Nuclear PP2A-Cdc55 prevents APC-Cdc20 activation during the spindle assembly checkpoint

The PP2A (protein phosphatase 2A) proteins are a conserved family of serine/threonine phosphatases that have many important roles in mitotic progression in eukaryotes (Shi, 2009). The heterotrimeric PP2A consists of a structural A subunit, a regulatory B subunit and a catalytic C subunit. B subunits bind to the AC heterodimer and regulate both the substrate specificity and the localization of the PP2A complexes. In budding yeast, Cdc55 (B) and Rts1 (B′) have been identified as B-regulatory subunits and bind to the core PP2A in a mutually exclusive manner (Jiang, 2006; Shu et al., 1997; Zhao et al., 1997). Recently, Zds1 (zillion different screens) and its paralog Zds2 have been found to specifically bind to PP2A-Cdc55, but not to PP2A-Rts1 (Queralt and Uhlmann, 2008; Wicky et al., 2010; Yasutis et al., 2010), and these two proteins regulate the nucleocytoplasmic distribution of PP2A-Cdc55 (Rossio and Yoshida, 2011), adding a new level of complexity.

PP2A-Cdc55 is involved in the stress response, polarized growth, meiosis and mitotic progression (Jiang, 2006). In the cytoplasm, PP2A-Cdc55, in complex with Zds1 or Zds2, promotes mitotic entry (Rossio and Yoshida, 2011). cdc55Δ and zds1Δ zds2Δ cells exhibit abnormally elongated cell morphology as a consequence of prolonged G2-phase delay (Bi and Pringle, 1996; Healy et al., 1991). The mitotic entry defect of cdc55Δ cells is due to inhibitory phosphorylation of Cdc28 (budding yeast Cdk1) on Tyr19 by the kinase Swe1 – the elongated morphology of cdc55Δ is rescued either by deletion of SWE1 (McMillan et al., 1999a; McMillan et al., 1999b; Rossio and Yoshida, 2011; Wang and Burke, 1997; Wicky et al., 2010; Yang et al., 2000) or by introduction of the cdc28-Y19F mutation.

In the absence of Zds1 and Zds2, PP2A-Cdc55 accumulates in the nucleus and prevents mitotic exit (Rossio and Yoshida, 2011). It is known that PP2A-Cdc55 inhibits mitotic exit because overexpression of Cdc55 is toxic to the mutants defective in mitotic exit (Wang and Ng, 2006) and because deletion of CDC55 rescues variety of mutants defective in mitotic exit (Clift et al., 2009; Queralt et al., 2006; Wang and Ng, 2006; Yellman and Burke, 2006). The target of PP2A-Cdc55 in mitotic exit is the phosphatase Cdc14, which is essential for mitotic exit in budding yeast (Visintin et al., 1998). Cdc14 is kept inactive by its inhibitor Net1 (also known as Cfi1) in the nucleolus (Shou et al., 1999; Straight et al., 1999; Visintin et al., 1999). Cdc14 release from the nucleolus requires Cdk-dependent phosphorylation of Net1 (Azzam et al., 2004). PP2A-Cdc55 counteracts Net1 phosphorylation and prevents mitotic exit (Queralt et al., 2006; Queralt and Uhlmann, 2008). Thus, it has been proposed that PP2A-Cdc55 is inactivated at anaphase onset either by nuclear exclusion (Rossio and Yoshida, 2011) or by inhibition of PP2A phosphatase activity (Calabria et al., 2012; Queralt and Uhlmann, 2008).

