indicating that Cla4 is necessary for regulation of the actin cytoskeleton when cells are confronted with replication stress. Interestingly, CLA4 also genetically interacts with RAD53 (see below). Bem1 is another component of the Cdc24–Cdc42 pathway. We found that bem1 mutants, like cla4 mutants, failed to depolarize actin upon HU treatment, resulting in elongated bud growth. We also tested strains lacking GTPase-activating proteins for the small Ras-like GTPases Cdc42 and Rho, which function in septin organization and bud growth (Caviston et al., 2003). Cells lacking the Bem2 and Bem3 failed to depolarize the actin cytoskeleton and elongated their buds after HU treatment, whereas lrg1 and rgd1 mutants responded to HU like wild-type cells (Fig. 5, B and C). This supports our finding that proper regulation of the Cdc42–Cla4 pathway is important for control of the actin cytoskeleton and cell morphology in response to HU. These results also suggest a possible involvement of specific bem2-controlled Rho pathways.

The kinases Elm1 and Gin4 function in septin ring assembly (Bouquin et al., 2000). The septin ring itself serves as a scaffold to bind various proteins, including Hsl1, Gin4, Elm1, and Hsl7. Hsl7 is an adaptor protein that helps recruit Swe1, bringing it into proximity of several kinases, including Hsl1, Clb2-Cdk1, Cla4, and Cdc5, which phosphorylate Swe1 to target it for degradation (Asano et al., 2005). Consistent with a role of Swe1 in regulating bud growth in response to HU, hsl1, hsl7, gin4, and elm1 mutants failed to depolarize their actin cytoskeleton and formed hyperpolarized buds after treatment with HU (Fig. 5, B and C, and not depicted).

We also found that cells lacking a variety of cell cycle regulators, such as the transcription factors Swi4, Swi6, and Fkh2, the F-box protein Grr1 (Fig. 5 C and not depicted), and the mitotic exit–regulating proteins Kel2 and Dbf2, hyperpolarized bud growth upon HU treatment. Swi4 and Swi6 are involved in transcriptional control of various genes, including Hsl1, Kcc4, and Gin4 (Spellman et al., 1998; Iyer et al., 2001; Flick and Wittenberg, 2005), which might explain why swi4 and swi6 mutants elongate their buds upon HU treatment. It is not clear why fkh2 mutants elongate their buds after treatment with HU, but it was previously found that a cla4 deletion synthetically interacts with fkh2, indicating a role for Fkh2 in the Cdc42–Cla4 pathway (Goehring et al., 2003). Kel2, Dbf2, and Grr1 regulate mitotic exit and actomyosin ring contraction. We are currently investigating why HU induces elongated bud growth in cells lacking these factors.

Bni1 is a member of the polarisome and functions together with Cla4 in septin organization. bni1 mutants did not develop elongated buds when challenged with HU, but instead appeared to undergo an extra round of cell growth, resulting in formation of “strings” of cells (graphically displayed in Fig. 5 D and quantified in Fig. 5 E). We also noticed that dbf2, swi4, and swi6 mutants form such strings of cells, whereas deletion of any of the Rho–GTPase-activating proteins or the septin-regulating kinases tested in our study had no such effect. Thus, Bni1, Dbf2, Swi4, and Swi6 also appear to assure that cells maintain proper morphology when challenged with prolonged replication stress.

If regulation of actin, septins, and cell morphology is part of the response to replication stress, one would predict that cells defective in these pathways are more sensitive to HU. Indeed, chronic treatment of strains lacking Cla4, Elm1, Hsl1, Hsl7, and Gin4 with HU caused reduced growth compared with wild-type cells (Fig. 6 A). Analysis of the colonies that formed on HU plates showed that these mutants formed elongated buds and clusters of cells with multiple unseparated cell bodies (unpublished data), indicating that cell proliferation may have slowed because of morphological aberrations rather than an inability to activate the replication checkpoint. All of these mutants, including cla4, recovered normally from a 4-h HU arrest, indicating that all mutants are checkpoint proficient, whereas additional deletion of RAD53 in these mutants resulted in a complete loss of viability (Fig. S4 A). swe1 mutants also recovered normally from HU arrest, whereas the swe1 rad53Δ double mutant did not recover, indicating that our finding that a swe1Δ mutation suppresses some of the morphological defects of a rad53Δ mutant is unrelated to an effect on checkpoint regulation of cell cycle arrest. These results suggest that septin/Swe1-regulating factors are unlikely to be directly involved in checkpoint control of cell cycle arrest, but instead have a different function in the cellular response to replication stress, such as ensuring proper bud growth.

