Budding Yeast Swe1 Is Involved in the Control of Mitotic Spindle Elongation and Is Regulated by Cdc14 Phosphatase during Mitosis*

Background: Timely mitotic spindle elongation ensures correct chromosome segregation, which is crucial for genetic stability. Results: The protein kinase Swe1 contributes to restraining anaphase progression and is regulated by the protein phosphatase Cdc14. Conclusion: Swe1 plays a role in the control of mitotic spindle elongation. Significance: Swe1 regulation by Cdc14 allows proper coordination of spindle elongation with mitotic progression. Cyclin-dependent kinase (Cdk1) activity is required for mitotic entry, and this event is restrained by an inhibitory phosphorylation of the catalytic subunit Cdc28 on a conserved tyrosine (Tyr19). This modification is brought about by the protein kinase Swe1 that inhibits Cdk1 activation thus blocking mitotic entry. Swe1 levels are regulated during the cell cycle, and they decrease during G2/M concomitantly to Cdk1 activation, which drives entry into mitosis. However, after mitotic entry, a pool of Swe1 persists, and we collected evidence that it is involved in controlling mitotic spindle elongation. We also describe that the protein phosphatase Cdc14 is implicated in Swe1 regulation; in fact, we observed that Swe1 dephosphorylation in vivo depends on Cdc14 that, in turn, is able to control its subcellular localization. In addition we show that the lack of Swe1 causes premature mitotic spindle elongation and that high levels of Swe1 block mitotic spindle elongation, indicating that Swe1 inhibits this process. Importantly, these effects are not dependent upon the role of in Cdk1 inhibition. These data fit into a model in which Cdc14 binds and inhibits Swe1 to allow timely mitotic spindle elongation.


Cyclin-dependent kinase (Cdk1) activity is required for mitotic entry, and this event is restrained by an inhibitory phosphorylation of the catalytic subunit Cdc28 on a conserved tyrosine (Tyr
). This modification is brought about by the protein kinase Swe1 that inhibits Cdk1 activation thus blocking mitotic entry. Swe1 levels are regulated during the cell cycle, and they decrease during G 2 /M concomitantly to Cdk1 activation, which drives entry into mitosis. However, after mitotic entry, a pool of Swe1 persists, and we collected evidence that it is involved in controlling mitotic spindle elongation. We also describe that the protein phosphatase Cdc14 is implicated in Swe1 regulation; in fact, we observed that Swe1 dephosphorylation in vivo depends on Cdc14 that, in turn, is able to control its subcellular localization. In addition we show that the lack of Swe1 causes premature mitotic spindle elongation and that high levels of Swe1 block mitotic spindle elongation, indicating that Swe1 inhibits this process. Importantly, these effects are not dependent upon the role of in Cdk1 inhibition. These data fit into a model in which Cdc14 binds and inhibits Swe1 to allow timely mitotic spindle elongation.
In all eukaryotic cells mitotic entry is allowed only in the presence of high Cdk1 5 activity. In budding yeast an important regulator of this process is the protein kinase Swe1 that inhibits entry into mitosis by phosphorylating the Tyr 19 residue of the catalytic subunit of Cdk1, Cdc28 (1). Cdk1 activation and entry into mitosis are allowed by the action of the protein phospha-tase Mih1 that removes this phosphorylation from Cdc28, thus promoting its activation (2). However, Tyr 19 phosphorylation and its subsequent dephosphorylation are not essential for M phase initiation in budding yeast (3); indeed the lack of Swe1 or Mih1 does not impair normal cell cycle progression in unperturbed condition. On the other hand, Cdc28 phosphorylation becomes essential for cell viability in response to perturbation in bud morphology or actin cytoskeleton (4). In these conditions, Cdc28 activation is prevented by the morphogenesis checkpoint, a pathway that inhibits Swe1 inactivation thus blocking entry into mitosis (5)(6)(7). Swe1 has therefore a critical role in coordinating cell morphogenesis with nuclear division, and it is subjected to multiple regulations that change its phosphorylation state (8), subcellular localization, and protein levels (9). During the S phase, Swe1 accumulates in the nucleus where it is phosphorylated by Clb-Cdc28 before being exported to the cytoplasm and then to the daughter side of the bud neck. Bud neck-localized Swe1 is phosphorylated by several protein kinases (10), and hyperphosphorylated Swe1 species are ubiquitylated (9,11). Subsequently, Swe1 is degraded via the proteasome, and this event allows mitotic entry (9). Little is known about the role of Swe1 after this point of the cell cycle, although it is reported that a pool of Swe1 persists in the cells after mitotic entry (11,12), suggesting that it might also play a role in mitotic progression.
