|Home | About | Journals | Submit | Contact Us | Français|
The phosphatase Cdc14 is required for mitotic exit in budding yeast. Cdc14 promotes Cdk1 inactivation by targeting proteins that, when dephosphorylated, trigger degradation of mitotic cyclins and accumulation of the Cdk1 inhibitor, Sic1. Cdc14 is sequestered in the nucleolus during most of the cell cycle but is released into the nucleus and cytoplasm during anaphase. When Cdc14 is not properly sequestered in the nucleolus, expression of the S-phase cyclin Clb5 is required for viability, suggesting that the antagonizing activity of Clb5-dependent Cdk1 specifically is necessary when Cdc14 is delocalized. We show that delocalization of Cdc14 combined with loss of Clb5 causes defects in DNA replication. When Cdc14 is not sequestered, it efficiently dephosphorylates a subset of Cdk1 substrates including the replication factors, Sld2 and Dpb2. Mutations causing Cdc14 mislocalization interact genetically with mutations affecting the function of DNA polymerase and the S-phase checkpoint protein Mec1. Our findings suggest that Cdc14 is retained in the nucleolus to support a favorable kinase/phosphatase balance while cells are replicating their DNA, in addition to the established role of Cdc14 sequestration in coordinating nuclear segregation with mitotic exit.
Phosphorylation controls many aspects of the eukaryotic cell cycle. In budding yeast, cyclin-dependent kinase 1 (Cdk1), in association with different activating subunits (cyclins), regulates cell cycle progression by phosphorylating specific substrates on serine or threonine residues (28, 37). A critical feature of the budding yeast cell cycle is the oscillatory activity of Cdk1 (reviewed in references 27 and 31), which is responsible, at least in part, for fluctuations in the phosphorylation status of numerous proteins. Presumably, phosphatases are also important for reversing Cdk1-mediated phosphorylation, although these events are much less well characterized. Cdc14 is a phosphatase that antagonizes Cdk1, in part by promoting the decline in Cdk1 activity that is needed for cells to exit mitosis and enter the subsequent G1 phase (reviewed in reference 29). Interestingly, Cdc14 has a preference for pSer/pThr-Pro motifs (13), which are also consensus sites for Cdk1. However, it is unclear if Cdc14 targets the bulk of Cdk1 substrates or if Cdc14 dephosphorylates a specific subset of Cdk1 substrates.
Cdk1 activity declines abruptly as mitosis is completed. This change in Cdk1 activity is accomplished by the degradation of mitotic Clbs and the accumulation of the G1-phase-specific Cdk inhibitor Sic1 (35, 47). In late mitosis, the anaphase-promoting complex (APC) bound to the adapter protein Cdh1 triggers the ubiquitination of mitotic Clbs. Dephosphorylation of Cdh1 by Cdc14 is required to activate APCCdh1 (16, 47). Sic1 accumulation is accomplished by both increased synthesis and decreased degradation. Both of these mechanisms require Cdc14 phosphatase activity (47). Cdc14 dephosphorylates and activates Swi5, a transcription factor for SIC1, and Cdc14 dephosphorylates Sic1 itself, which prevents its recognition by ubiquitin ligases that target it for proteolysis. Altogether, Cdc14 plays a major role in decreasing overall Cdk activity to coordinate mitotic exit by dephosphorylating three key substrates: Cdh1, Sic1, and Swi5, the major transcription factor for SIC1. cdh1Δ sic1Δ cells are nevertheless able to exit from mitosis, suggesting that Cdc14 likely dephosphorylates additional Clb-Cdk targets (50); other targets of Swi5 in addition to SIC1 could also contribute (but the involvement of the Swi5 target CDC6 in mitotic exit was found to be very minor) (3).
Cdc14 is sequestered in the nucleolus during most of the cell cycle by the nucleolar protein Net1 but is released during anaphase by two different signaling networks: the FEAR pathway (Cdc fourteen early anaphase release) and the MEN (mitotic exit network). It then diffuses throughout the nucleus and cytoplasm, presumably to dephosphorylate its targets in a timely manner (40, 48). The FEAR pathway operates in early anaphase and involves the functions of Esp1, the protease responsible for sister chromatid separation, as well as Cdc5, Slk19, which is an additional substrate of Esp1, and Spo12 (42). Although the FEAR network can trigger release of Cdc14 from the nucleolus in early anaphase and its accumulation at spindle pole bodies, it is not sufficient for exit from mitosis and it is possible that the MEN is required to sustain Cdc14 presence in the nucleus and cytoplasm (43). The MEN operates in late anaphase and is comprised of the protein kinases Cdc15, Dbf2/Mob1, and Cdc5 as well as the small G protein Tem1, the guanine nucleotide exchange factor Lte1, and Cdc14 itself. Mutations in any of these genes prevent exit from mitosis with cells arresting in late anaphase/telophase (reviewed in reference 52). Notably, Lte1 is localized in the bud cortex, while other MEN components are present at the spindle pole; this spatial restriction prevents MEN activation until the daughter spindle pole has entered the bud (4). Thus, mitotic exit is not initiated until the anaphase spindle is properly oriented and nuclear segregation is achieved, constituting the “spindle position checkpoint” (reviewed in reference 23).
