Kinetochore-associated Mps1 regulates the strength of kinetochore-microtubule attachments via Ndc80 phosphorylation

Dividing cells detect and correct erroneous kinetochore-microtubule attachments during mitosis, thereby avoiding chromosome mis-segregation. Most studies of this process have focused on the Aurora B kinase, which phosphorylates microtubule-binding elements specifically at incorrectly attached kinetochores, promoting their release and providing another chance for proper attachments to form. However, growing evidence suggests additional mechanisms, potentially involving Mps1 kinase, that also underlie error correction. Because these mechanisms overlap in vivo, and because both Mps1 and Aurora B function in numerous other vital processes, their contributions to the correction of erroneous kinetochore attachments have been difficult to disentangle. Here we directly examine how Mps1 activity affects kinetochore-microtubule attachments using a reconstitution-based approach that allowed us to separate its effects from Aurora B activity. When endogenous Mps1 that co-purifies with isolated kinetochores is activated in vitro, it weakens their attachments to microtubules via phosphorylation of Ndc80, a major microtubule-binding element of the outer kinetochore. Mps1 phosphorylation of Ndc80 appears to contribute to error correction because phospho-deficient Ndc80 mutants exhibit genetic interactions and segregation defects when combined with mutants in an intrinsic error correction pathway. In addition, Mps1 phosphorylation of Ndc80 is stimulated on kinetochores lacking tension. These data suggest that Mps1 provides an additional mechanism for correcting erroneous kinetochore-microtubule attachments, complementing the well-known activity of Aurora B.


Introduction
The equal segregation of duplicated chromosomes to daughter cells during cell division is fundamental to life. Segregation is mediated by interactions between dynamic microtubules and kinetochores, the megadalton protein complexes that assemble on the centromeres of each chromosome (Monda and Cheeseman, 2018;Musacchio and Desai, 2017). For accurate segregation, the sister kinetochores on each pair of chromatids must make bioriented attachments to microtubules emanating from opposite spindle poles.
When linked sister kinetochores achieve biorientation, they come under tension due to pulling forces exerted by the opposing microtubules. However, because kinetochoremicrotubule attachments initially form at random, erroneous connections lacking tension are often made. These must be detected and corrected to avoid mis-segregation. Tension appears to help cells distinguish correct from incorrect attachments, because attachments under tension are more stable than those lacking tension in vivo and in vitro (Akiyoshi et al., 2010;King and Nicklas, 2000;Nicklas and Koch, 1969).
A variety of error correction mechanisms help cells to make proper bioriented attachments. The most well-studied mechanism involves the conserved essential protein kinase Aurora B (Krenn and Musacchio, 2015). When a kinetochore attaches incorrectly, the lack of tension is believed to signal Aurora B kinase to phosphorylate kinetochore proteins, which weakens their grip on the microtubule, causing detachment, and giving the cell another chance to make a proper attachment (Biggins et al., 1999;Dewar et al., 2004;Hauf et al., 2003;Tanaka et al., 2002). When Aurora B is defective, cells are unable to make bioriented attachments (Biggins et al., 1999;Hauf et al., 2003;Tanaka et al., 2002). One of the major Aurora B substrates is Ndc80, a core component of the kinetochore that makes a significant contribution to kinetochore-microtubule coupling (Cheeseman et al., 2006;DeLuca et al., 2006). Ndc80 has two domains that mediate its interaction with the microtubule, a conserved calponin-homology 'head' domain and a disordered N-terminal 'tail' (Ciferri et al., 2008;Wei et al., 2007). The  tail contains multiple Aurora B consensus sites and is the site of Aurora B regulation (Akiyoshi et al., 2009;Cheeseman et al., 2006;Ciferri et al., 2008;DeLuca et al., 2006;Guimaraes et al., 2008;Miller et al., 2008;Wei et al., 2007). Additional kinetochore proteins are also phosphorylated by Aurora B to destabilize kinetochore-microtubule attachments but they vary depending on the organism (Cheeseman et al., 2002;Lan et al., 2004;Wordeman et al., 2007).
However, its role in regulating kinetochore-microtubule attachments and error correction has not been fully elucidated. In human cells, Mps1 regulates localization of the motor protein CENP-E and the Ska complex, which stabilize proper kinetochore-microtubule attachments (Espeut et al., 2008;Hewitt et al., 2010;Maciejowski et al., 2017;Stucke et al., 2004). In budding yeast, Mps1 is required for localization of the Dam1 complex (Meyer et al., 2018;Meyer et al., 2013), an ortholog of the Ska complex (van Hooff et al., 2017).
