The budding yeast Polo-like kinase Cdc5 is released from the nucleus during anaphase for timely mitotic exit

Abstract

Polo-like kinases are important regulators of multiple mitotic events; however, how Polo-like kinases are spatially and temporally regulated to perform their many tasks is not well understood. Here, we examined the subcellular localization of the budding yeast Polo-like kinase Cdc5 using a functional Cdc5-GFP protein expressed from the endogenous locus. In addition to the well-described localization of Cdc5 at the spindle pole bodies (SPBs) and the bud neck, we found that Cdc5-GFP accumulates in the nucleus in early mitosis but is released to the cytoplasm in late mitosis in a manner dependent on the Cdc14 phosphatase. This Cdc5 release from the nucleus is important for mitotic exit because artificial sequestration of Cdc5 in the nucleus by addition of a strong nuclear localization signal (NLS) resulted in mitotic exit defects. We identified a key cytoplasmic target of Cdc5 as Bfa1, an inhibitor of mitotic exit. Our study revealed a novel layer of Cdc5 regulation and suggests the existence of a possible coordination between Cdc5 and Cdc14 activity.

Keywords:

• Polo-like kinase
• Cdc5
• Cdc14
• Bfa1
• FEAR
• MEN
• mitosis
• mitotic exit

Abbreviations

FEAR	=	Cdc Fourteen Early Anaphase Release
MEN	=	Mitotic Exit Network

Introduction

Polo-like kinases are essential mitotic regulators that are highly conserved from yeast to humans. Since the first discovery of the polo gene in D. melanogaster,¹ Polo-like kinases have been identified to play many important roles in mitosis, including mitotic entry, spindle assembly, chromosome segregation, mitotic exit and cytokinesis.² How Polo-like kinases perform so many different functions coordinated both in space and in time is not fully understood.

In budding yeast, the Polo-like kinase Cdc5 plays important roles in many aspects of mitosis.³ First, Cdc5 promotes mitotic entry by promoting degradation of Swe1, an inhibitor of the mitotic Cdk.^4-6 In metaphase, Cdc5 facilitates chromosome segregation. Cdc5 phosphorylates the cohesin subunit Scc1 to allow its separase-dependent cleavage, which is essential for sister chromatid separation.^7-9 In early anaphase, Cdc5 promotes Cdc14 early anaphase release (FEAR) by phosphorylating the Cdc14 sequestering protein Net1/Cfi1 in the nucleolus.^10-13 The nucleolar release of Cdc14 phosphatase is essential for stabilizing the anaphase spindle,¹⁴ nucleolar segregation¹⁵ and primes mitotic exit by removing Cdk-dependent phosphorylation.^16-18 In addition to FEAR, Cdc5 also regulates chromosome condensation by phosphorylating condensin subunits Brn1, Ycg1, and Ycs4 in anaphase.¹⁹ Cdc5 is also an essential regulator of mitotic exit. A canonical signaling module called the mitotic exit network (MEN) promotes full activation of Cdc14 and thus is essential for mitotic exit in budding yeast.²⁰ Cdc5 activates the MEN by at least two pathways. First, Cdc5 phosphorylates and inactivates Bfa1, an inhibitor of MEN.^21-24 Cdc5 also phosphorylates Cdc15, a kinase that is essential for MEN.²³ Finally, Cdc5 is important for cytokinesis. Cdc5 phosphorylates Rho1 GEFs in anaphase and promotes the local activation of Rho1 at the cell division site for contractile actin ring assembly.²⁵ Thus, Cdc5 activity and functions need to be tightly regulated both in space and in time.

An important question is how Cdc5 can phosphorylate each substrate in a temporal order. Because kinase activity of Cdc5 is high from G2 phase to late mitosis,²⁶ it is likely that the interactions between Cdc5 and its substrates are spatially controlled. Supporting this idea, it has previously been reported that Cdc5 localizes to various subcellular structures/compartments during the cell cycle.^5,26-31 Cdc5 can localize to the SPBs, at the bud neck, and in the nucleus,²⁷ but the timing of Cdc5 localization to the specific subcellular compartments remains little explored. Cdc5 binding to its substrates depends on its polo-box domain (PBD)^32,33 and although previous studies have characterized Cdc5 PBD mutant function,^27,28 the role that Cdc5 substrate binding plays on its localization during mitosis has not been fully explored.

In this paper, we analyzed the spatial distribution of Cdc5 during the cell cycle. By taking advantage of both newly generated and previously published cdc5 alleles, we show that Cdc5 is exported from the nucleus to the cytoplasm in anaphase through an active mechanism. We further identify Cdc14 as a regulator of Cdc5 nuclear export and demonstrate that Cdc5 nuclear export is required for MEN activation and for timely mitotic exit. Our data suggest the existence of a possible positive feedback loop between Cdc5 and Cdc14 to regulate timely exit from mitosis.