The mitotic spindle is formed by sets of microtubules that are composed by heterodimers of ␣and ␤-tubulin assembled together in a head to tail fashion and that are nucleated by the spindle pole bodies. By the end of S phase, cells complete DNA synthesis and assemble a short bipolar spindle. During mitosis, chromosome segregation occurs in two steps: after cohesin removal, shortening of the kinetochore microtubules that link each sister chromatid to the correct spindle pole leads to chromosomes movement toward spindle poles (anaphase A). Afterward, pushing forces generated at the spindle midzone thanks to motor proteins cause the sliding of microtubules emanated from opposite poles, thus driving mitotic spindle elongation (anaphase B) (13). Spindle elongation is completed by the pull-ing forces generated by motor proteins associated with astral microtubules that link the spindle to the cell cortex (14).
Whereas metaphase is characterized by high Cdk1 activity, anaphase is marked by a reduction of Cdk1 activity also because of the activation of the phosphatase Cdc14. During most of the cell cycle, Cdc14 is kept inactive inside the nucleolus, and at anaphase onset, it is released and activated in two steps. First, the FEAR Cdc-Fourteen Early Anaphase Release pathway triggers a partial and transient release of Cdc14 from the nucleolus to the nucleus (15). However, the FEAR-controlled Cdc14 is not sufficient to promote mitotic exit; only the full release of Cdc14 allows it, by counteracting and inactivating Cdk1 activity. This full release is controlled by another regulatory pathway, the mitotic exit network that both promotes mitotic exit and induces cytokinesis (16).
Cdc14 release regulates several factors associated with the spindle, thus providing a link between anaphase onset and the changes in microtubule dynamics required for the completion of mitotic spindle elongation. FEAR-activated Cdc14 is involved in the control of microtubule dynamics by targeting several spindle-associated proteins that form the spindle midzone and/or that are required for the stabilization of microtubules in anaphase cells (17,18). In late anaphase, Cdc14 is able to dephosphorylate other proteins, thus helping mitotic spindle elongation (19).
In this study, we reveal that Swe1 plays a role during mitosis and is regulated by Cdc14; in particular, we demonstrate that Cdc14 is involved in Swe1 dephosphorylation during anaphase, which in turn controls its subcellular localization. This control might be important to ensure timely mitotic spindle elongation because we show an uncovered role for Swe1 in the inhibition of this process, which is correlated with its phosphorylation. Importantly, these effects are not dependent upon Swe1 role on Cdk1 inhibition.
Inhibition of Cdc15 kinase activity was performed as described (21). SWE1-PK3 cdc15-as1 and SWE1-PK3 CDC14-HA3 cdc15-as1 cells were grown at 25°C, prearrested with 2 g/ml ␣-factor, and released in the presence of 5 M 1NM-PP1 for 2 h to arrest them in late mitosis (88% budded cells with separated nuclei). The inhibition of Cdc28 kinase activity was performed as described (22). cdc14-3 cdc28-as1 cells were grown at 23°C and then shifted at 30°C for 3 h to arrest them in telophase (90% budded cells with divided nuclei); then the culture was split in two and treated with 1% DMSO or with 25 M 1NM-PP1 for 15 min.
Standard techniques were used for genetic manipulations (20,23). Gene deletions were generated by one-step gene replacement (24). One-step tagging techniques were used to create 3PK-tagged Swe1 variant and HA3-tagged Swe1 (25). Cdc14-HA3 expressing strain and cdc15-as1 mutant were a kind gift from Rosella Visintin. GAL1-CDC14 and GAL1-cdc14-C283A expressing strains were a kind gift from Elmar Schiebel. cdc28-E12K mutant was a kind gift from Daniel J. Lew. cdc28-as1 mutant was a kind gift from Gislene Pereira. All gene replacements and tagging were controlled by PCR based methods or Southern blot analysis. Fluorescence Microscopy and Live Imaging-In situ immunofluorescence was performed on formaldehyde-fixed cells and carried out as previously described (26). The nuclei were visualized by staining with 0.05 g/ml DAPI. To detect spindle formation and elongation, ␣-tubulin immunostaining was performed with the YOL34 monoclonal antibody (1:100, Serotec) followed by indirect immunofluorescence using rhodamineconjugated anti-rat antibody (1:500; Pierce). Swe1-3HA localization was observed in formaldehyde-fixed cells as described in Ref. 11. Digital images were taken with a Leica DC350F charge-coupled device camera mounted on a Nikon Eclipse 600 and controlled by the Leica FW4000 software or with the Meta-Morph imaging system software on a fluorescent microscope (Eclipse 90i; Nikon), equipped with a charge-coupled device camera (Coolsnap, Photometrics) with an oil 100ϫ 0.5-1.3 PlanFluor oil objective (Nikon).
For time lapse movies, cells were grown in synthetic complete medium ϩ 2% glucose then collected and imaged on agar in synthetic complete medium ϩ 2% glucose using a Delta Vision Elite imaging system (Applied Precision) based on an IX71 inverted microscope (Olympus) with a camera CoolSNAP HQ2 from Photometrics, and a UPlanApo 60ϫ (1.4 NA) oil immersion objective (Olympus). Every 3 min, four Z-stacks at 1.2-m intervals were taken for each fluorescent channel and projected onto a single image per channel using Fiji software (27). The timing of mitotic spindle elongation was scored in 62 wild type cells and 63 swe1⌬ cells. In Fig. 4B, time 0 is the time in which the spindle is correctly positioned and aligned at the bud neck.
