Cdk1 and SUMO Regulate Swe1 Stability

The Swe1/Wee1 kinase phosphorylates and inhibits Cdk1-Clb2 and is a major mitotic switch. Swe1 levels are controlled by ubiquitin mediated degradation, which is regulated by interactions with various mitotic kinases. We have recently reported that Swe1 levels are capable of sensing the progress of the cell cycle by measuring the levels of Cdk1-Clb2, Cdc5 and Hsl1. We report here a novel mechanism that regulates the levels of Swe1. We show that S.cerevisiae Swe1 is modified by Smt3/SUMO on residue K594 in a Cdk1 dependant manner. A degradation of the swe1K594R mutant that cannot be modified by Smt3 is considerably delayed in comparison to wild type Swe1. Swe1K594R cells express elevated levels of Swe1 protein and demonstrate higher levels of Swe1 activity as manifested by Cdk1-Y19 phosphorylation. Interestingly this mutant is not targeted, like wild type Swe1, to the bud neck where Swe1 degradation takes place. We show that Swe1 is SUMOylated by the Siz1 SUMO ligase, and consequently siz1Δ cells express elevated levels of Swe1 protein and activity. Finally we show that swe1K594R cells are sensitive to osmotic stress, which is in line with their compromised regulation of Swe1 degradation.

Small Ubiquitin-related MOdifier (SUMO, Smt3, 17% identical to ubiquitin) is conjugated to its targets by a system analogous to ubiquitin. Smt3 is activated in an ATP-dependent reaction by thioester bond formation with the E1 activator Aos1/Uba2 [11], transferred to the E2 ligase Ubc9 [12] and passed to a substrate lysine, usually in the sequence WKxD/E, where W is a hydrophobic amino acid, and x is any amino acid. There are four SUMO-E3 ligases in S. cerevisiae; Siz1, Siz2 [13], Mms21 [14], and the meiotically expressed Cst9 [15]. Siz1 is responsible for the majority of vegetative growth sumoylation, with Siz2 conducting most of the remainder [13]. Despite Smt3, Aos1, Uba2 and Ubc9 being essential genes [11,12], a siz1Dsiz2D strain is viable [13], albeit with a clonal lethality, manifested by a nibbled phenotype which is caused by the 2m plasmid [16]. In contrast, mms21D cells are not viable, though mutations in the RING finger domain that abolish its SUMO-ligase activity such as mms21 sp , mms21-11 or mms21 CH are not lethal [14,17,18], suggesting that Mms21 executes another, non-SUMO, essential function. Whereas other SUMO-E3 ligases are nuclear, Siz1 is additionally localized to the bud-neck [13,14,19,20]. Many proteins have been reported to be SUMOylated with effects being substrate dependent but including ubiquitin mediated proteolysis and re-localization. Different types of proteins are known to be SUMO substrates, many of them are involved in DNA replication stress response.

Yeast growth, synchronization and manipulation
Yeasts were transformed by the frozen lithium acetate method [21] and are listed in Table S1. Plasmids used are listed in Table  S2. Strains containing Cdk1 as1 , cdk1 Y19F and swe1 18A were kind gifts from D. Kellogg [6]. Strains in W303u lacking hex3 or slx8 were kind gifts from X. Zhao [22]. Strains in the JD52u background lacking SUMO E3 ligases were kind gifts from E. Johnson [13,17]. Mutagenesis of plasmids to introduce K594R and K328R mutations into Swe1 was performed using the Stratagene Quikchange kit and verified by sequencing. Swe1 was tagged with 6myc using pRS306-S6M or pRS306-S6M-K594R cut with ClaI, or by using pRS405-S6M cut with SnaBI. Taggings and knockouts were confirmed by PCR. Standard Yeast-Peptone and synthetic media (pH 5.8) supplemented with the appropriate carbon source (2%) was used throughout. Cells were grown at 30uC. S-phase arrest and release was achieved by releasing cells from G1 arrest (growth to saturation) for 1 hour, adding 0.2 M hydroxyurea (Sigma) for 2 h followed by three washes with DDW and release into media containing 5 mg/ml nocodazole (Sigma). For growth rates of cells under stress conditions, OD 600 was measured before and after 9 hours incubation in YPD with 0.75 M NaCl, 7.5 mM caffeine or water. For osmolarity experiments, cells were synchronized by two doses of alpha factor (5 mg/ml) and released for 1 hour prior to addition of 0.5 M NaCl. For pulse chases, Swe1-3myc was expressed for one hour following release from G1 arrest and glucose added to halt transcription, or was expressed for 1 hour in cdk1 as1 cells arrested in G2 with 0.5 mM 1NM-PP1 after which cells were released back into the cell-cycle. Cycloheximide was not added, to allow the cell cycle to continue.
For immunoprecipation, cells were released from G1 arrest for 1 hour and then 0.5 mM 1NM-PP1 (Toronto Research Chemicals) added for an additional 30 minutes. Cells were lysed in the presence of 10 mM NEM, yeast protease inhibitors (1006, Sigma) and yeast phosphatase inhibitor cocktails 1 and 2 (1006, Sigma). Extracts were incubated with mouse anti-c-myc (a kind gift from T. Ravid) overnight and with anti-Protein A beads for 2 hours before extraction into 26 sample buffer.
For Swe1-GFP experiments, Swe1-GFP was expressed for 90 minutes following release from G1 arrest and glucose added to halt transcription. Cells were immobilized using ConA on glass bottom tissue culture dishes (Mattek) and filmed using a Roper CCD Camera mounted on an Olympus IX70 microscope with a 606oil objective.

