Role of Kip2 during early mitosis – impact on spindle pole body separation and chromosome capture

The mitotic spindle is an elegant machine built by the cell to segregate chromosomes during mitosis. In the budding yeast Saccharomyces cerevisiae, SPBs act as microtubule-organizing centers, in a manner that is analogous to the centrosomes in mammalian cells (Jaspersen and Winey, 2004). The SPB is a dynamic organelle, with the daughter SPB assembling next to the mother SPB in G1 phase (Winey and Byers, 1993). The duplicated side-by-side SPBs are connected by a bridge at the end of G1, and this bridge is severed to allow SPB separation and formation of the two spindle poles (Adams and Kilmartin, 1999). Dynamic microtubules (MTs) from the SPBs capture duplicated chromosomes, with each SPB attached to one sister chromatid, thereby forming a bipolar spindle (Huang and Huffaker, 2006).

The budding yeast spindle is composed of three different classes of MTs, the kinetochore MTs (kMTs), which capture the chromosomes, the inter-polar MTs (ipMTs), which interdigitate with each other providing support to the spindle and the cytoplasmic MTs (cMTs), which are present outside the nucleus. cMTs, otherwise known as astral MTs (aMTs), are an integral component of the mitotic spindle and play an important role in nuclear migration from the mother to daughter cell and in spindle positioning (Shaw et al., 1997).

The dynamics of the mitotic spindle including its assembly, disassembly and positioning, are mediated by force generation from polymerization and depolymerization of MTs themselves, and from MT-associated motor proteins (McCarthy and Goldstein, 2006). MTs undergo assembly and disassembly to generate pushing and pulling forces, respectively (Dogterom et al., 2005). In vitro experiments have shown that MT assembly or disassembly by themselves can create forces strong enough to transport organelles even in the absence of motor protein-based force generation (Inoué, 1996). Nevertheless, the force generated by MT polymerization and depolymerization per se is insufficient to assemble a functional bipolar spindle.

The other major contributors to force generation are the motor proteins. Motor proteins act through their intrinsic plus-end or minus-end directed movement on the MTs (Yeh et al., 2000), and play a vital role in the regulation of MT dynamics. A key motor protein involved in cMT dynamics is Kip2p, a plus-ended kinesin. Kip2p acts as an exclusive MT polymerase that promotes the growth of cMTs (Hibbel et al., 2015) and stabilizes the cMTs by transporting the yeast CLIP170 homolog, Bik1p, towards the plus ends of cMTs (Carvalho et al., 2004). In addition, Kip2p is also essential for the maintenance of the normal number and length of cMTs (Cottingham and Hoyt, 1997; Huyett et al., 1998).

As previous studies have focused mostly on the role of cMTs during spindle positioning, not much is known about their function during early mitosis. Moreover, it is known that Kip2p is subjected to cell cycle control (Carvalho et al., 2004), with relatively low levels during G1 and higher Kip2p level during M phase, but the significance of its regulation is poorly understood. To understand the significance of the low Kip2p levels in G1, we sought to examine the effects of perturbing its abundance during G1 by using an overexpression system. Overexpression studies have a proven history of being utilized to determine the physiological significance of a protein as a complementary approach to traditional loss-of-function studies (Prelich, 2012). We found that when KIP2 is overexpressed in G1, aberrant cMT elongation resulted in SPB separation defects, leading to failure in chromosome capture and mitotic arrest due to spindle assembly checkpoint (SAC) activation. These defects could be rescued by inactivation of genes IPL1 and MAD2, which encode SAC components and by inducing expression of the gene CIN8, which encodes a motor protein that is involved in elongation of ipMTs. We propose that low Kip2p level maintains short cMTs early in the cell cycle, allowing timely SPB separation and chromosome capture prior to mitosis.

The spindle undergoes various morphological changes during the cell cycle, with relatively long cMTs and short ipMTs during G1, S and early mitosis, and shorter cMTs and long ipMTs during late mitosis (Shaw et al., 1997). In relation to this, the interplay between Kip2p levels and cMT length during the cell cycle has not been examined in depth.

To study this, we first examined spindle dynamics and Kip2p levels in GFP-TUB1 MYO1-GFP SPC42-eqFP611 KIP2-6HA PDS1-9MYC cells released from an α-factor (α-F) arrest. Spc42p (a central plaque component of SPBs) (Adams and Kilmartin, 1999; Jaspersen and Winey, 2004) fused to the red fluorescent protein eqFP611 (hereafter eqFP) was employed to visualize SPBs and Myo1p–GFP was utilized to show the bud neck. In G1 phase, the arrested cells exhibited long cMTs (Fig. S1A, pink arrows). As the cells progressed through mitosis, relatively short cMTs were observed in cells with elongated spindles at anaphase (Fig. S1A, blue arrow), as seen from the peak of large-budded cells (Fig. S1B, 90 min) and fluorescence-activated cell sorting (FACS) analysis (Fig. S1C, 90 min).