Statistical Analyses and Other Techniques-The experiments shown in Figs. 4A, 5, and 6 (A and B) were repeated three times, and for the graphs, at least 200 cells were counted for each time point for each strain. Standard deviations were calculated using Microsoft Excel software. The significance of the differences in mitotic spindle elongation was statistically tested by the Wilcoxon signed rank sum test using MATLAB software; samples are significantly different if p Ͻ 0.05. Flow cytometric DNA quantification was performed according to Ref. 30 on a Becton-Dickinson FACScan. Visualization and quantification of sister chromatid separation with Tet operators using GFP were performed as described in Ref. 30. Quantification of Swe1 localization shown in Fig. 1C was performed with Fiji software.

Characterization of Swe1 Protein Levels and Modifications after G 2 /M Transition-
We previously characterized Swe1 levels and phosphorylation in all phases of an unperturbed cell cycle, and we observed that, after mitotic entry, Swe1 is not completely degraded, and it is present in its dephosphorylated form in the following G 1 phase (11). However, little is known about Swe1 levels, modifications, and functions after G 2 /M transition. We therefore decided to investigate these issues and started by checking Swe1 levels in metaphase-arrested cells. The metaphase to anaphase transition is marked by the proteolitic cleavage of the cohesin subunit Scc1 by the separase Esp1, which during the previous phases of the cell cycle is kept inactive by the securin Pds1 (31); at this cell cycle point, Pds1 is ubiquitylated by the anaphase-promoting complex bound to its regulatory subunit Cdc20, to be degraded via the proteasome. Therefore, yeast cells can be arrested in metaphase by the conditional depletion of Cdc20. cdc20⌬ MET3-CDC20-HA3 SWE1-HA3 cells were grown in synthetic medium lacking methionine and shifted in YEPD medium containing methionine, to turn off Cdc20 expression. After 2 h, all the cells were arrested with a large bud, undivided nuclei, and short bipolar spindle, and importantly, Western blot analysis indicates the absence of Cdc20 and the accumulation of the mitotic cyclin Clb2 (Fig. 1A) consistent with a metaphase arrest. As expected, in these conditions Swe1 is still present in the cells (Fig. 1A) and appears as multiple bands, indicative of its phosphorylated state (7).
To carefully analyze the kinetics of Swe1 levels and modifications from metaphase to the following G 1 phase, we arrested cdc20⌬ MET3-CDC20-HA3 SWE1-HA3 PDS1-18myc cells in metaphase as described above and released them in synthetic medium lacking methionine to allow cell cycle progression and containing ␣-factor to arrest cells in the next G 1 phase. We monitored the securin Pds1 and the mitotic cyclin Clb2 levels to follow metaphase to anaphase transition and mitotic exit, respectively. After release, cells progress correctly into mitosis and, as expected, as soon as Cdc20 is produced, Pds1 is degraded, and Clb2 disappears after nuclear division (Fig. 1B). We observed that Swe1 is phosphorylated both in metaphase and in anaphase; its dephosphorylation starts concomitantly with nuclear division and Clb2 degradation, and as expected, the protein remains dephosphorylated in the following G 1 phase (Fig. 1B, 30, 45, and 60 min). Because Swe1 localization correlates with its modifications (10), we performed a similar experiment with cdc20::MET3-CDC20 SWE1-HA3 cells to analyze Swe1 localization, and we observed that it is mostly present in the cytoplasm of the mother cell during metaphase, and it spreads into the bud concomitantly with nuclear division (Fig.  1C). Indirect immunofluorescence quantification shows that in metaphase-arrested cells, the Swe1 levels in the mother are double compared with those found in the bud, whereas they are almost equal in early and late anaphase (Fig. 1C, right panel). Together, these data confirm that Swe1 is still present in the cells after G 2 /M transition and show that it is phosphorylated in metaphase and dephosphorylated concomitantly with nuclear division and that these modifications correspond to a change in its localization.
Cdc14 Controls Swe1 Dephosphorylation and Localization in Late Mitosis-Because our data indicated that Swe1 is dephosphorylated concomitantly with nuclear division, we hypothesize that Swe1 could be regulated by a protein phosphatase that is active during mitosis. Interestingly, a Cdc14 allele defective in its nuclear export was reported to allow mitotic exit but to accumulate a fraction of cells with elongated buds (32). This phenotype could be due to high Swe1 levels or unscheduled Swe1 activity, conditions that interfere with normal isotropic bud growth and cause hyperpolarization of the cells (33). Moreover, Cdc14 and Swe1 were found to physically interact in an high throughput screening (34). Among all the other candidates, we therefore decided to gain insight into the role of this phosphatase on Swe1 regulation. At first, we asked whether other conditional cdc14 mutants showed the accumulation of elongated cells and whether it was dependent upon Swe1. To this end, cells carrying two different CDC14 thermosensitive alleles, cdc14-1 and cdc14-3, with or without the deletion of SWE1, were grown at 25°C to mid log phase and then shifted to the restrictive temperature to analyze the morphology of the cells. As shown in Fig. 2A, after shift at 37°C, a fraction of cdc14-3 and cdc14-1 cells showed elongated buds, and this phenotype was at least partially reversed by deleting SWE1 ( Table  2), indicating that it was due to high Swe1 levels or activity (1). These data suggest that Cdc14 might be directly or indirectly involved in Swe1 inactivation.