Results

Cdc5 changes its localization during the cell cycle

To carefully investigate subcellular localization of Cdc5 during the cell cycle, we added a GFP tag to the carboxyl terminus of Cdc5 and expressed it from the endogenous locus. This CDC5-GFP construct is functional because the cells expressing CDC5-GFP as sole copy of CDC5 exhibited no obvious growth defects. In a log phase culture, Cdc5-GFP signal is absent in unbudded G1 cells (Fig. 1A) consistent with the fact that Cdc5 is targeted for degradation after mitosis by Cdh1.^34,35 In small to medium budded cells, which correspond to S or G2 phase, Cdc5-GFP starts to localize to the SPBs (Fig. 1A) as reported previously.^27,28,31 In large budded metaphase cells, we saw accumulation of Cdc5-GFP in the nucleus in addition to the SPBs. This is consistent with previous work showing that Cdc5 is transported into the nucleus upon mitotic entry.³⁶ In anaphase cells with fully segregated nuclei, Cdc5-GFP was localized at both SPBs and at the bud neck. The SPB signal of Cdc5-GFP was generally brighter in the daughter cell than mother cell in anaphase. We also noticed that nuclear Cdc5-GFP signal was significantly reduced in late mitotic cells compared to metaphase cells (Fig. 1A).

Figure 1. Cdc5-GFP localization during the cell cycle. (A) Representative images of live cells showing the localization of Cdc5-GFP expressed from the endogenous locus (SY1653) during the cell cycle. The arrow points to the elongated nucleus which has reduced Cdc5-GFP signal compared to metaphase. (B) Cdc5-GFP localization in cdh1Δ (SY1202) cells. The numbers indicate metaphase (1), anaphase (2), and post-mitotic (3) cells.

Because Cdc5 is degraded upon mitotic exit by an APC-Cdh1 dependent manner,^31,34 it is possible that the reduction of nuclear Cdc5-GFP signal in anaphase was due to degradation. In cdh1Δ, Cdc5-GFP is accumulated in the nucleus in metaphase (Cells 1 in Fig. 1B) and released from the nucleus and localized to the the SPBs and at the bud neck (Cells 2 in Fig. 1B) similar to wild type. Shortly after mitotic exit, as judged by loss of SPB and bud neck signals and segregated nuclei (Cells 3 in Fig. 1B), Cdc5-GFP accumulates back in the nucleus. The localization pattern of Cdc5-GFP in cdh1Δ suggests that there is an active nuclear transport mechanism for Cdc5 specifically during anaphase.

FEAR and Cdc14 activity are required for nuclear release of Cdc5 in anaphase

To identify the possible regulators that contribute for nuclear release of Cdc5 in anaphase, we examined the effects of gene deletion mutants implicated in late mitotic events (the gene deletions tested are listed in the Table 1). First, we examined the effect of the major mitotic cyclin Clb2. We found that Cdc5-GFP remains accumulated in the nucleus in anaphase in clb2Δ (Fig. 2A). This effect is specific to Clb2 because deletion of an S-phase cyclin CLB5 did not affect Cdc5 localization (Table 1). We also found a strong nuclear signal of Cdc5 in anaphase in slk19Δ and spo12Δ cells (Fig. 2B), which are defective in Cdc14 release.²⁰ Because Clb2 is also implicated in Cdc14 release,³⁷ we suspected that Cdc14 has a key role in Cdc5 localization in late mitosis. Indeed, in cdc14-1, we found that Cdc5 was enriched in the nucleus in anaphase (Fig. 2C).

Table 1. Analysis of Cdc5-GFP nuclear accumulation in anaphase cells in various mitotic mutants. Nuclear localization of Cdc5-GFP was judged ranging from no detectable nuclear Cdc5-GFP signal (−) to strong Cdc5-GFP nuclear signal (+++)

Mutant	Function	Cdc5-GFP anaphase nuclear accumulation
clb2Δ	Mitotic cyclin	+++
clb5Δ	S-phase cyclin	−
cdc14-1	Protein phosphatase	+++
spo12Δ	FEAR	++
slk19Δ	FEAR	+++
lte1Δ	MEN	−/+
cdc15-2	MEN	−
bfa1Δ	MEN inhibitor	−
rts1Δ	Protein phosphatase 2A	−
cdc55Δ	Protein phosphatase 2A	−
ste20Δ	p21-activated kinase	−
cla4Δ	p21-activated kinase	−/+
ase1Δ	Anaphase spindle elongation protein	−
hog1Δ	Mitogen activated protein kinase	−
nup57Δ	Nucleoporin	−
msn5Δ	Karyopherin	−
dma1/2Δ	Ubiquitin protein E3 ligase	−

Figure 2. Nuclear release of Cdc5 in anaphase depends on the mitotic cyclin Clb2 and the Cdc14 phosphatase. (A and B) Representative images of Cdc5-GFP in clb2Δ (SY689), slk19Δ (SY1387), and spo12Δ (SY1579) live cells cultured at 25°C. The arrows indicate nuclear accumulation of Cdc5-GFP in anaphase. (C) Representative image of Cdc5-GFP in WT (VB467) and cdc14-1 (SY1382) cultured at 30°C.