Cells that have a deletion of the gene encoding Net1 (net1Δ) are able to proceed through mitosis despite mutations in genes encoding components of the MEN, due to ectopic release of Cdc14 (40, 48). A Cdc14 mutant has been identified that allows cells to proceed through mitosis despite mutations in genes encoding components of the MEN. This point mutant (Pro116Leu), called TAB6 for telophase arrest bypass, binds less efficiently to Net1 and is therefore thought to be less tightly anchored to the nucleolus, eliminating the need for the MEN to transmit the signal necessary for Cdc14 release from Net1 (39). Interestingly, CDC14-TAB6 cells grow normally, suggesting that tight anchoring of Cdc14 by Net1 is not necessary for the function of the spindle position checkpoint and is not required for normal cell cycle progression. However, CDC14-TAB6 clb5Δ cells are inviable, while CDC14-TAB6 clb2Δ cells are viable, indicating that when Cdc14 localization, and presumably phosphatase activity, is not temporally regulated, the antagonizing Cdk1 activity promoted by Clb5 is required for cell cycle progression and growth (39). Importantly, Clb5 is an S-phase-promoting cyclin, while Clb2/Cdk1 regulates progression through M phase (reviewed in reference 27). It is still unclear how different Clb/Cdk complexes promote different aspects of the cell cycle. There is evidence that Clbs can functionally replace each other. However, Clb5-specific targets of Cdk1 have been identified in vitro, which include proteins involved in DNA replication (24, 25, 51), suggesting that there is substrate specificity in addition to temporal control of cyclin expression.
Cdk1-dependent phosphorylation is important for tight control of DNA replication. Cdk1 targets Cdc6, as well as components of the MCM complex and origin recognition complex (ORC), to prevent cells from replicating their DNA more than once in a single cell cycle (reviewed in reference 20). In addition to blocking rereplication, Cdk1 positively regulates initiation of DNA replication. Sld2 is an essential substrate of Cdk1 (25). When phosphorylated, Sld2 binds to Dpb11, and this complex recruits DNA polymerases to origins of replication. A subunit of DNA polymerase , Dpb2, is phosphorylated by Cdk1 as cells enter S phase. While phosphorylation of Dpb2 is not essential for DNA synthesis, mutations of the Cdk1 consensus sites are synthetic with the pol2-11 mutation in the catalytic subunit of DNA polymerase (21).
We initiated studies to understand the lethal consequences of mislocalized Cdc14 in the absence of Clb5. We found that when Cdc14 is mislocalized and Clb5-dependent Cdk1 activity is turned off, cells are unable to complete DNA replication. In addition, Cdc14-TAB6 targets specific substrates for dephosphorylation including the replication factors Sld2 and Dpb2. Moreover, CDC14-TAB6 interacts genetically with mutations that affect subunits of DNA polymerase and the S-phase checkpoint protein Mec1. We suggest that Cdc14 is sequestered in the nucleolus until anaphase not only to prevent premature mitotic exit but also to prevent excessive dephosphorylation of specific targets during S phase.
Yeast strains generated for this study are shown in Table Table1.1. Standard methods were used throughout. The GALS-CLB5 construct (RS304-GALS-CLB5) and the GAL-SIC1-Δ3P construct (46) were integrated by EcoRV digestion. All strains were in the w303 background, with the exception of TAY247 (MATα pol2-11) and TAY857 (MATa dpb2-CDKI-III::URA3) (21).
Protein extraction and immunoblotting were performed as previously described (50). Fluorescence-activated cell sorter (FACS) analysis was performed as previously described (10). Competition growth assays were performed as previously described (7).
Cells were arrested in yeast extract-peptone (YEP)-glucose containing alpha factor (final concentration of 0.1 μM) for 2.5 h (95% of cells were counted as unbudded). Cells were washed three times with cold YEP-glucose and then released into YEP-glucose.
Clb2-associated H1 kinase assays were as described previously (22, 50). In vitro phosphatase assays were carried out according to methods described in reference 17. Briefly, protein A-tagged Sld2, Orc6, and Sic1 were purified using rabbit immunoglobulin G (catalog no. 55346; ICN/Cappel) coupled to CNBr-activated Sepharose 4B (Amersham Biosciences), prepared according to the manufacturer's instructions. Beads were washed six times with lysis buffer and one time with phosphatase buffer (25 mM HEPES-NaOH, pH 7.4, 150 mM NaCl, 0.1 mg/ml bovine serum albumin). Immunoprecipitates were resuspended in phosphate buffer containing 2 mM MnCl2 and glutathione S-transferase (GST), GST-Cdc14, or λ phosphatase (New England Biolabs) in a total volume of 70 μl and incubated for 30 min at 30°C. GST-Cdc14 was prepared according to methods described in reference 15.
It has been demonstrated that Clb5 protein is required when Cdc14 is deregulated via the CDC14-TAB6 mutation (39). We also found that spores that have both NET1 and CLB5 deleted are inviable (unpublished results), suggesting that the antagonizing activity of Clb5 is required when Cdc14 is either partially delocalized via the CDC14-TAB6 mutation or fully delocalized via the NET1 deletion. These findings suggest that the lethality of CDC14-TAB6 clb5Δ cells is due to excessive dephosphorylation of mutual targets of Cdc14 and Clb5. To examine this phenomenon, we constructed CDC14-TAB6 clb5Δ strains and net1Δ clb5Δ strains that express Clb5 from a weakened galactose-inducible GAL1 promoter (GALS) (30), as overexpression of Clb5 from a full-length GAL1 promoter is toxic (9). The net1Δ clb5Δ GALS-CLB5 strains carried a centromeric plasmid for expression of RRN3 to rescue the nucleolar defects associated with net1Δ cells (39). As expected, both CDC14-TAB6 clb5Δ GALS-CLB5 cells and net1Δ clb5Δ GALS-CLB5 cells were inviable on glucose (GALS off) (Fig. (Fig.1A).1A). Additionally, net1Δ clb5Δ GALS-CLB5 cells had reduced viability on galactose compared to CDC14-TAB6 clb5Δ GALS-CLB5 cells, indicating that the effect of deletion of NET1 on Cdc14 function is more severe than that of the CDC14-TAB6 mutation.
