The mitotic exit network (MEN) is a spindle pole body (SPB)–associated, GTPase-driven signaling cascade that controls mitotic exit. The inhibitory Bfa1–Bub2 GTPase-activating protein (GAP) only associates with the daughter SPB (dSPB), raising the question as to how the MEN is regulated on the mother SPB (mSPB). Here, we show mutual regulation of cyclin-dependent kinase 1 (Cdk1) and the MEN. In early anaphase Cdk1 becomes recruited to the mSPB depending on the activity of the MEN kinase Cdc15. Conversely, Cdk1 negatively regulates binding of Cdc15 to the mSPB. In addition, Cdk1 phosphorylates the Mob1 protein to inhibit the activity of Dbf2–Mob1 kinase that regulates Cdc14 phosphatase. Our data revise the understanding of the spatial regulation of the MEN. Although MEN activity in the daughter cells is controlled by Bfa1–Bub2, Cdk1 inhibits MEN activity at the mSPB. Consistent with this model, only triple mutants that lack BUB2 and the Cdk1 phosphorylation sites in Mob1 and Cdc15 show mitotic exit defects.
Budding yeast Cdc14 belongs to a conserved class of dual-specificity protein phosphatases. During interphase and early mitosis yeast Cdc14 is kept inactive by entrapment in the nucleolus as a consequence of its association with Net1/Cfi1 (Shou et al., 1999; Visintin et al., 1999). However, upon anaphase onset Cdc14 is released from the nucleolus. The active Cdc14 then resides in the nucleoplasm, the cytoplasm, at the budding yeast centrosome (the spindle pole body [SPB]), and at the site of cytokinesis (Pereira et al., 2002; Yoshida et al., 2002; Stegmeier and Amon, 2004) where it dephosphorylates proteins that have previously been phosphorylated by cyclin-dependent kinase 1 (Cdk1). In addition, Cdc14 promotes inactivation of the Cdk1–Clb2 complex at the end of mitosis, thereby promoting the transition from mitosis to G1 phase of the cell cycle (Visintin et al., 1998).
Two pathways regulate Cdc14 localization. The Cdc14 early release pathway (FEAR) triggers a transient and partial release of Cdc14 at the beginning of anaphase (Pereira et al., 2002; Stegmeier et al., 2002). This short burst of Cdc14 activity is sufficient to promote the segregation of the rDNA locus, the targeting of the INCENP homologue Sli15 to the mitotic spindle, changes in microtubule (MT) dynamics, and spindle midzone assembly (Pereira and Schiebel, 2003; D’Amours et al., 2004; Lavoie et al., 2004; Sullivan et al., 2004; Higuchi and Uhlmann, 2005; Khmelinskii et al., 2007; Woodbury and Morgan, 2007). However, because the FEAR-activated Cdc14 is only transiently active and Cdk1–Clb2 activity is still high in early anaphase, the FEAR pathway does not induce mitotic exit.
Mitotic exit and cytokinesis require full activation of Cdc14 by the mitotic exit network (MEN), a GTPase-driven signaling cascade that is associated with the SPB (Shirayama et al., 1994; Luca and Winey, 1998; Cenamor et al., 1999; Gruneberg et al., 2000; Xu et al., 2000; Menssen et al., 2001; Pereira and Schiebel, 2001; Stegmeier and Amon, 2004). One of the most upstream MEN components is the Ras-like GTPase Tem1 that is controlled by the putative guanine nucleotide exchange factor Lte1 and the GTPase-activating protein (GAP) complex Bfa1–Bub2 (Shirayama et al., 1994; Bardin et al., 2000; Pereira et al., 2000; Geymonat et al., 2002). Tem1 interacts with the Pak-like kinase Cdc15 (Asakawa et al., 2001), which in turn activates the Dbf2–Mob1 kinase complex via phosphorylation of the kinase subunit Dbf2 (Mah et al., 2001). One function of the Dbf2–Mob1 complex is to phosphorylate Cdc14 at sites adjacent to its nuclear localization sequence, thereby retaining Cdc14 in the cytoplasm (Mohl et al., 2009).
The MEN activation scheme closely follows binding of MEN components to SPBs; SPB binding of Cdc15 requires Tem1 and Dbf2–Mob1 only associates with SPBs when Tem1 and Cdc15 are functional (Visintin and Amon, 2001). The close correlation between MEN activation and SPB localization indicates that MEN regulation occurs at SPBs. Consistently, mutants of the SPB component NUD1 are defective for MEN signaling (Gruneberg et al., 2000).
Growth by budding generates an inherent polarity in the yeast cell. This polarity is reflected in the two SPBs as they are functionally and biochemically distinct. The preexisting, older SPB is inherited by the daughter cell, the bud (Pereira et al., 2001). In addition, MEN proteins bind differently to the mSPB and dSPB. During an unperturbed cell cycle the inhibitory Bfa1–Bub2 GAP complex localizes preferentially at the dSPB, where it inhibits the MEN, until Cdc5 polo-like kinase inactivates the Bfa1–Bub2 complex in late anaphase (Bardin et al., 2000; Pereira et al., 2000; Hu et al., 2001; Caydasi and Pereira, 2009). Throughout the majority of anaphase Tem1 at the dSPB resides in a complex with the Bfa1–Bub2 GAP and therefore is probably in its inactive GDP-bound form (Pereira et al., 2000). The mSPB recruits only a small fraction of Tem1 (Molk et al., 2004). Cdc15 was reported to bind either first to the mSPB, the dSPB, or with equal timing to both SPBs (Cenamor et al., 1999; Xu et al., 2000; Menssen et al., 2001; Molk et al., 2004). These conflicting data are most easily explained by varying levels of the overexpressed Cdc15 protein (Cenamor et al., 1999). Finally, the Mob1 protein associates in early anaphase with the mSPB, well before the protein is recruited to the dSPB (Luca et al., 2001).
