An engineered minimal chromosomal passenger complex reveals a role for INCENP/Sli15 spindle association in chromosome biorientation

In eukaryotes, accurate segregation of the genetic material requires the products of DNA replication, paired sister chromatids, to connect with the correct geometry to spindle microtubules. Each chromatid must attach exclusively to microtubules from one spindle pole, and the two identical chromatids of a pair must connect to opposite poles. Erroneous connections that form on the path to this bioriented state are corrected by the kinase Aurora B. Aurora B phosphorylation of kinetochores lowers their affinity for microtubules, increasing turnover of kinetochore–microtubule attachments (Tanaka et al., 2002; Cheeseman et al., 2006; Sarangapani et al., 2013; Kalantzaki et al., 2015), which improves the likelihood that all chromosomes achieve the correct bioriented state (Lampson et al., 2004; Cimini et al., 2006). The budding yeast Saccharomyces cerevisiae is an important model to understand the mechanism of Aurora B action, as inhibition of Aurora B (Ipl1) in this species leads to very high (>90%) missegregation that arises from the inability to correct syntelic kinetochore–microtubule attachments, wherein the kinetochores of sister chromatids are attached to the same spindle pole (Biggins and Murray, 2001; Tanaka et al., 2002; Pinsky et al., 2006).

Aurora B/Ipl1 is a member of the four-subunit chromosomal passenger complex (CPC), along with Sli15/INCENP, Bir1/Survivin, and Nbl1/Borealin. The CPC has two conserved localizations: on chromatin and on microtubules (Cooke et al., 1987). In prometaphase and metaphase, the complex primarily localizes to chromatin at the inner centromere through either the Borealin or Survivin subunits, which form a three-helix bundle with the N-terminal CEN box of INCENP (Jeyaprakash et al., 2007; Kelly et al., 2010; Wang et al., 2010; Yamagishi et al., 2010; Cho and Harrison, 2011). Although in fission yeast and cancer cell lines, perturbation of the inner centromeric localization of the CPC results in misalignment and missegregation of chromosomes (Dai et al., 2005; Vader et al., 2006; Yamagishi et al., 2010), in chicken DT40 cells and in budding yeast, mutations that mislocalize the CPC away from the inner centromere are viable and do not exhibit significant chromosome missegregation (Yue et al., 2008; Campbell and Desai, 2013). In budding yeast, disruption of centromere localization is synthetic lethal with mutations affecting centromeric cohesion, suggesting that this localization may be relevant for a nonessential function in chromatid cohesion at centromeres (Campbell and Desai, 2013).

At anaphase onset, the CPC transitions to the microtubules between the separating chromatids. This localization relies on the central region of INCENP. Although the sequence is not well conserved, a coiled-coil is predicted in this region of INCENP from different species (Mackay et al., 1993; van der Horst and Lens, 2013). Although initially expected to be a coiled-coil, this region was later predicted to instead be a single α helix (SAH) that does not oligomerize (Peckham and Knight, 2009), and this prediction was subsequently verified (Samejima et al., 2015). The INCENP SAH binds directly to microtubules in vitro and the interaction requires positively charged residues of the predicted helix (Samejima et al., 2015; van der Horst et al., 2015).

The S. cerevisiae INCENP, Sli15, has a region adjacent to its putative SAH region (based on coiled-coil predictions), which is phosphorylated by both Cdk1/Cdc28 and Aurora B/Ipl1 and is dephosphorylated by the phosphatase Cdc14 at anaphase onset. We name this region, which has no structured domains, the phosphoregulated (PR) region (Fig. 1 A). Preventing Sli15 phosphorylation in this region causes it to localize prematurely to spindle microtubules in metaphase, demonstrating that dephosphorylation of this region regulates the transition to microtubules in anaphase. Nonphosphorylatable mutants of the PR region decrease the rate of anaphase spindle elongation but do not otherwise affect growth or viability (Pereira and Schiebel, 2003; Nakajima et al., 2011; Makrantoni et al., 2014). In contrast, deletions within the PR region are lethal and result in high chromosome missegregation rates (Sandall et al., 2006; Campbell and Desai, 2013).