Cdh1 and Sic1 are characterized substrates of both Cdc14 and Clb5 (11, 15, 24, 41, 46, 47, 54), and the rate of Clb5/Cdk1-mediated phosphorylation may exceed the rate of Cdc14-mediated dephosphorylation (38). Since dephosphorylation of Cdh1 by Cdc14 activates APCCdh1 for Clb2 degradation and dephosphorylation of Sic1 by Cdc14 stabilizes Sic1 (35, 47, 49), we looked at levels of Clb2 and Sic1 protein in CDC14-TAB6 clb5Δ GALS-CLB5 cells. Deregulated dephosphorylation of Cdh1 due to mislocalized Cdc14 and loss of Clb5-dependent Cdk1 activity in the CDC14-TAB6 clb5Δ GALS-CLB5 cells would be expected to reduce levels of Clb2. Surprisingly, we found that inactivation of Clb5 expression did not result in decreased Clb2 levels in CDC14-TAB6 clb5Δ GALS-CLB5 cells; indeed, Clb2 ultimately increased to a high level (Fig. (Fig.1B).1B). net1Δ clb5Δ GALS-CLB5 cells did not exhibit any decrease in Clb2 protein levels after Clb5 expression was turned off compared to wild-type controls (unpublished results). Expression of nonphosphorylatable Cdh1 from a weakened galactose-inducible promoter has been shown to have no effect on DNA replication (54). Together, these data suggest that the defects in DNA synthesis are not due to hypophosphorylation of Cdh1 in CDC14-TAB6 clb5Δ or net1Δ clb5Δ cells. Excessive dephosphorylation of Sic1 by delocalized Cdc14 would be expected to stabilize Sic1 protein; however, we found no effect on the levels or phosphorylation status of Sic1 when Cdc14 was delocalized in CDC14-TAB6 clb5Δ GALS-CLB5 cells compared to controls (Fig. (Fig.1B1B).
As an independent means to assess the possible involvement of Sic1 or Cdh1 in CDC14-TAB6 clb5Δ lethality, we sought to determine if deletion of SIC1 or CDH1 could rescue the inviability of CDC14-TAB6 clb5Δ cells in tetrad analysis; no rescue was detected (unpublished results). The inviability of sic1 cdh1 double mutants (35, 49) precludes determining if the double deletion rescues CDC14-TAB6 clb5Δ lethality.
We also looked at DNA content in CDC14-TAB6 clb5Δ GALS-CLB5 and net1Δ clb5Δ GALS-CLB5 cells by FACS analysis (Fig. (Fig.1C).1C). Expression of GALS-CLB5 resulted in a largely 2C DNA content for these strains at time zero, similar to previous results with GAL-CLB5 (14). Compared to controls, CDC14-TAB6 clb5Δ GALS-CLB5 cells exhibited enhanced accumulation of cells with sub-2C DNA content for at least 2 to 3 h after shutoff of Clb5, indicating difficulties in completing replication. net1Δ clb5Δ GALS-CLB5 cells exhibited more severe problems in DNA replication, arresting with DNA content between 1C and 2C even 4 h after shutoff of Clb5. These results indicate that clb5Δ cells that have Cdc14 delocalized, either via the CDC14-TAB6 mutation or deletion of NET1, have problems duplicating their genome properly. Given that levels of Sic1 are unaffected in clb5Δ cells that have delocalized Cdc14 (Fig. (Fig.1B),1B), it is unlikely that the defects in DNA replication are caused by delayed activation of Clb/Cdk1 complexes. This issue is addressed in more detail in the following section.
To more closely evaluate progression through S phase in cells in which Cdc14 is delocalized, we synchronized net1Δ cells, clb5Δ cells, and net1Δ clb5Δ cells and analyzed DNA content by FACS analysis. GALS-CLB5 cells, net1Δ GALS-CLB5 cells, clb5Δ GALS-CLB5 cells, and net1Δ clb5Δ GALS-CLB5 cells were arrested in G1 phase with alpha factor in the presence of glucose to repress expression of GALS-CLB5. Cells were released from alpha factor arrest into glucose-containing medium and collected at 10-min intervals. All of these strains budded at approximately the same time after release from alpha factor arrest (our unpublished results), indicating comparable release from the block. All of the strains initiated replication at about 30 min after release, as indicated by a shift from 1C to >1C (Fig. (Fig.2A).2A). Wild-type and net1Δ cells were able to complete DNA replication within about a 20-min interval, as indicated by nearly uniform 2C FACS peaks. Therefore, delocalization of Cdc14 alone does not have a significant effect on progression through S phase. clb5Δ cells required 10 to 20 min more to complete replication than wild-type cells, as previously reported (10). In striking contrast, net1Δ clb5Δ cells did not complete DNA replication even 120 min after release. In similar experiments, we found that CDC14-TAB6 clb5Δ cells were slower in S-phase progression than clb5Δ cells, although these cells eventually reached 2C DNA content (unpublished results). These results provide further evidence that delocalization of Cdc14 combined with loss of Clb5/Cdk1 activity results in defects in DNA replication, and this effect occurs within a single cell cycle of removal of Clb5.