We also used western blot analysis to examine the abundance of Kip2p over the cell division cycle. As markers for progression through cell cycle, we detected Pds1p, the yeast homolog of securin, and the protein Clb2p. Pds1p prevents chromosome segregation by inhibiting the protease separase, which is required for the cleavage of the cohesin Scc1p, which holds the sister chromatids together (Ciosk et al., 1998; Michaelis et al., 1997). Clb2p is the mitotic cyclin that activates the cyclin-dependent kinase Cdk1p to promote the transition from G2 to M phase (Fitch et al., 1992; Ghiara et al., 1991; Surana et al., 1991). Both proteins peak at mitosis and are degraded as cells exit from mitosis. As can be seen, when KIP2-6HA cells were released from α-F arrest, Kip2p levels gradually increased and peaked at M-phase (Fig. S1D, 90 min) [as evident from budding index (Fig. S1B, 90 min), FACS (Fig. S1C, 90 min) and the peak in the levels of the securin Pds1p and the mitotic cyclin Clb2p (Fig. S1D, 90 min)]. This was consistent with previous findings, where low Kip2p levels were observed in G1 and Kip2p levels peaked during M-phase (Carvalho et al., 2004).

These data indicate that Kip2p levels are tightly regulated during the cell division cycle. Moreover, there appears to be an inverse relationship between the levels of the motor protein Kip2p and the cMT length during the cell cycle, suggesting that the regulation of Kip2p levels might affect cMT dynamics in the cell.

The significance of the tight cell cycle regulation of Kip2p abundance was perhaps hinted at in a previous study where the overexpression of KIP2 in cells led to hyper-elongated cMTs (Carvalho et al., 2004). As the study was performed using asynchronous cultures with cells in different phases of the cell cycle, we sought to confirm the findings using synchronized cells (GAL-KIP2-TAP) with KIP2 driven by the GAL promoter, which is commonly used for strong induction of protein expression in budding yeast (Moriya, 2015). The cells were arrested in α-F after which they were released into medium containing galactose to overexpress KIP2 from G1 phase.

Upon release from G1 phase, the wild-type cells progressed through the S phase and reached M phase at 2 h (Fig. 1A, left panel) as evident from 99±1% cells that had large buds (n=314) (Fig. 1B, Wild-type) and FACS analysis (Fig. 1C, left panel) (all results in main text given as mean±s.d. unless otherwise stated). The cMTs at this time point were relatively short, and the SPBs had been segregated into the two daughter cells (Fig. 1A, left panel, pink arrows). At 3 h after G1 release, the percentage of large-budded cells decreased to 41.8±3.5% (n=325) (Fig. 1B, Wild-type), while the percentage of cells that did not have a bud (‘unbudded’) or those that only had a small bud was increased. Moreover, the FACS profile showed a mixed population of 1N and 2N cells (Fig. 1C, left panel). In addition, western blot analysis showed that Pds1p levels at 2 h were low (Fig. 1D). At 3 h, cells had progressed to the subsequent round of the cell cycle (Fig. 1B,C).

By contrast, cells overexpressing KIP2 arrested mostly as large-budded cells (Fig. 1A, right panel) 3 h after α-F release (99±0.1%, n=347, Fig. 1B) and exhibited hyper-elongated MTs (Fig. 1A, right panel, blue arrows). FACS analysis indicated that cells had 2N DNA content (Fig. 1C), while western blots showed Pds1p and Clb2p levels remained high even 3 h after α-F release (Fig. 1D) suggesting that chromosomes remained unsegregated and cells were arrested in M phase. Strikingly, at 3 h after release from α-F, more than 60% of the cells displayed single SPB spots (Fig. 1A, right panel, pink arrows, Fig. 1E).

It was unclear whether the single SPB spots observed in KIP2-overexpressing cells were due to a problem with SPB duplication or failure in SPB separation. To determine which of the two possibilities could explain the KIP2 overexpression phenotype, we performed time-lapse experiments with wild-type and GAL-KIP2-TAP cells, to examine the dynamics of the SPBs. We assumed that if the SPBs failed to undergo duplication, only a single spot would be observed in the KIP2-overexpressing cells. However, if the SPBs were duplicated but failed to separate completely, we posited that it was possible to observe ‘breathing’ of the SPBs, akin to centromere breathing (Chai et al., 2010) due to a cycle of spindle formation followed by spindle collapse.

From our observations, wild-type cells released from G1 arrest separated their SPBs to opposite ends of dividing cells as spindles elongated normally. However, GAL-KIP2-TAP cells with long cMTs had SPBs that were separated at one time point but had then collapsed at the next time point (Fig. S2B, blue arrows). This phenomenon of SPB breathing was observed in 62.5±2.4% of cells (n=104). This confirmed our hypothesis that the SPBs had indeed duplicated, but were not fully separated in GAL-KIP2-TAP cells.

We next sought to confirm whether the hyper-elongated MTs observed (Fig. 1A, right panel) were indeed cMTs. Bik1p and Dyn1p, known cargoes of Kip2p, that are transported to the plus-ends of cMTs (Carvalho et al., 2004; Markus and Lee, 2011; Roberts et al., 2014) were employed as markers. We examined the localization of Bik1p–3GFP and Dyn1p–3GFP, respectively, in cells harboring GAL-KIP2-TAP mRuby2-TUB1. Upon overexpression of KIP2, Bik1p–3GFP spots were observed to colocalize with the hyper-elongated cMTs (Fig. S3A). Dyn1p–3GFP spots were also seen at the end of cMTs touching the cell cortex (Fig. S3B), corroborating our idea that the overexpression of KIP2 indeed led to the hyper-elongation of cMTs (as seen in Fig. 1A, right panel, blue arrows).