At the C terminus of INCENP, distal to the SAH, is the conserved IN box that binds to and activates Aurora B kinase (Sessa et al., 2005). Aurora B is also activated by physical clustering of the CPC, e.g., by localization to the microtubule surface or by antibody-mediated cross-linking, which promotes transphosphorylation of neighboring molecules (Kelly et al., 2007; Tseng et al., 2010). We previously showed that a Sli15 deletion mutant that prevents localization to the inner centromere leads to both its and Ipl1’s increased enrichment on metaphase spindles, which led to the model that kinase activation by clustering on microtubules was sufficient for chromosome biorientation in the absence of inner centromere localization (Campbell and Desai, 2013).

In this study, to test the clustering-based kinase activation model for the role of microtubule binding by Sli15, we used truncations and artificial dimerization to engineer a minimal CPC, comprised of Ipl1 and a dimerized SAH–IN box segment of Sli15, that we show is sufficient for the CPC’s essential biorientation function. Although dimerization compensated for the removal of the PR region of Sli15, it did not compensate for removal of the putative SAH. Notably, analysis of the minimal CPC that requires dimerization for function did not reveal a role for clustering-based kinase activation and instead suggested that spindle association, likely mediated by the SAH, is essential to target substrates at the kinetochore in order to correct misattachments and promote biorientation.

To determine whether specific parts of the PR region of Sli15 are required for chromosome segregation, we engineered sequential deletion constructs (Fig. 1 A). All five tested deletions, including the smallest deletion (Δ2–300) extending past the region previously demonstrated to be dispensable (Δ2–228; Campbell and Desai, 2013), resulted in lethality, as assayed by the inability of mutant spores to form colonies after tetrad dissection (Fig. 1 A). Immunoblots using an antibody raised against the C terminus of Sli15 showed that the truncated constructs were expressed at levels equal to or greater than endogenous Sli15 (Fig. 1 B). A construct that deletes only the PR region (Δ228–450) was also lethal, indicating that the lethality was not a synthetic effect of deleting both the PR region and the CEN box.

As biorientation is the only known essential function of the CPC in budding yeast, these results indicate that the PR region contributes to chromosome segregation. We tested this directly by monitoring the missegregation of chromosome IV in the first cell division after depletion of full-length Sli15 expressed under the control of a galactose-inducible promoter (Fig. 1 C). In the absence of any rescuing Sli15, the missegregation rate for chromosome IV was ∼60% per division. The Δ2–300 and Δ2–350 deletions resulted in missegregation rates of ∼5% for chromosome IV per division, which is consistent with the observed lethality. Larger deletions resulted in higher missegregation rates of ∼30% for the marked chromosome, indicating that these deletions were significantly compromised in their ability to correct syntelic kinetochore–microtubule attachments. Thus, the central PR region of Sli15 plays an essential role in chromosome segregation.

Based on the role of clustering in Aurora B kinase activation, a likely explanation for why deletions in the microtubule-binding region are lethal is that they limit kinase activation by reducing clustering on microtubules. To test this possibility, we artificially dimerized Sli15 through an N-terminal fusion with the obligate dimer GST. Addition of GST rescued the lethality of all of the PR region deletions (Fig. 2 A). Growth of even the largest deletion construct (GST-Δ2–500) was indistinguishable from full-length Sli15, as assayed by both colony size and dilution series on rich media (Fig. 2, A and B). When challenged with a moderate concentration of the microtubule-depolymerizing drug benomyl, only GST-Δ2–500 displayed increased drug-sensitivity relative to full-length (Fig. 2 B). Direct measurement of chromosome missegregation revealed that fusion to GST lowered missegregation rates of PR region deletions to levels indistinguishable from GST-fused, full-length Sli15 (Fig. 2 C). Collectively, these results indicate that the essential function or functions of the PR region of Sli15 can be compensated for by fusion to the exogenous dimer, GST.