The FEAR pathway, the MEN, and Cdk1 are all linked by complex interdependencies. FEAR activation promotes phosphorylation of Net1 by Cdk1–Clb2 in early anaphase, which is a prerequisite for the FEAR-dependent release of Cdc14 (Azzam et al., 2004; Queralt et al., 2006). In addition, Cdk1 phosphorylation activates the FEAR pathway component Spo12 (Tomson et al., 2009). In contrast, Cdk1 negatively regulates the function of the MEN component Cdc15 (Jaspersen and Morgan, 2000). FEAR-released Cdc14 eventually dephosphorylates Cdc15 in early anaphase, which, in a positive feedback loop, further activates Cdc14 (Jaspersen and Morgan, 2000; Menssen et al., 2001; Pereira et al., 2002; Stegmeier et al., 2002). It is therefore puzzling that a mutant version of Cdc15, which is no longer regulated by Cdk1, does not show mitotic exit defects (Jaspersen and Morgan, 2000).
Here, we describe mutual regulation between Cdk1 and MEN proteins. Activation of the MEN component Cdc15 in early anaphase recruits Cdk1 kinase to the mSPB, whereas Cdk1 reduces mSPB binding of Cdc15. In addition, Cdk1 inhibits the activity of the Mob1–Dbf2 kinase complex through Mob1 phosphorylation. We suggest that the Cdk1–Cdc15/Mob1 regulation loop restrains full activation of the MEN at the mSPB in early anaphase. The MEN at the dSPB is inhibited by the Bfa1–Bub2 GAP complex until Cdc5 polo kinase phosphorylates Bfa1 (Hu et al., 2001).
Cdk1 binds to the mSPB in anaphase
Previously, we have shown that Cdk1 binds to the dSPB in G1/S phase of the cell cycle from where it translocates to the plus end of cytoplasmic MTs (Maekawa et al., 2003; Maekawa and Schiebel, 2004; Fig. 1 A, −8 and −6, yellow arrows). We now describe that Cdk1 also changes SPB localization in anaphase. In time-lapse experiments of early anaphase cells, Cdk1-GFP was detected at the SPB in the mother cell but not at the SPB in the bud (Fig. 1 A, 4–9, white arrows). In some cells, a second, dot-like, Cdk1-GFP signal appeared later at the SPB in the daughter cell (Fig. 1 A, 11–13, white arrows). With mitotic exit Cdk1 was no longer present at either of the two SPBs (Fig. 1 A, 14–24). This result was confirmed using α-factor–synchronized CDK1-3GFP SPC42-eqFP611 cells in which the length of the mitotic spindle was used to monitor progression through anaphase (Fig. 1 B). In cells with a short metaphase spindle Cdk1 was not enriched at SPBs. Cdk1-GFP was found in the majority of early-mid anaphase cells (Fig. 1 B, >70%; 3–8-µm spindle length) in a polar manner at the mSPB. In late anaphase cells (8–10 µm) Cdk1-3GFP dissociated from SPBs. Thus, in early anaphase Cdk1 binds preferentially to the mother spindle pole. This recruitment suggests that Cdk1 is likely to be performing a novel, uncharacterized function at this location.
As yeast kinetochores cluster close to SPBs for the majority of the cell cycle (Jin et al., 1998), the pole signal may arise from the binding of Cdk1 to SPBs, kinetochores, or both. Conditional lethal ndc10-1 cells lack functional kinetochores (Goh and Kilmartin, 1993). A persistence of the pole signal in ndc10-1 cells therefore suggests that the Cdk1 signal arises from its recruitment to SPBs. In ndc10-1 cells Cdk1-GFP associated with the pole marker Spc42-RFP (Fig. 1 C), indicating that it is at SPBs in anaphase.
We next asked whether Cdk1 binding to SPBs requires one of the four mitotic cyclins, Clb1–Clb4. Cdk1 localization at SPBs was not grossly perturbed in clb1Δ, clb2Δ, clb3Δ, or clb4Δ cells. Neither was it altered in the clb3Δ clb4Δ double mutant cells (unpublished data). However, in clb1Δ Gal1-CLB2 cells, from which Clb2 had been depleted (resulting in cell cycle arrest in anaphase; Azzam et al., 2004), Cdk1-GFP failed to bind to the mSPB and dSPB (Fig. 1 D). It is therefore likely that it is Cdk1 in a complex with either Clb1 or Clb2 that associates with anaphase SPBs. This result is consistent with the reported SPB localization of Clb2 in anaphase cells (Hood et al., 2001; Bailly et al., 2003) and supports the notion that the functional Cdk1–Clb2 kinase complex binds to the mSPB in anaphase.
The MEN regulates binding of Cdk1 to SPBs
The MEN is an SPB-associated signaling cascade that becomes active in anaphase (Stegmeier and Amon, 2004). MEN activation could promote the recruitment of Cdk1 to SPBs. To test this hypothesis, we analyzed the localization of Cdk1 in cells lacking either the KAR9 or DYN1 genes. In kar9Δ or dyn1Δ cells, ∼10–20% of anaphase spindles become misaligned in the mother cell body (Li et al., 1993; Miller and Rose, 1998). The MEN of kar9Δ and dyn1Δ cells is inactive when the anaphase spindle is mispositioned, but active when the spindle is correctly aligned along the mother–bud axis (Bardin et al., 2000; Pereira et al., 2000). In kar9Δ cells with a correctly aligned spindle, Cdk1-3GFP bound in early anaphase to the mSPB (Fig. 2 A, kar9Δ, enlargement 4; and Fig. 2 B) in the same way as it did in wild-type cells (Fig. 2 A, KAR9, enlargement 1; and Fig. 2 B). In contrast, Cdk1 was not detectable at either of the two SPBs in kar9Δ cells with a misaligned spindle (Fig. 2 A, kar9Δ, asterisks and enlargements 2 and 3; and Fig. 2 B). In these cells Cdk1 was distributed throughout the nucleus. Similar data were obtained for dyn1Δ cells (Fig. S1, A and B). Thus, MEN activity may regulate the SPB recruitment of Cdk1.