To better understand whether Cdc14 could be directly involved in Swe1 regulation, we performed an immunoprecipitation experiment preparing total protein extracts from cells expressing Swe1-PK3 and Cdc14-HA3 and arrested in late mitosis because of cdc15-as1 allele, an ATP analog-sensitive allele of CDC15 (21). Importantly, Swe1 could specifically immunoprecipitate Cdc14, suggesting that Cdc14 might be the protein phosphatase that dephosphorylates Swe1 in late mitosis (Fig. 2B).
Swe1 levels and activity are regulated by phosphorylation (10), and we asked whether Cdc14 inactivation could cause changes in Swe1 phosphorylation state. For this purpose, we analyzed Swe1-3HA phosphorylation in wild type and in cdc14-3 mutant cells after a shift at 34°C, a temperature that is sufficient to inactivate Cdc14 as indicated by the fact that, after 2 h of incubation, 88% of cdc14-3 cells were arrested in

Swe1 Role in Mitotic Spindle Elongation
telophase as large budded cells with divided nuclei. In wild type cells, Swe1 phosphorylation state did not change after the shift at the restrictive temperature (Fig. 2C, left panel), whereas cdc14-3 cells accumulated hyperphosphorylated Swe1 in the same experimental conditions (Fig. 2C, right panel). The same results were obtained by using cdc14-1 cells (our observation), suggesting that Cdc14 could be involved in Swe1 dephosphorylation. Swe1 is phosphorylated by different kinases, including Cdc28 (22). Because Cdc14 is involved in Cdc28 inactivation (35), the accumulation of Swe1 phosphorylated forms in the absence of Cdc14 activity could be due to hyperactivation of Cdc28 rather than to a direct role of Cdc14 in Swe1 dephosphorylation. To investigate this possibility, we took advantage of the Cdc28-as1 variant, whose kinase activity is blocked by the ATP analog 1NM-PP1 (36). SWE1-HA3 cdc14-3 cdc28-as1 cells were shifted at the restrictive temperature, and then we added 1NM-PP1 to inactivate Cdc28, or DMSO, as a control. We used 30°C as restrictive temperature because in these experimental conditions 1NM-PP1 is able to inhibit Cdc28 (22); this temperature is also sufficient to inactivate cdc14-3 because after 3 h 90% of SWE1-HA3 cdc14-3 cdc28-as1 cells were arrested in telophase. As indicated in Fig. 2D, Swe1 phosphorylated forms do not disappear after inactivation of Cdc28 kinase activity (lane 4), indicating that they are likely due to the lack of Cdc14 phosphatase activity rather than to the lack of Cdc28 inhibition. In this experiment, the electrophoretic mobility of Swe1 is slightly different from the one shown in Fig. 2C, because of the different experimental conditions.
To further analyze Cdc14 role on Swe1 phosphorylation, we analyzed the changes in Swe1 electrophoretic mobility after overproduction of wild type Cdc14 or of a phosphatase-dead Cdc14 variant (Cdc14-C283A). We performed this experiment in metaphase-arrested cells because in this phase Swe1 is present in its phosphorylated form (Fig. 1A). We arrested SWE1-HA3 MET3-CDC20-HA3 GAL1-CDC14 and SWE1-HA3 MET3-CDC20-HA3 GAL1-cdc14-C283A cells in metaphase by the depletion of Cdc20, and then we added galactose to induce Cdc14 overproduction. Samples were taken to check cell morphology and to analyze Cdc20, Cdc14 and Swe1 levels and modifications by Western blot. As shown in Fig. 3A, cycling cells contain Cdc20 and Swe1 in its phosphorylated forms (lanes 1 and 3), whereas in metaphase-arrested cells Cdc20 is undetectable and Swe1 is still present and phosphorylated (lanes 2 and 4). Importantly, when wild type Cdc14 is overproduced, Swe1 phosphorylated forms disappear (lane 5); conversely, when the phosphatasedead version of Cdc14 is overproduced, Swe1 phosphorylated forms persist (lane 6). These data indicate that Cdc14 phosphatase activity is required for Swe1 dephosphorylation.