The mechanism by which Cdc14 regulates Cdc5 localization

Cdc14 is a phosphatase that dephosphorylates Cdk1 substrates.¹⁷ To understand the mechanism by which Cdc14 regulates the localization of Cdc5 in anaphase, we considered three possibilities. First, Cdc14 could directly dephosphorylate Cdc5 because Cdc5 is a substrate of Cdk1.³⁸ Second, Cdc14 could dephosphorylate Cdc5 anchoring proteins in the nucleus and thereby promote the liberation of Cdc5 from the nucleus, because Cdc5 localization often depends on docking of the PBD to phospho-proteins.³² Third, Cdc14 could indirectly regulate Cdc5 localization by dephosphorylating regulators of Cdc5 nucleocytoplasmic transport.

To test the first hypothesis, we have introduced mutations into the Cdk-dependent phosphorylation sites on Cdc5. Cdc5 is known to be phosphorylated by Cdk1/Clb2 at 5 residues: T23, T70, T242, S419, and S657.³⁸ Because phosphorylation of T242 is essential for Cdc5 kinase activity, it is unlikely that T242 is dephosphorylated in vivo in anaphase because Cdc5 kinase activity remains high until mitotic exit.^26,38 Therefore, we mutated the remaining 4 residues to unphosphorylatable alanine (cdc5-4A-GFP) or phosphomimetic glutamate (cdc5-4D-GFP). We found that both cdc5-4A-GFP and cdc5-4D-GFP expressed from a low-copy centromeric plasmid were able to complement the viability of a strain which had endogenous Cdc5 depleted (Fig. 3A). If Cdc14-dependent dephosphorylation of Cdc5 promotes its cytoplasmic localization, Cdc5-4A-GFP should be readily excluded from the nucleus. For this experiment, we slightly overexpressed CDC5-GFP from a low copy number plasmid. In this condition, both Cdc5-GFP and Cdc5-4A-GFP were enriched in the nucleus even in anaphase (Fig. 3B), possibly due to over-dosage. Because there was no obvious change in Cdc5-4A-GFP nuclear signal compared to Cdc5-GFP in this experiment, it is unlikely that Cdc14 is regulating Cdc5 localization by its direct dephosphorylation.

Figure 3. Involvement of Cdc5 phosphorylation by Cdk1 or the PBD of Cdc5 in nuclear release of Cdc5. (A) cdc5-4A-GFP and cdc5-4D-GFP are functional and complement depletion of CDC5. SY1619 (cdc5Δ::GAL1-CDC5), whose sole CDC5 gene is under glucose-repressible GAL1 promoter, was transformed with a low copy number plasmid containing either cdc5-4A-GFP (pVR35) or cdc5-4D-GFP (pVR36) and was tested for growth in galactose (GAL) or glucose (GLU) containing media for 2 days at 30°C. (B) Localization of Cdc5-4A-GFP expressed from a low copy number plasmid in a wild type strain (PY3295) at room temperature. Note that expression of Cdc5-GFP from a plasmid (pVR13) resulted in nuclear accumulation in anaphase most likely because of overexpression. Scale bar: 5 μm. (C) Localization of Cdc5 (pVR13) and of PBD mutants Cdc5-16-GFP (pVB38) and Cdc5-FAA-GFP (pVB39) expressed from a low copy number plasmid in a wild type strain (PY3295) at room temperature. Although the SPB and the bud neck signals of Cdc5-16-GFP and Cdc5-FAA-GFP were significantly impaired, their nuclear localization pattern was indistinguishable from wild type Cdc5-GFP. Scale bar: 5 μm.