Given the extensive elongation of cMTs in KIP2-overexpressing cells, we wondered whether there were effects on the viability of the cells. We assessed the viability of the cells by performing spotting assays using agar plates containing raffinose (Raf) and galactose (Gal) as a means of inducing prolonged overexpression of KIP2 over a period of 2 days. Gal was added to induce the expression of the GAL promoter, whereas Raf is used in the place of glucose (Glu), since Glu represses the expression of the GAL promoter. Thus, the addition of Raf and Gal to the yeast-extract peptone (YP) agar plates mimic the conditions used in the imaging experiment (YP/Raf/Gal). Wild-type cells grew well on YP/Raf/Gal plates, whereas the growth of GAL-KIP2-TAP cells was severely compromised (Fig. 1F), indicating that overexpression of KIP2 was lethal. Our data was in agreement with a previous large-scale study demonstrating that KIP2 overexpression caused cell death (Sopko et al., 2006).

Collectively, our data showing the presence of hyper-elongated cMTs in cells overexpressing Kip2p from G1 phase were consistent with data in a recent report demonstrating Kip2p as a MT polymerase and catastrophe inhibitor (Hibbel et al., 2015). More significantly, the hyper-elongated cMTs correlated with an absence of proper SPB separation (Fig. 1A, right panel, 3 h, Fig. 1E) and a failure to degrade Pds1p and Clb2p (Fig. 1D). The mitotic arrest and lethality (Fig. 1F) upon prolonged KIP2 overexpression hinted at the physiological importance of maintaining a low level of Kip2p early on in the cell division cycle to regulate cMT lengths and ensure progression through mitosis.

The persistence of Pds1p and Clb2p levels in large-budded cells was reminiscent of cells arrested in mitosis during checkpoint activation. We next tested whether the hyper-elongated cMTs could have affected chromosome capture by examining the centromeres of chromosome V (CenV) through labeling with GFP (Tanaka et al., 2000) in KIP2-overexpressing cells.

Wild-type and the KIP2-overexpressing CenV–GFP cells (GAL-KIP2-TAP tetR-GFP 1.4 kb left of CENV::tetO2X112 SPC42-eqFP611; Table S1) were arrested in α-F and released into fresh medium containing galactose for 3 h. The association of CenV–GFP with SPBs was analyzed in mitotic cells with SPBs separated by <6 µm. Representative wild-type and GAL-KIP2-TAP cells displaying association of centromeres with SPBs are shown (Fig. 2Ai). In wild-type cells, all the CenV–GFP signal was associated with SPBs (n=318). Strikingly, 19.6±1.9% of the CenV–GFP signal in KIP2-overexpressing cells failed to associate with SPBs (n=336, Fig. 2Aii).

Our findings thus far suggest that the hyper-elongated cMTs resulting from KIP2 overexpression might affect SPB separation (Fig. 1E), which could eventually interfere with chromosome capture during mitosis (Fig. 2A). This could explain the reduced frequency of SPBs observed to be associated with CenV, as well as the mitotic arrest in KIP2-overexpressing cells.

The loss of association of CenV–GFP with SPBs in 20% of the KIP2-overexpressing cells hinted to us that the SAC could be activated, leading to the mitotic arrest (Fig. 1A). This prompted us to examine the localization of the SAC components in these cells.

To test our hypothesis, we examined the localization of the checkpoint protein Mad2p–GFP with reference to SPBs and kinetochores. Mad2p is an integral component of the SAC and localizes to the kinetochore of uncaptured chromosomes, preventing cell cycle progress until the chromosomes are bi-oriented (Musacchio and Salmon, 2007). We utilized Spc42p–eqFP (Fig. 2B) and Ndc80p–tdTomato [a component of the kinetochore-associated Ndc80p complex (Cheeseman et al., 2006; DeLuca et al., 2006), Fig. 2C] as markers for SPBs and kinetochores, respectively.

As a positive control, we arrested the wild-type cells with the MT poison Nocodazole (Noc) to activate the SAC and visualize the localization of Mad2p–GFP (Fig. 2B,C top panels, Noc treated). As anticipated, upon treatment with Noc, more than 50% of cells displayed a Mad2p–GFP spot (Fig. 2B,C top panels, blue arrows), which colocalized with both the SPB (Fig. 2B, top panel) and kinetochore (Fig. 2C, top panel, pink arrow).

We next examined whether the mere overexpression of KIP2 would lead to the activation of the SAC. Wild-type and GAL-KIP2-TAP cells released from α-F arrest into fresh YP/Raf/Gal medium, and localization of Mad2p-GFP with respect to Spc42p–eqFP was observed (Fig. 2B, bottom panel). After 3 h, very few wild-type cells had Mad2p–GFP spots (5.8±0.1%, n=308, Fig. 2D). By contrast, the frequency of appearance of Mad2p–GFP spots in GAL-KIP2-TAP cells was significantly higher than in wild-type cells (34.3±4.7%, n=347, Fig. 2D).