PP2A-Cdc55 is also an essential regulator of the spindle assembly checkpoint (SAC) (Clift et al., 2009; Evans and Hemmings, 2000; Koren et al., 2004; Minshull et al., 1996; Wang and Burke, 1997; Wang and Ng, 2006; Yellman and Burke, 2006) although the target of PP2A that is relevant for the SAC is not known. cdc55Δ cells, when challenged with microtubule-depolymerizing drugs, fail to arrest the cell cycle in mitosis and re-enter next round of the cell cycle and die. In cdc55Δ cells, the securin Pds1 is degraded in an anaphase-promoting complex (APC)-dependent manner and sister chromatids separate as in other well-characterized SAC mutants (Minshull et al., 1996; Wang and Ng, 2006; Yellman and Burke, 2006), such as mad1Δ or mad2Δ (Alexandru et al., 1999). Interestingly, in cdc55Δ cells, the mitotic cyclins Clb2 and Clb3 remain stable, but Cdk1 activity declines and cells exit from mitosis (Chiroli et al., 2007; Minshull et al., 1996; Yellman and Burke, 2006). Recent papers (Wang and Ng, 2006; Yellman and Burke, 2006) proposed that the target of Cdc55 in the SAC is Cdc14, because Cdc14 is precociously released in cdc55Δ, and the Cdc14 activator Tem1 is dephosphorylated in a manner dependent on Cdc55. However, activation of Cdc14 does not explain why cdc55Δ cells undertake Pds1 degradation and show precautious sister chromatid separation upon spindle damages.

Here, we propose that the major target of PP2A-Cdc55 in the SAC is the APC bound to its regulatory subunit Cdc20 (APC-Cdc20). It is known that Cdk1-dependent phosphorylation of at least three APC subunits (Cdc16, Cdc23 and Cdc27) is required for optimal activity of APC-Cdc20 (Rudner and Murray, 2000) but the counteracting phosphatase has not been identified. The early finding by Wang and Burke (Wang and Burke, 1997) that cdc55 deletion rescues the temperature-sensitive growth defect of cdc20-1 cells, and the finding that Tpd3, an A subunit of PP2A, can be immunoprecipitated with APC (Boronat and Campbell, 2007; Kornitzer et al., 2001) suggest that PP2A-Cdc55 has a direct role in the regulation of APC-Cdc20 activity.

By isolating novel cdc55 mutants specifically defective in the SAC, but not the other cell cycle steps, we found that mutation of cdc55 can rescue growth defects in a large variety of APC mutants. We also demonstrate that the APC subunits Cdc16 and Cdc27 are hyperphosphorylated in the absence of Cdc55, and that unphosphorylatable APC mutants rescued the SAC defects of cdc55Δ, supporting the hypothesis that APC-Cdc20 is a direct target of Cdc55 in the SAC. In addition, we show evidence that dephosphorylation of APC-Cdc20 is taking place within the nucleus. When treated with nocodazole, both Cdc55 and APC are in the nucleus and APC is inhibited. When nuclear accumulation of Cdc55 is perturbed, as in the cdc55-101 mutation or upon Zds1 overexpression, APC-Cdc20 is activated and cells become SAC defective. Our results are the first demonstration that APC is a functional target of PP2A-Cdc55 and confirm that regulation of nucleocytoplasmic distribution of PP2A-Cdc55 is important for proper cell cycle control.

tmr4 mutants have been identified as recessive-revertant suppressor mutants of the temperature-sensitive mitotic growth arrest seen in the tom1 mutant (Sasaki et al., 2000; Utsugi et al., 1999). On the basis of the benomyl sensitivity of tmr4 mutants, we screened plasmids from a genomic library for the ability to complement the benomyl sensitivity of tmr4 mutants. Next, we verified whether the plasmids inhibited the growth of each tmr4 tom1-2 double mutant at 35°C. By sub-cloning and sequence analysis, we identified that tmr4 is allelic to CDC55 and confirmed the mutant alleles by backcrossing. We then mapped the mutation sites of tmr4 mutants (hereafter we designate them cdc55-101, cdc55-102 and cdc55-103) and found that cdc55-101 has the single mutation Gly43 to Asp, cdc55-102 has the mutation Asp319 to Asn and cdc55-103 has the mutation Gly47 to Arg. These mutated residues are highly conserved in the PP2A-B55 family from yeast to human (Fig. 1A). On the basis of a recent structural study of the PP2A complex, Gly43 and Gly47 are located in the β1B sheet and Asp319 is in the β5C sheet (Li and Virshup, 2002; Xu et al., 2008) (Fig. 1B). Because these mutation sites are far away from binding sites between the B-subunit and the AC complex and from putative substrate-binding regions, they are unlikely to be affecting the structural integrity of PP2A complex or enzymatic activity.