To directly confirm a role of the MEN in the localization of Cdk1 to the anaphase SPB, we inactivated the polo-like kinase Cdc5 and Tem1 that are both essential for MEN activity (Shirayama et al., 1998; Shou et al., 1999; Hu et al., 2001). We used Gal1-CDC5 and Gal1-UPL-TEM1 cells (TEM1 fused to a degron element) for this experiment. Gal1-expressed Cdc5 and UPL-Tem1 are both rapidly depleted from cells after their expression is repressed by the addition of glucose to the growth medium. Cells then arrested in late anaphase as a consequence of the lack of MEN activity (Shirayama et al., 1998; Shou et al., 1999). In such repressed Gal1-CDC5 and Gal1-UPL-TEM1 cells, Cdk1 was no longer detectable at SPBs at any stage of anaphase (Fig. 2 C), indicating that an active MEN is required for the binding of Cdk1 to SPBs.
We next addressed how inactivation of the MEN inhibitor Bfa1 affects the recruitment of Cdk1 to SPBs. In contrast to wild-type cells, which showed a polar SPB binding of Cdk1 (Fig. 2 D, enlargements 1 and 2), Cdk1 of bfa1Δ cells bound with equal intensity and timing to both SPBs in early-mid anaphase (Fig. 2 D, enlargements 3 and 4; and Fig. 2 E). Thus, in wild-type cells the Bfa1–Bub2 GAP complex inhibits the binding of Cdk1 to the dSPB.
The FEAR pathway regulates release of Cdc14 in early anaphase and could be essential for the binding of Cdk1 to the mSPB (Stegmeier et al., 2002). To test this possibility, we analyzed binding of Cdk1-GFP to the mSPB in FEAR-defective spo12Δ or slk19Δ cells. In addition, we compared timing of binding of Cdk1 to the mSPB of spo12Δ cells with the release of Cdc14-3mCherry from the nucleolus as an indication for full activation of the MEN. Time-lapse data showed that Cdk1 bound in mid-anaphase to the mSPB of FEAR-defective spo12Δ cells (Fig. S1 C) and slk19Δ cells (not depicted). In spo12Δ cells, Cdk1-GFP bound to the mSPB 6–8 min (n = 5) before Cdc14-3mCherry was released from the nucleolus (Fig. S1 C). These data suggest that the FEAR network is not important for Cdk1 binding to the mSPB.
SPB localization of Cdk1 requires Cdc15 kinase activity but persists in DBF2 and MOB1 mutants
Our analysis already demonstrated the essential role of the GTPase Tem1 in recruiting Cdk1 to SPBs (Fig. 2 C). We now asked whether the MEN kinase Cdc15 was also required to promote the binding of Cdk1 to the dSPB. The role of CDC15 was first analyzed in cells expressing the cdc15-as1 allele that can be inhibited by the ATP analogue “PP1 analog 8” (D’Aquino et al., 2005). Inhibition of cdc15-as1 kinase activity strongly reduced the efficiency of Cdk1-GFP binding to the mSPB in any stage of anaphase (Fig. 3, A and B). This failure in Cdk1 recruitment was not caused by a deficiency in the binding of the cdc15-as1 protein to SPBs. On the contrary, cdc15-as1-GFP accumulated to a much higher level at SPBs than Cdc15-GFP (Fig. 3 C), even though the cellular levels of Cdc15 and cdc15-as1 were identical as shown by immunoblot (Fig. 3 D). The dependency of SPB localization upon Cdc15 activity was confirmed in cdc15-1 cells (Fig. S2, A and B).
In contrast to cdc15 cells, Cdk1-GFP still associated with the mSPB of dbf2-2 and mob1-67 cells (Fig. 3, E and F). This could mean that it is only the most upstream branch of the MEN, consisting of Tem1 and Cdc15, that directs Cdk1 to SPBs. Alternatively, Cdk1 could bind to the inactive dbf2-2 and mob1-67 proteins at SPBs. Analysis of the localization of GFP-tagged mob1-67 and dbf2-2 showed that the proteins were no longer at SPBs when cells were incubated at the restrictive temperature (Fig. S2, C and D). In addition, Cdk1 was still targeted to the mSPB in dbf2-2 dbf20Δ cells (not depicted), indicating that the Dbf2 paralogue Dbf20 could not substitute for this function of Dbf2. In conclusion, the Dbf2–Mob1 complex is not essential for the mSPB binding of Cdk1.
Mutual regulation of Cdc15 and Cdk1 at the mSPB
What function does Cdk1 execute at anaphase SPBs? Cdk1 has been shown to negatively regulate the MEN by phosphorylating the MEN component Cdc15 at seven sites (Jaspersen and Morgan, 2000; Stegmeier et al., 2002). The function of this modification has remained obscure because CDC15-7A cells do not show growth or mitotic exit defects and the specific activity of the Cdc15 kinase is not altered by Cdk1 phosphorylation (Jaspersen and Morgan, 2000; Xu et al., 2000). We wondered whether phosphorylation by Cdk1 may be regulating the binding of Cdc15 to the mSPB. Conflicting data on the cell cycle–dependent binding of overexpressed Cdc15 to SPBs have been reported (Cenamor et al., 1999; Xu et al., 2000; Menssen et al., 2001), and the analysis of the localization of nonoverexpressed Cdc15-GFP was only qualitative (Molk et al., 2004). We therefore quantified SPB binding of the functional CDC15-GFP gene product that was expressed as the sole CDC15 gene copy at the native CDC15 locus from its native promoter in wild-type cells and in td-cdc14 degron cells. td-cdc14 degron cells lack Cdc14, the phosphatase that dephosphorylates Cdc15 in early anaphase, and have high Cdk1–Clb2 activity. Cdc15 therefore remains hyperphosphorylated in anaphase of td-cdc14 cells (Jaspersen and Morgan, 2000; Pereira and Schiebel, 2003).