To test whether dimerization is the basis for the rescue of PR region deletion by GST fusion, we fused Sli15Δ2–450 to FKBP–Rapamycin binding (FRB) and FKBP12 domains, coexpressed the fusions in the same cell, and inducibly heterodimerized them by addition of the drug Rapamycin (Fig. 2 D). We observed a rescue of the lethality of Sli15Δ2–450 only in the presence of Rapamycin, providing strong support for the conclusion that dimerization is the property that rescues the lethality associated with the removal of the PR region. One explanation for rescue observed by GST fusion or inducible dimerization mediated by Rapamycin is that they compensate for dimerization mediated by the PR region. However, we were unable to detect association between coexpressed Venus- and 6HA-tagged Sli15 via immunoprecipitation from yeast extracts, suggesting that Sli15 is not dimeric in vivo (Fig. S1 A), similar to what has been shown for INCENP in human cells (Klein et al., 2006).

To confirm that the PR region contributes to microtubule localization in vivo, we generated Venus fusions of GST-dimerized truncations and imaged them in asynchronous cells or cells arrested in metaphase by Cdc20 depletion (Figs. 2 E and S1 B). As shown previously (Campbell and Desai, 2013), deletion of the CEN box caused an increase in CPC localization to the metaphase spindle (Figs. 2 E and S1 B), potentially because of a decreased phosphorylation of the PR region. Consistent with this, progressive deletions of the PR region reduced the elevated localization on the metaphase spindle. The decrease in spindle localization upon PR region deletion was not caused by differences in protein expression; however, the decrease in spindle localization for full-length Sli15 upon the addition of GST likely was caused by decreased expression (Fig. S1 C). For mutants with the largest deletions, GST-Sli15Δ2–450 and GST-Sli15Δ2–500, preanaphase spindle fluorescence was not abolished but was similar to full-length Sli15 (Fig. S1 B). Thus, the deletions of the PR region that are lethal unless rescued by dimerization contribute to preanaphase spindle association in vivo.

We conclude that artificial dimerization mediated by two different means rescues the lethality and substantial chromosome missegregation observed after deletion of the PR region of the Sli15 microtubule-binding domain.

We next wanted to test the model that dimerization rescued deletions within the microtubule-binding domain by promoting transphosphorylation-based kinase activation (Fig. 3 A). For this purpose, we used the chemically inducible dimerization system to engineer heterodimers of Sli15Δ2–450, in which one of the two copies of Sli15 harbored two mutations in the IN box (W646G and F680A; Fig. 3 B). W646 and F680 are ubiquitously conserved IN box residues that contact two locations on the Aurora B kinase N-lobe surface and contribute significant binding affinity (Sessa et al., 2005). Consistent with their conservation and importance for kinase binding, full-length Sli15 with W646G and F680A mutations failed to rescue viability after depletion of endogenous Sli15 (Fig. S2 A).

Rapamycin-induced heterodimers engineered to bind only one copy of Ipl1 rescued the Δ2–450 PR region deletion (Fig. 3 C), demonstrating that transphosphorylation-based activation is not critical for the phenotypic suppression observed after inducible dimerization. This rescue required the presence of both the FRB and FKBP12 constructs (Fig. S2 B), suggesting that dimerization rescues deletion of the PR region through a mechanism that is independent of clustering-based kinase activation.

To confirm the aforementioned result using an independent assay, we measured phosphorylation of the conserved Aurora B substrate histone H3 on serine 10. H3S10 phosphorylation is ubiquitous throughout mitosis irrespective of the attachment state of the kinetochores and serves as a readout for global kinase activity. Depletion of full-length Sli15 using a promoter shutoff decreased H3S10 phosphorylation (Fig. 3 D). In contrast, Sli15Δ2–450 exhibited H3S10 phosphorylation equal to that of full-length Sli15 (Fig. 3 D), and fusion of this fragment to GST did not elevate H3S10 phosphorylation but instead appeared to modestly decrease it. In addition, chromosome missegregation or spindle checkpoint–mediated arrest phenotypes associated with increased Ipl1 kinase activity after overexpression (Muñoz-Barrera and Monje-Casas, 2014) were not observed after dimerization (not depicted).