It is known that cdc55Δ cells are defective in mitotic entry and have abnormally elongated morphology. In a clear contrast to the cdc55Δ cells, the newly isolated cdc55 mutants did not show elongated morphology or cold sensitivity (Fig. 1C), suggesting that these mutants are not defective in mitotic entry. However, all three cdc55 mutants showed clear sensitivity to the microtubule-depolymerizing drug benomyl that was almost comparable to cdc55Δ (Fig. 1C). The benomyl sensitivity of cdc55Δ was associated with misregulation of catalytic activity of PP2A because overexpression of any of the catalytic subunits, PPH21, PPH22 and PPH3, partially rescued the benomyl sensitivity (Fig. 1D). We also found that overexpression of SIT4, a homolog of the phosphatase PP6, can rescue the benomyl sensitivity of cdc55Δ although the functional relationship between Cdc55 and Sit4 is not clear.

The benomyl sensitivity of cdc55-101, cdc55-102 and cdc55-103 mutants prompted us to examine the functionality of SAC in these mutants. Indeed, they failed to maintain mitotic arrest in the presence of nocodazole, and rebudded and lost viability with similar kinetics to mad1Δ (a known SAC mutant) (Hoyt et al., 1991; Li and Murray, 1991) and cdc55Δ (Fig. 2A,B). In addition, we observed that the cdc55-101 mutant failed to maintain sister chromatid cohesion after nocodazole-induced arrest, as with mad1Δ (Fig. 2C) (Hoyt et al., 1991; Stearns et al., 1990; Straight et al., 1996). Pleiotropic cell cycle defects associated with cdc55Δ were one reason that analysis of Cdc55 function in the SAC has been difficult. Thus, we took advantage of our newly isolated cdc55 mutants that were only defective in the SAC but not mitotic entry. Because all three cdc55 mutants showed similar SAC defects, we focused our analysis on cdc55-101.

To further characterize the SAC defects of cdc55-101, we synchronized yeast cells in G1 phase by use of mating pheromone and released them into the medium containing nocodazole. In wild-type cells, the securin Pds1, S-phase cyclin Clb5 and mitotic cyclin Clb2 were all stabilized (Fig. 2D; supplementary material Fig. S1) owing to SAC-dependent inhibition of APC-Cdc20 (Visintin et al., 1997). In the SAC-defective mad2Δ strain, Pds1 is precociously degraded (Fig. 2D). In addition, Clb2 is gradually degraded as a consequence of mitotic exit and activation of APC-Cdh1 (Zachariae et al., 1998). In cdc55-101, as in mad2Δ, Pds1 and Clb5 were precociously degraded. Consistent with previous reports (Chiroli et al., 2007; Minshull et al., 1996; Yellman and Burke, 2006), we did not see clear effects on Clb2 degradation (Fig. 2D) suggesting that Cdc55-101 has a specific effect on APC-Cdc20 but not APC-Cdh1. We also confirmed that the timing of Pds1 and Clb2 degradation in cdc55-101 is very similar to that of cdc55Δ but distinct from that of mad2Δ (Fig. 2D). Thus, the cdc55-101 mutant is completely defective in Cdc55 function in the SAC.

It is known that the cdc20-1 temperature sensitivity is suppressed by cdc55Δ (Wang and Burke, 1997). We found that cdc55-101, cdc55-102 and cdc55-103 also rescued the temperature sensitivity of cdc20-3 (Fig. 3A). The suppression of cdc20-3 is not simply due to a SAC defect because neither mad1Δ nor bub2Δ rescued cdc20-3 (Fig. 3A). Furthermore, not only cdc20-3, but also several other APC mutants, including cdc16-1, cdc26Δ and apc1-1 were rescued by cdc55-101 mutation (Fig. 3B). These genetic data suggest that Cdc55 has a specific function in controlling APC-Cdc20 activation or localization.