To confirm that the Mad2p signals were associated with kinetochores, we also examined the localization of the Mad2p signals relative to all the kinetochores marked with Ndc80p–tdTomato. Wild-type and GAL-KIP2-TAP cells arrested in α-F were released into fresh YP/Raf/Gal medium for 3 h (Fig. 2C, bottom panel). Consistently, we observed a significantly higher frequency of Mad2p–GFP spots in GAL-KIP2-TAP cells. In wild-type cells, only 8.5±0.8% displayed bright Mad2p–GFP spots (n=332, Fig. 2D). By contrast, 36±4.5% of GAL-KIP2-TAP cells (n=317, Fig. 2D) showed a Mad2p–GFP spot (Fig. 2C, lower panel, blue arrows) that colocalized with the Ndc80p–tdTomato foci (Fig. 2C, lower panel, pink arrow). The recruitment of Mad2p–GFP to the kinetochore in KIP2-overexpressing cells indicates the presence of unattached kinetochores in these cells, which results in SAC activation.

The SAC prevents anaphase onset in mitotic cells until all chromosomes are attached to the spindle (Musacchio and Salmon, 2007). Key components of the SAC include Mad2p and Ipl1p. While the checkpoint protein Mad2p monitors the attachment of chromosomes to MTs, the aurora kinase Ipl1p plays a major role in sensing the tension generated in the spindle upon proper bi-orientation of chromosomes and is necessary for proper kinetochore–MT attachment (Carmena et al., 2012; Musacchio, 2015). To further validate our observations that overexpression of KIP2 led to unattached kinetochores, we inactivated these checkpoint components individually and in combination to test whether the mitotic arrest phenotype observed in GAL-KIP2-TAP cells could be rescued.

α-F arrested cells were released into medium containing galactose at 32°C (semi-permissive temperature), to partially inactivate the ipl1-321 allele (Fig. 3A). In GAL-KIP2-TAP cells, 94±4.1% arrested as large-budded cells (n=315, Fig. 3B), whereas the proportion of large-budded GAL-KIP2-TAP mad2Δ and GAL-KIP2-TAP ipl1-321 cells was significantly lower at 52±3.6% (n=342) and 66.7±4.3% (n=318), respectively. Moreover, the proportion of large-budded GAL-KIP2-TAP mad2Δ ipl1-321 cells was further reduced to 39.5±8.5% (n=307). The FACS profiles showed an increase in 1N population, suggesting that mitotic arrest was partially rescued (Fig. 3C). The study of CenV association with SPBs had revealed that ∼20% of GAL-KIP2 cells showed a failure in chromosome capture (Fig. 2A), providing an opportunity for some of the remaining cells to properly segregate their chromosomes when the SAC was abrogated.

Consistent with the imaging and FACS results, Pds1p and Clb2p levels were also decreased in the mad2Δ ipl1-321 double mutant cells (Fig. 3D). Interestingly, Kip2p levels in ipl1-321 and mad2Δ ipl1-321 mutant cells were decreased, but this was not the case in the mad2Δ cells. In addition, the migration of the Kip2p band was altered in ipl1-321 mutant cells, indicating a possible role for Ipl1p in the post-translational modification of Kip2p. Further investigation is required to understand the potential role of Ipl1p in Kip2p stability.

Strikingly, the mad2Δ and mad2Δ ipl1-321 cell lines exhibited cells with multiple buds (Fig. 3A, blue arrow, Fig. 3B) and DNA content >2N (Fig. 3C), suggesting that cells had entered the next cycle and that the rescue is not due to a delay in cell cycle progression. Moreover, ipl1-321 and mad2Δ ipl1-321 mutants grew on YP/Raf/Gal agar plates, and had thereby partially alleviated the lethality of KIP2 overexpression (Fig. 3E).

These findings indicate that proper separation of SPBs is required to ensure timely chromosome capture and bi-orientation in the spindle. Indeed, it has been reported that maintaining the right distance between SPBs is essential to facilitate proper chromosome–MT interaction (Liu et al., 2008). Thus, timely SPB separation safeguards faithful chromosome capture, possibly by regulating cMT lengths.

Our findings thus far reflect the significance of low Kip2p levels in G1 (Fig. S1), since by inducing KIP2 expression starting from G1 (Fig. 1), we not only caused defects in SPB separation, but also reduced chromosome capture and activated the SAC in mitosis. These data imply that hyper-elongated cMTs prevented SPB separation indirectly, perhaps through pushing against the cortex and preventing extension of the spindle (Hildebrandt and Hoyt, 2000). To test this notion, we compared the dynamics of spindle elongation in the presence or absence of KIP2 in cells that were entering mitosis. If indeed cMTs have a role to play in opposing the extension of the mitotic spindle, we expected spindle elongation to be faster in kip2Δ cells, which have short or absent cMTs throughout the cell cycle (Miller et al., 1998).

