Phospho-regulation of kinesin-5 during anaphase spindle elongation

Evolutionary conserved homotetrameric kinesin-5 motor proteins have major roles in mitotic spindle morphogenesis (Blangy et al., 1995; Goshima and Vale, 2005; Hagan and Yanagida, 1992; Heck et al., 1993; Hoyt et al., 1992). These motors crosslink and slide antiparallel microtubules (Gheber et al., 1999; Kapitein et al., 2005) originating from opposite spindle poles, thereby performing their mitotic functions (Kashina et al., 1997). In addition to their well-established roles in spindle assembly (Hoyt et al., 1992), kinesin-5 motor proteins also have a role in anaphase B spindle elongation in the budding yeast Saccharomyces cerevisiae (Gerson-Gurwitz et al., 2009; Movshovich et al., 2008; Saunders et al., 1995; Straight et al., 1998) and other organisms (Sharp et al., 1999; Touitou et al., 2001). Although kinesin-5 motor proteins perform essential functions at several stages of spindle dynamics, and their levels are regulated in a cell-cycle-dependent manner (Gordon and Roof, 2001; Hildebrandt and Hoyt, 2001; Spellman et al., 1998), regulation of their function is not well understood. Previous reports have indicated that kinesin-5 motors are phosphorylated by mitotic kinases (Blangy et al., 1995; Chee and Haase, 2010; Garcia et al., 2009; Giet et al., 1999; Sawin and Mitchison, 1995; Sharp et al., 1999), but the mechanism by which phosphorylation regulates kinesin-5 functions is not fully understood. Here we examine the role of phosphorylation in controlling the intracellular function of the S. cerevisiae kinesin-5 homologue Cin8.

To examine the phosphorylation of Cin8 as a function of cell cycle progression, we used cells expressing Myc-tagged Cin8 that were arrested at different stages of the cell cycle. Fractionation of the cell extracts by SDS-PAGE followed by western blot analysis demonstrated that in cells arrested at late anaphase by any one of several temperature-sensitive mutations [cdc15-2, cdc14-1 (Jaspersen et al., 1998) and cdc5-7 (Park et al., 2003)], Cin8 exhibits a slow-migrating form (Fig. 1A), indicative of protein phosphorylation. This band-shift was observed for Cin8 produced either from a CEN (centromere) or 2 μm plasmid tagged with the Myc epitope either at its N- or C-terminus (Fig. 1) and for Cin8–BCP (Gheber et al., 1999) (data not shown). The slow-migrating band of Cin8 was not present in cells arrested in S-phase or in metaphase (Fig. 1B), and was abolished by treatment of the cell extracts with phosphatase (Fig. 1C). These results indicate that Cin8 is differentially phosphorylated during anaphase.

The amino acid sequence of Cin8 contains five putative phosphorylation sites for the conserved cyclin-dependent kinase 1 Cdk1 (Cdc28 in S. cerevisiae) ([S/T]-Px-[R/K] or [S/T]-P-[R/K]) (Langan et al., 1989; Shenoy et al., 1989). Three sites are located in the motor domain and two are in the stalk and tail. Mutation of all five amino acids to alanine (Cin8-5A) or of the three located in the motor domain (Cin8-3A) but not of the two in the stalk and tail (Cin8-2A), almost completely abolished the slow-migrating band of Cin8 at the end of anaphase (Fig. 1D). To examine whether Cin8 is phosphorylated by Cdk1, we performed an in vitro phosphorylation assay using a bacterially expressed truncated form of Cin8, which contains its catalytic motor domain (Cin8-590) (Hildebrandt et al., 2006). Cin8-590 was indeed phosphorylated by Cdk1, and the mitotic complex Clb2–Cdk1 showed higher specificity relative to the earlier complexes Clb3–Cdk1 and Clb5–Cdk1 (Fig. 1E). Moreover, alanine replacement of the three Cdk1-specific sites that abolish the slow-migrating band of Cin8 during anaphase (Fig. 1D) completely eliminated the phosphorylation of Cin8-3A-590 by the Clb2–Cdk1 in vitro (Fig. 1E). Consistent with these findings, it has been recently reported that Cin8 is a substrate of Clb2–Cdk1 both in vivo and in vitro (Chee and Haase, 2010). However, this recent report (Chee and Haase, 2010) does not demonstrate specifically that sites in the Cin8 motor domain are targets for Cdk1 phosphorylation. Our results indicate that during anaphase, Cin8 is phosphorylated by Clb2–Cdk1 at site(s) located within its motor domain. Moreover, in agreement with previous reports that a low Clb2–Cdk1 activity is maintained throughout anaphase (Queralt et al., 2006), our data suggest that this activity is sufficient for accumulation of the phosphorylated form of Cin8 at the end of anaphase. Alternatively, during anaphase, the same sites might be phosphorylated by an additional kinase.