It is known that activation of APC-Cdc20 requires Cdk-dependent phosphorylation (Rudner et al., 2000; Rudner and Murray, 2000). At least three subunits of APC (Cdc16, Cdc23 and Cdc27) are phosphorylated both in vitro and in vivo by Cdk, and mutating these phosphorylation sites (Cdc16-6A, Cdc23-A and Cdc27-5A) impairs APC functions in vivo (Rudner and Murray, 2000). Because Cdk1 and PP2A counteract each other, we hypothesized that PP2A-Cdc55 inhibits APC-Cdc20 through dephosphorylation. We found that Cdk-dependent phosphorylation of Cdc16 and Cdc27 is controlled by Cdc55. Cdc16, Cdc23 and Cdc27 are known to be phosphorylated by Cdk and migrate slowly in the SDS-PAGE after Cdk-dependent phosphorylation (Rudner and Murray, 2000). The amount of the slow-migrating species of Cdc16–HA and Cdc27–HA seen in western blots is more abundant in cdc55Δ when cells were arrested by nocodazole, resulting in SAC activation (Fig. 4A), indicating that PP2A-Cdc55 is required for dephosphorylation of Cdc16 and Cdc27 in this condition. We also found that the slow-migrating form of Cdc16–HA accumulated in cdc55-101, similar to in cdc55Δ after nocodazole treatment (Fig. 4C). We also tested Cdc23 in the same conditions, but were not able to see an obvious mobility shift in the cdc55Δ cells (data not shown), probably because there are fewer phosphorylation sites on Cdc23 than on Cdc16 and Cdc27 (Rudner and Murray, 2000). To confirm that the SAC defect of cdc55Δ is due to hyperphosphorylation of APC, we combined cdc55 deletion with the unphosphorylatable APC mutants (Cdc16-6A, Cdc23-A and Cdc27-5A) and found that the SAC defects of cdc55 were almost completely rescued by these mutations (Fig. 4B). This genetic data suggests that the major target of PP2A-Cdc55 in the SAC is the APC.

We also examined localization of the APC subunit Cdc23. As it has been reported previously, Cdc23–GFP is constitutively in the nucleus and sometimes localized to the spindles (Jaquenoud et al., 2002; Melloy and Holloway, 2004) in both wild-type and in cdc55Δ cells (supplementary material Fig. S4). Activation of the SAC, by addition of nocodazole, had no clear effect on general nuclear Cdc23–GFP localization (supplementary material Fig. S2). Thus, phosphorylation of APC subunits does not affect APC localization. This is consistent with previous data that indicated that localization of Cdc23 is not altered in a strain with all Cdk consensus sites mutated in the APC/C subunits Cdc23, Cdc27 and Cdc16 (Melloy and Holloway, 2004).

It is formally possible that the APC is not the only target of Cdc55 in the SAC. To examine the contribution of Cdc14 release by phosphorylation of Net1 in the SAC defect of cdc55Δ, we tested whether unphosphorylatable Net1 (NET1-6cdk-9myc) (Azzam et al., 2004) could rescue the SAC defects of cdc55Δ. Indeed, NET1-6cdk-9myc partially rescued the SAC defects of cdc55Δ, however, NET1-9myc, which was used as a negative control, also rescued the SAC defects of cdc55Δ (supplementary material Fig. S3). This result suggests that 9xmyc tag of Net1 has an unexpected dominant role in delaying mitotic exit and the contribution of Cdk-dependent phosphorylation of Net1 in the SAC remains unclear.

Cdc55 localizes to the polarized growth sites, as well as the cytoplasm and nucleus (Gentry and Hallberg, 2002). We have recently shown that Cdc55 has compartment-specific functions (Rossio and Yoshida, 2011); for example, cytoplasmic Cdc55 promotes mitotic entry, whereas nuclear Cdc55 delays mitotic exit. It is known that APC-Cdc20 and its substrates are in the nucleus (Jaquenoud et al., 2002; Melloy and Holloway, 2004); hence, we anticipated that nuclear Cdc55 is responsible for APC dephosphorylation. Because Cdc55-101 is competent for mitotic entry (Fig. 1C) but defective in the SAC (Fig. 2) we hypothesized that Cdc55-101 was excluded from the nucleus.