Because Swe1 localization is coupled with its regulation (37), in the same experiment described above we also analyzed Swe1 subcellular localization. In metaphase-arrested wild type cells, Swe1 was present and mostly confined in the mother, but upon wild type Cdc14 overproduction, Swe1 spread also in the daughter cytoplasm (Fig. 3B, left panel). It is noteworthy that this change in Swe1 subcellular localization did not occur if the Cdc14 phosphatase-dead version was overproduced, indicating that it requires Cdc14 activity (Fig. 3B, right panel). Collectively, our data indicate that Cdc14 interacts with Swe1 and is involved in its dephosphorylation and localization, suggesting that Swe1 could be a target of Cdc14.  The Lack of Swe1 Induces Precocious Mitotic Spindle Elongation-Our data led us to hypothesize a role for Swe1 after G 2 -M transition. To shed light on this issue, we decided to carefully analyze cell cycle progression of swe1⌬ cells released from a metaphase arrest. Samples were taken at different time points after release to monitor nuclear division and mitotic spindle dynamics by immunofluorescence. We observed that wild type and swe1⌬ cells assemble a bipolar spindle with similar kinetics (Fig. 4A, left panel), but then swe1⌬ cells elongate the mitotic spindle 10 min earlier than wild type cells, indicating that the lack of Swe1 caused precocious spindle elongation (Fig. 4A, right panel). These data suggest that Swe1 could act as an inhibitor of mitotic spindle elongation.
Because single cell analysis allows to point out even subtle differences, to confirm our population data, we performed time lapse analysis of mitotic spindle dynamics in wild type and swe1⌬ cells expressing Tub1-GFP variant. Consistent with previous findings, we observed that the rate of bipolar spindle assembly and the metaphase spindle length are almost identical in wild type and swe1⌬ cells (Refs. 38 and 39 and supplemental Movies S1 and S2). Interestingly, after the mitotic spindle was correctly positioned at the bud neck, in swe1⌬ cells it reached the anaphase morphology faster than in wild type cells (6.8 Ϯ 1.5 min versus 9.7 Ϯ 2.9 min) ( Fig. 4B and supplemental Movies S1 and S2). The difference between the two strains is significant (p value obtained by Wilcoxon rank sum test p ϭ 2.6 ϫ 10 Ϫ12 ) and indicates that Swe1 act as an inhibitor of mitotic spindle elongation, in line with the result of the experiment above. Swe1 inhibits Cdk1 activity by phosphorylating the Tyr 19 residue of its catalytic subunit, Cdc28 (1). To exclude that the effect that we observed in swe1⌬ cells was due to precocious Cdc28 activation, we performed the same experiment in cells carrying the nonphosphorylatable version of Cdc28, cdc28-Y19F (3). If the effect of SWE1 deletion was due to a precocious activation of Cdc28, we would expect to see a speed up of spindle elongation in a strain expressing a Cdc28 version lacking Tyr 19 phosphorylation site. Interestingly, we observed that not only the time of spindle elongation is not shorter (11.8 min Ϯ 3.2, 59 cells), but it is also significantly longer than in wild type cells (p value obtained by Wilcoxon rank sum test p ϭ 1.3 ϫ 10 Ϫ5 ), according with the notion that Cdk1 controls the phosphorylation of several microtubule binding proteins and therefore the lack of Cdc28 inhibition might interfere with correct spindle dynamics. These data indicate that the precocious mitotic spindle elongation that we observed in swe1⌬ cells is not due to precocious Cdc28 activation but likely other regulator(s) of

Swe1 Role in Mitotic Spindle Elongation
spindle elongation acting in a parallel pathway with respect to Cdc28 are misregulated in the absence of Swe1. The Nonphosphorylatable cdc28-Y19F Allele Does Not Restore Proper Spindle Elongation in SWE1-overexpressing Cells-To confirm Swe1 role in mitotic spindle elongation, we decided to carefully analyze the phenotype of cells overexpressing SWE1. Swe1 overexpression from the GAL1 promoter (GAL1-SWE1) is lethal for cells because it totally inhibits Cdc28, thus blocking both switch from polar to isotropic bud growth and spindle pole body separation; as a result, cells arrest with elongated buds, undivided nuclei, and without a mitotic spindle (39). As already reported, we observed that cell lethality induced by GAL1-SWE1 is suppressed by the presence of a nonphosphorylatable version of Cdc28 (cdc28-Y19F) (3) or by the presence of cdc28-E12K, a variant that could not physically interact with Swe1 and therefore could not be phosphorylated and inhibited by Swe1 (40) (our observation). However, cdc28-Y19F and cdc28-E12K alleles are not able to restore normal cell cycle progression in Swe1-overexpressing cells (see below and our observation), further indicating that Swe1 could have other targets besides Cdc28 in controlling mitotic progression. Therefore we carefully characterized the phenotype of GAL1-SWE1 cdc28-Y19F cells in galactose containing medium. For this purpose, wild type, GAL1-SWE1, cdc28-Y19F, and GAL1-SWE1 cdc28-Y19F cells were inoculated in YEPR, arrested in G 1 with ␣-factor and released in galactose-containing medium (YEPRG) to induce Swe1 overproduction. Samples were taken at different time points after release to monitor the kinetics of DNA replication by FACS