CLA4 has been found to genetically interact with a variety of DNA replication, repair, and checkpoint factors (Fig. S4 B; Goehring et al., 2003; Pan et al., 2006). In line with this, we found that a rad53Δ mutation caused synthetic fitness defects when combined with mutations in CLA4, ELM1, and GIN4 (Fig. 6 B) and caused temperature-sensitive growth when combined with a mutation in SHS1 (Fig. 6 C). These results indicate that Rad53 may cooperate with the pathway involving Cla4, Elm1, Gin4, and Shs1. Therefore, we tested whether Rad53 is involved in septin organization. We studied septin ring organization by immunofluorescence rather than expression of Cdc12-GFP because we found that Cdc12-GFP weakly interfered with septin organization (unpublished data). Interestingly, whereas rad53Δ mutants had no noticeable septin defect, deletion of RAD53 augmented the septin defects of elm1, gin4, and cla4 mutants (Fig. 6 D; see Fig. S4 C for examples of septin phenotypes). These data indicate that Rad53 has a redundant function in the organization of septin rings.

Finally, we observed that a rad53Δ deletion, as well as mutations in upstream regulators of Rad53, often lead to formation of small aggregates of cells (Fig. 1 A), indicating that Rad53 may also be involved in regulation of cytokinesis. Therefore, we treated cells with zymolase to digest the cell wall, which results in removal of the buds of cells that have successfully completed cytokinesis, but preserves the buds of cells that have not completed cytokinesis, as these cells contain buds that are still connected through the plasma membrane. We found that log-phase cultures of wild-type cells contained ∼40% of cells with a single bud and 7% of cells with more than one bud after zymolase treatment (Fig. 6 E). In contrast, zymolase treatment of rad53Δ mutants, as well as mrc1Δ rad9 mutants, left 36 and 41% of cells with two or more buds, respectively, indicating that these cells failed to complete cytokinesis (Fig. 6 E). Treatment of cells with HU modestly increased the number of multibudded cells in mrc1Δ rad9 mutants and did not increase the number of multibudded cells in wild-type and rad53Δ cultures. In addition, rad53-KD mutants also had cytokinesis defects (unpublished data). Based on these results, we conclude that Rad53 kinase activity is required for the successful completion of cytokinesis and separation of mother and daughter cells.

During replication stress, cells of most wild-type S. cerevisiae strains (see second to last paragraph of this section) arrest with spherically shaped buds approximately the same size as the mother cell, indicating that mechanisms exist to monitor and control cell growth and morphology when DNA replication has halted. We observed that checkpoint mutants were frequently misshapen, had defective cell walls, and displayed hyperphosphorylation of Slt2. This is in line with a recent report showing genetic interactions between Rad9 and Slt2 (Queralt and Igual, 2005). Checkpoint proteins were also found to support proper bud site selection, indicating an involvement in maintenance of cell polarity and control of the actin cytoskeleton (Pruyne and Bretscher, 2000a,b). Indeed, we found that checkpoint proteins, including Rad53, are involved in the removal of Cdc24 and Sec4 from the bud tip during replication stress, resulting in depolarization of the actin cytoskeleton, and that failure to do so resulted in the formation of elongated buds. We identified several additional pathways involved in preventing elongated bud growth during replication stress, and at least some of these pathways (e.g., the Cla4 pathway) genetically interact with the Rad53 pathway. Our results indicate that checkpoints and additional pathways cooperate to support cell morphology and cytokinesis during chronic replication stress.