Cdc55–GFP was localized to both in the nucleus and the cytoplasm, as judged by Nup159–mCherry staining of nuclear envelope (Fig. 5A), in contrast Cdc55-101–GFP was excluded from the nucleus throughout the cell cycle (Fig. 5A). Loss of nuclear signal was not due to reduced expression or instability of Cdc55-101–GFP protein because Cdc55-101–GFP was expressed to similar levels as wild-type Cdc55–GFP as judged by western blotting (Fig. 5B). Similar nuclear exclusion was observed with Cdc55-103–GFP (supplementary material Fig. S4). Cdc55-102–GFP was also partially excluded from the nucleus, but the effect was not as strong as Cdc55-101 and Cdc55-103, suggesting that this mutant might only show a minor localization defect. Importantly, when the SAC is activated by nocodazole, Cdc55 was found both in the nucleus and in the cytoplasm (Fig. 5C). By contrast, Cdc55-101 barely accumulated in the nucleus (3.5% of the cells have Cdc55-101 in the nucleus compared with 89% of the control). These results are consistent with our hypothesis that nuclear Cdc55 is responsible for its role in the SAC.

To test whether nuclear exclusion of Cdc55 results in the activation of APC-Cdc20 and the SAC defect, we examined the effect of Zds1 overexpression. Zds1 forms a stoichiometric complex with PP2A-Cdc55 (Queralt and Uhlmann, 2008; Wicky et al., 2010; Yasutis et al., 2010) and enhances the cytoplasmic functions of Cdc55 but inhibits the nuclear functions of Cdc55 (Rossio and Yoshida, 2011).

We first confirmed that Zds1 is a cytoplasmic protein (Bi and Pringle, 1996; Rossio and Yoshida, 2011) and that it is always excluded from the nucleus by using the nuclear envelope marker Nup159–mCherry (supplementary material Fig. S5A). We also confirmed that Cdc55 was excluded from the nucleus after ZDS1 overexpression (Rossio and Yoshida, 2011) by using Nup159–mCherry (supplementary material Fig. S5B).

We found that the temperature-sensitive growth defect of cdc20-3 was effectively suppressed by overexpression of ZDS1 (Fig. 6A). Thus, Zds1 overexpression caused a similar effect to the cdc55-101 mutation or to cdc55Δ. To assess the effect of ZDS1 overexpression on Cdc55 in the SAC, we used zds1Δc800, which lacks Cdc55-binding domain (CBD) as a negative control. Note, we found highly conserved residues within the CBD of fungal Zds1 family proteins and the vertebrate cortactin-binding proteins CTTNBP2 and CTTNBP2NL (Fig. 6B), which have been recently identified as PP2A-B55 (PP2A-Striatin)-binding proteins (Goudreault et al., 2009) and a crucial regulator of PP2A-B55 localization to the dendritic spines (Chen et al., 2012).

Overexpression of ZDS1 but not zds1Δc800 caused sensitivity to benomyl (Fig. 6C). Cells overexpressing ZDS1 were not able to arrest in nocodazole and rebudded, which is characteristic of SAC mutants, and this effect was dependent on the ability of Zds1 to bind to Cdc55 because overexpression of zds1Δc800 did not cause these defects (Fig. 6C,E), although Zds1Δc800 was expressed to a similar level to that of the full-length Zds1 (Fig. 6D). Thus, overexpression of Zds1 has a similar effect to loss of Cdc55 and causes SAC defects that are probably due to depletion of nuclear Cdc55.

To further test whether nuclear Cdc55 is responsible for the SAC arrest, we took advantage of engineered cdc55 mutants, where Cdc55 is either predominantly cytoplasmic (cdc55-NES) or predominantly nuclear (cdc55-NLS) (NES, nuclear export signal; NLS, nuclear localization signal) (Rossio and Yoshida, 2011). We also tested the zds1Δzds2Δ mutant, where Cdc55 is predominantly nuclear (Rossio and Yoshida, 2011). First, we found that the cdc55-NES strain is sensitive to benomyl, like cdc55Δ (Fig. 7A). Note that cdc55-NES is not simply a loss of function because cdc55-NES has a dominant function in the cytoplasm and can induce mitotic entry in zds1Δzds2Δ (Rossio and Yoshida, 2011).

cdc55-NES was indeed defective in the SAC. The cdc55-NES strain rebudded in the presence of nocodazole, like cdc55Δ (Fig. 7C). In contrast, zds1Δzds2Δ was competent in the SAC (Fig. 7C). We also confirmed the SAC defects of cdc55-NES by FACS analysis of the DNA contents and found that cdc55-NES reduplicated its DNA in the presence of nocodazole, as does cdc55Δ (Fig. 7B). In these assays we deleted the SWE1 gene to exclude the potential G2 phase cell cycle delay in cdc55 or zds1Δzds2Δ strains.