Swe1 is SUMOylated on K594 in a Cdk1 dependent manner
Ubiquitin mediated degradation of Swe1 is one of the major modes of its cell cycle specific activity. The regulation of this degradation has been found to be exceptionally complex and dependent on a multitude of cellular inputs. Cdk1-Clb and Cdc5 activities are both required for destruction of Swe1, however interaction between Swe1 and Clb2 is dispensable [6,10]. As Cdk1 activity is not required for Cdc5 to be active [5] we considered other potential mechanisms as to how Cdk1 regulates elimination of Swe1. The SUMO E3 ligase Siz1 is a Cdk1 substrate [23] whose stability is negatively regulated by the replication-fork poison hydroxyurea and by inhibition of Cdk1 [24]. We explored the possibility that SUMO is involved in Swe1 regulation.
Immuno-precipitation of a 6myc tagged Swe1 shows Swe1 to be SUMOylated ( Figure 1A left lane). Tagged Swe1 was also precipitated from extracts of cdk1 as1 cells, which expresses cdk1 that is sensitive to the ATP analogue 1NM-PP1 [25]. In the presence of the inhibitor the precipitated Swe1 was not SUMOylated ( Figure 1A middle lane), suggesting that sumoylation of Swe1 depends on Cdk1 activity.
Analysis of the Swe1 sequence with the SUMO site prediction program SUMOplot (Abgent) found a high probability (0.73) SUMOylation site at K594, which is within the Swe1 kinasedomain. This lysine is conserved from yeasts to man, with the SUMOylation site present in sensu stricto fungi (Table 1). Figure 1B shows that a prominent band of 110 kDa was detected when whole-cell extracts were probed with anti-Smt3. Strikingly this band was significantly reduced in extracts of cells expressing Swe1 K594R as their sole copy of Swe1 and almost absent from swe1D cells. Figure 1C shows that upon treatment with 0.2 M hydroxyurea, the levels of Smt3 at the 110 kDa band declines, whilst, as previously reported, Swe1 accumulates [26]. Together these observations suggest that SUMOylation of Swe1 on K594 is Cdk1-Clb dependent.
We used the elongated-bud phenotype that occurs upon overexpression of wild type Swe1 [1] to determine whether the swe1 K594R mutant could inhibit Cdk1-Clb2. Overnight expression of a single copy of swe1 K594R from a GAL promoter resulted in elongated buds, similarly to overexpression of wild-type Swe1 ( Figure 1D), indicating that this mutant is active. In contrast, overexpression of swe1 K328R , whose mutation is in the region involved in interaction with Hsl1 [8] did not produce elongated buds.