To identify the checkpoint pathways that restrict bud growth during replication stress, we tested mutants lacking a wide range of checkpoint factors. Mutations in SGS1 and RFA1, but not in other genes (i.e., RAD24, RFC5, or DBP11) thought to encode upstream-acting checkpoint factors, caused a significant defect in control of cell morphology during replication stress (Fig. 3 C). Although a rad24 single mutant had no defects, sgs1 rad24 double mutants had a more severe phenotype than sgs1 single mutants, in line with previous reports showing that sgs1 rad24 double mutants have increased sensitivity to HU and methylmethanesulfonate (Frei and Gasser, 2000). Consistent with this, Rad53 and its downstream-acting factor Dun1 were involved in controlling cell morphology during replication stress. Mec1 is generally thought to function upstream of Rad53 in checkpoint signaling. Although mec1 and tel1 single mutants did not have a strong morphological phenotype, mec1 tel1 double mutants had morphological aberrations that were comparable to those of rad53Δ and rad53-KD mutants. This fits with the view that Mec1 and Tel1 can have redundant functions, and is reminiscent of the Mre11 complex–mediated S phase checkpoint response that is Tel1-dependent, but is predominantly seen in the absence of Mec1 (Usui et al., 2001). Indeed, we found that the MRX complex is important for maintaining proper morphology in response to HU treatment. In addition, we found that Mrc1 and Rad9 together were required for the morphological response to replication stress, which is consistent with the finding that these proteins are redundant activators of Rad53 (Alcasabas et al., 2001). Defects in Chk1 did not cause morphological defects, which is not surprising because chk1 mutations do not cause sensitivity to HU (Sanchez et al., 1999). Finally, mutations in TOF1 and CSM3 did not cause morphology defects, suggesting that there is redundancy among these genes or that they act in different aspects of checkpoint responses. Overall, the results presented in this study link replication stress checkpoint functions to the control of cell morphology.

A recent study has shown that mec1 and rad53Δ mutants show premature spindle elongation during HU-induced replication stress because of failure to inhibit the microtubule-associated proteins Cin8 and Stu2, resulting in precocious chromosome segregation (Krishnan et al., 2004). This raises the possibility that the defects in regulation of morphology and spindle dynamics in checkpoint mutants may be related. Premature spindle elongation is not sufficient for HU-induced bud elongation in checkpoint mutants because mec1 mutants, which elongate their spindles prematurely (Krishnan et al., 2004), do not show HU-induced elongated bud growth. Furthermore, nocodazole treatment, which inhibits spindle elongation, did not block HU-induced elongated bud growth in rad53Δ mutants (Fig. S4 D), indicating that premature spindle elongation is not required for elongated bud growth induced by HU treatment of checkpoint mutants.

In contrast to S. pombe and higher eukaryotes, S. cerevisiae does not target Cdk1 to block cell cycle progression during replication stress, but instead directly targets processes such as late origin firing and chromosome segregation (Santocanale and Diffley, 1998; Sanchez et al., 1999). Rather, S. cerevisiae Cdk1 has taken on a different role; it is essential for the morphogenetic switch from polar to isotropic bud growth, and mutants that fail to degrade Swe1 have a delayed morphogenetic switch, resulting in elongated bud growth. We found that a Rad53-dependent pathway is important for timely degradation of Swe1 during replication stress, and showed genetically that it is this failure of checkpoint mutants to degrade Swe1 that results in elongated bud growth. However, it is currently unknown how Rad53 controls Swe1. Swe1 recruitment to septin rings was unaffected in rad53Δ mutants, which is consistent with our finding that septin organization is normal in these mutants. In S. pombe, the Rad53 homologue Cds1 may directly phosphorylate Wee1 to control its activity (Boddy et al., 1998), raising the possibility that in S. cerevisiae Rad53 also directly phosphorylates Swe1. However, it is also possible that Rad53 indirectly controls Swe1 levels. For instance, Rad53 may target kinases that in turn directly phosphorylate Swe1, like Cdc5 and Cla4 (Lee et al., 2005). Indeed, Rad53 has been suggested to control Cdc5 activity (Sanchez et al., 1999).

We found that RAD53 genetically interacts with a variety of upstream regulators of septins, as well as with the nonessential septin SHS1. Furthermore, replication stress checkpoint mutants showed a cytokinesis defect, which is often linked to septin malfunction. Therefore, we investigated the possibility that Rad53 may affect septin organization. We found that Rad53 by itself is not essential for the organization of septin rings, but has a redundant role with upstream septin regulators like Cla4, Gin4, and Elm1 in septin organization and cell growth. Consistent with this, defects in septin-regulating factors caused sensitivity to HU and HU-induced bud elongation. Although the molecular nature of the Rad53 interactions with septins is not yet clear, interestingly, Rad53 can directly phosphorylate Shs1 in vitro, and a pool of Rad53 may localize to the septin ring in vivo (see Smolka et al. on p. 743 of this issue). It is possible that defects in this cooperation could explain the deregulation of Swe1 that occurs in checkpoint mutants in response to HU. Further studies are needed to unravel the function of Rad53 in control of septin organization.