Thus, loss of Cdc55 from the nucleus by cdc55-NES causes a similar effect to the cdc55-101 mutation and to cdc55Δ or to ZDS1 overexpression.

In this study, we have shown that the crucial target of Cdc55 in the SAC is APC-Cdc20 because the phosphorylation status of APC subunits are controlled by Cdc55 in vivo and the SAC defects of cdc55 are largely rescued by preventing APC phosphorylation.

The targets of Cdc55 in the SAC have long been debated. Given that the SAC defect is accompanied by mitotic exit, recent studies suggested that Cdc14 activation is a key step that is prevented by Cdc55 during SAC arrest (Wang and Ng, 2006; Yellman and Burke, 2006). However, we favor APC-Cdc20 activity rather than Cdc14 activity as a key target of Cdc55 phosphatase in the SAC because Cdc14 activates APC-Cdh1, not APC-Cdc20 (Jaspersen et al., 1999; Visintin et al., 1998; Zachariae et al., 1998). The fact that APC-Cdc20 activity but not APC-Cdh1 activity is promoted by Cdk1-dependent phosphorylation (Rudner and Murray, 2000), and our data showing that there is preferential degradation of the APC-Cdc20 substrates Pds1 and Clb5, but not Clb2, a major target of APC-Cdh1, in cdc55 is consistent with APC-Cdc20 being under the control of Cdc55.

It is important to mention that Cdc14 activation and subsequent APC-Cdh1 activation is under the control of APC-Cdc20. Thus Cdc14 and APC-Cdh1 should be eventually activated in the cdc55 mutants arrested in spindle damages. Unfortunately, we were not able to address the direct contribution of Cdc14 inhibition by Cdc55 in the SAC because the effects of the 9×Myc tag on NET1, not the mutation in the Cdk1-dependent phosphorylation sites of Net1, caused suppression of the SAC defects of cdc55Δ.

Although deletion of NET1 rescues various mitotic exit mutants, similar to deletion of CDC55 (Straight et al., 1999; Visintin et al., 1999; Yellman and Burke, 2006), deletion of NET1 does not suppress cdc20-1 (Visintin et al., 1999). Highly specific suppression of APC-Cdc20 function by cdc55 mutations is consistent with our idea that Cdc55 not only regulates Net1 for Cdc14 release but also APC-Cdc20 for sister chromatid separation.

The roles of PP2A in the regulation of APC and SAC are currently not clear in animal cells because of redundancy of the PP2A subunits (Shi, 2009; Virshup and Shenolikar, 2009) and the pleiotropic defects associated with PP2A inhibition (Janssens and Goris, 2001). Several studies imply that APC activity is negatively regulated by PP2A in animal cells. In Drosophila, the metaphase-like arrest of mks (a Cdc27 homolog) mutants can be suppressed by mutations in the twins (CDC55 homolog) gene, but not by mutations in SAC genes (Deak et al., 2003). In human, the Cdc55 homolog PRB2B regulates the phosphorylation status of Cdc27 and affects the localization of Cdc27 to the mitotic spindle (Torres et al., 2010). Thus it is highly likely that regulation of APC by PP2A is an evolutionarily highly conserved process.

We have isolated novel cdc55 mutants as revertant suppressors of the tom1-2 mutant. tom1 mutants arrest the cell cycle in G2/M phase, but the substrates and targets of Tom1 important for mitotic progression have not been identified (Sasaki et al., 2000). In contrast to cdc55Δ, which shows an abnormally elongated morphology and a cold-sensitive phenotype in addition to the SAC defect, the only phenotype associated with cdc55-101, cdc-102 and cdc-103 was the SAC defect. Because Cdc55-101 protein was expressed at the wild-type level and because mutation sites are not located in the binding surface to the PP2A AC subunits, it is unlikely that cdc55-101 is impaired in PP2A complex assembly or PP2A catalytic activity. The absence of apparent morphological defects in cdc55-101 also suggests that PP2A activity is not severely impaired. It is formally possible that Cdc55-101 protein has lost the ability to interact with specific substrates such as APC, but we favor the idea that Cdc55-101 is specifically defective in its nuclear localization because the phenotype of cdc55-101 was very similar to that of cdc55-NES strain and that of the cells overexpressing ZDS1.