We then tested our next hypothesis: the involvement of the PBD in nuclear localization. Because Cdk1 creates a phosphorylation-dependent docking site for the PBD of Cdc5,³² Cdc14 could release Cdc5 from the nucleus via removing these docking sites by dephosphorylation. To test if Cdc5 localizes to the nucleus due to anchoring via the PBD, we used previously described PBD mutants of Cdc5. First, we used cdc5-16,^28,32,39 which is defective in binding to the proteins primed by phosphorylation and cdc5-FAA,²⁷ which is completely defective in PBD function. If Cdc5 is anchored by PBD-dependent binding, we expect to see a reduction of nuclear Cdc5 signal in the PBD mutants. Consistent with previous reports,^27,28 both Cdc5-16-GFP and Cdc5-FAA-GFP failed to localize to the SPBs and bud neck even when slightly overexpressed as an exogenous copy (Fig. 3C). However, nucleocytoplasmic distribution of these mutants was comparable to wild type in this condition. Therefore, it is unlikely that removal of Cdc5 docking sites by Cdc14 is solely responsible for Cdc5 nuclear delocalization.

Taken together, the molecular mechanism by which Cdc14 promotes Cdc5 delocalization from the nucleus remains elusive. However, our analysis suggests that Cdc14-dependent nuclear release of Cdc5 is unlikely through direct regulation of Cdc5 but rather yet to be identified nuclear export or nuclear import factors are involved.

Release of Cdc5 from the nucleus in anaphase is important for mitotic exit

To understand the physiological relevance of Cdc5 delocalization from the nucleus in anaphase, we artificially enriched Cdc5 in the nucleus by adding a strong nuclear localization signal (NLS) of SV40 antigen to the carboxyl terminus of CDC5-GFP to generate CDC5-GFP-NLS (Fig. 4A). For comparison, we also created a CDC5-GFP-NES in a similar manner. Cells expressing cdc5-GFP-NLS or cdc5-GFP-NES from the endogenous CDC5 locus as the only copy of Cdc5 were viable, however, the cdc5-GFP-NLS strain showed mild growth defects at 37°C (Figure 4B). As predicted, Cdc5-GFP-NLS remained accumulated in the nucleus throughout the cell cycle including in anaphase (Fig. 4C). Consistent with this, no bud neck signal was observed in cdc5-GFP-NLS in anaphase. Cdc5-GFP-NES was excluded from the nucleus throughout the cell cycle including in metaphase when wild-type Cdc5-GFP is accumulated in the nucleus (Fig. 4D).

Figure 4. Generation of Cdc5 mutants with altered nucleocytoplasmic distributions. (A) Schematic representation of Cdc5 mutant constructs. (B) cdc5Δ::CDC5-GFP (VB467), cdc5Δ::cdc5-GFP-NES (VB468), and cdc5Δ::cdc5-GFP-NLS (VB543) are viable both at 25° and 37°C. (C) Localization of Cdc5-GFP (VB813), Cdc5-GFP-NES (VB918), and Cdc5-GFP-NLS (VB815) in metaphase and anaphase cells. The nuclear pore protein Nup159-mCherry was used as a nuclear marker. (D) Localization of Cdc5-GFP-NES (VB918) in metaphase compared to Cdc5-GFP (VB813). The arrow points to the nucleus of Cdc5-GFP-NES, which has no detectable GFP signal.

We monitored the cell cycle kinetics of CDC5-GFP, cdc5-GFP-NLS, and cdc5-GFP-NES cells by staining their mitotic spindle and nuclei after release from G1 (Fig. 5D). In cdc5-GFP-NES strain, we noticed no or little defect in the timing of spindle assembly (Fig. 5B); however, there was a slight delay in the timing of spindle elongation suggesting that nuclear accumulation of Cdc5 is important for mitotic entry and/or anaphase onset.^8,36 cdc5-GFP-NES strain exists from mitosis with a 10 minutes delay, but this is likely reflected by the delay in the initiation of mitosis (Fig. 5B).

Figure 5. Release of Cdc5 from the nucleus in anaphase is important for timely mitotic exit. Cell cycle progression of CDC5-GFP (VB467) (A), cdc5-GFP-NES (VB468) (B), and CDC5-GFP-NLS (VB543) (C) were analyzed after G1 release at 25°C. Cell cycle progression was judged by the budding index and by the length of mitotic spindle immunostained by anti-alpha tubulin antibody. (D) A representative image of tubulin and DAPI staining at the time point 100 min of (A) is shown. Scale bar: 5 μm.

The cdc5-GFP-NLS strain has a slight delay (maybe 10-minutes) in early mitosis. This is possibly due to delay in Swe1 degradation, which occurs in the cytoplasm.⁶ Strikingly, we found a 20–30 min delay in mitotic exit (Fig. 5C). This result strongly suggests that Cdc5 release from the nucleus in anaphase is required for timely mitotic exit.

cdc5-GFP-NLS is synthetic lethal with FEAR mutants slk19Δ and spo12Δ

The mitotic exit defects of cdc5-GFP-NLS prompted us to test if this mutant has problems in a specific mitotic exit pathway. Cdc5 promotes mitotic exit through 2 distinct pathways. First, Cdc5 promotes Cdc14 release from the nucleolus as a part of the FEAR network.¹⁶ Second, Cdc5 activates the cytoplasmic MEN by inactivating the Bfa1/Bub2 complex^22,24,30 and by activating the Cdc15 kinase.²³ Because Cdc5-GFP-NLS is accumulated in the nucleus in anaphase, we hypothesized that cdc5-GFP-NLS is defective in the MEN pathway.