In this experiment, Sic1 degradation was delayed by approximately 10 min in net1Δ cells, clb5Δ cells, and net1Δ clb5Δ cells compared to wild-type cells (Fig. (Fig.2B).2B). However, since Sic1 protein was not more stabilized in net1Δ clb5Δ cells than in cells carrying either of the single deletions, the defect in DNA replication cannot be explained by an inability to degrade Sic1 in a timely manner. Clb2 accumulation was delayed by about 10 min in clb5Δ cells compared to controls and was more significantly delayed in both net1Δ cells and net1Δ clb5Δ cells (Fig. (Fig.2B).2B). Clb3 protein synthesis occurred in control and clb5Δ cells between 40 and 50 min following release from alpha factor; this accumulation was delayed in cells lacking Net1, occurring between 50 and 60 min in net1Δ cells and between 60 and 70 min in net1Δ clb5Δ cells (unpublished results). The delay in Clb2 and Clb3 protein accumulation in net1Δ cells is likely due to hyperactivity of APCCdh1 resulting from Cdc14 delocalization.
In clb5Δ cells, replication is known to be driven primarily by Clb6/Cdk1 (36). It is likely that Clb6/Cdk1 is sufficient as well to drive initiation of DNA replication in clb5Δ cells in which Cdc14 is delocalized, since additional deletion of CLB6 delays replication considerably in CDC14-TAB6 clb5Δ cells, using the assay described for Fig. Fig.11 (unpublished results). Therefore, timely initiation of replication in net1Δ clb5Δ cells is probably due to Clb6-Cdk1 activation. DNA replication is completed in control cells, net1Δ cells, and clb5Δ cells prior to significant Clb2 and Clb3 protein accumulation (Fig. (Fig.2).2). In contrast, replication in net1Δ clb5Δ cells initiates but fails to complete even after Clb2 and Clb3 accumulation. Although we cannot rule out hyperactive APCCdh1 as a contributing factor, it is unlikely that failure of Clb-Cdk1 activation accounts for failure to complete replication in net1Δ clb5Δ cells. Instead, we consider it likely that dephosphorylation of additional factors by delocalized Cdc14 contributes to the replication defects observed in net1Δ clb5Δ cells.
Since CDC14-TAB6 clb5Δ cells exhibited problems with DNA replication, we tested whether cells were dependent on Mec1-dependent checkpoint pathways. Mec1 kinase is the yeast homolog of human ATM/ATR and acts as the primary effector for replication and DNA damage checkpoint pathways (34). Mec1 slows down S-phase progression to allow replication fork maintenance and to prevent premature onset of mitosis. Mec1 also performs an essential function, but deletion of SML1 alleviates the need for this role and allows for the investigation of Mec1 checkpoint functions exclusively (55). There was no notable synthetic phenotype between CDC14-TAB6 and mec1Δ sml1Δ in the presence of CLB5 (unpublished results). To determine if replication is at all impaired in CDC14-TAB6 cells, which presumably would cause an increased dependence on Mec1 function, we performed a sensitive growth competition assay. We found that CDC14-TAB6 mec1Δ sml1Δ cells exhibited a 10% selective disadvantage compared to wild-type cells, while mec1Δ sml1Δ cells alone showed only a 2% selective disadvantage (Fig. (Fig.3A).3A). The selective disadvantage of CDC14-TAB6 mec1Δ sml1Δ cells may reflect a decrease in doubling time due to slower progression through S phase or may reflect a fraction of the population dying due to unrepaired DNA damage. Importantly, CDC14-TAB6 alone showed no disadvantage in competition growth assays (Fig. (Fig.3A).3A). This suggests that CDC14-TAB6 cells achieve a normal growth rate only through reliance on an active Mec1-dependent checkpoint. Notably, this result demonstrates a selective disadvantage for CDC14-TAB6 mec1Δ sml1Δ cells even though they possess wild-type CLB5. This suggests that Cdc14 delocalization causes DNA replication difficulties or DNA damage even in the presence of Clb5/Cdk1, although loss of Clb5 function significantly potentiates this effect (Fig. (Fig.1).1). We speculate that CDC14-TAB6 cells may activate a Mec1-dependent checkpoint during DNA replication. However, deletion of MRC1, the replication checkpoint-specific effector of Mec1 (1, 44), resulted in a 17% selective disadvantage compared to wild-type cells, and an identical selective disadvantage was observed in CDC14-TAB6 mrc1 cells (unpublished results). Therefore, we cannot at present assign the need for Mec1 in a CDC14-TAB6 background specifically to a requirement for the replication checkpoint.
CDC14-TAB6 clb5Δ GALS-CLB5 cells are largely inviable on glucose (Fig. (Fig.1A),1A), but some residual growth was detected at long incubation times (Fig. (Fig.3A);3A); this residual growth was eliminated upon deletion of MEC1. Importantly, deletion of MEC1 alone had no effect on cell viability in clb5Δ cells carrying wild-type CDC14. FACS analysis revealed that MEC1 deletion caused more severe defects in DNA replication in the CDC14-TAB6 clb5Δ GALS-CLB5 strain, with cells unable to reach a 2C DNA content 4 h following shutoff of Clb5 (Fig. (Fig.3B).3B). These results are consistent with the conclusions above that CDC14-TAB6 may have an increased requirement for Mec1 due to difficulties in completing replication and that these difficulties are enhanced by deletion of CLB5. Arguing against this interpretation, though, we were unable see an effect on the downstream targets of Mec1 in CDC14-TAB6 clb5Δ cells, such as Rad53 phosphorylation or Ddc2 foci (unpublished results).