Cycling kip2Δ GFP-TUB1 cells were imaged at 30 s intervals using time-lapse microscopy. First, we established that there was no significant change in the relative lengths of the spindle in wild-type and kip2Δ cells, measured as the ratio of the maximum spindle length to the length of the cell when the spindle was at its maximum length. For instance, the spindle-to-cell length ratio for wild-type cells was 1.04±0.01 (n=51) and that of kip2Δ cells was 1.01±0.01 (n=52, Fig. 4A).

Next, we determined the rate of spindle elongation based on the measured spindle length obtained at each 30-s time point from the time-lapsed images. The rate of spindle elongation of a representative cell from wild-type and kip2Δ cells was measured (Fig. 4B). The spindle elongation rate of the kip2Δ cell was higher than in the wild-type cell, during both the fast and slow phase of anaphase, as evidenced from the steeper slopes of the kip2Δ cell (Fig. 4B).

We noted that the average rate of spindle elongation in kip2Δ cells was higher than in wild-type cells (Fig. 4C). The rate of spindle elongation during the fast phase in wild-type cells was 9.30±0.51 nm/s (n=14) whereas in kip2Δ cells it was 14.89±1.60 nm/s (n=16, Fig. 4C). During slow phase, the rate of elongation in wild-type cells was 1.58±0.11 nm/s (n=14) and that of kip2Δ cells was 2.10±0.14 nm/s (n=16, Fig. 4C).

The faster elongation rate in kip2Δ cells suggested unrestrained spindle elongation in the absence of cMTs. As such, the cMTs in wild-type cells could provide an inward pushing force at the SPBs to regulate spindle elongation. This idea is consistent with the role of cMTs in generating pushing forces to orient the spindle (Adames and Cooper, 2000; Shaw et al., 1997). Growing cMTs exert a pushing force on the cortex, which, in turn, produces a counteracting force, pushing back the cMT (Dogterom et al., 2005), which is then transmitted, via cMT minus-ends, to the SPBs. In KIP2-overexpressing cells, this counteracting force would be magnified owing to hyper-elongated cMTs, and could be the key factor in preventing SPB separation.

In the GAL-KIP2-TAP cells (Fig. 1), it was possible that an overwhelming amount of pushing force generated by the cMTs due to the high levels of Kip2p prevented SPB separation. The notion that the SPB separation defect could be due to an imbalance of forces acting on the spindle by the hyper-elongated cMTs implied that increasing expression of motor proteins needed for mitotic spindle extension could rescue the SPB separation defect. We therefore tested whether the overexpression of a motor protein, such as Cin8p, which has been shown to separate SPBs when overexpressed (Crasta et al., 2006) through its sliding action that pushes interdigitating ipMTs apart (Straight et al., 1998), could abrogate the GAL-KIP2 phenotype.

We utilized the Tet-off system where CIN8 was placed under the control of Tet promoter in the GAL-KIP2-TAP strain background to provide us control of the expression of KIP2 and CIN8 independently of each other (GAL-KIP2-TAP tTA, TetR'-SSN6 TetO₂-CIN8-3HA; Table S1). These cells were arrested in G1 phase and released into fresh YP/Raf/Glu (1%) and YP/Raf/Gal (1%) media, each with or without 20 µg/ml doxycycline (Tanaka et al., 2015).

In control cells in which both KIP2 and CIN8 were repressed (1%Glu+Dox, Fig. 5Ai), SPB separation of >6 μm was found in 5.1±3.8% cells (n=392) (Fig. 5Bi, 150 min, indicated by asterisk). In comparison, when CIN8 was de-repressed, a higher percentage of cells had SPBs fully separated (Fig. 5Aii, blue arrows), with 19.3±4.1% of cells (n=357) having SPBs separated by >6 µm (Fig. 5Bii, 150 min, indicated by asterisk).

When cells were induced for KIP2 with CIN8 repressed (Fig. 5Aiii), only 14.6±1.1% of cells (n=247) had SPBs >6 µm apart (Fig. 5Biii, 180 min, indicated by asterisk). However, upon de-repression of CIN8 when KIP2 was overexpressed, the proportion of cells with SPBs fully separated (Fig. 5Aiv, blue arrows) was significantly higher with 26.8±4.6% of cells (n=173) having SPBs separated by >6 µm (Fig. 5Biv, 180 min, indicated by asterisk). These cells with de-repressed CIN8 in the presence of high Kip2p levels also had a significantly higher percentage of cells with >6 μm SPB separation at 120 min and 150 min (Fig. 5Biv).

Taken together, the data show that the force exerted by ipMT elongation induced by Cin8p was indeed able to counteract the forces generated by Kip2p-induced hyper-elongated cMTs. Our findings further support the notion that the hyper-elongated cMTs in GAL-KIP2 cells indeed exerted a pushing force on the SPBs that prevents proper separation. These results further imply that having a lower Kip2p level during early cell division could prevent the excessive pushing force being exerted on the SPBs by long cMTs, allowing timely SPB separation to be mediated by the extending spindle during early mitosis.