In synchronized CDC15-GFP cells, Cdc15 showed a weak association with both the mSPB and dSPB in early anaphase (Fig. 4 B, yellow bar, 4–6 µm) and, as reported (Visintin and Amon, 2001), this association was enhanced in mid- to late anaphase (Fig. 4, A and B, blue bar, >6 µm). Importantly, Cdk1 regulated SPB binding of Cdc15 as shown in td-cdc14 degron cells. In td-cdc14 cells the hyperphosphorylated Cdc15-GFP did not bind strongly to SPBs, even after the spindle had extended to lengths exceeding 6 µm (Fig. 4, A and B, red bars). Cdk1 phosphorylation of Cdc15 may therefore inhibit SPB association. To confirm this, we quantified SPB binding of the nonphosphorylated Cdc15-7A-GFP in td-cdc14 cells. Cdc15-7A-GFP bound strongly to SPBs nearly independently of anaphase progression (Fig. 4 C, red bars).
The data above suggest that phosphorylation by Cdk1 reduces the binding of Cdc15 to the mSPB. However, Cdc15 activity is required to recruit Cdk1 to SPBs (Fig. 3 A). This mutual regulation between Cdc15 and Cdk1 should limit accumulation of both cell cycle regulators at the mSPB. Disrupting the negative regulation between Cdk1 and Cdc15 should result in the hyperaccumulation of proteins at the mSPB. This is the case for Cdc15, as cdc15-as1-GFP accumulates to higher levels at SPBs than Cdc15 (Fig. 3 C). To test this further, we analyzed SPB localization of Cdk1-GFP in CDC15-7A cells. Cdk1-GFP at the mSPB was clearly increased in CDC15-7A cells compared with CDC15 wild-type cells (Fig. 4, D and E). Cdk1-GFP protein levels were the same in both cell types (Fig. 4 F). Taken together, our data suggest that Cdk1 negatively regulates binding of Cdc15 to the mSPB, whereas Cdk1 requires the activity of Cdc15 to bind to SPBs. This interdependency restricts binding of both Cdk1 and Cdc15 to the mSPB in early anaphase.
Phosphorylation of Mob1 by Cdk1
CDC15-7A cells do not have a mitotic exit defect (Jaspersen and Morgan, 2000). Cdk1 may regulate additional MEN proteins. Mob1 is a good candidate because it is phosphorylated by Cdk1 in vitro (Holt et al., 2007). Analysis of the amino acid sequence of Mob1 identified two full Cdk1 consensus sites (Fig. 5 A; S/T-P-x-K/R). Mutagenesis of these two sites, S36 and T85, to alanine reduced the phosphorylation of Mob1-2A by Cdk1–Clb2 to 11% of wild-type Mob1 (Fig. 5 B). Mutagenesis of the five remaining S/T-P sites (Mob1-7A) further diminished the ability of Cdk1–Clb2 to phosphorylate Mob1 (Fig. 5 B). Thus, Cdk1–Clb2 phosphorylates Mob1 in vitro predominantly at two sites, S36 and T85.
We next asked whether Mob1 is phosphorylated in vivo. Gal1-CDC20 MOB1-6HA cells were arrested in metaphase by depletion of Cdc20. Mob1 appeared as multiple bands in SDS-PAGE (Fig. 5 C). Treatment of Mob1 with lambda phosphatase caused the collapse of the Mob1 bands to a single faster migrating species. Adding the inhibitor EDTA prevented this collapse, suggesting that Mob1 is a phosphoprotein in vivo.
α-Factor–synchronized cells were analyzed in order to monitor changes in Mob1 phosphorylation that accompany cell cycle progression. In G1 phase, Mob1 migrated as a single band corresponding in mobility to dephosphorylated Mob1 protein (Fig. 5 D). 20–30 min after release of the G1 cell cycle block, as soon as cells started to develop a bud, and well before the start of anaphase (Fig. 5 D, 50 min), the upshift of the protein band indicated that a fraction of Mob1 became phosphorylated (Fig. 5 D, 20 min). Dephosphorylation of Mob1 coincided with mitotic exit when the levels of Clb2 decreased and the Sic1 protein accumulated (Fig. 5 D, 60–70 min). In contrast, Mob1-2A showed strongly reduced phosphorylation at all stages of cell cycle progression (Fig. 5 D, Mob1-2A), indicating that S36 and T85 are the two major in vivo Cdk1 phosphorylation sites.
Cdk1 inhibits activity of the Dbf2–Mob1 kinase complex
A genetic approach was used to address whether Cdk1 phosphorylation activates or inhibits Mob1 function. Conditional lethal dbf2-2 cells were combined with MOB1, MOB1-S36A, MOB1-T85A, MOB1-2A, or MOB1-7A alleles. All strains grew equally well at 23°C (Fig. 6 A). However, at 37°C MOB1-T85A, MOB1-2A, and MOB1-7A were clearly fitter than wild-type MOB1 cells, indicating that nonphosphorylated mutant Mob1 proteins possessed MEN-activating function. The overall impact of the mutations was insensitive to the additional deletion of DBF20 with the exception that dbf2-2 dbf20Δ cells were more temperature sensitive than dbf2-2 cells (Fig. S3 A). This indicates that the suppression of the dbf2-2 growth defect by the MOB1-2A allele does not require DBF20 function. Moreover, MOB1-2A did not suppress the growth defect of tem1-3, cdc15-1, and cdc14-1 cells (unpublished data). Thus, phosphorylation of Mob1 by Cdk1 inhibits the MEN at the level of the Dbf2–Mob1 complex.