Cdc14 has a preference for phospho-Ser/Thr-Pro sites, and its characterized substrates are known targets of Cdk1. However, it is unclear whether Cdc14 efficiently dephosphorylates all substrates of Cdk1. To examine the specificity of Cdc14, we developed an assay that allows us to examine the phosphorylation status of Clb/Cdk1 substrates in vivo in cells containing wild-type CDC14 or CDC14-TAB6. To block the activity of Clb/Cdk1 specifically, undegradable Sic1 was expressed from a galactose-inducible promoter (GAL-SIC1-Δ3P) (46). This method allowed us to detect, by Western blot analysis, the phosphorylation status of putative substrates under conditions with varying Cdc14 release from the nucleolus without the complication of competing Clb/Cdk1 activity. Clb/Cdk1 activity is required for mitotic spindle formation that precedes activation of the MEN and Cdc14 release, and accordingly, we observed that green fluorescent protein (GFP)-tagged Cdc14 remained nucleolar following GAL-SIC1-Δ3P induction (unpublished results). Therefore, we anticipated that, in cells containing wild-type CDC14, specific substrates of Cdc14 would remain phosphorylated after Clb/Cdk1 inactivation. In contrast, when Cdc14 localization is deregulated via the CDC14-TAB6 mutation, specific substrates of Cdc14 should be dephosphorylated after Clb/Cdk1 inactivation. Substrates that are regulated by other phosphatases should be progressively dephosphorylated over time, independently of Cdc14 status.
As shown in Fig. Fig.4A,4A, in cells containing either wild-type CDC14 or CDC14-TAB6, Clb2-associated kinase activity decreased rapidly upon expression of GAL-SIC1-Δ3P, being strongly reduced after 15 min and virtually undetectable after 1 h of growth in galactose-containing medium. We crossed protein A-tagged or six-myc-tagged versions of potential substrates into these strains and examined their levels of phosphorylation in a time course following Clb/Cdk1 inactivation. We found that the phosphorylation status of some, but not all, of the substrates analyzed was reproducibly dependent on Cdc14 status (Fig. (Fig.4B)4B) (our unpublished results). In particular, Sld2 and Dpb2, proteins that are involved in DNA replication were less phosphorylated when Cdc14 localization was deregulated. In cells carrying the CDC14-TAB6 mutation, Sld2 was less phosphorylated at time zero and was dephosphorylated more rapidly upon inactivation of Clb/Cdk1. Although Dpb2 remained partially phosphorylated after inactivation of Clb-dependent Cdk1, in cells carrying CDC14-TAB6, Dpb2 was less phosphorylated overall. This residual Dpb2 phosphorylation was likely due to persisting phosphorylation by still active Cln-Cdk1 complexes (21). In contrast, the phosphorylation of Orc6 and Ace2 was much less or not at all dependent on Cdc14 status.
After Clb/Cdk1 inactivation, Orc6, Sld2, and Ace2 are all progressively dephosphorylated, even in CDC14 wild-type cells, suggesting that additional phosphatases can regulate these proteins, since Cdc14 is nucleolar under these conditions (unpublished results) and therefore is presumably both sequestered and inhibited by nucleolar Net1 binding (45). These results suggest that, for many substrates, phosphatases other than Cdc14 are sufficient for dephosphorylation. However, the observation that Sld2 and Dpb2 are less phosphorylated when Cdc14 is not sequestered in the nucleolus suggests that only a subset of Clb/Cdk1 substrates, including proteins involved in DNA replication, are efficiently and specifically targeted by delocalized Cdc14.
Next, we looked at the phosphorylation status of the Cdk1 substrates, Sld2, Dpb2, and Orc6, in CDC14-TAB6 clb5Δ GALS-CLB5 cells and net1Δ clb5Δ GALS-CLB5 cells (Fig. (Fig.5).5). This allowed us to focus on the antagonism between Cdc14 and Clb5 specifically, since in these cells, only Clb5-dependent Cdk1 is ablated, while all other Clb/Cdk complexes remain active. Sld2 was completely dephosphorylated 1 to 2 h following shutoff of GALS-CLB5 by growth in glucose-containing medium in CDC14-TAB6 clb5Δ GALS-CLB5 cells and net1Δ clb5Δ GALS-CLB5 cells. In contrast, Sld2 remained partially phosphorylated 4 h after shutoff of Clb5 expression in clb5Δ GALS-CLB5 cells carrying wild-type CDC14. Similar to our observation in Fig. Fig.4B,4B, Dpb2 remained partially phosphorylated following shutoff of Clb5 expression. However, Dpb2 was less phosphorylated at time zero and was dephosphorylated more rapidly in cells where Cdc14 was delocalized due to the CDC14-TAB6 mutation or deletion of NET1. The phosphorylation status of Orc6, however, was independent of Cdc14 status. Of note, Sld2, Dpb2, and Orc6 were shown to be preferred substrates of Clb5/Cdk1 in vitro (24), but our results indicate that only Sld2 and Dpb2 are specifically targeted by Cdc14. Interestingly, Sld2 was rapidly dephosphorylated despite the presence of alternate active Clb/Cdk1 complexes. This indicates that Sld2 is a specific mutual target for Clb5/Cdk1 (24, 25) and delocalized Cdc14.