To support our idea that an imbalance of forces acting on the spindle due to hyper-elongated cMTs leads to chromosome mis-capture and SAC activation, we first tested whether destabilizing the MTs in KIP2-overexpressing cells and allowing normal re-growth of all MTs in the cells could allow chromosome re-capture and inactivation of the SAC. If in fact that were possible, cells would progress normally through the cell cycle once the hyper-elongated cMTs were depolymerized.

CenV–GFP cells with overexpression of Kip2p (GAL-KIP2-TAP tetR-GFP 1.4 kb left of CENV::tetO2X112 SPC42-eqFP611; Table S1) were first arrested in M-phase. We then destabilized the MTs with Noc in YPD (to shut down the GAL promoter), followed by release into fresh YPD, allowing cells to enter the subsequent cell division cycle (Fig. 6Ai). We then examined whether the cells were able to reverse the chromosome attachment defect and progress through mitosis.

In control cells, the percentage of cells with unattached chromosomes remained unchanged (Fig. 6Aii, YPD+DMSO to YPD). However, in Noc-treated cells, the percentage of cells with chromosomes that were not associated with SPBs decreased from 21.1±2.5% at 0 min (n=168) to 7.1±1.1% of cells at 30 min (n=155) after release into YPD (Fig. 6Aii, YPD + Noc to YPD). Thus, in cells with overexpressed KIP2, upon dissolution of MTs using Noc, there were more cells with their centromeres associated with the SPBs, as compared to the untreated control cells. Our results indicate that a balance of forces acting on the spindle is indeed required for proper chromosome capture given that, in KIP2-overexpressing cells, the hyper-elongated cMTs interfered with the capture of chromosomes and resulted in SAC activation.

However, although Noc treatment could rescue the chromosome attachment and mitotic arrest phenotype imposed by heightened level of Kip2p, Noc causes the dissolution of all types of MTs. To further support the notion that hyper-elongated cMTs are responsible for the mitotic arrest in KIP2-overexpressing cells, we specifically depolymerized cMTs by utilizing the auxin-inducible degron (AID) system for rapid depletion of ectopically expressed Kip2p (Kubota et al., 2013; Nishimura and Kanemaki, 2014).

GAL-KIP2-3xMini-AID cells arrested in α-F were released into galactose-containing medium for 3 h and then treated with 1 mM 1-naphthalenacetic acid (NAA) to deplete Kip2p-3xMini-AID (Fig. 6Bi). At 90 min after addition of NAA, cMTs were depolymerized (Fig. 6Bi, lower right panel, blue arrows). The percentage of large-budded cells in the control set at 90 min was 92±2.7% (n=315); however, the percentage of large-budded cells in the NAA-treated set at this time point was significantly lower at 50.6±11.5% (n=368, Fig. 6Bii).

FACS analysis at this time point showed that most cells were in haploid state (1N), suggesting that cells had been rescued from mitotic arrest (Fig. 6Biii). A western blot showed that the level of Kip2p-3xMini-AID was rapidly reduced as early as 30 min after addition of NAA [Fig. 6B(iv)]. The Clb2p levels also were decreased (Fig. 6Biv, red box), signifying that the cells had exited mitosis. At 120 min, most of the cells progressed to the next cell cycle and exhibited elongated spindles (Fig. 6Bi, pink arrow) and elevated Clb2p levels (Fig. 6Biv). Thus, depletion of Kip2p-3xMini-AID rescued the mitotic arrest of KIP2-overexpressing cells.

Our data suggest that the selective degradation of cMTs in cells utilizing auxin-degradable Kip2p is sufficient to rescue the cells from the KIP2 overexpression phenotype and allowed them to progress to the subsequent cell cycle. This portrays the significance of cMT dynamics during mitosis, not only for the proper positioning of the spindle, but also for the timely separation of SPBs, the capture of chromosomes and hence the assembly of a bipolar spindle.

The assembly of a bipolar mitotic spindle is crucial to ensure bi-orientation and faithful segregation of chromosomes to the two daughter cells. The spindle pole body constitutes the MT-nucleating center of the spindle and must be duplicated and separated to opposite ends of the nucleus for the formation of a bipolar spindle. While the process of SPB duplication is well studied, the mechanism of SPB separation is not as clearly understood. In this study, we show how the motor protein Kip2p plays a role in SPB separation, when low Kip2p levels are maintained during early phases of the cell cycle. Other motor proteins that have been previously demonstrated to play an important role in the timely separation of spindle poles are Cin8p and its functionally redundant homolog Kip1p (Crasta et al., 2006). Interestingly, the stability of these proteins was regulated by Cdk1p during SPB separation. It is noteworthy that the regulation of motor protein levels affects not only MT dynamics but also the separation of spindle poles.

From our findings, we propose that the forces acting on the SPBs during early mitosis play a crucial role in SPB separation. Regulation of cMT lengths during these early phases is essential to permit separation of the spindle poles and bi-orientation of chromosomes (Fig. 7A). An unbalanced exertion of forces on the spindle poles, such as through the hyper-elongated cMTs in KIP2-overexpressing cells, prevented the timely separation of SPBs (Fig. 7B), pushing the poles into closer proximity. This could have affected the angle of chromosome capture by the kMTs, hampering the attachment of sister chromatids to the SPBs. Consequently, the SAC was activated and cells were unable to progress through mitosis and showed reduced viability with prolonged overexpression of KIP2. This phenotype suggests that there is a window during the cell division cycle when a minimum distance between the SPBs needs to be established for proper capture of the kinetochores.