To address the effect of Sli15 truncations on the phosphorylation of kinetochore substrates of Ipl1, we used an antibody raised against a peptide containing phosphorylated serine 37 of the kinetochore protein Ndc80. Immunoblots revealed a similar decrease in signal after the depletion of Sli15 (Fig. 3 D) or in a temperature-sensitive Ipl1 mutant (Fig. S2 C). No signal was detected when seven Ipl1 consensus sites in Ndc80, including serine 37, were mutated to alanine (Akiyoshi et al., 2009), confirming the phosphospecificity of the antibody (Fig. S2 C). In contrast to H3S10 phosphorylation, Ndc80 serine 37 phosphorylation was not restored by the Sli15Δ2–450 mutant but was restored by fusion of GST to Sli15Δ2–450. Thus, deletion of the majority of the PR region does not affect global kinase activity, but it reduces the phosphorylation of a kinetochore substrate, and consistent with the rescue of chromosome segregation and viability, fusion of the truncated Sli15 to GST restores kinetochore substrate phosphorylation.

Collectively, these results indicate that dimerization of the Sli15 microtubule-binding domain deletion mutants does not act by promoting global kinase activation but restores targeting of kinetochore substrates, which is likely the basis for the rescue of viability.

As the largest GST-fused PR region deletions exhibited microtubule localization similar to full-length Sli15 during anaphase (Fig. S1 B), we next wanted to test whether artificial dimerization acted by promoting microtubule localization. For this purpose, we added the mNeonGreen fluorophore to the C terminus of FKBP12-Sli15Δ2–450 and compared the effects of removing or adding back Rapamycin on localization (Fig. 4 A). In the presence of Rapamycin and FRB-Sli15Δ2–450, we observed robust mNeonGreen signal at both anaphase and preanaphase spindles (the signal was stronger with the mNeonGreen fusion relative to the Venus fusions likely because of its superior fluorophore properties in vivo; Fig. 4 B). Spindle localization decreased significantly when cells were switched to media without Rapamycin (Fig. 4, B and C). When Rapamycin was returned to the media, mNeonGreen signal rapidly and fully returned to the spindles (Fig. 4, B and C). Expression of the mNeonGreen-conjugated fusion was not altered by the presence or absence of Rapamycin (Fig. 4 D). These data indicate that induced dimerization increases the spindle association properties of either the C-terminal part of Sli15 or of Ipl1 that is bound to the Sli15 IN box. Furthermore, the results suggest a correlation between spindle association (Fig. 4, B and C) and the ability to promote chromosome biorientation and support viability (Fig. 2 D).

The aforementioned results suggest that dimerization promotes microtubule association of a region distal to amino acid 500 of Sli15 or of Ipl1 bound to Sli15. The SAH of INCENP, incorrectly predicted as a coiled-coil that ends ∼50 amino acids before the IN box in vertebrates, is required for enrichment of the CPC on anaphase spindles in human cancer cell lines and binds directly to microtubules (Samejima et al., 2015; van der Horst et al., 2015), making it a good candidate for a microtubule-binding domain whose affinity is promoted by dimerization. In Sli15, which is not dimeric (Fig. S1 A), coiled-coil predictions identify residues 520–565, and we therefore refer to these residues as the SAH (Fig. 5 A). Similar to the results with Rapamycin-induced dimerization, addition of GST to a Venus-tagged Sli15 construct that lacks the N-terminal 500 amino acids increased the localization to the spindle in the first cell division after depletion of full-length Sli15 (Fig. 5 B). To determine whether the SAH is required for this spindle localization, we deleted an additional 74 amino acids that include the putative SAH domain. Venus–GST-Sli15Δ2–574 exhibited strongly decreased spindle association both before and after anaphase onset (Fig. 5 B). Thus, artificial dimerization of a Sli15 fragment lacking the first 500 amino acids promotes spindle association, and this association requires the SAH that immediately follows the truncated region.