We compared the phosphatase activity of Cdc14 toward Sld2, Orc6, and Sic1 in vitro to determine if Sld2 and Orc6 are direct substrates for Cdc14. Sld2 and Orc6 were phosphorylated in vivo by expression of GALS-CLB5, and Sic1 was phosphorylated and stabilized by expression of GAL-CLN1 and inactivation of CDC4 (cdc4-1). Immunoprecipitated Sld2, Orc6, and Sic1 were then subjected to in vitro phosphatase assays (Fig. (Fig.6).6). Sld2 and Orc6 were efficiently dephosphorylated by GST-Cdc14. Interestingly, Sic1 was not sensitive to treatment with GST-Cdc14 under our conditions, although its dephosphorylation has been shown previously (47). Our inability to dephosphorylate Sic1 suggests that our phosphatase assay is not highly efficient; nevertheless, we find that, in vitro, Sld2 and Orc6 are substrates for Cdc14. Our observation that Orc6 phosphorylation is not dependent on Cdc14 status in vivo (Fig. (Fig.4B4B and and5)5) suggests that Orc6 may be posttranslationally modified or associated with other proteins that prevent it from being an efficient physiological substrate for delocalized Cdc14; it may also be largely dephosphorylated by other phosphatases in vivo.
We generated conditional Clb5 strains in which Cdc14 or Cdc14-TAB6 was tagged at the C terminus with five copies of GFP (5GFP) (53) to examine the localization of Cdc14 following shutoff of Clb5. Surprisingly, CDC14-TAB6-5GFP was not synthetically lethal with CLB5 deletion (Fig. (Fig.7A).7A). We also found that CDC14-TAB6-5GFP did not bypass mutations in the MEN components TEM1 and MOB1 (unpublished results) as CDC14-TAB6 had been previously shown to do (39; unpublished results). We reasoned that the addition of 5GFP generated a hypomorphic allele of CDC14-TAB6 such that, even though Cdc14 is delocalized, its phosphatase activity is abrogated and substrates are not efficiently targeted. (The creation of a hypomorphic allele in SFP1 by the addition of an epitope tag has been previously described ).
We analyzed the DNA content of the conditional Clb5 strains in which Cdc14 or Cdc14-TAB6 was untagged or tagged with five copies of GFP by FACS analysis. In cells with wild-type Cdc14, the addition of the five-GFP tag had no effect on the cell cycle profile following shutoff of Clb5 (unpublished results). However, the DNA replication defects observed in CDC14-TAB6 clb5Δ GALS-CLB5 cells were rescued by the five-GFP tag. Additionally, we found that, in net1Δ clb5Δ GALS-CLB5 cells, DNA replication defects were partially rescued when Cdc14 carried a five-GFP tag (Fig. (Fig.7B),7B), although viability was not rescued (Fig. (Fig.7A7A).
To determine if Sld2 phosphorylation correlated with viability of clb5Δ cells with this hypomorphic CDC14 allele, we examined the phosphorylation status of Sld2 in conditional Clb5 strains in which Cdc14 or Cdc14-TAB6 was untagged or tagged with five copies of GFP. In contrast to the rapid dephosphorylation of Sld2 that we observed in CDC14-TAB6 clb5Δ GALS-CLB5 cells following shutoff of Clb5, Sld2 remained partially phosphorylated for 4 h following shutoff of Clb5 in CDC14-TAB6-5GFP clb5Δ GALS-CLB5 cells (Fig. (Fig.7C).7C). In net1Δ clb5Δ GALS-CLB5 cells, Sld2 was still completely dephosphorylated 2 h after expression of Clb5 was turned off when Cdc14 was tagged with five copies of GFP, which may account for the inability to fully rescue DNA replication defects; however, the phosphorylated form of Sld2 persisted slightly longer than in net1Δ clb5Δ GALS-CLB5 cells with untagged Cdc14.
Given that delocalized Cdc14 resulted in excessive dephosphorylation of DNA polymerase -associated replication factors, and that this correlated with defects in DNA synthesis, we tested if CDC14-TAB6 augmented the effect of a temperature-sensitive mutation in the catalytic subunit of DNA polymerase , pol2-11 (6). Colonies carrying both CDC14-TAB6 and pol2-11 exhibited reduced viability, and viable colonies grew much more slowly than wild-type or single-mutant colonies at the permissive temperature of 23°C (Fig. (Fig.8A).8A). This synthetic interaction between CDC14-TAB6 and pol2-11 was observed despite the presence of active Clb5/Cdk1 complexes. This is consistent with our results with mec1Δ CDC14-TAB6 (Fig. (Fig.3C)3C) in showing that Cdc14-TAB6 affects DNA replication even when CLB5 is wild type. Interestingly, the growth defect of pol2-11 is also exacerbated either by a temperature-sensitive mutation in SLD2 (sld2-6) (19) or by mutation of three Cdk1 sites in DPB2 (dpb2-CDKI-III) (21). Thus, our results are consistent with the idea that dephosphorylation of Sld2 and Dpb2 by Cdc14-TAB6 decreases the efficiency of the DNA polymerase complex. Accordingly, the modest viability of the CDC14-TAB6 clb5Δ GALS-CLB5 cells on glucose was eliminated when these cells carried the pol2-11 mutation (Fig. (Fig.8B8B).