To test this hypothesis, we tested the genetic interaction between cdc5-GFP-NLS and FEAR and MEN mutants. We found that cdc5-GFP-NLS was synthetic lethal with the FEAR mutants slk19Δ and spo12Δ (Figs. 6A and 6B). This genetic interaction strongly suggests that cdc5-GFP-NLS is impaired in activation of MEN.

Figure 6. Genetic interaction of cdc5-GFP-NLS with mutants defective in mitotic exit. cdc5-GFP-NLS cells harboring a complementing CDC5 plasmid (with a URA3 selective marker) were crossed with slk19Δ, spo12Δ, and lte1Δ. After tetrad dissection, relevant double mutants were spotted onto SC-URA or SC with 5-fluororotic acid (5-FOA which counter selects for URA3+) plates together with each single mutants (VB677, VB683, VB657, VB659, VB661) at 25°C. cdc5-GFP-NLS slk19Δ (VB597, VB598), cdc5-GFP-NLS spo12Δ (VB622, VB624) and cdc5-GFP-NLS lte1Δ (VB645, VB646) all failed to lose the complementing plasmid indicating synthetic lethality.

We also found that cdc5-GFP-NLS is synthetic lethal with deletion of LTE1 (Fig. 6C). Because Lte1 is a non-essential component of MEN,^40-43 synthetic lethality between lte1Δ and cdc5-GFP-NLS further suggests reduced activity of MEN in the cdc5-GFP-NLS mutant.

Cdc5-dependent Inhibition of Bfa1 is Required for Timely Mitotic Exit

Bfa1 is the critical target of Cdc5 in mitotic exit.²² In anaphase, Cdc5 phosphorylates and thereby inactivates Bfa1, allowing MEN to be switched on.^21,22,24 Because our cell cycle and genetic analysis strongly suggest that cdc5-GFP-NLS is defective in activation of MEN, and because Bfa1 localizes to the cytoplasmic surface of the SPB, we hypothesized that cdc5-GFP-NLS is impaired in Bfa1 phosphorylation.

To monitor the phosphorylation status of Bfa1-3HA, we arrested the cells in late anaphase by the cdc14-1 mutation for 3 hours. Consistent with the previous reports,^22,30 in the wild-type control (CDC5-GFP), slower migrating bands of Bfa1-3HA were dominant by Western blotting (Fig. 7A). This slow migration is a result of Cdc5-dependent phosphorylation because it is largely reduced in the cdc5-2 mutant (Fig. 7A). Consistent with our hypothesis, Bfa1-3HA was migrating faster in cdc5-GFP-NLS compared to wild type. This reduced phosphorylation of Bfa1-3HA was specific to cdc5-GFP-NLS because cdc5-GFP-NES did not show this pattern.

Figure 7. Bfa1 is the major cytoplasmic target of Cdc5. (A) Phosphorylation status of Bfa1-3HA in cdc5-GFP-NLS or cdc5-GFP-NES mutant cells. These strains (VB807, VB800, VB915) were arrested in telophase by cdc14-1 mutation at 37° for 3 hours and cell lysates were subjected for the Western blotting of Bfa1-3HA. cdc5-2 BFA1-3HA (VB725) was used as a negative control. Tubulin was used as a loading control. (B) Scc1 phosphorylation is not defective in cdc5-GFP-NLS in metaphase. CDC5-GFP (VB749), cdc5-GFP-NLS (VB753) and cdc5-2 (VB808) cells were arrested in G2/M phase with 15 μg/ml nocodazole for 3 hours at 25°C and shifted to 37°C for 2 hours (to inactivate cdc5-2). The cell lysates were subjected for the Western blotting of Scc1-3HA. Tubulin was used as a loading control. (C) Deletion of BFA1 (VB808) suppresses the temperature sensitivity of cdc5-GFP-NLS (VB572) at 37°C. Indicated strains were streaked to single colonies on YEPD plates at 25°C and 37°C for 2 days. (D) Deletion of BFA1 (VB920) rescues the synthetic lethality of cdc5-GFP-NLS spo12Δ (VB622). Indicated strains harboring a complementing CDC5-URA3 plasmid were spotted on plates either containing 5-FOA or SC-URA at room temperature for 4 days.