Swe1 K594R cells express elevated levels of Swe1 protein and activity
Wild-type and swe1 K594R cells were released from hydroxyurea arrest into nocodazole (see [26]), and levels of Swe1 and Cdc5 as well as phosphorylation of Cdk1-Y19 were monitored. Figure 2A and Figure S1 show that swe1 K594R cells displayed prolonged phosphorylation of Cdk1-Y19 and a concomitant delay in Cdc5 accumulation and Swe1 destruction. Pulse-chase upon release from G2 arrest of Swe1-3myc expressed from a GAL promoter as the only source of Swe1 in the cell likewise showed swe1 K594R activity to be prolonged with a concurrent delay in Cdc5 accumulation and Swe1 destruction ( Figure 2B). This novel mechanism to regulate Swe1 level could be one of the ways that Cdk1 regulates Swe1 destruction, since direct interaction and phosphorylation of Swe1 by Cdk1-Clb2 is not a prerequisite for Swe1 destruction [6,10].
As bud-neck localization is involved in Swe1 destruction [4,5], the localization of swe1 K594R -GFP was examined. Whereas wildtype Swe1-GFP showed bud-neck localization prior to its disappearance, swe1 K594R -GFP did not accumulate at the budneck and was diffuse throughout the cell ( Figure 2C). It is thus conceivable that the elevated levels of swe1 K594R and its delayed degradation are due to the lack of its targeting to the bud neck.
The increased stability of swe1 K594R does not depend on its kinase activity or its interaction with Cdk1 Increased Swe1 activity could delay entry into G2 and subsequent Swe1 destruction, but it is also possible for the increased, persistent Swe1 activity to be a consequence of higher intrinsic Swe1 levels in the swe1 K594R mutant cells. Indeed, swe1 K594R is present at higher levels than wild type Swe1 during log-phase growth ( Figure 1B) or when cells are arrested in S-phase ( Figure 2A). To determine whether swe1 K594R levels are linked to Swe1 activity, swe1 K594R levels were examined in cells expressing cdk1 Y19F , which lacks the Swe1 phosphorylation site on Cdk1. Figure 3A shows that in these mutants, in which Swe1 activity could not phosphorylate Cdk1, Swe1 levels remained higher in Swe1 K594R cells but without a delay in the initiation of its destruction.
To eliminate the possibility that Swe1 is affecting Cdk1 by binding to it we tested the effect of the K594R mutation on the accumulation of swe1 18A , in which all 18 Cdk1 phosphorylation sites were mutated to alanine resulting in a loss of Swe1 kinase activity and lack of interaction with Cdk1-Clb2 [6]. Figure 3B shows that also in this case, swe1 K594R was present at higher levels in S-phase arrested cells and again the initiation of its destruction was not delayed. Therefore, the increased stability of swe1 K594R is independent of its kinase activity and its interaction with Cdk1-Clb2.

Siz1 is the SUMO E3 ligase of Swe1
Of the three SUMO E3-ligases expressed during a mitotic cell cycle, the bulk of SUMOylation is carried out by Siz1 [13]. Levels of Swe1 in cells with knocked-out or inactive SUMO E3-ligases were examined under unstressed conditions or upon arrest in Sphase. Figure 4A shows that Swe1 levels and activity were elevated in siz1D cells. In contrast, Swe1 levels were not elevated above WT levels in siz2D or mms21 sp cells, even though mms21 deficient cells are hypersensitive to hydroxyurea [27] whereas siz1D cells are not sensitive [13,28] (and E Johnson, personal communication). Further knockout of SIZ2 did not significantly elevate Swe1 levels in siz1D cells. As expected deletion of the meiotic SUMO E3 ligase Cst9 also did not increase Swe1 abundance ( Figure S3A). Figure 4B shows that indeed the 110 kDa SUMO band corresponding to Swe1 was absent in siz1D cells.
Pulse chase of Swe1-3myc in wild-type and siz1D cells shows Swe1 to be degraded slower in siz1D cells ( Figure 4C). Similarly to swe1 K594R , Swe1 levels were elevated in siz1D cells when growing in mid-log phase or when arrested in S-phase. When released from hydroxyurea arrest, siz1D cells showed a considerable delay in dephosphorylation of Cdk1-Y19 and a concomitant lag in Cdc5 accumulation and Swe1 destruction ( Figure 4D and Figure S2). Although the effects of the Siz1 deletion could be through lack of SUMOylation of other proteins, the congruence of phenotypes with swe1 K594R suggests Siz1 does indeed SUMOylate Swe1 on K594. Furthermore, Siz1 is the only SUMO E3-ligase that colocalizes with Swe1 at the bud-neck, suggesting that the bud-neck may serve as an organizing platform for components of the Swe1 regulatory pathway additionally to Hsl1/7, Cdc5 and Clb2. SUMOylation of Swe1 by Siz1 is thus a novel mechanism by which Cdk1-Clb regulates Swe1 stability.
A subset of SUMOylated proteins are ubiquinylated by Hex3/ Slx8 or Ris1 Figures S3A and S3B) [29,30]. Cdc5 accumulation following release from S-phase arrest was delayed in hex3D or slx8D cells, probably as a consequence of the sensitivity of these strains to hydroxyurea [30] but without affecting Swe1 abundance. This observation indicates that SUmoylation is unlikely to be required for Swe1 degradation per se.