CDC14-TAB6 pol2-11 also exhibited slow growth at 30°C compared to cells carrying only the pol2-11 mutation (Fig. (Fig.8C).8C). An increased dosage of DPB11, which encodes a protein that binds to Sld2 and recruits DNA polymerase to origins, has been shown to rescue the temperature sensitivity of pol2-11 (2). We found that increased dosage of DPB11 rescued the growth defect of CDC14-TAB6 pol2-11 cells similarly to cells carrying the pol2-11 mutation alone at 30°C (Fig. (Fig.8C).8C). We also tested if increased dosage of DPB11 had an effect on the phenotypes observed in CDC14-TAB6 clb5Δ GALS-CLB5 cells and net1Δ clb5Δ GALS-CLB5 cells. A 2μm plasmid carrying DPB11 was introduced into CDC14-TAB6 clb5Δ GALS-CLB5 and net1Δ clb5Δ GALS-CLB5 strains, and DNA content was measured by FACS analysis following shutoff of conditional Clb5 (unpublished results). We observed a small yet reproducible effect on DNA replication, with more CDC14-TAB6 clb5Δ GALS-CLB5 and net1Δ clb5Δ GALS-CLB5 cells approaching a 2C DNA content and a shift of the 2C peak to the right in the presence of extra copies of DPB11. Increased DPB11, however, had no effect on the viability of CDC14-TAB6 clb5Δ or net1Δ clb5Δ cells (unpublished results).
We also looked for a genetic interaction between CDC14-TAB6 and dpb2-CDKI-III cells. Colonies carrying the double mutation were viable and grew normally (unpublished results). However, we found that introduction of the dpb2-CDKI-III mutation reduced the viability of CDC14-TAB6 clb5Δ cells (Fig. (Fig.8D).8D). As described above, Dpb2 was less phosphorylated in CDC14-TAB6 clb5Δ GALS-CLB5 cells than in clb5Δ GALS-CLB5 cells after shutoff of Clb5 but still remained partially phosphorylated. This genetic interaction suggests that complete abrogation of Dpb2 phosphorylation by removal of the Cdk1 consensus sites together with excessive dephosphorylation of Sld2 (and perhaps other DNA replication proteins) in CDC14-TAB6 clb5Δ GALS-CLB5 cells sufficiently impairs DNA synthesis to strongly reduce viability.
In our study of the antagonism between Clb5 and Cdc14, we have identified some unexpected targets for deregulated Cdc14. The Cdc14 substrates Cdh1, Sic1, and Swi5, when dephosphorylated, lower the activity of Cdk1 to promote mitotic exit (15, 47). Given that Cdh1 is a preferred substrate of Clb5 (24) and that Clb5/Cdk1 is also capable of phosphorylating Sic1 and Swi5 (11, 24, 41, 46), mislocalized Cdc14 in the absence of Clb5/Cdk1 activity could be predicted to result in hypophosphorylated Cdh1, Sic1, and Swi5. This should lead to excessive degradation of Clb2 and stabilization of Sic1, which would cause an arrest in the G1 phase of the cell cycle. Instead, we found that, in clb5Δ cells in which Cdc14 was delocalized by the TAB6 mutation or by deletion of NET1, S phase was initiated, Clb2 protein accumulated, and Sic1 protein was not stabilized (Fig. (Fig.1).1). Moreover, in synchronized net1Δ clb5Δ cells, Sic1 degradation occurred in a timely manner, and while Clb protein accumulation was somewhat delayed, this delay cannot account for the DNA replication defects (Fig. (Fig.2;2; see above). These results show that Sic1 and Cdh1 can still be inactivated when Cdc14 is not resequestered in the nucleolus during G1, despite lower Cdk activity due to deletion of CLB5. These results strongly suggest that targets of Cdc14 other than Cdh1, Sic1, and Swi5 exist, which contribute to efficient DNA replication when phosphorylated. We have found that replication factors Sld2 and Dpb2 are substrates for deregulated Cdc14, and their dephosphorylation may contribute to the inability of these cells to complete DNA replication.
We also sought to determine if Cdc14 dephosphorylates the bulk of Cdk1 substrates to act as a general antagonist of Cdk1 or, alternatively, if Cdc14 has more-specific targets by looking at the phosphorylation status of known Cdk1 substrates under conditions of different Cdc14 activity/localization. To remove competing Clb-dependent Cdk1 activity, we expressed undegradable Sic1 from a GAL1 promoter. An alternative method for Cdk1 inactivation is to use the cdc28-as1 allele, which selectively renders Cdk1 sensitive to kinase inhibitors (5). Although the cdc28-as1 technique is both rapid and reversible, Cdc28-as1 is compromised in the absence of inhibitor, having a longer doubling time than cells with wild-type CDC28 (5). In addition, our use of an inducible undegradable Sic1 expression system allowed us to turn off only Clb/Cdk1, which permitted us to examine Cdc14-Clb antagonism specifically. In our system, only a subset of substrates is dephosphorylated more rapidly when Cdc14 is mislocalized (Fig. (Fig.4).4). It will be of interest to ascertain if there are determinants that contribute to Cdc14 substrate specificity. Cdk1 phosphorylation is dependent on the presence of a consensus site (S/TPXK/R) in the substrate, while phosphorylation by Clb5/Cdk1 specifically is enhanced by the presence of the hydrophobic patch of Clb5 and an RXL motif in the substrate (8, 24). Cells carrying both CDC14-TAB6 and a hydrophobic patch mutant of Clb5 (CLB5-hpm) were slow growing but still viable (unpublished results). However, the mutation of the Clb5 hydrophobic patch does not completely eliminate binding to or phosphorylation of Clb5 substrates (51), which likely accounts for the viability of CDC14-TAB6 CLB5-hpm cells. Likewise, Cdc14 may recognize features on its targets in addition to its preferred pSer/pThr-Pro motifs (13). Our results also corroborate the idea that cyclins exhibit substrate specificity. We show that Sld2 is jointly regulated by delocalized Cdc14 and Clb5-dependent Cdk1, specifically, and can be converted to the hypophosphorylated form even in the presence of other active Clb/Cdk1 and Cln/Cdk1 complexes (Fig. (Fig.5).5). Another substrate that is mutually regulated by Cdc14 and a specific cyclin-Cdk complex was previously identified: Cdc14 and Clb6-Cdk1 control the phosphorylation status of Swi6, a subunit of the SBF and MBF transcription factors to regulate Swi6 localization (12). This gives further credence to the idea that distinct cyclin-Cdk complexes exhibit functional specificity.