analysis, budding, nuclear division, and spindle assembly and elongation. Wild type and cdc28-Y19F cells showed very similar profiles of DNA replication, budding, and nuclear division (Fig. 5, A-C); GAL1-SWE1 cells, as expected, replicated their DNA but failed to divide it, and, after 240 min, accumulated with 2C DNA content, undivided nuclei, and elongated buds and were unable to assemble a mitotic spindle (Fig. 5). Interestingly, GAL1-SWE1 cdc28-Y19F cells assembled a bipolar spindle with kinetics similar to cdc28-Y19F cells, but then spindle elongation was defective, because cells accumulated with the spindle stretched at the bud neck (Fig. 5, C and D). Accordingly, GAL1-SWE1 cdc28-Y19F cells were about 30 -45 min delayed with respect to wild type cells in reaccumulating the 1C DNA peak (Fig. 5A); this difference was always observed in different GAL1-SWE1 cdc28-Y19F segregants during independent repeats of the same experiment (our observations). Interestingly, we also obtained the same results using the cdc28-E12K allele (data not shown). Therefore, the presence of Cdc28 variants that cannot be inhibited by Swe1 does not restore proper mitotic spindle elongation in Swe1 overproducing cells, further suggesting that the protein kinase Swe1 likely affects this process by acting on other target(s) besides Cdc28.

The Spindle Elongation Defect of GAL1-SWE1 cdc28-Y19F Cells Is Not Due to a Defect in Sister Chromatids Cohesion
Removal-During anaphase, to allow sister chromatid separation, cohesin has to be removed because it is a structural barrier that opposes the pulling forces of the microtubules. After cohesin removal, shortening of the kinethocore microtubules that link each sister chromatid to the correct spindle pole leads to chromosomes movement to spindle poles and afterward pushing forces drive mitotic spindle elongation (13). Therefore, for proper spindle elongation, a timely removal of the cohesion between sister chromatids is required. We therefore wondered whether the spindle elongation defect of GAL1-SWE1 cdc28-Y19F cells could be due to a delay in cohesin removal, rather than to a delay in spindle elongation. To address this point, wild type, GAL1-SWE1, cdc28-Y19F, and GAL1-SWE1 cdc28-Y19F cells, all carrying the TetO/TetR system to visualize sister chromatid separation (31), were inoculated in YEPR, arrested in G 1 with ␣-factor, and released in YEPRG to induce Swe1 overpro-duction. Samples were taken at different time points after release to monitor the kinetics of budding, nuclear division, and sister chromatid separation (Fig. 6A), visualized as fluorescent GFP dots. All the strains progressed through the cell cycle as described in Fig. 5. Interestingly, GAL1-SWE1 cdc28-Y19F cells started to separate sister chromatids with a kinetic similar to wild type and cdc28-Y19F cells (Fig. 6A), indicating that the spindle elongation defects of these cells are not due to a persistence in sister chromatids cohesion but to a defect in spindle dynamics, further supporting the idea that Swe1 plays a role in anaphase progression. In agreement with our previous experi-

Swe1 Role in Mitotic Spindle Elongation
ments, we also observed that SWE1 overexpression from the GAL1 promoter completely impairs spindle elongation in cells released both from a metaphase arrest and from S phase arrest (our observation and see below). Cdc14-dependent Swe1 Dephosphorylation Induces Spindle Elongation in SWE1-overexpressing Cells-We showed that the protein phosphatase Cdc14, which is involved in anaphase progression, is likely involved in Swe1 inactivation (Figs. 1-3). Because Swe1 seems to act as an inhibitor of spindle elongation, a hyperactive Cdc14 variant might allow spindle elongation in Swe1-overexpressing cells. To test this hypothesis, we took advantage of CDC14 TAB6Ϫ1 dominant allele (41), and we analyzed mitotic spindle dynamics in wild type, GAL1-SWE1, and GAL1-SWE1 CDC14 TAB6Ϫ1 cells released from an hydroxyurea block, a condition in which cells arrest with separated spindle pole bodies and short bipolar spindle. Wild type cells elongate the spindle 90 min after the release, whereas Swe1 overproduction completely blocks this process (Fig. 6B), consistent with our previous data (Fig. 5). Interestingly, the introduction of CDC14 TAB6Ϫ1 allele allows spindle elongation in a significant fraction of SWE1-overexpressing cells (Fig. 6B), confirming an important role of Cdc14 in Swe1 regulation. However, the presence of the CDC14 TAB6Ϫ1 allele cannot restore wild type kinetics of spindle elongation, but this is not surprising because the phosphatase activity of the mutant enzyme is reduced to 75% of wild type levels (41) and because Swe1 is overexpressed in these cells. To test whether spindle elongation in the presence of CDC14 TAB6Ϫ1 cells is related to Swe1 dephosphorylation, we analyzed mitotic spindle dynamics and Swe1 modifications in a similar experiment. GAL1-SWE1-PK3 and GAL1-SWE1-PK3 CDC14 TAB6Ϫ1 cells were released from an hydroxyurea block and after 180-min samples were taken to prepare protein extracts. As shown in Fig. 6C, Swe1 phosphorylated forms are reduced in GAL1-SWE1-PK3 CDC14 TAB6Ϫ1 mutant with respect to GAL1-SWE1-PK3, supporting the hypothesis that Swe1 dephosphorylation correlates with mitotic spindle elongation.