Two of the three sites that abolish the Cin8 phosphorylation, S277 and T285, are located within the uniquely large loop 8 (aa 255–353) of Cin8 motor domain (Hoyt et al., 1992). The third, S493 is only 13 residues away from loop 12 and α-helix 5 in the kinesin motor domain, which are important for the interaction with microtubules (Kull et al., 1996; Sablin et al., 1996; Woehlke et al., 1997). Interestingly, S493 itself is conserved among the kinesin family members and, as in Cin8, is part of a consensus sequence for phosphorylation by Cdk1 also in the second S. cerevisiae kinesin-5 homologue Kip1.

Because Cdk1 is activated before anaphase (Grandin and Reed, 1993), the finding that phosphorylated Cin8 is accumulated only during anaphase (Fig. 1A,B) suggests that prior to anaphase, it is dephosphorylated by one or more phosphatases. Therefore, we examined whether the separase-dependent Ser/Thr-specific PP2A^Cdc55 phosphatase, which dephosphorylates Cdk1 sites before anaphase onset (Queralt et al., 2006), affects the phosphorylation of Cin8. Indeed, a reproducible slow-migrating band of Cin8 was observed for cdc55Δ but not for WT cells arrested in S-phase with short spindles (Fig. 1F). This band was abolished when the phosphorylation-deficient mutant Cin8-3A was expressed in cdc55Δ cells (Fig. 1F). These results indicate that the PP2A^Cdc55 phosphatase is involved, at least in part, in Cin8 dephosphorylation before anaphase onset. A similar regulation mechanism was previously demonstrated for inhibition of Cdk1-dependent Net1 phosphorylation before anaphase by the PP2A^Cdc55 phosphatase (Queralt et al., 2006). Our data therefore suggest that inactivation of separase-dependent PP2A^Cdc55 promotes phosphorylation of a number of Cdk1 targets, including Cin8, whose phosphorylation is required for anaphase progression.

To examine the possibility that the conserved Cdc14 phosphatase (Stegmeier et al., 2002) dephosphorylates Cin8 during anaphase, the band-retardation assay was performed in cells synchronized in late anaphase by cdc15-2 and then Cdc14 was overexpressed from a galactose-inducible promoter, while holding the late-anaphase arrest. Overexpression of Cdc14 reduced the appearance of the slow-migrating band of Cin8 in a time-dependent manner (Fig. 1G). Quantitative analysis indicates that without Cdc14 overexpression, the ratio of intensities between the slow-migrating (phosphorylation) and the low bands of Cin8 is ~1, whereas after overexpression of Cdc14 for 2 hours, this ratio drops to ~0.5.

To further explore the involvement of Cdc14 in Cin8 dephosphorylation, we examined Cin8 phosphorylation during anaphase progression in WT cells versus cells deleted for SPO12, a member of the early-anaphase Cdc14 activation pathway (Stegmeier et al., 2002). Seventy minutes after release from S-phase arrest, when ~60% of cells are in anaphase (not shown), a reproducible slow-migrating band appeared in both WT and spo12Δ cells (Fig. 1H). Quantitative analysis showed that in the absence of Spo12, the ratio between the slow-migrating and the low bands of Cin8 increased by 40% (Fig. 1H), indicating the deletion of SPO12 enhances Cin8 phosphorylation. Taken together, these results show that Cdc14 is involved in Cin8 dephosphorylation during anaphase.