It also phosphorylates the Ndc80 protein, but this phosphorylation was reportedly involved in spindle checkpoint signaling and not in error correction (Kemmler et al., 2009).
Recently, it was reported that Mps1 regulates biorientation via phosphorylation of the Spc105 protein (Benzi et al., 2020). While Mps1 targets many key microtubule-binding kinetochore elements, a unified view of how it participates in error correction remains elusive, in part because it is challenging to disentangle this function from the similar function of Aurora B and the well-established role for Mps1 in checkpoint signaling.
To directly study the effect of Mps1 activity on kinetochore-microtubule attachments, we took advantage of a yeast reconstitution system where the strength of attachment between individual isolated kinetochores and single microtubules can be measured in vitro (Akiyoshi et al., 2010;Sarangapani et al., 2013). Using isolated kinetochore particles that copurify with Mps1 kinase but lack other kinase activity (London et al., 2012), we found that Mps1 phosphorylation of Ndc80 directly weakens kinetochoremicrotubule attachments, similar to the function of Aurora B. Moreover, phosphorylation of Ndc80 in vivo by Mps1 occurs during mitosis when kinetochores are prevented from coming under tension. Cells containing mutations in the Mps1-targeted phosphorylation sites on Ndc80 exhibit genetic interactions and chromosome segregation defects when combined with inhibitory mutations in another, intrinsic error correction pathway (Akiyoshi et al., 2010;Miller et al., 2018;Miller et al., 2016). Taken together, our data suggest that Mps1-mediated phosphorylation of Ndc80 provides an additional mechanism for specifically weakening kinetochore-microtubule attachments that lack tension, complementing the Aurora B and intrinsic error correction pathways, thereby helping to release erroneous attachments and ensure the accuracy of chromosome segregation.

Results
Copurifying kinase activity weakens the attachment of isolated kinetochores to microtubules We previously found that native kinetochore particles purified from budding yeast lack Aurora B kinase activity and that the major copurifying kinase activity is due to Mps1 (London et al., 2012). To test whether the copurifying Mps1 activity directly affects interactions between kinetochores and microtubules, we modified our previously developed approach for measuring the strengths of individual kinetochore-microtubule attachments in vitro (Akiyoshi et al., 2010), by adding ATP to activate any copurifying kinase. We isolated kinetochores via anti-Flag immunoprecipitation of the Dsn1 protein (Dsn1-6His-3Flag) (Supplemental Figure S1A). After elution with Flag peptide, we linked the native kinetochore particles to polystyrene microbeads, mixed them with ATP to activate the copurifying kinase, and immediately introduced them into a flow chamber containing dynamic microtubules grown from coverslip-anchored microtubule seeds ( Figure 1A). During the time required to seal the slide chamber and mount it onto the laser trap (~10 min), some kinetochore-decorated beads attached spontaneously to the sides of coverslip-anchored microtubules. The laser trap was used to bring these beads that were laterally attached to microtubules to the plus end tips and then apply gradually increasing force until the kinetochores ruptured from the tips ( Figure 1B and 1C). Rupture strengths for many individual attachments were collected in the presence of ATP, or ADP, or without adenosine, until 90 min after sealing the slide. For each population, the median rupture force was calculated and the fraction of attachments that survived up to a given level of force was also plotted. Similar to our previous work (Akiyoshi et al., 2010;Sarangapani et al., 2013), kinetochore particles not exposed to ATP ruptured over a range of forces, with a median strength of 9.8 piconewtons (pN) in the absence of adenosine, or 8.7 pN in the presence of ADP ( Figure 1D). However, when ATP was included, the rupture force distribution was shifted to lower values and the median strength was only 5.3 pN, suggesting that phosphorylation by a kinetochore-associated kinase decreased the strength of kinetochore-microtubule attachments. Consistent with this interpretation, if λ-phosphatase was included together with the ATP, then the ATPdependent weakening was reduced and the kinetochores maintained a median strength of 7.5 pN, similar to untreated and ADP-treated kinetochores ( Figure 1D and Supplemental Figure S1B). We monitored the rupture strength as a function of time, which indicated that the ATP-dependent weakening reaction was completed during the ~10 min slide preparation, with no significant weakening thereafter (Supplemental Figure S1C).