The targets of Tom1 that are crucial for mitotic progression are not understood yet, but it is most likely to be APC-Cdc20 activity. Interestingly, in the same screen as that which identified tmr4/cdc55, tmr1/cyr1 and tmr2/sch9 were also identified (Sasaki et al., 2000). Cyr1 encodes an adenylate cyclase and Sch9 encodes a kinase antagonizing the protein kinase A (PKA) pathway. In addition, suppression of the PKA pathway, either by overexpression of BCY1, a regulatory subunit of PKA or by overexpression of PDE2, a phosphodiesterase, can also rescue the growth defect of tom1 mutants (Sasaki et al., 2000). It is known that PKA activity inhibits APC-Cdc20 activity by phosphorylating Cdc20 (Anghileri et al., 1999; Searle et al., 2004). Because inhibition of PKA activity rescues mitotic defects associated with tom1 mutation, we suspect that activation of APC-Cdc20 is the key process compromised in the tom1 mutant.

We showed that both Cdc55 and APC subunits are in the nucleus in nocodazole-arrested cells. Nuclear localization of APC during mitosis is consistent with the fact that Cdc20, as well as its substrates Pds1 and Clb5, is predominantly nuclear (Jaquenoud et al., 2002).

Three lines of evidence suggest that stable SAC arrest and inhibition of APC-Cdc20 activity requires PP2A-Cdc55 activity in the nucleus: first, a SAC defective Cdc55-101 mutant protein is excluded from the nucleus; second, retention of Cdc55 in the cytoplasm by overexpression of ZDS1 impaired the SAC; and, third, forced exclusion of Cdc55 from the nucleus by adding a strong nuclear export signal (cdc55-NES) results in SAC defects. In these cases, the temperature-sensitive growth defect of cdc20-3 was also rescued.

In the previous study, we reported that Cdc55 localization changes during the cell cycle and that nuclear Cdc55 signal is high in G1 or small budded cells but is reduced in mitotic cells in a Zds1- or Zds2-dependent manner (Rossio and Yoshida, 2011). Exclusion of Cdc55 from the nucleus during mitosis is consistent with our model that nuclear Cdc55 interferes with APC-Cdc20 activity (this study) and with Cdc14 activation (Rossio and Yoshida, 2011) (Fig. 8).

The mechanism by which Zds1 and/or Zds2 exports the Cdc55-containing PP2A complex and how interaction between Zds1 and/or Zds2 and Cdc55 is regulated during the cell cycle and upon SAC activation is a very important question. We did not clearly see an obvious change in the nuclear localization of Cdc55 upon SAC activation (data not shown), suggesting that PP2A-Cdc55 is not specifically responding to spindle damage, but rather setting a high threshold for Cdk1-dependent activation of APC to prevent hyperactivation of APC-Cdc20.

Although Zds1 is highly conserved in almost all fungal species, no obvious homolog is found in animals or plants. By carefully examining the CBD sequence of Zds1, we found some conversation between this region and the PP2A-binding region of human CTTNBP2NL and CTTNBP2. Importantly, CTTNBP2NL and CTTNB2 specifically bind to the regulatory B-subunit of PP2A in the region with high similarity to the CBD of Zds1 and this interaction is essential for PP2A complex to localize to the dendritic spines (Chen et al., 2012). Thus Zds1 and CTTNBP2 are functioning in a similar manner to recruit PP2A complex to specific subcellular components. We propose that spatial control of PP2A complex by Zds1-like regulatory proteins is a widely conserved mechanism used to compartmentalize PP2A activity.