We performed coimmunoprecipitation experiments to show that Mob1 phosphorylation by Cdk1 does not regulate complex formation between Mob1 and Dbf2 (Fig. S3, B and C). Instead, Cdk1 controls the kinase activity of the Dbf2–Mob1 complex as shown by two approaches. First, we reconstituted the MEN pathway using the purified Cdc15 and Dbf2–Mob1 proteins (Mah et al., 2001; Geymonat et al., 2007). Cdk1–Clb2 was purified from yeast cells (Loog and Morgan, 2005). As a substrate for Dbf2–Mob1 kinase, we used a C-terminal fragment of Cdc14 (C-Cdc14) that was readily phosphorylated by Dbf2–Mob1 but not by Cdk1–Clb2 kinase (Fig. 6 B; unpublished data; Mohl et al., 2009). The purified Dbf2–Mob1 complex bound to glutathione beads was preincubated with Cdk1–Clb2 and Cdc15 as outlined in Fig. 6 C. The glutathione beads were than washed and incubated with γ-[32P]ATP and C-Cdc14 to determine the Dbf2–Mob1 kinase activity. This activity was corrected for the amount of Dbf2–Mob1 in the kinase reaction to generate the read-out of specific activity (Fig. 6 C, top graph). The Dbf2–Mob1 kinase complex was only active when preincubated with Cdc15 (Fig. 6 C, compare lanes 1, 2, and 6; Mah et al., 2001). Importantly, incubation of Dbf2–Mob1 with Cdk1–Clb2 led to a roughly twofold reduction in the specific activity of the Dbf2–Mob1 kinase complex (Fig. 6 C, compare lanes 2 with 4, and 5 with 6). The inhibition of Dbf2–Mob1 kinase activity by Cdk1–Clb2 was seen irrespective of whether Cdc15 was first incubated with Dbf2–Mob1 followed by Cdk1–Clb2 or vice versa (Fig. 6 C, lanes 2 and 4 vs. 5 and 6). Thus, Cdk1–Clb2 inhibits the activity of the Dbf2–Mob1 kinase complex.
Second, to verify that Cdk1–Clb2 inhibits the MEN at the level of Cdc15 and Mob1, we overexpressed a stable version of CLB2 (CLB2-ΔDB) in synchronized yeast cells and allowed cell cycle progression into anaphase. Immunoprecipitation of Dbf2 showed that the specific kinase activity of the Dbf2–Mob1 complex was low in the presence of high Cdk1–Clb2 activity (Fig. 6 D; Stegmeier et al., 2002). Importantly, the specific kinase activity of Dbf2–Mob1 was elevated between 4- and 10-fold in MOB1-2A, CDC15-7A, and MOB1-2A CDC15-7A cells. These data indicate that Cdk1 kinase inhibits cellular Dbf2–Mob1 kinase activity via phosphorylation of both Cdc15 and Mob1.
Cdc14 dephosphorylates Mob1
Abrupt dephosphorylation of Mob1 coincided with mitotic exit (Fig. 5 D). The abruptness of this transition suggests that the phosphatase Cdc14 may be driving this dephosphorylation. Three lines of evidence support this possibility. Mob1 was dephosphorylated by recombinant wild-type Cdc14 but not by the catalytically inactive Cdc14-C283S mutant protein (Fig. 7 A). Premature activation of CDC14 by overexpression from the Gal1 promoter led to Mob1 dephosphorylation in metaphase-arrested cells (Fig. 7 B). Finally, synchronized td-cdc14 cells showed that Mob1 dephosphorylation depended on Cdc14 activity (Fig. 7 C). Together, these data suggest that Cdc14 directly dephosphorylates Mob1 during mitotic exit.
Cdk1 together with the Bfa1–Bub2 complex inhibits mitotic exit of cells
Deletion of BUB2 or the introduction of the CDC15-7A mutation hardly influenced cell cycle progression (Fraschini et al., 1999; Jaspersen and Morgan, 2000). However, the function of Bfa1–Bub2 GAP complex in inhibiting mitotic exit becomes apparent when combined with mutants like kar9Δ that cause spindle misalignment (Pereira et al., 2000). Similarly, CDC15-7A and MOB1-2A showed genetic interactions with kar9Δ. The CDC15-7A MOB1-2A kar9Δ triple mutant was no longer able to inhibit mitotic exit when the spindle was misaligned (Fig. S4, A and B). These data suggest that dual inhibition of the MEN by Bfa1–Bub2 and Cdk1 is essential when the spindle is misaligned. However, this MEN inhibition by Cdk1 probably does not take place at the SPB because kar9Δ cells with misaligned spindles do not accumulate Cdk1 at SPBs (Fig. 2, A and B).
During normal cell cycle progression, inhibition of the MEN by either Cdk1 or the Bfa1–Bub2 GAP complex may be sufficient to prevent premature mitotic exit. This possibility was tested using combinations of single, double, and triple mutants of bub2Δ, CDC15-7A, and MOB1-2A. Analysis of the growth of cells at different temperatures revealed that only the triple mutant bub2Δ CDC15-7A MOB1-2A cells grew slower at 23°C and showed an elevated growth defect at 37°C (Fig. 8 A).
We analyzed the MT cytoskeleton and the Cdc14-GFP distribution of cells grown at 23°C. Wild-type, single, and CDC15-7A MOB1-2A and bub2Δ MOB1-2A double mutants had normal proportions of cells in G1, S, and mitotic phases of the cell cycle (Fig. 8, B and C; unpublished data). Synchronization of these cells by α-factor block and release did not show obvious mitotic exit defects (not depicted). In contrast, ∼8% of log phase bub2Δ CDC15-7A cells exhibited the abnormal phenotypes of premature Cdc14 release (i.e., released Cdc14 in cells with a short spindle), cytokinesis failure (cells with multiple buds), and defects in nucleolar division (cells with multiple Cdc14 nucleoli foci) (Fig. 8 C). These defects were strongly enhanced in bub2Δ CDC15-7A MOB1-2A triple mutant cells (Fig. 8, B and C). Moreover, analysis of α-factor–synchronized bub2Δ CDC15-7A MOB1-2A cells confirmed the premature release of Cdc14 from the nucleolus in about half of cells with a short metaphase spindle (Fig. 8 B, top bub2Δ CDC15-7A MOB1-2A cell). In contrast, deletion of the spindle orientation checkpoint gene KIN4 in CDC15-7A MOB1-2A cells did not cause mitotic exit defects (not depicted). Kin4 blocks mitotic exit of cells with a misaligned anaphase spindle by keeping the Bfa1–Bub2 complex active (D’Aquino et al., 2005; Pereira and Schiebel, 2005; Caydasi and Pereira, 2009; Monje-Casas and Amon, 2009). Thus, the Bfa1–Bub2 GAP complex together with Cdk1 phosphorylation of Cdc15 and Mob1 inhibit mitotic exit during an unperturbed cell cycle.