Monitoring of the localization of Cin8–3GFP to the anaphase spindles in unsynchronized WT cells revealed that in early-mid anaphase, when spindles are shorter than 4.5–5 μm, WT Cin8 binds to a large portion of the spindle, as well as near the spindle poles (Fig. 2Aa,Ba). As spindles elongate, Cin8 was concentrated at a decreasing portion of the middle region. At late anaphase, when spindles are longer than ~8 μm, Cin8 was barely detectable at the middle spindle and was diffusively localized near the spindle poles (Fig. 2Aa). Consistent with the localization pattern in cells arrested in late anaphase by the cdc15-2, cdc14-1 and cdc5-7 mutations, when significant phosphorylation of Cin8 is detectable (Fig. 1A), Cin8 localized diffusively near the spindle poles and was not focused at the midzone (supplementary material Fig. S1).

The Cin8-2A mutant, which does not affect Cin8 phosphorylation during anaphase (Fig. 1D), exhibited similar spindle localization to that of WT Cin8 (Fig. 2Ab). Compared with WT Cin8 and Cin8-2A, the phosphorylation-deficient Cin8-3A mutant (Fig. 1D) remained localized near the spindle poles, occupied a larger portion of the middle spindle during mid-late anaphase and remained localized to this region for a longer period of time (Fig. 2Ac,Bb). Similarly to the Cin8-3A mutant, in cells that lack the Cdk1 cyclin Clb2 (clb2Δ), which is involved in Cin8 phosphorylation (Fig. 1E), WT Cin8 occupied a large portion of the anaphase spindles (supplementary material Fig. S2B). By contrast, the phosphomimic Cin8-3D, where Cdk1 sites in the motor domain are replaced by aspartic acid, failed to bind to the spindle and exhibited diffusive localization at early anaphase (Fig. 2Ad). Also, in cdc55Δ cells, where Cin8 phosphorylation is increased (Fig. 1F), partial diffusive localization of Cin8 in early anaphase was observed (supplementary material Fig. S2C). Thus, taken together, our results indicate that Cin8 dephosphorylation at early anaphase is required for its attachment to the spindle, whereas phosphorylation of Cin8 leads to its detachment from the spindle at late anaphase.

Next, viability of cin8Δ kip1Δ cells was examined in which CIN8 shuffle plasmid (Gheber et al., 1999) was replaced by a centromeric plasmid expressing CIN8, cin8-3A or cin8-3D. Cin8 and Cin8-3A cells remained viable at all examined temperatures, but cells producing Cin8-3D were not viable at 37°C (supplementary material Fig. S3A). At room temperature, ~65% of the cin8Δ kip1Δ cells expressing cin8-3D were large-budded with monopolar spindles, compared with only ~10% of the cells expressing CIN8. At 37°C, the cells producing Cin8-3D arrested as large-budded cells with collapsed spindles (supplementary material Fig. S3B). These findings indicate that in the absence of Kip1, the reduced binding of Cin8-3D to the spindle interferes with spindle assembly and suggest that dephosphorylation of Cin8 is required to assemble mitotic spindles and maintain their bipolar structures before anaphase.

The phosphorylation-dependent affinity of Cin8 to the spindles might be explained by two mechanisms. The additional negative charge caused by phosphorylation (or mutation to aspartic acid) of residues in the motor domain might directly reduce the affinity of Cin8 for MTs, causing its detachment from the spindle. This is opposed to what has been shown thus far for several kinesin-5 family members, namely that a C-terminal tail domain phosphorylation promotes localization to the mitotic spindle (Blangy et al., 1995; Cahu et al., 2008; Sawin and Mitchison, 1995; Sharp et al., 1999). Here, we show for the first time that phosphorylation of the kinesin-5 Cin8 negatively regulates its localization to the mitotic spindle: its prevention increases Cin8 binding to the spindle. A similar effect of phosphorylation has been shown for the Caenorhabditis elegans kinesin-6 homologue ZEN-4: the motor domain phosphorylation by Cdk1 reduced the affinity of ZEN-4 for MTs (Mishima et al., 2004). In addition, phosphorylation of the conserved serine residue in the motor domain of the Drosophila melanogaster kinesin-13, KLP10A, alters the interaction of KLP10A with the MTs and diminishes its MT-depolymerizing activity (Mennella et al., 2009).