DISCUSSION
Even if the regulation of Saccharomyces cerevisiae Swe1 and its role in controlling mitotic entry have been very well described by previous studies, much of the evidence collected in this research indicates that other undiscovered Swe1 targets might exist and that it also might play a role during mitosis. Swe1 protein levels and activity are finely controlled during the cell cycle, because several protein kinases phosphorylate it, thus regulating its activity and the association with its substrates (10,22). To allow mitotic entry, Swe1 is partially degraded during G 2 phase in unstressed cells (9); however, the molecular mechanisms that inactivate Swe1 are not fully uncovered. Accordingly, the isolation of Swe1 mutant variants that are more stable but do not show a defect in their interaction with any of the known Swe1 regulators (9) indicates the existence of other proteins that control Swe1 activity and abundance. In accordance with other studies (7,12), we previously reported that Swe1 is not completely degraded after mitotic entry during an unperturbed cell cycle (11), and in this paper we confirm that Swe1 is present in metaphase-arrested cells, consistent with a physiological role for Swe1 after mitotic entry.
. Swe1-overexpressing cells are able to separate sister chromatids but cannot complete anaphase, and this defect is partially suppressed by a hyperactive Cdc14 variant. A, exponentially growing cultures of wild type, GAL1-SWE1, cdc28-Y19F, and GAL1-SWE1 cdc28-Y19F cells were arrested in G 1 by ␣-factor in YEPR and released in YEPRG at 25°C (time 0) to induce SWE1 overexpression. At the indicated times after release, cell samples were taken for scoring budding, nuclear division, and sister chromatid separation. B, exponentially growing cultures of wild type, GAL1-SWE1 and GAL1-SWE1 CDC14 TAB6Ϫ1 cells were arrested in S phase in YEPR by hydroxyurea (HU) and released in the cell cycle in YEPRG medium at 25°C to induce SWE1 overexpression. Cell samples were taken at the indicated time points for scoring mitotic spindle elongation by immunofluorescence using anti-tubulin antibodies. C, GAL1-SWE1-PK3 and GAL1-SWE1-PK3 CDC14 TAB6Ϫ1 cells were treated as in B. 180 min after release, samples were taken to analyze Swe1 phosphorylation using anti-PK antibodies. Pgk1 was used as a loading control. Size markers on the side of the gel blots indicate kilodaltons.
In addition, here we show that Swe1 is phosphorylated and mostly localized in the nucleus and in the cytoplasm of the mother cell during metaphase, and then, during anaphase and concomitantly with nuclear division, it is dephosphorylated and spread also into the bud. Interestingly, we observed that inactivation of the protein phosphatase Cdc14 leads to the formation of cells with elongated buds and that this phenotype is reversed by SWE1 deletion, indicating that it is likely due to altered Swe1 regulation, which is known to cause hyperpolarized bud growth (9). In agreement with published data (34), our further analysis showed that Cdc14 can bind Swe1 in vivo, and we collected evidence that it is involved in its dephosphorylation. Indeed, Swe1 phosphorylated forms accumulate in the absence of Cdc14 function and disappear upon CDC14 overexpression. The accumulation and disappearance of phosphorylated proteins is driven by a balance of kinase and phosphatase activities. It is known that Cdc14 is involved in Cdk1 inactivation (35), and Cdk1 is implicated in Swe1 phosphorylation (22); however, we demonstrated that the accumulation of Swe1 phosphorylated bands upon Cdc14 inactivation is not caused by high Cdk1 activity. In addition, we showed that Swe1 phosphorylated forms completely disappear upon overproduction of Cdc14 but persist in the presence of high levels of a phosphatase-dead Cdc14 variant. Our data suggest that the complete disappearance of Swe1 phosphorylated forms upon Cdc14 overexpression is more likely due to a direct Cdc14-dependent dephosphorylation rather than to Cdk1 inactivation that also occurs in these experimental conditions (35) and that has been reported to only partially decrease Swe1 hyperphosphorylated forms in mitosis (22). Consonant with the correlation between Swe1 phosphorylation and localization, we also observed that Cdc14-dependent dephosphorylation induces changes in Swe1 subcellular localization. Altogether, our data led us to envisage a role for Cdc14 on Swe1 inhibitory regulation during mitosis.