Indeed, there is some support for this notion from a previous study on SGO1 that is required to sense a lack of tension on mitotic chromosomes (Indjeian et al., 2005). In the study, wild-type and mutant sgo1-100 cells released from G1 into media containing MT poisons showed collapsed SPBs and a low frequency of bi-orientation of chromosomes after removal of the poisons. However, in cells that separated SPBs before kinetochores were captured, bi-orientation occurred normally. The authors proposed that this could be due to the tendency for chromosomes to bi-orientate due to an intrinsic geometric bias. Here, we propose that the intrinsic ability of chromosomes to bi-orient could in part be contributed by short cMTs during early cell division. When cMTs are short, SPB separation occurs in an unhindered manner. This allows for kMTs to elongate with varying degrees of freedom both in terms of length and angles about the SPBs, so as to capture kinetochores that are scattered in the cytoplasm and on opposite sides of chromatid pairs. In cells where SPBs are collapsed, the constraints on the kMTs might prevent proper attachments and affect kinetochore bi-orientation.

It remains to be seen at what stage the SPB separation failed in GAL-KIP2-TAP cells. In budding yeast cells, the duplicated SPBs start moving apart when the bridge connecting them is severed (Jaspersen and Winey, 2004). Our results show that SPBs are duplicated and show ‘breathing’ (Fig. S2) as they move apart and then collapse. As the SPBs breathe, it might be possible that spindle MTs assemble when the SPBs move apart and then collapse when the SPBs do so, as these MTs are dynamically unstable and continue to grow and shorten until they capture the kinetochore (Heald and Khodjakov, 2015). Another possible cause of spindle collapse could be the depletion of the tubulin pool of the cell (Woodruff et al., 2012) upon continuous growth of hyper-elongated cMTs after KIP2 overexpression. However, if the latter were true, the growth of all MTs should have completely stopped, whereas we observed continuous growth of cMTs even up to after 5 h of KIP2 overexpression. Therefore, although the exact timing of failure in SPB separation is not clear, the accompanying collapse of the spindle highlights the significance of timely SPB separation.

The significance of separation of spindle poles is also evident from studies in higher eukaryotes. During cell division, the microtubule organizer that plays the role of the SPB, the centrosome, is also duplicated and separated once in a timely manner that is tightly coordinated with the cell cycle machinery (Elserafy et al., 2014). In early embryonic cells of C. elegans and Drosophila, it has been shown that centrosomes are completely separated before nuclear envelope breakdown (Lee et al., 2000; Rosenblatt, 2005). The significance of timely separation of centrosomes, before chromosomes can be captured and a bipolar spindle is assembled, is revealed by a study in cultured mammalian cells (Silkworth et al., 2012). Interestingly, cells with incomplete centrosome separation displayed higher frequencies of unattached kinetochores and mis-segregated chromosomes as compared to cells with proper centrosome separation. This is consistent with our findings on how incomplete SPB separation leads to unattached kinetochores in KIP2-overexpressing cells. Additionally, the improper timing of centrosome separation has been attributed as a cause of aneuploidy in human cancers (Nam et al., 2015; Zhang et al., 2012), highlighting the importance of timely spindle pole separation during cell division.

Given the role of Kip2p in chromosome capture, there could be speculation that it has a direct role at the kinetochore and on nuclear MTs. However, Kip2p has so far only been known to play a role in the cytoplasm, regulating the dynamics of cMTs. Deletion of KIP2 affects cMT length and number (Huyett et al., 1998), and Kip2p localizes exclusively on cMTs (Miller et al., 1998). However, in mammalian cells, unlike budding yeast, the nuclear envelope breaks down and open mitosis occurs, removing the barrier between the cytoplasm and the nucleus. The motor with the most closely related domain to Kip2p in mammalian cells is CENP-E (Miki et al., 2001), a plus-end-directed mitotic kinesin that promotes MT elongation (Sardar et al., 2010) and is involved in the attachment of kinetochores to MTs (Gudimchuk et al., 2013). Disruption of CENP-E function affects chromosome behavior and SAC signaling, leading to errors in chromosome segregation (Wood et al., 2008). Thus, both Kip2p and CENP-E play a role in ensuring faithful chromosome segregation, indicating that the significance of Kip2p levels might hold true in higher eukaryotes as well.

In animal cells, where aMTs play a crucial role in the separation of centrosomes during spindle assembly (Rosenblatt et al., 2004). The aMTs undergo rapid catastrophe and rescue, to provide ‘polar ejection forces’ (PEFs) to push away cargo (Brouhard and Hunt, 2005; Rieder and Salmon, 1994). PEFs are thought to be generated by the action of chromokinesins on kMTs (Cane et al., 2013), although an earlier study had proposed that these forces could also arise due to the impact of a growing MT plus-end against chromosomes (Rieder and Salmon, 1994). In our study, similar forces could be generated upon the impact of growing cMTs on the cell cortex, pushing the SPBs closer together, preventing their separation. Taken together, our data implicate the maintenance of Kip2p levels in ensuring proper cMT length in early cell cycle to safeguard SPB separation and faithful chromosome capture. Our study therefore provides additional insights into the complex dynamics of cMT lengths, mitotic spindle elongation and chromosome capture.