An alternative role for Cin8 phosphorylation might be to reduce interactions between Cin8 and accessory spindle binding protein(s). It has recently been shown that Cin8 interacts directly with the conserved MT bundling and midzone-organizing protein Ase1, an interaction that is dependent on Ase1 dephosphorylation (Khmelinskii et al., 2009): delivery of Cin8 to the spindle seems to be controlled by dephosphorylation of Ase1 (Khmelinskii et al., 2009). Our data strongly indicate that spindle detachment of Cin8 is regulated by its own phosphorylation. It is possible that the interaction between Cin8 and Ase1 depends on the phosphorylation state of both proteins, as is the case for the fission yeast kinesin-6 and Ase1 (Fu et al., 2009), and that Cin8 phosphorylation reduces its interaction with Ase1.

In S. cerevisiae cells, anaphase B consists of two phases, a fast phase followed by a slow one, with Cin8 contributing mainly to the elongation during the fast phase (Straight et al., 1998). To examine the role of phosphorylation in Cin8 function on the spindle, we monitored the anaphase spindle elongation kinetics and the spindle morphology. In Cin8-3D cells (Fig. 2Ad), as well as in cdc55Δ cells (supplementary material Fig. S2C), where Cin8 exhibits diminished binding to the spindle compared with WT cells, the spindle elongation rate was considerably slower (Fig. 3A, Table 1 and supplementary material Fig. S2C). However, in Cin8-3A and clb2Δ cells, where Cin8 occupies a larger portion of the middle spindle (Fig. 2Ac,Bb and supplementary material Fig. S2B), the spindle elongation rate was considerably faster (Fig. 3A, Table 1 and supplementary material Fig. S2B). Also, in Cin8-3A cells, the overlapping region of antiparallel interpolar MTs (iMTs) is significantly decreased (Fig. 3B, arrowheads and Table 1) and the spindle disassembly is delayed by ~3 minutes (Table 1). Finally, Cin8-3A cells exhibited deformed, bent and asymmetrical spindles (Fig. 3B). Thus our data indicate that phosphorylation of Cin8 controls anaphase spindle elongation.

The higher spindle elongation rate in cells expressing the phosphorylation-deficient Cin8-3A is likely to cause the shorter iMT overlapping region in these cells (Table 1). Because this region serves as a binding site for proteins that control spindle stabilization and iMT plus-end dynamics during anaphase (Buvelot et al., 2003; Fridman et al., 2009; Higuchi and Uhlmann, 2005; Khmelinskii et al., 2007; Schuyler et al., 2003; Thomas and Kaplan, 2007), the decreased size of this region is likely to reduce the stabilization of the iMT plus-ends and thus cause the deformed and asymmetrical spindle morphologies observed in these cells. Hence, our results indicate that the decreased amount of Cin8 at the midzone region at the end of anaphase is required for slowing down the rate of spindle elongation and maintaining the correct structure of the midzone and anaphase spindle.

To summarize, we propose that the phosphorylation–dephosphorylation of Cdk1 sites in Cin8 motor domain controls the binding of a correct amount of Cin8 to the spindle during mitosis. Before the onset of anaphase, Cin8 is dephosphorylated to promote its attachment to the spindle and facilitate its function in spindle assembly. At early–mid anaphase, the activation of Cdc14 phosphatase by the FEAR (Cdc14 early anaphase release) pathway (Chiroli et al., 2009; Ross and Cohen-Fix, 2004; Stegmeier et al., 2002) dephosphorylates Cin8, thus inducing binding of Cin8 to the short anaphase spindle and promoting fast spindle elongation. At late anaphase, when Cdc14 is no longer present at the spindle midzone (Stegmeier et al., 2002) (supplementary material Fig. S4), Cin8 phosphorylation becomes dominant and it detaches from the spindle. This detachment is required for slowing down the rate of spindle elongation to maintain the correct structure of the spindle.