As the SAH of Sli15 is required for spindle localization, we next determined its contribution to chromosome segregation and viability. Spores containing deletions of Sli15 that remove the SAH alone or together with the CEN box and the PR region failed to form colonies, even when fused to GST, indicating that the SAH is essential for viability, is not interchangeable with the PR region, and cannot be compensated for by artificial dimerization (Figs. 6 A and S3, A and B). Consistent with this, Rapamycin-induced dimerization of Sli15 missing only the SAH (Δ520–574) also failed to rescue viability (Fig. 6 B). A heterodimer consisting of one Sli15 molecule missing the SAH and another Sli15 with point mutants in the IN box was also inviable, suggesting that the SAH and the IN box cannot act in trans (Fig. 6 B). Furthermore, the chromosome missegregation rate of a GST-fused mutant that deletes both the PR and SAH regions (GST-Δ2–574) was comparable with Sli15 depletion (Fig. 6 C; compare with Fig. 2 C).

To determine whether removal of the SAH leads to a penetrant Sli15 loss-of-function phenotype because it is required to activate Ipl1 in vivo, we analyzed H3S10 phosphorylation, which is significantly reduced by Sli15 depletion (Figs. 3 D and 6 D). The GST-Δ2–574 mutant exhibited H3S10 phosphorylation close to wild-type levels (Fig. 6 D), indicating that the extreme C-terminal 124 amino acids of Sli15, which include the IN box, are sufficient for kinase activation. In contrast to H3S10, phosphorylation of the kinetochore substrate Ndc80 at serine 37 was reduced in the GST-Δ2–574 mutant to a comparable level as Sli15 depletion (Fig. 6 D). We confirmed a reduced phosphorylation of kinetochore substrates in this mutant by analyzing the electrophoretic mobility of Dam1, a subunit of the kinetochore-localized microtubule-binding Dam1 complex that is targeted by Ipl1. A shift in Dam1 mobility is detectable that is sensitive to phosphatase treatment, to temperature-sensitive inactivation of Ipl1, and to depletion of Sli15 (Figs. 6 D and S3 C; Storchová et al., 2011; Muñoz-Barrera and Monje-Casas, 2014). In the GST-Sli15Δ2–574 mutant, the shift in Dam1 mobility was reduced to the same degree as Sli15 depletion (Fig. 6 D). Thus, phosphorylation of two major Ipl1 targets at the kinetochore, Ndc80 and Dam1, requires the SAH of Sli15, irrespective of artificial dimerization. We conclude that the putative SAH of Sli15 is dispensable for the global activation of Ipl1 but is required for the kinase activity to target substrates at the kinetochore, and this requirement cannot be superseded by artificial dimerization.

In budding yeast, the Aurora kinase Ipl1 promotes biorientation by correcting the naturally high frequency of syntelic attachments that are caused by centromere replication and kinetochore formation before maturation of the daughter spindle pole body (Tanaka et al., 2002). As in other species, Ipl1 is present in a complex with the conserved CPC subunits Sli15/INCENP, Bir1/Survivin, and Nbl1/Borealin. In this study, we define a minimal biorientation module that is sufficient for viability and contains only the C-terminal region of Sli15/INCENP together with Ipl1 (Fig. 7 A). The engineered Sli15 in this module lacks ∼70% of its primary amino acid sequence, including the centromere/chromatin localization domain that is required to bind to Bir1/Survivin-Nbl1/Borealin in vivo (Campbell and Desai, 2013) as well as a large segment of its central region that contributes to spindle association and is phosphorylated by Cdk1 and Ipl1.

This minimal CPC is only functional in combination with induced dimerization (Fig. 7 A). Although we and others have been unable to identify an oligomeric form of INCENP in vivo (Fig. S1 A; Klein et al., 2006), there is evidence from human cell culture that Borealin dimerization contributes to the spindle checkpoint function of the CPC (Bekier et al., 2015). Notably, dimerization-based rescue of lethality by the minimal CPC is not caused by the promotion of transphosphorylation-based activation of the kinase, a prominent model for how dimerization/clustering influences CPC function (Kelly et al., 2007; Tseng et al., 2010). Instead, artificial dimerization promotes localization to the spindle and restores the phosphorylation of kinetochore substrates, suggesting that preanaphase spindle association is important for Ipl1 to execute its biorientation function.