Our observation that many Cdk1 substrates are rapidly dephosphorylated following shutoff of Clb/Cdk1, even when Cdc14 is sequestered in the nucleolus (Fig. (Fig.4),4), suggests that phosphatases other than Cdc14 are sufficient to dephosphorylate many Cdk1 substrates. This supports our hypothesis that Cdc14-specific targets remain to be identified, since Cdc14 and the proteins that contribute to its release from the nucleolus are essential for mitotic exit and Cdh1 and Sic1 dephosphorylation does not account for this requirement (50). (While we propose Sld2 and Dbp2 as candidate Cdc14 targets here, dephosphorylation of these proteins is unlikely to contribute to mitotic exit.) There also remains a possibility that there are specific targets of Cdc14 that are able to shuttle in and out of the nucleolus, making the localization of Cdc14 irrelevant. This, however, is unlikely, as Net1 has been shown to be a competitive inhibitor of Cdc14 activity (45). We tried to address this by testing if Cdk1 substrates remained phosphorylated in cells carrying a temperature-sensitive allele of CDC14 (cdc14-1) following expression of GAL-SIC1-Δ3P. However, we observed that elevated temperatures affected substrate phosphorylation and made the results difficult to interpret (unpublished results).
In the course of this study, we identified CDC14-TAB6-5GFP as a hypomorphic allele of CDC14-TAB6. CDC14-TAB6-5GFP did not show synthetic lethality with clb5Δ, nor did it bypass mutations in components of the mitotic exit network despite being fully functional for viability. The five-GFP tag appears to affect the catalytic activity of Cdc14, since Sld2 remains partially phosphorylated in CDC14-TAB6-5GFP clb5Δ cells (Fig. (Fig.7C).7C). Thus, a decrease in Cdc14 activity can rescue the DNA replication defects associated with clb5Δ and deregulated Cdc14, and the phosphorylation status of Sld2 correlates with the ability of cells to progress through S phase.
We have found that when Cdc14 is not spatially and temporally sequestered, progression through the S phase of the cell cycle is compromised. We have provided evidence that, in the absence of competing Clb5/Cdk1 activity, mislocalized Cdc14 prevents cells from efficiently completing DNA replication. This block in DNA synthesis correlates with the dephosphorylation of Sld2 and Dpb2. Importantly, phosphorylation of Sld2 as cells enter S phase promotes complex formation with Dpb11, which then recruits DNA polymerase to origins of replication (19, 26). Similarly, Dpb2 is phosphorylated in a cell cycle-regulated manner, and although this modification is not essential, it may facilitate formation of the DNA polymerase holoenzyme (21).
Cdk1 has been shown to phosphorylate origin-associated replication factors (the ORC, members of the Mcm complex, Cdc6) to prevent cells from replicating their DNA more than once in a single cell cycle (32). However, origins can be reloaded in cells in which cdc14-1 is inactivated at 37°C and Clb/Cdk1 activity is inhibited by expressing nondegradable Sic1, suggesting that phosphatases other than Cdc14 may dephosphorylate these origin-associated proteins (33). Consistent with this, we found that delocalized Cdc14 does not seem to target Orc6, as its phosphorylation status is independent of Cdc14 localization. In addition, no predisposition of CDC14-TAB6 clb5Δ or net1Δ clb5Δ cells to rereplicate their DNA was observed. Moreover, the ability of increased dosage of DPB11 to suppress the DNA replication defects of CDC14-TAB6 clb5Δ cells and the synthetic phenotype caused by TAB6 pol2-11 double mutation reinforces the idea that mislocalized Cdc14 affects proteins that promote DNA replication at an origin-loading step.
Sequestration of Cdc14 in the nucleolus has been shown to be a key element in a nuclear positioning surveillance mechanism, ensuring that mitotic exit is restrained until the daughter spindle pole body has entered the bud (4). Our results strongly suggest that an additional reason for sequestration of Cdc14 in the nucleolus for most of the cell cycle is to maintain appropriate balance between kinase and phosphatase activity during S phase.
We thank H. Araki, R. Deshaies, A. Toh-e, and C. Wittenberg for providing reagents. We thank V. Archambault, V. Campbell, and L. Schroeder for their contribution to this work and B. Drapkin, B. D. Lindenbach, and Y. Lu for comments on the manuscript.
This work was supported by a postdoctoral fellowship (PF-06-017-01-CCG) from the American Cancer Society, NIH training grant T32 CA009673-29, and The Rockefeller University's Women and Science Fellowship to J.B. and PHS grant GM047238 to F.R.C.
Published ahead of print on 20 November 2006.