It is a formal possibility that cdc5-GFP-NLS has reduced kinase activity. To confirm this that cdc5-GFP-NLS retains kinase activity in vivo, we monitored the phosphorylation status of nuclear Cdc5 substrate Scc1. In nocodazole-arrested metaphase cells, Scc1-3HA was hyper-phosphorylated in both CDC5-GFP and in cdc5-GFP-NLS strains but not in cdc5-2 (Fig. 7B). This result confirms that cdc5-GFP-NLS is able to efficiently phosphorylate its nuclear substrates. Thus, impaired phosphorylation of Bfa1 is due to its spatial separation from cdc5-GFP-NLS and not due to reduced kinase activity.

Lastly, to confirm the hypothesis that cdc5-GFP-NLS is specifically defective in activation of MEN, we asked if deletion of BFA1 was sufficient to rescue the defects associated with cdc5-GFP-NLS. We found that deletion of BFA1 rescued the slow growth phenotype of cdc5-GFP-NLS at the high temperature at 37°C (Fig. 7C). Furthermore, synthetic lethality of cdc5-GFP-NLS spo12Δ was rescued by the deletion of BFA1 (Fig. 7D). These data confirm that cdc5-GFP-NLS is defective in MEN activation and that Bfa1 is a critical cytoplasmic target of Cdc5 important for timely mitotic exit.

Discussion

Cdc5 localization during the cell cycle

In this study, we revisited Cdc5 localization during the cell cycle using a functional CDC5-GFP construct expressed at the endogenous locus. Consistent with previous reports,^5,27-31,36 we found that Cdc5-GFP localizes to the SPBs, nucleus, and bud neck during mitosis. Cdc5-GFP accumulates in the nucleus at the G2/M phase as has been previously described.^28,36 However, we found that Cdc5-GFP signal is lost from the nucleus in anaphase. Because this loss of nuclear Cdc5-GFP signal is not a result of Cdh1-dependent degradation, we suspect that Cdc5 is exported from the nucleus in anaphase. Cdc5-GFP localizes to the bud neck only in anaphase when nuclear Cdc5-GFP signal is reduced. It has been reported that Cdc5 tagged c-terminally with 5 copies of GFP and expressed from the endogenous locus can localize to the bud neck after prolonged nocodazole arrest.²⁷ However, our Cdc5-GFP construct, which contains just one copy of GFP, when expressed at the endogenous level localizes to the bud neck in an unperturbed cell cycle only during anaphase. Our data suggests that a robust bud neck localization of Cdc5 requires nuclear release. Because Cdc5 is a key regulator of cytokinesis, it is of interest to study the correlation of the timing of Cdc5 localization to the bud neck and cytokinetic actin ring assembly.

Possible mechanism of Cdc5 nuclear export

It is widely accepted that Polo-like kinase localization is dependent on its PBD. By binding to phospho-proteins primed by mitotic kinases, the PBD dictates the subcellular localization of Polo-like kinases.⁴⁴ Because FEAR and Cdc14 phosphatase are required for Cdc5 nuclear release in anaphase, we hypothesized that removal of Cdk1-dependent phosphorylation from Cdc5 docking proteins in the nucleus triggers Cdc5 release. However, because Cdc5 PBD mutants defective in substrate binding were not readily released from the nucleus, the PBD is not essential for nuclear retention of Cdc5. Because Cdc5 is known to be phosphorylated by Cdk1 during mitosis,³⁸ an alternative explanation for Cdc5 nuclear release in anaphase could be direct dephosphorylation by Cdc14. However, because the unphosphorylatable cdc5-4A-GFP localized similarly to wild type Cdc5-GFP, it is unlikely that Cdc14 controls Cdc5 localization directly. The caveat of these analysis was that we analyzed the localization of Cdc5-GFP exogenously expresses from a low-copy number plasmid. As has been previously reported,²⁷ when expressed ectopically, Cdc5-GFP can accumulate in the nucleus even in anaphase (Figs. 3B and 3C). We did not observe any noticeable difference in nuclear localization among Cdc5-GFP, Cdc5-16-GFP, Cdc5-FAA-GFP, and Cdc5-4A-GFP in anaphase.

One of the alternative mechanisms by which Cdc14 can regulate Cdc5 nuclear release is via regulating exportins or importins for Cdc5. The identification of the exportins for Cdc5 is an important future question but because Cdc5 is released from the nucleus only during anaphase when Cdc14 is active, Cdc14 might be activating the exportin for Cdc5 in a cell cycle stage specific manner.