Deletions within the microtubule-binding domain of Sli15 severely disrupt the chromosome segregation function of the CPC and are lethal. This is in contrast to mutants that prevent or mimic phosphorylation in this region, which display little to no growth defects (Pereira and Schiebel, 2003; Nakajima et al., 2011; Makrantoni et al., 2014). Based on this difference, we suggest that phosphorylation of this region of Sli15 allows for weak spindle association, which is enhanced upon dephosphorylation in anaphase. Localization of the CPC to preanaphase spindles has been previously reported in budding yeast (Buvelot et al., 2003; Nerusheva et al., 2014), Xenopus laevis egg extracts (Tseng et al., 2010), and HeLa cells (Banerjee et al., 2014). Our emphasis on the importance of this preanaphase spindle association in biorientation comes from the key features of the minimal biorientation module: the putative SAH, the kinase-binding IN box, and the catalytic Ipl1 kinase subunit. The IN box and Ipl1 are both essential for kinase activity; in contrast, the SAH is not required for global kinase activation but is required to target kinetochore substrates. Based on the correlation between dimerization-based rescue and spindle association, we propose that the SAH by itself has low affinity for microtubules, but the PR domain, via its independent weak affinity, generates sufficient preanaphase microtubule-associated, IN box–bound, and activated Ipl1 to correct syntelic attachments (Fig. 7 B). This model suggests why removal of the PR region can be compensated for by artificial dimerization, as it brings together two SAHs, and the consequent increased avidity enhances microtubule binding. This proposal is supported by the ability of the SAH from INCENP to directly bind microtubules (Samejima et al., 2015; van der Horst et al., 2015) and by the negative effect of SAH deletion on INCENP localization to microtubules (van der Horst et al., 2015). Additionally, an INCENP mutant lacking the SAH fails to activate the spindle checkpoint in response to taxol treatment in human U2OS cells (Vader et al., 2007). One potential reason for why microtubule association is important is that it may lead to a reduction in dimensionality, increasing the activity of Aurora B toward microtubule-bound substrates (Noujaim et al., 2014). Microtubule association may also help preferentially target incorrectly microtubule-bound versus unattached kinetochores (Fig. 7 B). This view is in agreement with analysis in human cells that showed a graded degree of Ndc80 phosphorylation with kinetochores in prometaphase cells exhibiting the highest phosphorylation, followed by unattached kinetochores in nocodazole-treated cells, and finally kinetochores of metaphase cells (DeLuca et al., 2011).

In contrast to the PR region, truncations removing the SAH of Sli15 cannot be compensated for by dimerization. Thus, the PR region and the SAH are not interchangeable, and the SAH is essential for biorientation regardless of dimerization (Fig. 7 B). In preliminary work, lethality caused by point mutants that decrease the spindle localization of Sli15’s SAH is suppressed by mutations that increase the spindle binding of the PR domain, supporting the notion that the SAH contributes to chromosome biorientation at least in part through spindle association (unpublished data). The fact that the SAH has specific physical characteristics suggests that microtubule association may not be its only activity relevant for biorientation. Our attempts to replace the SAH with microtubule-binding regions from three different yeast microtubule-associated proteins have not been successful to date (Stu2, Irc15, and She1; unpublished data), which indicates that the SAH may serve an additional function. This view is supported by the inability of the SAH to function with the IN box in trans, suggesting that the two domains need to be adjacent on the same polypeptide. The special property of the SAH could be a particular mode of microtubule binding, tethering of the activated kinase at a specific distance from the microtubule surface based on its predicted extended structure, or an additional, as yet unidentified, function, such as mediating binding to a site at the kinetochore. Our definition of a minimal module that provides the essential biorientation function of the CPC in budding yeast should help address the fascinating and challenging question of how cells detect and correct errors in chromosome–spindle attachments to ensure accurate segregation of the genome.