Altogether, these data indicate that phosphorylation of native kinetochores by a copurifying kinase activity reduces the strength of their attachments to microtubules.
Mps1 activity is required for the ATP-dependent weakening of isolated kinetochores Because Mps1 was the only kinase activity we detected on purified kinetochores (London et al., 2012), we next tested whether Mps1 mediates the ATP-dependent weakening of kinetochore particles using an Mps1 inhibitor called reversine . We first verified that reversine inhibits Mps1 activity on the particles by incubating them with radioactive ATP in the presence or absence of the drug (Figure 2A).
Autoradiography revealed phosphorylation on Spc105, Ndc80 and Dsn1 when reversine was omitted (Figure 2A), in agreement with our previous work (London et al., 2012). When reversine was added, total phosphorylation levels on all the substrates decreased by 69%, confirming significant inhibition. We then performed rupture force assays, mixing kinetochore-decorated microbeads with reversine plus ATP, or with the vehicle DMSO plus ATP, ADP, or no adenosine as controls. The median strength of kinetochores measured in the presence of DMSO was 8.2 pN without adenosine and fell to 5.0 pN in the presence of ATP. This loss of strength was completely blocked by reversine ( Figure 2B), indicating that the ATP-dependent weakening of isolated kinetochore particles requires Mps1 activity.

ATP-dependent weakening depends on phosphorylation of Ndc80 not Spc105
We next sought to identify the key substrate(s) whose phosphorylation by Mps1 causes the weakening of reconstituted kinetochore-microtubule attachments. Mps1mediated phosphorylation of Spc105 is vital for initiating spindle checkpoint signaling (London et al., 2012;Shepperd et al., 2012;Yamagishi et al., 2012), and recent work suggests it also promotes kinetochore biorientation through recruitment of the Bub1 protein (Benzi et al., 2020;Storchova et al., 2011). To test whether Spc105 is the relevant substrate underlying ATP-dependent kinetochore weakening in vitro, we purified kinetochores from phospho-deficient spc105-6A cells, which carry alanine substitutions to block phosphorylation at all six Mps1 phosphorylation sites (the 'MELT' motifs) within the disordered N-terminal region of Spc105 (London et al., 2012) (Figure 3A and Supplemental Figure S2A). In rupture force assays, the median strength of phosphodeficient Spc105-6A kinetochores was 8.3 pN without adenosine and fell to 3.2 pN in the presence of ATP ( Figure 3B). We were surprised that the Spc105-6A kinetochores were weaker after ATP addition than their wildtype counterparts, so we analyzed the levels of copurifying Mps1. The phospho-deficient Spc105-6A kinetochores retained more Mps1 than wildtype (Supplemental Figure S2B), potentially explaining their enhanced sensitivity to ATP. While the reason for this higher retention of Mps1 on Spc105-6A kinetochores remains unclear, the rupture strength data nevertheless indicate that Spc105 is not the key Mps1 target underlying ATP-dependent weakening of the kinetochores in vitro.
We next focused on the major microtubule binding component of the kinetochore, Ndc80c (Akiyoshi et al., 2010;Cheeseman et al., 2006;DeLuca et al., 2006). Previous work identified a total of fourteen sites on the Ndc80 protein that are phosphorylated by Mps1 (Kemmler et al., 2009) ( Figure 3C). However, three of these overlap with known Aurora B phospho-sites (Akiyoshi et al., 2009), so we tested whether the remaining eleven are involved in the ATP-dependent weakening. We purified kinetochore particles from phospho-deficient ndc80-11A cells, which carry alanine substitutions at all eleven of the Mps1 phosphorylation sites, to prevent their phosphorylation (Supplemental Figure   S2C) (Kemmler et al., 2009). Similar levels of Mps1 copurified with both Ndc80-11A and wild type kinetochores (Supplemental Figure S2D), but the kinase activity had no effect on the rupture strengths of Ndc80-11A kinetochores. The median strength of Ndc80-11A kinetochores was 8.7 pN in the absence of adenosine, similar to wild type kinetochores ( Figure 3D). Upon exposure to ATP, the median strength of Ndc80-11A kinetochores remained high, 8.5 pN, and statistically indistinguishable from the strength measured without adenosine ( Figure 3D and Supplemental Figure S2E). These data indicate that one or more of the Mps1 phosphorylation sites on Ndc80 are required for the decreased attachment strength.