Physiological relevance for Cdc5 nuclear release in anaphase

We revealed that Cdc5 release from the nucleus in anaphase is required for timely mitotic exit. Cdc5-GFP-NLS, which is artificially trapped in the nucleus in anaphase (Fig. 4C), showed a delay in spindle disassembly (Fig. 5C) and exhibited synthetic lethality with mutants defective in mitotic exit (Fig. 6). We further showed that Cdc5-GFP-NLS is defective in phosphorylation of MEN inhibitor Bfa1 (Fig. 7A). Indeed, the mitotic exit defect of Cdc5-GFP-NLS can be explained by Bfa1 activity because deleting BFA1 rescued the temperature sensitive growth defect of cdc5-GFP-NLS (Fig. 7C) and synthetic lethality of cdc5-GFP-NLS spo12Δ (Fig. 7D).

It is well established that Cdc5 triggers Cdc14 release from the nucleolus and thereby activates Cdc14.^10-12,20 The requirement of Cdc14 in Cdc5 release suggests a potential positive feedback loop between Cdc14 and Cdc5. The precise mechanism by which Cdc5 is regulated awaits future investigation. However, our study uncovered a novel spatial layer of Cdc5 regulation that is important for timely mitotic exit.

Materials and Methods

Yeast genetics and plasmid construction

All yeast strains in this study were isogenic or congenic to BY4741 (MATa leu2Δ0 his3Δ1 met15Δ0 ura3Δ0). The strains and plasmids are listed in Tables 2 and 3, respectively. Standard yeast genetics was used to generate the strains. PY and SY strains were gifts from D. Pellman (Dana-Farber Cancer Institute, Boston, MA). PJH plasmids were gifts from J. Haber (Brandeis University, Waltham, MA). Gene deletions or modifications were performed using pFA6a vectors provided by J. Pringle (Stanford University, Stanford, CA).⁴⁵ The NLS sequence taken from the SV40 antigen (AAAPKKKRKVG) and the NES sequence (AAALALKLAGLNI) were previously described.^46,47 The NLS or NES sequences were introduced directly after the GFP sequence of CDC5-GFP by PCR-mediated one-step gene replacement.⁴⁵ Alpha-factor was used at 2 μg/ml. Nocodazole was used at 15 μg/ml. All cultures were grown in YPED unless otherwise noted.

Table 2. Strains used in this study

Strain name	Genotype	Mating type
PY3295	leu2Δ0 his3Δ1 met15Δ0 ura3Δ0	a
SY1653	cdc5::CDC5-GFP-HIS3	a
VB467	cdc5::CDC5-GFP::HIS3	a
SY1202	cdc5::CDC5-GFP::HIS3 cdh1Δ::KAN	a
SY689	cdc5::CDC5-GFP::HIS3 clb2Δ::KAN	a
SY685	cdc5::CDC5-GFP::HIS3 clb5Δ::KAN	a
SY1382	cdc5::CDC5-GFP::HIX3 cdc14-1	a
SY1579	cdc5::CDC5-GFP::HIS3 spo12Δ::KAN	a
SY1387	cdc5::CDC5-GFP::HIS3 slk19Δ::KAN	a
SY1204	cdc5::CDC5-GFP::HIS3 lte1Δ::KAN	a
SY1383	cdc5::CDC5-GFP::HIS3 cdc15-2	a
SY1532	cdc5::CDC5-GFP::HIS3 bfa1Δ::KAN	a
SY1536	cdc5::CDC5-GFP::HIS3 rts1Δ::KAN	a
SY1574	cdc5::CDC5-GFP::HIS3 cdc55Δ::KAN	a
SY1682	cdc5::CDC5-GFP::HIS3 ste20Δ::KAN	a
SY1679	cdc5::CDC5-GFP::HIS3 cla4Δ::KAN	a
SY1386	cdc5::CDC5-GFP::HIS3 ase1Δ::KAN	a
SY1540	cdc5::CDC5-GFP::HIS3 hog1Δ::KAN	a
SY1573	cdc5::CDC5-GFP::HIS3 nup57Δ::KAN	alpha
SY1647	cdc5::CDC5-GFP::HIS3 msn5Δ::KAN	a
SY1572	cdc5::CDC5-GFP::HIS3 dma1Δ::KAN dma2Δ::KAN	alpha
VB543	cdc5::cdc5-GFP-NLS::KAN	a
VB468	cdc5::cdc5-GFP-NES::KAN	a
SY1589	cdc5::cdc5-2::NATR	alpha
VB815	cdc5::cdc5-GFP-NLS::KAN NUP159-mCherry::HIS3	a
VB918	cdc5::cdc5-GFP-NES::KAN NUP159-mCherry::HIS3	a
VB813	cdc5::CDC5-GFP::HIS3 NUP159-mCherry::HIS3	a
VB677	cdc5::CDC5-GFP::HIS3 + pJH1966	a
VB683	cdc5::cdc5-GFP-NLS::KAN + pJH1966	a
VB657	cdc5::CDC5-GFP::HIS3 slk19Δ::KAN + pJH1966	alpha
VB597	cdc5::cdc5-GFP-NLS::KAN slk19Δ::KAN + pJH1966	alpha
VB598	cdc5::cdc5-GFP-NLS::KAN slk19Δ::KAN + pJH1966	alpha
VB659	cdc5::CDC5-GFP::HIS3 spo12Δ::KAN + pJH1966	a
VB622	cdc5::cdc5-GFP-NLS::KAN spo12Δ::KAN + pJH1966	alpha
VB624	cdc5::cdc5-GFP-NLS::KAN spo12Δ::KAN + pJH1966	alpha
VB661	cdc5::CDC5-GFP::HIS3 lte1Δ::KAN + pJH1966	a
VB645	cdc5::cdc5-GFP-NLS::KAN lte1Δ::KAN + pJH1966	alpha
VB646	cdc5::cdc5-GFP-NLS::KAN lte1Δ::KAN + pJH1966	alpha
VB807	cdc5::CDC5-GFP::HIS3 BFA1-3HA::HIS3 cdc14-1	a
VB915	cdc5::cdc5-GFP-NLS::KAN BFA1-3HA::HIS3 cdc14-1	a
VB800	cdc5::cdc5-GFP-NES::KAN BFA1-3HA::HIS3 cdc14-1	alpha
VB725	cdc5-2::URA3 BFA1-3HA::HIS3	alpha
VB749	cdc5::CDC5-GFP::HIS3 SCC1-3HA::HIS3	alpha
VB753	cdc5::cdc5-GFP-NLS::KAN SCC1-3HA::HIS3	a
VB808	cdc5-2::URA3 SCC1-3HA::HIS3	a
VB920	cdc5::cdc5-GFP-NLS::KAN spo12Δ::KanMX6 bfa1Δ::HIS3 + pJH1966	alpha
VB572	cdc5::cdc5-GFP-NLS::KAN	alpha
VB821	cdc5::cdc5-GFP-NLS::KAN bfa1Δ::KAN	alpha
SY1589	cdc5-2::NATR from y7092 (Brenda Andrews, University of Toronto)	alpha
SY1619	cdc5Δ::GAL1-3HA-CDC5::KAN	a