Bud morphogenesis during replication stress has been studied previously. Several studies have shown that replication stress induced filamentous differentiation strongly in one wild-type strain (Σ1278b), weakly in two wild-type strains (W303 rad5-535 and A374), and not at all in another wild-type strain (S288c; Jiang and Kang, 2003; Liu and Wang, 2006). Induction of this phenotype in Σ1278b cells appeared to require Mec1, Rad53, and Swe1, but not Sgs1 or Dun1. The HU-induced bud elongation phenotype we have studied does not occur in our wild-type strain (S288c) and is induced in various checkpoint mutants, and thus appears to be a different phenotype. To better understand the differences between wild-type strains, we tested seven different laboratory wild-type strains collected over the years and identified one strain showing a strong bud elongation phenotype (DBY745), two strains showing a weaker phenotype (W303 RAD5 and W303 rad5-535), and five strains (MGD, BY4741, L2955, Y55, and JKM139) that did not show HU-induced bud elongation. Because DBY745 was derived by crossing markers into S288c (unpublished data, Botstein, D., personal communication) and JKM139 was derived by crossing markers from Y55 into DBY745 (unpublished data, Haber, J., personal communication), it seems likely that the bud elongation resulted from a mutation introduced during the construction of DBY745. Similarly, the bud elongation phenotype of the W303 strains may have resulted from the introduction of a mutation during the intercrossing of the S288c derivatives used to construct W303 (for detailed information on W303 see the Saccharomyces Genome Database; www.yeastgenome.org). Further analysis will be required to understand the exact genetic basis for the replication stress–induced bud elongation and filamentous differentiation that some wild-type strains show.

In conclusion, we found that replication stress checkpoint proteins like Rad53 function together with additional pathways to promote the timely degradation of Swe1, thereby relieving inhibition of Cdk1. Cdk1 then induces the switch from polar to isotropic bud growth, thus preventing elongated bud growth and contributing to cell viability because such elongated buds are susceptible to cell wall stress. Therefore, replication stress checkpoint–mediated control of bud morphology is part of the response to replication stress and contributes to cell survival.

S. cerevisiae strains were grown in standard YPD medium or synthetic complete medium lacking the appropriate amino acid. Strains were directly derived from the S288c strain RDKY3615 using either standard gene replacement methods or intercrossing (Table S1). To construct cells harboring the mrc1-AQ allele, a TRP1 cassette was inserted into a PacI site 250 bp downstream of the mrc1-AQ allele in plasmid pAO138 (gift from S. Elledge, Harvard Medical School, Boston, MA). Subsequently, mrc1-AQ.TRP1 was PCR amplified and used to replace a URA3 cassette previously inserted at the MRC1 locus. All other strains were obtained from the systematic deletion project (BY4741; derived from the same parental strains as RDKY3615, i.e., S288c) and were only used when specifically stated in the figure legends. Plasmids pRS414-ADH-CDC24-GFP and pRS415-ADH-CDC24-GFP were provided by M. Peter (Swiss Federal Institute of Technology, Zurich, Switzerland), and pRS414-GFP-SEC4 was obtained from S. Emr (University of California, San Diego, La Jolla, CA). The YepT-SWE1-GFP plasmid was obtained from K. Lee (National Institutes of Health, Bethesda, MD).

Live cells (expressing either Cdc24-GFP, Sec4-GFP, or Swe1-GFP) were imaged at room temperature in synthetic complete medium with an inverted microscope (Eclipse TE300; Nikon) equipped with a 100×/1.40 NA Plan Apo objective lens (Nikon), using a charge-coupled device camera (Orca-ER; Hamamatsu) and MetaMorph software (Universal Imaging Corp.), and images were processed using Photoshop and Illustrator software (both Adobe). Alternatively, cells were fixed with 3.7% formaldehyde and either stained with 50 μg/ml Calcofluor white or rhodamine-phalloidin according to manufacturer's instructions (Invitrogen) or processed for septin immunofluorescence, as previously described (Pringle, 1991). Rabbit anti-Cdc11p antibodies (Santa Cruz Biotechnology, Inc.) were used at 1:10 dilution, followed by TRITC–conjugated goat anti–rabbit secondary antibody at a 1:50 dilution (Jackson ImmunoResearch Laboratories). Fixed cells were embedded in Vectashield HardSet mounting medium (Vector Laboratories) to reduce photobleaching. At least 100 cells were counted per strain and per treatment. For analysis of bud scar positioning, only cells with at least three bud scars were counted. Scars were scored as “normal” when all scars on a cell were located at the same pole. When one or more scars deviated from that pole, bud scar positioning was scored as “abnormal.”