Mps1 phosphorylation of Ndc80 is not required for spindle assembly checkpoint signaling A previous report suggested that Mps1-mediated phosphorylation of Ndc80 regulates the spindle assembly checkpoint and does not affect kinetochore-microtubule attachments or chromosome segregation (Kemmler et al., 2009), a conclusion that seemed inconsistent with our in vitro results. We therefore re-examined the phenotypes of mutant ndc80-14A cells, with alanine substitutions at all fourteen Mps1 target sites included for consistency with the earlier study. To test whether ndc80-14A mutants are defective in the spindle assembly checkpoint, we released wild type and ndc80-14A cells from G1 into the microtubule depolymerizing drug nocodazole, which generates unattached kinetochores that normally trigger the spindle assembly checkpoint, and monitored cell cycle progression by analyzing levels of the anaphase inhibitor Pds1/securin (Cohen-Fix et al., 1996). Pds1 levels accumulated and then stabilized in both wild type and ndc80-14A cells as they progressed through the cell cycle and then arrested ( Figure 4A). The stabilization of Pds1 in both wild type and ndc80-14A cells depended on the spindle assembly checkpoint, because it was eliminated in both strains by deletion of the checkpoint gene MAD2 ( Figure 4A). These observations indicate that blocking the phosphorylation of all known Mps1 target sites on the Ndc80 protein does not lead to a defective spindle assembly checkpoint as previously reported (Kemmler et al., 2009). It was also reported that replacing all fourteen Mps1 target residues with phospho-mimetic aspartic acid residues was lethal due to constitutive activation of the spindle assembly checkpoint (Kemmler et al., 2009). To test this, we performed a plasmid shuffle assay using strains expressing a wild-type copy of NDC80 from a plasmid that also contains the URA3 gene, which renders cells susceptible to the cytotoxic drug 5-Fluoroorotic acid (5-FOA) (Boeke et al., 1987). Cells carrying a functional chromosomal copy of NDC80 can spontaneously lose the NDC80-URA3 plasmid and are therefore able to grow on media containing 5-FOA, whereas cells with a non-functional chromosomal allele require the plasmid for viability and are killed by 5-FOA. Using this assay, we confirmed that ndc80-14D is recessive lethal ( Figure 4B). However, in contrast to the previous report (Kemmler et al., 2009), deleting the spindle assembly checkpoint gene MAD2 did not rescue this lethality ( Figure 4B). Notably, spurious colonies appeared after the ndc80-14D mutants were grown for longer times, suggesting the existence of spontaneous suppressor mutations (Supplemental Figure S3). While our data indicate that mutation of the previously identified Mps1 phosphorylation sites on Ndc80 does not cause defects in spindle assembly checkpoint signaling, they confirm the previously reported lethality of ndc80-14D (Kemmler et al., 2009). Therefore, one or more of the Mps1 phosphorylation sites have a critical cellular function.

Weakening occurs via phosphorylation of Mps1-specific targets in N-terminal tail of Ndc80
Our observation that Mps1-mediated phosphorylation of Ndc80 weakens reconstituted kinetochore-microtubule attachments is similar to the well-documented function of Aurora B-mediated phosphorylation of Ndc80, which inhibits kinetochoremicrotubule attachments in vivo and in vitro (Cheeseman et al., 2006;Ciferri et al., 2008;DeLuca et al., 2006;Sarangapani et al., 2013;Wei et al., 2007). Eight of the eleven Mps1specific target sites on Ndc80 fall within the disordered N-terminal tail domain, which is also where the key Aurora B target sites are located ( Figure 3C). To analyze the contribution of the Mps1-specific sites within the Ndc80 tail ( Figure 5A), we generated a mutant with alanine substitutions at just these eight sites (Nc80-8A). We initially included an epitope tag (-3HA) to allow direct immunoprecipitation of the phospho-deficient Ndc80-8A, or of wild type Ndc80 as a control. Mps1 copurified with wild type Ndc80, as previously demonstrated (Kemmler et al., 2009), and we found that a similar level of Mps1 copurified with Ndc80-8A ( Figure 5B). After incubating both immunoprecipitations with radioactive ATP, autoradiography showed 49% less phosphorylation on Ndc80-8A relative to wild type ( Figure 5B), confirming that Mps1 phosphorylates Ndc80 at one or more of the eight Mps1-specific target sites within its N-terminal tail domain.