Swe1 K594R cells are osmosensitive
Swe1 plays a central role in stress response and adaptation, consequently perturbations in Swe1 regulation often lead to stress Figure 2. Swe1 K594R is more stable and expressed in higher levels then wild type Swe1. A. Swe1 activity is more persistent with a delay in Cdc5 accumulation and destruction of Swe1 in swe1 K594R cells. W303a cells were arrested for 2 hours with 0.2 M hydroxyurea and released into fresh media containing 5 mg/ml nocodazole. B. Swe1 activity is more persistent with a delay in Cdc5 accumulation and destruction of Swe1 when swe1 K594R -3myc is expressed from a GAL promoter for 1 hour in cdk1 as1 swe1D cells. Cells were released into media containing 5 mg/ml nocodazole and 4% glucose to shut-off Swe1-3myc expression. C. Localisation of Swe1-GFP and swe1 K594R -GFP expressed from a GAL promoter for 90 minutes. doi:10.1371/journal.pone.0015089.g002  sensitivity. We stressed wild type and swe1 K594R cells with caffeine and with elevated levels of NaCl and determined their growth rate. Caffeine activates the cell-wall integrity (CWI) pathway and leads to dual-phosphorylation of the Slt2 MAPK [31]. Hyperosmolarity activates the Hog1 MAPK [32] pathway. Figure 5A shows that, in comparison to wild type cells, swe1 K594R cells are sensitive to osmotic stress but not to caffeine. Cells were synchronized with afactor, released into S-phase and the phosphorylation of Cdk1-Y19 and dual-phopshorylation of Hog1 monitored upon addition of 0.5 M NaCl. Phosphorylation of Cdk1-Y19 peaked at 40 minutes following stress, but was stronger and more persistent in the swe1 K594R cells ( Figure 5B and C). Hog1 was equally dually-phosphorylated after 5 minutes in both wild-type and swe1 K594R cells, but dual-phosphorylation of Hog1 attenuated more rapidly in swe1 K594R cells ( Figure 5B and C). Together, this increased G2 delay and reduced Hog1 signalling duration could account for the osmo-sensitivity of swe1 K594R cells.
Why would increasing Swe1 abundance through the swe1 K594R mutation affect sensitivity to hyperosmotic stress but not to CWI stress? Whereas the only effector of hyperosmotic stress signalling upon Cdk1-Clb2 activity is through regulation of Swe1 destruction [33] CWI stress regulates Cdk1-Clb2 activity through multiple mechanisms [31] and thus the stress response is probably more buffered against increases in Swe1 abundance. Figure 5. Swe1 K594R cells are hypersensitive to high osmolarity but not to caffeine. A. Growth rates of cells under stress conditions, OD 600 was measured before and after 9 hours incubation in YPD with 0.75 M NaCl, 7.5 mM caffeine or water. For osmolarity experiments, cells were synchronized by two doses of alpha factor (5 mg/ml) and released for 1 hour prior to addition of 0.5 M NaCl. Growth error bars are two standard deviations from three independent experiments. B and C. Swe1 K594R activity is elevated and prolonged upon exposure to 0.5 M NaCl. Cells were synchronized in late G1 with alpha-factor (5 mg/ml) and released for 1 hour prior to addition of 0.5 M NaCl. The difference between the blots is the different time points. doi:10.1371/journal.pone.0015089.g005

Conclusions
Although Cdk1-Clb activity is essential for Swe1 destruction, activation of Cdc5 [5], Cdk1-Clb2 activity or interaction with Swe1 are dispensable for Swe1 degradation [6,10]. We have found that Swe1 is SUMOylated on K594 and that this SUMOylation is prevented by inhibition of Cdk1-Clb activity or by knockout of SIZ1. Mutation of Swe1 to swe1 K594R or knockout of SIZ1 results in a greater abundance of Swe1 and consequently Swe1 destruction is delayed. Siz1 abundance is negatively regulated by hydroxyurea and inhibition of Cdk1 [24]. This provides a mechanism through which Cdk1-Clb activity indirectly decreases Swe1 abundance, and may serve to limit Swe1 levels during an unperturbed cellcycle or following recovery from stress.  Figure S3 Cst9, ris1, hex3 and slx8 do not seem to be involved in swe1 SOMOylation. A. Deletion of cst9 or ris1 does not increase Swe1 abundance. Cells were released from G1 arrest (saturation) for 90 minutes, or synchronized in S-phase with 0.2M hydroxyurea for 90 minutes following release from saturation and then released into fresh media containing 5mg/ml nocodazole. B. Deletion of hex3 (upper) or of slx8 (lower) does not increase Swe1 abundance. W303ua cells were arrested for 2 hours with 0.2M hydroxyurea and released into fresh media containing 5mg/ml nocodazole. (TIF) Table S1

Supporting Information
(DOC)