See Table S1 for all yeast strains and plasmids used in this study. Strains were grown in either yeast extract/peptone supplemented with 40 µg/ml adenine-HCl (YPA) or synthetic media with 2% sugars. Rapamycin (Santa Cruz Biotechnology, Inc.) was used at a concentration of 100 ng/ml. Cultures were incubated at 30°C unless otherwise indicated. Epitope and fluorescent tags were inserted at the C terminus of genes at their native loci as described previously (Longtine et al., 1998). Gene truncations and point mutants of plasmids were made using Gibson assembly (Gibson et al., 2009). PCR products were purified from agarose gels and combined 1:1 with Assembly Master Mix (1 U/µl T5 exonuclease, 33 U/ml Phusion DNA polymerase, 5 U/µl Taq DNA ligase, 13 mM magnesium chloride, 13 mM dithiothreitol, 8 µM deoxynucleotides, 1.3 mM nicotinamide adenine dinucleotide [NAD⁺; New England Biolabs, Inc.], 67 mg/ml polyethylene glycol, and 133 mM Tris, pH 7.5), incubated for 1 h at 50°C, and transformed into chemically competent bacteria. Plasmids were integrated either at their native loci or at the URA3 or HIS3 loci by digesting the plasmids with the StuI, EcoRV, or SnaBI restriction enzymes. All plasmids and integrations were checked by PCR and sequencing.

To immunoprecipitate Sli15-6HA, 50 ml cells in exponential growth were pelleted and resuspended in 600 µl immunoprecipitation buffer with protease inhibitors (50 mM Tris, pH 7.6, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 2 mM phenylmethylsulphonyl fluoride, 4 mM benzamidine, and cOmplete EDTA-free protease inhibitor cocktail [Roche]) and vortexed for 45 min with 400 µl glass beads at 4°C. Lysates were cleared at 18,000 g for 10 min, transferred to a new tube, and centrifuged again. A total of 25 µl of antibody–bead slurry (anti-HA monoclonal clone 3F10; Roche) was combined with 500 µl of cleared lysate and rotated at 4°C for 1 h. Beads were washed once with immunoprecipitation buffer and three times with Tris-buffered saline, pH 7.4, and then were resuspended in 100 µl protein sample buffer. Whole-cell extracts were prepared for Western blotting by pelleting 2 ml of cultures in exponential growth and resuspending them in 100 µl sample buffer and 30 µl glass beads. Samples were then vortexed at 4°C for 30 min and centrifuged at 18,000 g for 5 min. Extracts for phosphoprotein analysis were prepared by pelleting 2 ml of cells, resuspending them in 100 µl of 5% trichloroacetic acid, and incubating them at room temperature for 10 min. The cells were then washed once with water and resuspended in 100 µl lysis buffer (50 mM Tris, pH 7.4, 1 mM EDTA, 50 mM DTT, cOmplete EDTA-free protease inhibitor cocktail, and PhosSTOP [Roche]) plus glass beads. Samples were vortexed at 4°C for 30 min, followed by addition of 33 µl of 4× sample buffer.

For immunoblots, samples were separated on 10% or 15% acrylamide gels, transferred to nitrocellulose membranes, and blotted with one of the following antibodies: Sli15–rabbit polyclonal raised against the C terminus (Sandall et al., 2006), HA–clone 3F10 (Roche), Myc–mouse monoclonal 4A6 (EMD Millipore), H3–rabbit polyclonal 9715 (Cell Signaling Technology), Phospho-H3S10–rabbit polyclonal D2C8 (Cell Signaling Technology), Pgk1–mouse monoclonal 22C5D8 (Thermo Fisher Scientific), and Ndc80 phospho-S37–rabbit polyclonal (a gift from S. Biggins, Fred Hutchinson Cancer Research Center, Seattle, Washington). The Ndc80 antibody was incubated with unphosphorylated competitor peptide at 5× concentration by mass. The membranes were then probed with horseradish peroxidase–conjugated secondary antibodies. Intensities of the blots were quantified using ImageJ (National Institutes of Health).