The MEN is depicted as a linear SPB-associated pathway that is regulated by the inhibitory Bfa1–Bub2 GAP complex and, in an undefined way, by Cdk1 (Stegmeier and Amon, 2004). In one report it was suggested that Cdk1 inhibits the relocalization of Cdc15 from the mSPB to the dSPB (Menssen et al., 2001). However, the overexpressed Cdc15 construct used in this previous study already bound to the SPB of G1 cells, which is in clear contrast to all other reports on the localization of Cdc15 (Cenamor et al., 1999; Xu et al., 2000; Visintin and Amon, 2001; Molk et al., 2004) including the data presented here using a fully functional, nonoverexpressed CDC15-GFP gene fusion. In this study, we revise the view of how the MEN is regulated and provide evidence that the MEN is inhibited concretely by the Bfa1–Bub2 GAP and the phosphorylation of Cdc15 and Mob1 by Cdk1.
Regulation of Cdc14 and Mob1 by Cdk1
Recent models on MEN regulation suggest that inhibition of the MEN by the Bfa1–Bub2 GAP complex at the dSPB and the binding of Lte1 to the cortex of the daughter cell could be sufficient to restrain MEN activation (Bardin et al., 2000; Pereira et al., 2000). However, we consider this relatively simple model as unlikely because inactivation of the Bfa1–Bub2 GAP does not cause premature mitotic exit (Fraschini et al., 1999; Pereira et al., 2000).
How is the MEN regulated throughout the cell cycle? In S phase and metaphase the MEN is kept inactive by the inhibitory Bfa1–Bub2 GAP complex residing at SPBs (Fraschini et al., 1999; Pereira et al., 2000). At this phase of the cell cycle Cdc15 and Mob1 are not enriched at SPBs (Visintin and Amon, 2001); however, Cdk1 phosphorylates both proteins (Fig. 5; Jaspersen and Morgan, 2000). This suggests that Cdk1 in the cytoplasm is able to phosphorylate Cdc15 and Mob1 to reduce MEN activity (Jaspersen and Morgan, 2000; Pereira et al., 2000; Menssen et al., 2001; Figs. 5, 6, and 8).
At the metaphase–anaphase transition APCCdc20 induces a decline of mitotic Cdk1 activity by degrading about half of the mitotic cyclin Clb2 (Shirayama et al., 1999; Yeong et al., 2000). In addition, activation of separase leads to a drop of PP2A activity, allowing the remaining Cdk1–Clb2 together with FEAR network components to trigger the transient release of Cdc14 from the nucleolus (Stegmeier et al., 2002; Azzam et al., 2004; Queralt et al., 2006). These changes in Cdc14 phosphatase and Cdk1 kinase activities and the movement of the spindle toward the MEN activator Lte1 that is associated with the cortex of the bud promote activation of Tem1 and at least partial dephosphorylation of Cdc15 and Mob1 (Bardin et al., 2000; Jaspersen and Morgan, 2000; Pereira et al., 2000; Stegmeier et al., 2002; Fig. 7).
In early anaphase the Bfa1–Bub2 GAP complex probably still inhibits the binding of active Tem1-GTP to the dSPB (Pereira et al., 2002). Active Tem1 may bind to the mSPB devoid of the Bfa1–Bub2 GAP (Molk et al., 2004). In addition, FEAR-released Cdc14 is already sufficiently high to dephosphorylate to some degree cytoplasmic Cdc15 and Mob1, which occurs simultaneously to the degradation of the first wave of Clb2 (Fig. 5, and Fig. 9 for model; Jaspersen and Morgan, 2000; Asakawa et al., 2001; Visintin and Amon, 2001; Stegmeier et al., 2002; Molk et al., 2004). Tem1 then recruits nonphosphorylated Cdc15 to the mSPB (Fig. 4 C). Cdc15 in turn facilitates binding of the Dbf2–Mob1 kinase complex and Cdk1 to the mSPB (Fig. 3 and Fig. S4 C). How Cdk1 binds to the mSPB is presently unclear. However, it is interesting to note that the fission yeast Cdc15 homologue Cdc7 can phosphorylate the SPB scaffold protein Nud1, named Cut11 in fission yeast (Krapp et al., 2003).
Binding of active Cdc15 to the SPB should be sufficient to activate the MEN (Visintin and Amon, 2001). However, measurement of the timing of Cdk1 binding to the mSPB as an indication of Cdc15 activity and of Cdc14 release from the nucleolus as an indicator of full MEN activity in FEAR-defective spo12Δ cells suggests a delay of 6–8 min (anaphase in yeast is only 15 min) between Cdc15 activation at the mSPB and release of Cdc14 from the nucleolus (Fig. S1 C). This delay indicates a transient inhibition of full MEN activation (as measured by Cdc14 release) in early anaphase.
The finding that increased levels of active Cdc15 at the mSPB (Cdc15-7A) also increases Cdk1 at the mSPB (Fig. 4 D) is consistent with mutual regulation of both proteins at this pole. In our model (Fig. 9), the proximity of Cdk1, Cdc15, and Mob1 at the mSPB in early anaphase allows efficient phosphorylation of Cdc15 and Mob1 despite decreasing Cdk1 activity in the cytoplasm and nucleoplasm. Phosphorylated Cdc15 then binds with reduced affinity to the SPB, thus diminishing its binding and subsequently also that of Cdk1. In addition, phosphorylation of Mob1 reduces Dbf2–Mob1 kinase activity (Fig. 6). The consequence of the Cdk1–Cdc15/Mob1 regulation loop is the self-inhibition of the MEN in early anaphase. This model is supported by the observations that interfering with the feedback loop by inhibition of Cdc15 kinase activity (Fig. 3 C) or blocking the ability of Cdk1 to phosphorylate Cdc15 (Fig. 4 C; Cdc15-7A) leads to hyperaccumulation of Cdc15 at the mSPB. Moreover, Cdk1 accumulated more strongly at the mSPB in CDC15-7A cells than in CDC15 cells (Fig. 4, D and E).