To test whether these Mps1-specific sites in the Ndc80 tail are involved in ATPdependent weakening of kinetochore-microtubule attachments, we isolated kinetochores from strains containing wild type or Ndc80-8A protein (via anti-Flag-based immunoprecipitation of Dsn1-6His-3Flag). Based on silver-staining after SDS-PAGE, the composition of the purified Ndc80-8A kinetochore was similar to wild type (Supplemental Figure S4A). In the rupture force assay, the Ndc80-8A kinetochores had a median strength of 9.4 pN when no adenosine was included, and a strength of 8.6 pN in the presence of ADP, values similar to wild type kinetochores ( Figure 5C). However, unlike wild type kinetochores, the Ndc80-8A kinetochores were unaffected by exposure to ATP, maintaining a high median rupture strength of 9.4 pN ( Figure 5C). To further test the importance of the Mps1-specific tail sites, we also purified Ndc80-8D kinetochores carrying aspartic acid substitutions at all eight sites, to mimic their phosphorylation. The phospho-mimetic Ndc80-8D kinetochores had a composition similar to wild type kinetochores (Supplemental Figure S4A), but their rupture strength was constitutively low, with a median rupture strength of only 4.8 pN in the absence of adenosine ( Figure 5C). Altogether, these observations suggest that Mps1 phosphorylation of the Ndc80 tail is necessary and sufficient for directly weakening the isolated kinetochores upon exposure to ATP.

Mps1 phosphorylation of Ndc80 regulates kinetochore function in vivo
To determine whether Mps1 phosphorylates Ndc80 in vivo, we set out to generate an antibody that recognizes an Mps1 site. Although we were not able to produce one against a site specific to Mps1 phosphorylation, we succeeded in generating a polyclonal antibody that recognizes Ndc80 carrying a phosphate modification at Thr-74, a site that is phosphorylated by both Ipl1 and Mps1 (Akiyoshi et al., 2009;Kemmler et al., 2009). On immunoblots of purified kinetochore complexes, this Ndc80-T74P antibody detected a signal corresponding to the Ndc80 protein that was diminished when the kinetochores were treated with λ-phosphatase ( Figure 6A), confirming the phospho-specificity of the antibody. To check whether Mps1 contributes to the phosphorylation of this site, we purified kinetochores from mutant mps1-1 cells, in which Mps1 is specifically inactivated when the cells are grown at a non-permissive temperature ( Figure 6B). The Ndc80 signal detected in wild type kinetochore particles by the Ndc80-T74P antibody was reduced when mutant mps1-1 kinetochores were probed. To confirm that Aurora B also phosphorylates this residue, we purified kinetochores from ipl1-321 cells that lack Aurora kinase activity and also found a reduction in phosphorylation of T74 on Ndc80 (Biggins et al., 1999) (Figure 6C). Taken together, these data show that both Mps1 and Aurora B contribute to Ndc80-T74 phosphorylation on kinetochores in vivo.
A lack of tension on erroneous kinetochore-microtubule attachments is thought to promote their release by triggering Aurora B-mediated kinetochore phosphorylation (Biggins and Murray, 2001;Nicklas and Koch, 1969;Stern and Murray, 2001;Tanaka et al., 2002). To test whether a lack of tension can also trigger Mps1-mediated kinetochore phosphorylation in vivo, we analyzed the phosphorylation of Ndc80-T74 on kinetochores in a mutant mcd1-1 strain, which cannot generate tension on its kinetochores due to a defect in sister chromatid cohesion (Stern and Murray, 2001;Tanaka et al., 2000). Wild type and mcd1-1 mutant cultures were released from G1 to the non-permissive temperature and kinetochores were purified at various time points as the cells progressed through the cell cycle. The cells had MAD3 deleted to ensure equivalent progression through the cell cycle, which we verified by analyzing Pds1 levels in the lysates ( Figure 6D). Immunoblotting confirmed that kinetochores purified from either wild type or mcd1-1 cells contained similar amounts of Ndc80 protein. In the mad3∆ background, we detected weak Ndc80-T74 phosphorylation on wild type kinetochores at time points corresponding to mitosis. However, kinetochores purified from the cohesion-deficient mcd1-1 cells showed an enrichment of phosphorylation at Ndc80-T74 ( Figure 6D), indicating that this phosphorylation is enhanced in vivo under conditions where kinetochores lack tension. To test whether Mps1 contributes to this phosphorylation, we repeated the experiment and compared mcd1-1 mad3∆ cells to mcd1-1 mad3∆ mps1-1 cells. Strikingly, the phosphorylation that occurs in the cohesion mutant was reduced in the absence of Mps1 activity ( Figure 6E), suggesting that Mps1 contributes to Ndc80 phosphorylation in vivo in response to tension defects.