S. cerevisiae strains used are listed in supplementary material Table S1.

Cell cycle was arrested after overnight growth to mid-log phase and transfer into the arrest conditions for 3-4 h. The following reagents were applied: for G1 arrest, 5 μg/ml α-factor (Sigma); for S phase, 0.1 M hydroxyurea (Sigma); for metaphase, 10 μg/ml nocodazole (Sigma). To release cells from the S-phase arrest, cells were washed and resuspended in fresh medium lacking hydroxyurea. For late anaphase arrest, cells expressing the ts mutations cdc15-2, cdc14-1 or cdc5-7 were shifted to 37°C. The ability of Cin8 phosphorylation mutants to complement a CIN8 deletion allele was tested in a strain with deletions in both CIN8 and KIP1. Since cin8Δ kip1Δ double mutants are not viable, the viability of the tester strain was maintained by a CIN8 plasmid carrying CYH2 as well. Despite the presence of a recessive cycloheximide resistance allele cyh2 in the genome, this strain is sensitive to cycloheximide because of the dominant CYH2 allele carried on the plasmid (Gheber et al., 1999). At 26°C, transformation with a second plasmid carrying CIN8, cin8-3A or cin8-3D, however, allowed cells to live without the CIN8-CYH2 plasmid and therefore to grow on cycloheximide-containing medium (supplementary material Fig. S1). Treatment of extracts with CIP phosphatase was previously described (Cohen-Fix and Koshland, 1997). Whole-cell protein extracts were obtained as desribed (Hildebrandt et al., 2006). For 3HA-Cdc14 overexpression during anaphase arrest, cdc15-2 cells expressing Cin8-13Myc and P_GAL1-HA-Cdc14 were grown in 2% raffinose medium to mid-log phase and shifted to 37°C. 2% galactose was added to the medium for 1 or 2 hours, whilst holding the cells at 37°C. The total incubation time at 37°C was 6 hours. For the band-retardation assay, to visualize the slow-migrating band of Cin8 (Fig. 1H), whole-cell extracts were fractionated on a 10% SDS PAGE for 7–9 hours and subjected to western blot analysis. The antibodies used were mouse anti-HA (12CA5, Abcam), mouse anti-Myc (9E10, Santa Cruz) and goat anti-mouse IgG HRP conjugate (W402B, Promega).

The TAP method was applied for purification of cyclin Cdc28 complexes from sic1ΔP_GAL-CLB2-TAP, sic1ΔP_GAL-CLB3-TAP, and sic1ΔP_GAL-CLB5-TAP strains, as described previously (Puig et al., 2001; Ubersax et al., 2003). Purification of CIN8-590–TEV–EGFP–6xHIS and Cin8-3A-590–TEV–EGFP–6xHIS was performed using standard cobalt affinity chromatography, and the elution was accomplished with 200 mM imidazole. For the phosphorylation assays of Cin8, about 0.1 mg/ml purified Cin8 was used. The basal composition of the assay mixture contained 50 mM HEPES, pH 7.4, 150 mM NaCl, 80 mM imidazole, 2% glycerol, 2 mM EGTA, 0.25 mg/ml BSA, 80 μg/ml Cks1, and 500 μM ATP [with added [γ-³²P]ATP (Perkin Elmer)]. Around 50 nM of the purified kinase complex was used and the reaction was stopped with SDS-PAGE sample buffer.

Real-time imaging of GFP-tagged proteins and measurements of the spindle elongation kinetics were previously described (Fridman et al., 2009; Gerson-Gurwitz et al., 2009; Movshovich et al., 2008). In brief, time-lapse z-stacks images were acquired with a Zeiss Axiovert 200M-based Nipkow spinning-disc confocal microscope (UltraView ESR, Perkin Elmer, UK) equipped with an EMCCD camera, with 0.2 μm separation between planes and 4–60 second time intervals. Image processing and spindle length measurements (in 3D) were performed using ImageJ and MetaMorph software.

The significance of the differences between the average values was determined using Student's t-test. *P<0.05; **P<0.01; ***P<0.001.