Down-regulation of the MEN by Cdk1 continues until, in mid/late anaphase, the Cdc14/Cdk1 ratio is shifted in favor of the phosphatase. Cdc15 and Mob1 are then completely dephosphorylated by Cdc14 (Fig. 7; Jaspersen and Morgan, 2000), causing the collapse of the regulation loop at the mSPB concomitant with the full activation of the MEN at the mSPB. Eventually, Cdc5 phosphorylates Bfa1 in late anaphase/telophase to inactivate the Bfa1–Bub2 GAP complex (Hu et al., 2001; Pereira et al., 2002).
In cells with a misaligned anaphase spindle Cdk1 does not bind to SPBs (Fig. 2 and Fig. S1). However, Cdk1 contributes to the inhibition of mitotic exit through phosphorylation of cytoplasmic Cdc15 and Mob1 (Fig. S4). In this respect, it is interesting that in case of spindle alignment defects Bfa1–Bub2 activity becomes transferred from the SPB into the cytoplasm (Caydasi and Pereira, 2009).
The checkpoint kinase Kin4 (D’Aquino et al., 2005; Pereira and Schiebel, 2005; Caydasi and Pereira, 2009; Monje-Casas and Amon, 2009) does not have a function in an unperturbed cell cycle because kin4Δ CDC15-7A MOB1-2A cells have the same mitotic exit phenotype as CDC15-7A MOB1-2A cells (unpublished data). Thus, Kin4 is only important when the anaphase spindle is misaligned to prevent Cdc5 polo kinase to phosphorylate Bfa1 (Hu et al., 2001; D’Aquino et al., 2005; Pereira and Schiebel, 2005; Maekawa et al., 2007; Caydasi and Pereira, 2009).
Opposing functions of Cdk1 kinase in anaphase
Cdk1 kinase has a dual but antagonistic role in the regulation of Cdc14 phosphatase. As part of the FEAR pathway Cdk1 activates the release of Cdc14 in early anaphase (Azzam et al., 2004). Simultaneously, Cdk1 inhibits activation of the MEN (Fig. 6). This dual action of Cdk1 may help to regulate the timing of events in anaphase. Recruitment of Cdk1–Clb2 to the mSPB is an inhibitory mechanism that ensures that the MEN does not become active until the FEAR-released Cdc14 has accomplished its functions in spindle midzone formation and rDNA segregation (Pereira and Schiebel, 2003; D’Amours et al., 2004; Lavoie et al., 2004; Sullivan et al., 2004; Higuchi and Uhlmann, 2005; Khmelinskii et al., 2007). Genetic data impressively demonstrate the importance of the concerted action of MEN inhibition by the Bfa1–Bub2 complex and by Cdk1 for the coordination of anaphase events. bub2Δ CDC15-7A MOB2-2A cells are compromised for growth, and even at the permissive temperature, prematurely release Cdc14 from the nucleolus and exhibit defects in both cytokinesis and the segregation of their nucleoli (Fig. 8).
The MEN, the septation initiation network (SIN) of Schizosaccharomyces pombe, and the Hippo pathway of Drosophila and vertebrates have, in Cdc15- and Dbf2–Mob1-like kinases, common signaling elements. Like the MEN, the SIN pathway is inhibited by Cdk1 activity (McCollum and Gould, 2001). However, a recent chemical genetic analysis concluded that the relevant Cdk1 targets whose phosphorylation prevents activation of the SIN remain to be identified (Dischinger et al., 2008). Little is known about cell cycle control of the Hippo pathway (Pan, 2007). It is, however, interesting that the Dbf2-related kinase Lats1 was originally described as a partner of Cdk1 (Tao et al., 1999). Thus, our analysis on the interplay between Cdk1 and MEN components will help to understand the regulation of MEN-related pathways in other organisms.
Materials and methods
Yeast strains and plasmids
Yeast strains and plasmids are listed in Table I. Yeast strains were derivatives of S288c unless stated otherwise. Yeast strains were constructed by PCR-based methods (Janke et al., 2004). The red fluorescent eqFP611 (Wiedenmann et al., 2002) was used to mark SPBs through a fusion with SPC42. The Cherry-TUB1 construct was described previously (Khmelinskii et al., 2007).
Construction of MOB1 mutants
The MOB1 gene was cloned into yeast integration vector pRS305H (Taxis and Knop, 2006). Mutations in MOB1 were introduced by PCR-directed mutagenesis and confirmed by DNA sequencing. Serine or threonine residues of two full Cdk1 consensus sites and five minimal Cdk1 consensus sites were mutated to alanine. The MOB1-7A mutant resulted in the exchange of S10, S36, T41, S80, T85, T105, and T180 to alanine. All MOB1 constructs were functional because they rescued the lethality of a MOB1 deletion.
Cell cycle analysis and growth conditions
For synchronization, yeast cells were grown in yeast extract adenine dextrose (YPAD) medium and arrested in G1 by treatment with α-factor (10 µg/ml) for 2.5 or 2 h at 23 or 30°C, respectively, until >95% of cells showed a mating projection. Cells were then washed with prewarmed growth medium to remove α-factor and resuspended in YPAD medium at the indicated temperatures. Cells with CDC5 or UPL-TEM1 under the control of the Gal1 promoter were grown in YPA with 3% raffinose and 2% galactose (YPAG/R). For the depletion of Cdc5 and Tem1, Gal1-CDC5 and Gal1-UPL-TEM1 cells in YPAG/R were synchronized with α-factor. G1-arrested cells were then released into YPAD to repress the Gal1 promoter.