Overnight cultures were diluted 100-fold and grown out of stationary phase for 4–5 h. Cells were pelleted by brief centrifugation, washed once and resuspended in water, placed on 1% agarose pads supplemented with complete synthetic media, covered with a coverslip, and sealed around the edges with VALAP (a 1:1:1 mixture of petroleum jelly [Vaseline], lanolin, and paraffin [Thermo Fisher Scientific] by weight). Images were collected on a DeltaVision microscopy system (Applied Precision, Ltd.) at 22°C using a 60×, 1.35 NA Olympus U-Plan Apochromat objective and a CoolSNAP charge-coupled device camera (Photometrics). 14 z-sections were taken with 0.5-µm steps and deconvolved with softWoRx software (Life Sciences Software). Further image analyses, including maximum-intensity projections and contrast adjustments, were performed in ImageJ. Images within each figure were all collected on the same day and contrast adjusted identically. Spindles were determined to be preanaphase if they were >2 µm in length and had not yet aligned with the bud neck. Anaphase spindles were categorized as those that had passed through the bud neck but had not yet reached the ends of the mother/bud. Spindles of similar length were used for quantification. Quantification of spindle intensities was performed by first taking a line scan in ImageJ perpendicular to the spindle with a 5-pixel-wide line. The intensity profile was then analyzed using a custom python script that fits a Gaussian curve to the line scan and calculates the area under the Gaussian fit (Text S1).

For chromosome segregation assays, Sli15 expressed from the Gal10-1 promoter was depleted by first synchronizing cells for 90 min in YPA plus 1% galactose and 1% raffinose (YPAGR) and 10 µM α-factor. The cultures were then switched to YPA plus 2% glucose (YPAD) and 10 µM α-factor for 90 min to deplete Sli15. To release from G1 arrest, cells were washed five times with fresh YPAD and incubated for an additional 90 min before imaging. For metaphase arrest with Cdc20 depletions, asynchronous cultures in YPAGR were washed three times and switched to YPAD media for 2.5 h before imaging.

Diploid strains were sporulated by first growing them on YPAD plates overnight at 30°C and then transferring them to sporulation plates (8.2 mg/ml sodium acetate and 16 mg/ml agar) and incubating at 23°C for 2–3 d. Spores were then washed with water and digested with 1 mg/ml zymolyase (Zymo Research) for 5–10 min at 30°C and then dissected onto YPAD or YPAD plus Rapamycin plates. Genotyping was performed by replica plating onto selective media.

Fig. S1 shows additional experiments and controls to accompany Fig. 2, including an immunoprecipitation experiment with two differently labeled forms of Sli15, localization of Sli15 variants at different cell cycle stages, and quantification of Sli15 mutant expression by Western blotting. Fig. S2 contains controls for experiments in Fig. 3. Fig. S3 shows controls and example images for experiments in Fig. 6. Table S1 contains a list of all of the yeast strains and plasmids used in this study and includes the figures in which the corresponding strains were used. Text S1 shows code for a python script used to measure fluorescence intensity on the mitotic spindle.

The authors thank the Desai, Oegema, and Campbell laboratories for helpful discussions and for comments on the manuscript. We thank Sue Biggins for the gifts of yeast strains, peptides, and the Ndc80 phosphoantibody.

This work was supported by National Institutes of Health grant GM074215 to A. Desai and Vienna Science and Technology Fund grant VRG14-001 and Austrian Science Fund grant Y 944-B28 to C.S. Campbell. A. Desai receives salary and other support from the Ludwig Institute for Cancer Research.

The authors declare no competing financial interests.

Author contributions: S. Fink performed experiments and edited the manuscript. K. Turnbull performed experiments. A. Desai conceived experiments and edited the manuscript and figures. C.S. Campbell conceived, designed, and performed experiments, wrote the manuscript, and created the figures.