The N-terminal 6-Myc tag and the C-terminal 3HA tags of Cin8 were previously described (Hildebrandt et al., 2006). A PCR-generated strategy was used to replace CDC55 open reading frame with SpHIS5 (Delneri et al., 2000) using (f) 5′-TCGATTACGTCAATTAGGCTCTCTTCTATATTTTAGTTCAGATCCGCTAGGGATAACAGG-3′ and (r) 5′-TTTCAATTAAAACAGTAGTAGTATGTGGGGAAGATATGGGTCATCGATGAATTCGAGCTC-3′ primers and the pKT101 plasmid from EUROSCARF collection (Sheff and Thorn, 2004) as a template. The C-terminal 13-Myc tag was constructed using the pKT233 plasmid from EUROSCARF collection (Sheff and Thorn, 2004). The 13Myc and ADH1 termination region were PCR-amplified from the pKT233 plasmid using (f) 5′-CGCGAGCGGCCGCCCCGGGTTAATTAACGGTGAAC-3′ and (R) 5′-GAGCGTCTAGACCCTAGCGGATCTGCCGGTAGAG-3′ primers and replacing the NotI-XbaI fragment of pTK47, creating pRM39. Then, the PstI-XbaI fragment was transferred from pRM39 to pTK49 (Hildebrandt et al., 2006), creating pRM40. Finally, the SalI-XbaI fragment of CIN8-13Myc was transferred to the pRS316 vector, resulting in pJKY1. The C-terminal 3GFP tag of Cin8 was constructed by PCR amplification of the 3GFP sequence using the pB1963 plasmid as a template [a gift from David Pellman's lab (Buttery et al., 2007)] and the (f) 5′-TCAAGCGGCCGCTCCTGCAGCCCGGGGGATCCA-3′ and (r) 5′-ACCGCGGTGGCGGCCGCT-3′ primers. The PCR-amplified sequence was introduced into the C-terminal NotI site of CIN8 (Hildebrandt et al., 2006). For genomic integration of CIN8-3GFP, CIN8-3GFP was subcloned into the pRS305 vector (Sikorski and Hieter, 1989) and integrated into the LEU2 locus. Plasmid containing the mutations of the five Cdc28 phosphorylation sites (cin8-5A) in pTK49 (Hildebrandt et al., 2006), pYLCIN8 1-5, was a gift from Gavin Sherlock (Cold Spring Harbor Laboratories, Cold Spring Harbor, NY). To construct 6Myc-cin8-5A, AatII-Sph1 fragment of cin8-5A sequence was transferred from the pYLCIN8 1-5 plasmid to pTK103 (Hildebrandt et al., 2006). The separation between the Cdc28-kinase phosphorylation sites in the motor domain and the tail of Cin8 was done using the unique PstI site within the CIN8 sequence. The phosphomimic cin8-3D mutations (D277, D285, D493) were generated by PCR site-directed mutagenesis (Ko and Ma, 2005), the following primers were used: (f) 5′-ATGCAAGCTTTGTCGACGTTTTCGCCCCAGGTT-3′, (r) 5′-ATGCGGTCTCAATGACCTAGGGTCACTGTTCCTAGAATTACTCCTGGAACTACTAGC-3′, (f) 5′-ATGCGGTCTCATCATTAAATGATCTGGATCCTAAAGCTGCTCTATTAAGAAAAAGG-3′, (r) 5′-ATGCCTGCAGGTTTCTTCAGAAGTTACCTTTGCAGGATCGATAGTAGCAATTAGTGCGGTTTTCGT-3′. The pRM59 plasmid containing the 3HA-CDC14 coding sequence under the PGAL1 promoter was constructed by PCR-amplifying the 3HA-CDC14 sequence from the p194 plasmid (a gift from Angelica Amon, MIT Cambridge, MA), using (f) 5′-CGTCGTCTAGACGCTTATAAAAAAAAAAAAAAAATGCG-3′ and (r) 5′-CGGTGAAGTTATTCCTAGGTACCAGG-3′ primers and sub-cloning into the KpnI-XbaI sites of p416-GAL1 (Mumberg et al., 1994). For bacteria expression of Cin8 motor domain, first 590 residues of (CIN8-WT) and (cin8-3A) were PCR amplified and cloned into NdeI and SphI sites of modified pRSET-B plasmid that contain TEVEGFP–6HIS at the C-terminus, creating pAG26 and pJKY34, respectively. The following primers were used: (f) 5′-AGTCCATATGGTATGGCCAGAAAGTAACGTTGAG-3′, (r) 5′-CTCGAGGCGGCCGCGCGCATGCCTTGCATTTTTCGATGTCAAACTTTCAAT-3′.

We thank David Pellman, Angelica Amon and Gavin Sherlock for providing plasmids and Monica Einav for technical assistance. This work was supported in part by ISF grant no. 1043/09 awarded to L.G. and by a BSF grant 2003141 awarded to L.G. and M.A.H. M.L. was supported by Wellcome Trust International Fellowship 079014/Z/06/Z and a grant from Estonian Science Foundation, ETF 6766. Deposited in PMC for immediate release.