Standard molecular biology and molecular genetic techniques such as PCR-based tagging of endogenous genes and tetrad dissection were used to construct strains with various genotypes (Table S1). The plasmids for the green fluorescent protein (GFP), red fluorescent protein (eqFP611), 9Myc, TAP and 6HA cassettes were obtained from EUROSCARF (Frankfurt, Germany) (Janke et al., 2004). Plasmid construct for pKAN tdtomato cyc loxP KAN loxP was Addgene plasmid 35193, deposited by Robert Singer (Larson et al., 2011). The construct for pHIS3p:mRuby2-TUB1+3'UTR::HPH was Addgene plasmid 50633 (Markus et al., 2015) and pBS-3GFP was Addgene plasmid 50659 (Lee et al., 2003), deposited by Wei-Lih Lee. The constructs for the AID system - 3xMini-AID and TDH3-yeOSTIR1, were provided by the National BioResource Project (NBRP) of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. Details of the primers used for the strain construction will be provided upon request.

The wild-type haploid W303 S. cerevisiae strain was used in this study. Yeast-extract peptone (YP) medium supplemented with 2% glucose (Glu) was used to grow cells at 24°C. Galactose (Gal) induction was performed by growing cells in YP supplemented with 2% raffinose (Raf) followed by addition of 2% Gal, unless otherwise mentioned.

Synchronization of cell cultures was performed as described previously (Chin et al., 2012). Exponential phase cells were diluted to 10⁷ cells/ml in growth medium at 24°C. For a typical G1 arrest, cells were treated with α-factor (US Biological, Swampscott, MA) at 0.4 μg/ml for 3 h. For a Noc arrest, cells were arrested with 7.5 μg/ml of Noc (US Biological) for 2 h, followed by the further addition of 7.5 μg/ml for another 2 h. Arrested cells were then washed to remove the drug, followed by re-suspension in growth medium according to the experimental conditions described.

Western blot analysis was performed as described previously (Chai et al., 2010). Anti-TAP antibody (Bethyl Laboratories, Inc., TX) was used at a dilution of 1:10,000, anti-Myc, anti-HA and anti-Clb2p antibodies (Santa Cruz Biotechnology, CA) were used at dilutions of 1:1000, anti-AID (MBL, Japan) was used at 1:5000 and anti-Pgk1p antibody (Invitrogen, CA) was used at a dilution of 1:100,000. Clarity™ Western Enhanced Chemiluminiscence kit (Bio Rad Laboratories, Inc., CA) was used according to the manufacturer's recommendations. Image Lab software (Bio Rad Laboratories, Inc., CA) was used to capture and analyze image data.

Cells expressing fluorescence (GFP, mRuby, tdTomato and eqFP611)-tagged proteins were collected at each time point, washed with sterile water and observed directly under the microscope. For time-lapse microscopy, cells were washed with complete synthetic medium (SC) containing 2% Glu and mounted onto SC-Glu agarose (5%) pads on glass slides. The protocol for confocal microscopy was adapted from previous reports (Chin et al., 2016). An Olympus IX81 epifluorescence microscope (Melville, NY), with a 60× oil lens (NA 1.4) and a 1.0× Opitvar was used to observe the cells. Sapphire LP 488 nm and 561 nm solid-state lasers (Coherent) were used for sample excitation. Images were captured by using the Photometrics 512EM-CCD attached behind the Yokogawa CSU22 (Tokyo, Japan) connected to the microscope. GFP and mRuby/eqFP/tdTomato images were captured simultaneously using a Dual-View image-splitter (Optical Insights, Tucson, AZ). Metamorph software (Molecular Devices, Sunnyvale, CA) was used to control image acquisition. Typically, the exposure time for the acquisition of the images was 0.2–1 s for GFP and 0.2–1 s for mRuby/eqFP/tdTomato per plane. Nine optical z-sections at 0.5-μm intervals were obtained for each time point, and the images shown are average projections of the z-sections. Images were processed using ImageJ (National Institutes of Health, Bethesda, MA) and Adobe Photoshop (San Jose, CA) were used for the production of the montages and figures. Statistical analysis was performed using GraphPad Prism. Unless otherwise stated, P-values were determined using Student's t-tests.

We thank Uttam Surana (Institute of Molecular and Cell Biology, Agency for Science, Technology and Research, Singapore), Orna Cohen-Fix (The Laboratory of Molecular and Cellular Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, USA) and David Pellman (Department of Pediatric Oncology, Dana-Farber Cancer Institute, USA) for their generous gift of yeast strains. We also appreciate the technical assistance of Yuan Yuan Chew. We are grateful to the anonymous reviewers for their comments and suggestions for improving the manuscript.