Yeast cells with CDK1-GFP, CDK1-3GFP, CDC15-GFP, cdc15-as1-GFP, MOB1-GFP, mob1-67-GFP, DBF2-GFP, dbf2-2-GFP, and CDC14-3mCherry were analyzed by fluorescence microscopy without washing or fixation with the exception of CDK1-3GFP, MOB1-GFP, and CDC14-GFP cells in Figs. 1 (B and C), 2, 3 E, 8 B, S1 A, S2 A, and S4 C. These cells were analyzed after fixation with 4% paraformaldehyde in 150 mM K-phosphate buffer, pH 6.5, for 10 min at 20°C. Cells were incubated in PBS containing DAPI to visualize DNA. Z series of images of 0.35-µm steps were captured with a microscope (Axiophot; Carl Zeiss, Inc.) that was equipped with a 100x NA 1.45 oil immersion objective (Plan-Fluar; Carl Zeiss, Inc.) and a camera (Cascade:1K; Photometrics). Pictures were quantified or processed with MetaMorph software (Universal Imaging Corp.; Figs. 1 [C and D], 2 [B and C], 3 [A and C], 4 [A–C], 8 B, S1 A, S2 [C and D], and S4 C). A microscope (Deltavision; Applied Precision) equipped with GFP and TRITC filters (Chroma Technology Corp.), a 100x NA 1.4 oil immersion objective (PlanApo, IX70; Olympus), and a camera (CoolSNAP HQ; Photometrics) was used for pictures in Figs. 1 (A and B), 2 (A and D), 3 E, 4 D, S1 C, and S2 A. Pictures were processed with SoftWoRx 3.5.0 software (Applied Precision). Fluorescence intensity was measured in one plane with the SPB in focus.
Time-lapse experiments were performed (Fig. 1 A and Fig. S1 C) on Con A–coated glass-bottom dishes (MatTek) using the Deltavision microscope at 30°C. Z series at 0.35-µm steps (2 × 2 binning) were acquired every 1 min (SHM1757) or 2 min (CKY949). Deconvolution was performed using SoftWoRx 3.5.0 software (Applied Precision) with default settings.
Adobe Photoshop and ImageJ were used to mount the images and to produce merged color images. No manipulations other than contrast and brightness adjustments were used.
In vitro kinase assays
Clb2 was expressed from a 2-µm Gal-CLB2-TAP plasmid (Übersax et al., 2003). Kinase complex was purified via the TAP tag. Cdk1 kinase assays with GST-Mob1 as substrate were performed as described previously (Loog and Morgan, 2005), except that the reaction buffer was 25 mM Hepes, pH 7.4, 150 mM NaCl, 10% glycerol, 10 mM MgCl2, 10 µM ATP, 5 mM β-glycerophosphate, and 5 µCi γ-[32P]ATP in a 20-µl reaction volume.
The GST–Mob1–Dbf2 complex of Fig. 6 was affinity purified from yeast strains expressing GST-MOB1 and DBF2 under the control of the pGal1-10 promoter without eluting the complex from the glutathione affinity beads (Geymonat et al., 2002). Co-purification of Dbf2 was confirmed by MALDI-TOF/TOF. In Fig. 6 C, the beads with ∼80 ng Dbf2–Mob1 were incubated in 40 µl Cdc15/Cdk1 kinase buffer (50 mM Hepes-KOH, pH 7.5, 100 mM NaCl, 10 mM MgCl2, 2.5 mM MnCl2, 5 mM β-glycerophosphate, and 1 mM DTT) containing 5 mM ATP and complete EDTA-free protease inhibitor cocktail (Roche) for 1 h at 30°C with 3 ng of Cdc15 or Cdk1–Clb2 or without kinase (first reaction). The beads were washed three times with Cdc15/Cdk1 kinase buffer. The second reaction was performed as the first reaction. One eighth of the beads were incubated in Dbf2 kinase buffer (50 mM Tris, pH 7.4, 100 mM NaCl, 10 mM MgCl2, 5 mM β-glycerophosphate, 1 mM DTT, and 5 µCi γ-[32P]ATP) with 1 µg GST-C-Cdc14 at 20°C for 10, 20, and 30 min. The kinase reaction was stopped by the addition of 5 µl 6x sample buffer and heating the samples at 95°C for 5 min. Samples were loaded onto a 10% SDS-PAGE gel. Gels were stained with SimplyBlue Safe Stain (Invitrogen) and radioactivity was detected by a PhosphorImager (FLA-300; Fujifilm) and quantified with Image Gauge v3.45 (Fujifilm).
TAP-Dbf2 immunoprecipitations were performed in 25 mM K-phosphate, pH 7.6, 150 mM KCl, 1% NP-40, 10% glycerol, 5 mM MgCl2, 1 mM EDTA, complete EDTA-free protease inhibitor cocktail (Roche), and 80 mM β-glycerophosphate using rabbit IgG coupled to M-270 Epoxy Dynabeads. Half of each reaction was used for immunoblot analysis and the other half for kinase assays in Dbf2 kinase buffer using GST-C-Cdc14 as substrate as in Fig. 6 C with incubation for 15 min at 30°C.
Online supplemental material
Fig. S1: Cdk1 localization in dyn1Δ cells and time-lapse analysis of CDK1-GFP CDC14-3mCherry cells. Fig. S2: Cdk1 localization in cdc15-1 cells and localization of mob1-67 and dbf2-2 proteins. Fig. S3: suppression analysis in dbf2-2 dbf20Δ cells and binding of Mob1-2A to Dbf2. Fig. S4: genetic interaction of CDC15-7A MOB1-2A with kar9Δ and localization of Mob1 in cdc15-as1 cells.
We thank D. Morgan and S. Sedgwick for yeast strains, M. Knop for plasmids, and M. Fischer for anti-GFP antibody. Members of the Schiebel laboratory, I. Hagan, and G. Pereira are acknowledged for their comments on the manuscript.
C. König and H. Maekawa contributed equally to this paper.
H. Maekawa’s present address is Kobe Advanced ICT Research Center, National Institute of Information and Communications Technology, 588-2 Iwaoka, Nishi-Ku, Kobe 651-2492, Japan.