Log-phase cells were treated with HU as indicated, pelleted, and resuspended in hot (95°C) Laemmli sample buffer supplemented with protease inhibitors and boiled for 5 min, after which glass beads were added and cells were vortexed for 15 min at 4°C. After centrifugation, soluble proteins were analyzed by SDS-PAGE and Western blot using phospho-Cdc2 antibodies (New England Biolabs) to analyze Cdk1 phosphorylation. Total Cdk1 levels were determined using PSTAIRE antibodies (Millipore). Antibodies against Swe1 were provided by D. Kellogg (University of California, Santa Cruz, Santa Cruz, CA). Phosphorylated Slt2 was analyzed with phospho-specific p42/44 MAPK antibodies (Cell Signaling Technology), and pan-Slt2 was detected using Slt2 antibodies (Santa Cruz Biotechnology, Inc.). Detection was performed using the SuperSignal West Femto Detection kit according to the manufacturer's instructions (Pierce Chemical Co.).

10 ml of cells were grown to OD₆₀₀ of 1 and quickly washed with 15 ml H₂O. Cells were then resuspended in 10 ml hypotonic buffer (10 mM Hepes, pH 7.5) supplemented with 20 U/ml lytic enzyme (Puregene) and incubated at 30°C with occasional agitation. OD₆₀₀ was measured every 15 min.

100 μl of 1–2 M HU in H₂O was added to 1 ml log-phase cells (in YPD), followed by 8–16 h at 30°C while shaking. The fraction of cells with elongated buds was determined as the percentage of large-budded cells only (unbudded and small-budded cells were ignored). A bud was scored as “elongated” when the length of the bud was at least twice its width.

1 ml of log-phase cells was washed with PBS and incubated at 30°C in 25 mM Hepes, pH 7.5, containing 1 M sorbitol (for osmotic support), and 5 mg/ml zymolase. After 30 min, cells with no buds, one bud, or two or more buds were counted. At least 100 cells were counted per treatment.

Halo assays were performed as previously described (Ayscough et al., 1997). In brief, 250 μl of a log-phase YPD culture was diluted in 3.5 ml YPD, after which 3.5 ml molten Agarose (1% wt/vol) in YPD (cooled to ∼50°C) was added. The cell suspension was then poured onto a YPD plate. Paper disks (6 mm diam; Becton Dickinson) were placed on top of the plates, and either 10 μl latrunculin A or 10 μl DMSO were spotted onto the center of the disks. Plates were inverted and incubated at 30°C until halos were visible.

Overnight cultures (prestationary phase) were diluted in 50 ml fresh, prewarmed YPD to an OD₆₀₀ of 0.05. Cells were then incubated at 30°C while shaking. 1-ml samples were taken every hour for 12 h, and doubling times were determined using Excel (Microsoft).

Table S1 shows S288c-derived strains used in the described studies. Fig. S1 shows analysis of the sensitivity of various S phase checkpoint mutants to Calcofluor white, images of the actin cytoskeleton in mec1, tel1, sgs1, and chk1 mutants upon treatment with HU, and the failure of rad53Δ mutants to remove Sec4 from the bud tip during replication stress. Fig. S2 shows analysis of metabolic activity of HU-treated wild-type and rad53Δ mutants during prolonged replication stress, images of SWE1-dependent HU-induced bud elongation in rad53-KD and sgs1 cells. Fig. S3 shows an analysis of the budding index of wild-type, rad53Δ, mrc1Δ rad9, and cla4 mutants that shows that none of these mutants reenter the cell cycle during HU arrest, and a Western blot demonstrating hyperphosphorylation of Y19 of Cdk1 during HU arrest in elm1 and hsl1 mutants. Fig. S4 shows that upstream regulators of septins are replication checkpoint proficient, extensive genetic interactions between CLA4 and DNA replication, repair, checkpoint genes, examples of various septin phenotypes, and that HU-induced bud elongation in rad53Δ mutants is not blocked by nocodazole.

We thank members of the Kolodner laboratory, D. Kellogg, and Matt Gentry for insightful discussions and helpful suggestions, Ellen Kats for construction of strains, and Hilde Abrahamsen and Kristina Schmidt for critical reading of the manuscript.

This work was supported by National Institutes of Health grants GM26017 (R.D. Kolodner) and HG002604 (H. Zhou). J.M. Enserink is a recipient of a fellowship from the Dutch Cancer Society (Koningin Wilhelmina Fonds/Kankerbestrijding).

The authors have no commercial affiliations or conflicts of interest.