What could be the physiological role for Swe1 after mitotic entry? Meticulous analyses of swe1⌬ phenotype revealed that the lack of Swe1 induces premature mitotic spindle elongation with respect to wild type cells, indicating that Swe1 might play a physiological role in this process. Interestingly, this phenotype is not due to unscheduled Cdc28 activation but to misregulation of regulator(s) of spindle elongation that act in parallel with respect to Cdc28. Premature mitotic spindle elongation is tolerated by otherwise wild type cells likely because, in normal conditions, a slight advance in anaphase progression does not affect cell viability. Instead, we observed that the lack of Swe1 increases the amount of binucleate cell bodies, arising from anaphase occurring in the mother cell, in spindle positioning mutants such as kar9⌬ and dyn1⌬ (our observation and Ref. 42). We also observed that the lack of Swe1 causes a slight sensitivity to benomyl (our observation), a compound that interferes with microtubules polymerization, further supporting the notion that Swe1 is involved in spindle dynamics.
In addition and in agreement with previous studies, we show that Cdc28 variants that cannot be inhibited by Swe1 are not able to completely revert the phenotype caused by a large amount of Swe1 (39). Indeed, SWE1-overexpressing cells accumulate with elongated buds and undivided nuclei and are unable to assemble a mitotic spindle; the presence of cdc28-Y19F and cdc28-E12K mutant proteins allows mitotic spindle formation but does not restore proper kinetics of spindle elongation and nuclear division in these cells. These results indicate that some players of anaphase progression are not active in these conditions and that Swe1 could inhibit this process by acting on other protein(s) besides Cdc28. We also show that the spindle elongation defect of GAL1-SWE1 cdc28-Y19F cells is not due to faulty sister chromatid separation, thus supporting the notion that Swe1 plays a role in anaphase progression after cohesin removal likely by playing a role in mitotic spindle dynamics. The idea that Our data support a model in which Cdc14 regulates Swe1 dephosphorylation and localization, thus controlling mitotic spindle elongation and Swe1 distribution between mother and daughter cell (see text for details).
Swe1 might phosphorylate protein(s) important for this process is also supported by the finding that its homolog in Drosophila melanogaster (dwee1) is involved in preserving mitotic spindle integrity (43). Importantly, we showed that an hyperactive Cdc14 variant (Cdc14 TAB6Ϫ1 ) could allow Swe1 dephosphorylation and spindle elongation in a considerable fraction of Swe1-overexpressing cells, thus confirming that Cdc14 controls Swe1 to allow timely anaphase progression.
Altogether, our results fit into the model depicted in Fig. 7. In metaphase, Swe1 is fully phosphorylated, and Cdc14 is sequestered in the nucleolus; in early anaphase, the FEAR pathway allows Cdc14 release from the nucleolus into the nucleus and to the spindle midzone to dephosphorylate specific targets to drive mitotic spindle elongation. After mitotic exit, the mitotic exit network pathway allows complete activation of Cdc14 that in turn localizes in the cytoplasm and at the bud neck to induce cytokinesis completion (16). In this scenario, we propose that Cdc14 binds and dephosphorylates Swe1, thus causing its inactivation and its distribution in the cytoplasm of the daughter cell in anaphase (Fig. 7). Because we could never visualize Swe1 at the spindle, we hypothesize that its dephosphorylation likely occurs in the nucleoplasm after Cdc14 has been released from the nucleolus. This regulation contributes to release the block of mitotic spindle elongation so that this process can take place. During early anaphase, Swe1 dephosphorylation could either inhibit Swe1 kinase activity or destroy the complex of Swe1 with its target(s). Both mechanisms are plausible because a nonphosphorylatable Swe1 variant was described to display reduced kinase activity in vitro (22), indicating that Swe1 dephosphorylation may lead to its inactivation. However, overproduction of a kinase-dead Swe1 version causes the same effects on spindle dynamics as the overproduction of the wild type protein (our observation), indicating that Swe1 might play its inhibitory role independently of its kinase activity. In anaphase, released Cdc14 could dephosphorylate the remaining Swe1, thus ensuring that a pool of Swe1 is translocated to the daughter cell and is present during the following G 1 phase. Our study discloses new aspects of the control of anaphase progression that reveal Swe1 as a new player in this process that is regulated in a Cdc14dependent manner and that allows proper coordination of important cellular processes such as nuclear division and mitotic spindle elongation. We hypothesize that Swe1 contributes to restraining mitotic spindle elongation until anaphase and that its inactivation by dephosphorylation is required to allow timely mitotic spindle elongation, thus ensuring the maintenance of genomic stability.
An interesting question that remains unanswered is which protein is kept inactive or activated by Swe1 to block spindle elongation, and further studies will be required to shed light on this issue. We hypothesize that Swe1 could bind a factor that is localized at the mitotic spindle; however, we could never observe Swe1 localization at the spindle, suggesting that it might form a complex with a diffusible protein that binds microtubule-associated proteins or the spindle only after Swe1 inactivation.