Table 3. Plasmids used in this study

Plasmid name	Vector	Insert
pSY222	pRS415	CDC5-GFP
pVR35	pRS415	cdc5-4A-GFP
pVR36	pRS415	cdc5-4D-GFP
pVR32	pRS315	empty
pVB38	pRS415	cdc5-16-GFP
pVB39	pRS415	cdc5-FAA-GFP
pJH1966	pRS316	CDC5 via pKL2690^5

For construction of the pSY222 plasmid, CDC5-GFP fragment containing 500 bps of 5’UTR was PCR amplified using genomic DNA of SY1653 as a template and was inserted into pRS415 vector using KpnI and HindIII. pVR35, pVR36, pVB38, and pVB39 were derived from pSY222 by site-directed mutagenesis using QuikChange II XL Site-Directed Mutagenesis Kit (Agilent; 20051).

Biochemistry

For Western blotting, the cell lysates were prepared by trichloroacetic acid (TCA) method.⁴⁸ Mouse monoclonal anti-HA antibody (16B12, Covance) and rat anti-α tubulin monoclonal antibody (MCA78G, AbD Serotec) were used at 1: 3,000 dilution in TBST containing 5% skim milk. Horseradish peroxidase (HRP)-conjugated mouse secondary antibody (Millipore) was used at 1:10,000 dilution in TBST containing 5% skim milk. The signal was detected by Amersham ECL Prime Western Blotting Detection Reagent (GE Healthcare Life Sciences; RPN2232).

Fluorescence microscopy

Images were acquired using the Eclipse E600 (Nikon) epifluorescence microscope equipped with a charge-coupled device camera (DC350F; Andor) with a 60×, NA 1.4× oil objective. The images were captured and analyzed with NIS-Elements software (Nikon).

Cells for live imaging were cultured in synthetic complete (SC) or SC-Leu medium. Cell preparation for immunofluorescence was performed as previously described.⁴⁷ The mitotic spindle was visualized with a rat anti-α tubulin rat monoclonal antibody (MCA78G, AbD Serotec) conjugated with an Alexa Fluor 568 goat anti-rat antibody (Molecular Probes; A-11077). Nuclei were visualized with DAPI.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Drs. Jim Haber, Kyung Lee, John Pringle, David Pellman, Yoshiaki Kamada, and Brenda Andrews for the gifts of reagents. We are grateful to members of the Yoshida and Haber labs for discussion and sharing reagents, especially Yana Ostrovsky for her technical help.