All yeast strains used in this study were isogenic or congenic to BY4741 (MATa leu2Δ0 his3Δ1 met15Δ0 ura3Δ0, obtained from Thermo Fisher Scientific) or to W303 (Mata, ade2-1, trp1-1, leu2-3,112, his3-11,15, ura3, ssd1 obtained from Thermo Fisher Scientific). Standard yeast genetics was used to generate the strains. Yeast strains are listed in supplementary material Table S1. PY3295 and SY strains were gifts from D. Pellman (Dana-Farber Cancer Institute, Boston, MA). D. Lew (Duke University, Durham, NC) provided a SWE1 gene knockout plasmid. Gene deletions or modifications were performed with PCR-mediated one-step gene replacement using pFA6a vectors provided by J. Pringle (Stanford University, Stanford, CA) (Longtine et al., 1998) and were confirmed by PCR. The LacO/LacI system for monitoring sister chromatid cohesion was a gift from D. Koshland (UC Berkeley, CA). The ZDS1 plasmid was purchased from the National Bio-Resource Project (NBRP). α-factor was used at 2 µg/ml. Nocodazole was used at 15 µg/ml. Benomyl was used at 12.5 µg/ml, 7.5 µg/ml or 15 µg/ml as indicated in the figure legends. For galactose induction, 2% galactose was added to the medium.

Protein extracts were prepared by trichloroacetic acid (TCA) precipitation as previously described (Piatti et al., 1996). Mouse anti-HA (16B12, Roche and Covance), mouse anti-GFP antibody (Millipore), mouse anti-Myc (9E10, Wako) and anti-PSTAIRE (sc-53, Santa Cruz Biotechnology) antibodies were used. Horseradish peroxidase (HRP)-conjugated secondary antibodies were obtained from Millipore, and proteins were detected by an enhanced chemiluminescence system (ECL Prime; GE Healthcare).

For FACS (fluorescence-activated cell sorting) analysis, cells were collected by centrifugation and then fixed in 70% ethanol. Cells were then washed once with 1 ml of 50 mM Tris-HCl pH 7.5, and the pellet was then resuspended in 0.5 ml of 50 mM Tris-HCl pH 7.5 containing 1 mg/ml RNase. After incubation overnight at 37°C, cells were collected by centrifugation and were then washed once with 1 ml of FACS buffer (200 mM Tris-HCl pH 7.5, 200 mM NaCl, 78 mM MgCl₂) and resuspended in the same buffer containing 50 µg/ml propidium iodide.

Logarithmically growing cells were collected and resuspended in YPD containing nocodazole (15 µg/ml). At total of 200 cells were plated on a YPD plate at each time point after nocodazole addition. Once colonies had formed, viability was calculated by dividing the number of colonies formed at the different time points by the number of colonies formed at time 0. The rate of death in the presence of nocodazole is an excellent indicator of checkpoint deficiency (Hoyt et al., 1991).

Wild-type cells arrest in response to SAC activation as large budded cells in mitosis. The checkpoint mutants continue through the cell cycle and they eventually pass through the subsequent G1 generating a new bud. Logarithmically growing cells were treated with nocodazole (15 µg/ml). At each indicated time point, cells were fixed in 70% ethanol at room temperature. Cells were then washed two times in PBS, vortexed vigorously, examined by brightfield microscopy and categorized into G1 (unbudded), dumbell (large budded) and rebudded cells (more than one bud). Sister chromatid separation was monitored by visualizing the right arm of chromosome XV using the LacO/green fluorescent protein (GFP)-LacI system (Straight et al., 1996).

Fluorescence images were acquired with a fluorescence microscope (Eclipse E600; Nikon) equipped with a charge-coupled device camera (DC350F; Andor) with 100×, 1.45 NA, or 60×, 1.4 NA, oil objectives. The images were captured and analyzed with NIS-Elements software (Nikon), and the figures were processed and assembled in Photoshop (Adobe) or Image J. Fluorescence images of Zds1–GFP in the presence of Nup159–mCherry (supplementary material Fig. S2) were acquired with a spinning disk confocal microscope (Zeiss AxioObserver) with a Plan Apochromat 63×, 1.4 NA oil M27 lens with a DIC prism. The images were captured, analyzed with SlideBOOK^TM software and the figures were processed and assembled in ImageJ or Photoshop.

We thank Andrew Murray, Akio Toh-e, David Pellman, Daniel Lew, John Pringle, Raymond J. Deshaies, Kim Nasmyth, Jim Haber, Thomas Eng, Douglas Koshland and Simonetta Piatti for the gift of the reagents.