While Aurora B and Mps1 both regulate the N-terminus of Ndc80, mutating their phosphorylation sites in the Ndc80 tail is not lethal ( Figure 6F and (Akiyoshi et al., 2009)), consistent with additional pathways contributing to error correction. One such mechanism is the direct stabilization of kinetochore-microtubule attachments by tension (Akiyoshi et al., 2010), an intrinsic property of kinetochores that requires kinetochore-bound Stu2 (Miller et al., 2018;Miller et al., 2016). We previously found that mutants in Aurora B exhibit genetic interactions with stu2 cc∆ , an allele of Stu2 that specifically causes defects in kinetochore biorientation (Miller et al., 2018). We therefore tested whether the Mps1dependent error correction pathway we identified also exhibits growth defects in combination with stu2 cc∆ . Although the ndc80-8A cells carrying phospho-blocking substitutions at Mps1-specific target sites are viable, combining ndc80-8A with stu2 cc∆ led to a growth defect ( Figure 6F). Thus we analyzed chromosome segregation in these cells by generating ndc80-8A stu2 cc∆ strains with a fluorescent marker on a single chromosome and with the endogenous wild type STU2 gene under control of a conditional auxininducible degron (stu2-AID) to maintain viability. Because stu2 cc∆ causes a spindle assembly checkpoint-dependent arrest, we also deleted the MAD3 gene to enable the cells to progress to anaphase. After arresting the cells in G1, we released them into growth media with auxin to repress the endogenous stu2-AID gene and then analyzed chromosome segregation in anaphase. Consistent with our prior work (Miller et al., 2018), chromosome mis-segregation was increased in stu2 cc∆ mad3∆ cells relative to mad3∆ control cells (6.5% in STU2 WT mad3∆ cells versus 19.4% in stu2 cc∆ mad3∆; Figure 6G).
The addition of ndc80-8A strongly exacerbated the chromosome mis-segregation defect (30.3%), indicating that Mps1 phosphorylation of Ndc80 contributes to accurate chromosome segregation.

Discussion
To study the effects of kinetochore-associated Mps1 on kinetochore-microtubule attachments independently of other cellular kinases or pathways, we devised a reconstitution approach where the endogenous Mps1 that copurifies with isolated kinetochores is activated via exposure to ATP immediately before laser trap strength measurements. Our data show that Mps1-mediated phosphorylation of the N-terminal tail of Ndc80 causes a direct weakening of the kinetochore-microtubule interface in vitro. This function was previously difficult to identify in vivo due to the presence of Aurora B that can also phosphorylate Ndc80. Our in vitro approach allowed us to separate their functions. We further show that Mps1 phosphorylation of Ndc80 occurs in vivo, and is enhanced when kinetochores lack tension due to defective sister chromatid cohesion.
Moreover, specifically blocking this Mps1-mediated phosphorylation exacerbates chromosome mis-segregation when combined with a stu2 mutant defective in another kinase-independent, intrinsic error correction pathway (Miller et al., 2018;Miller et al., 2016). Together, these observations suggest that Mps1 contributes directly to the release of erroneous kinetochore-microtubule attachments.
A previous report identified a total of fourteen Mps1 target sites on Ndc80 and suggested that their phosphorylation was important for regulating the spindle assembly checkpoint without affecting kinetochore-microtubule attachments (Kemmler et al., 2009).
While we confirmed the lethality of the phospho-mimetic ndc80-14D mutant, its lethality was not suppressed in our hands when the checkpoint was eliminated, nor was our phospho-deficient ndc80-14A strain defective in checkpoint signaling as reported (Kemmler et al., 2009). These differences in our findings may be due to different strain backgrounds or the possibility that suppressors can arise more easily in the absence of the checkpoint. The ndc80-8D strain with phospho-mimetic substitutions at only the eight Mps1-specific target sites within the Ndc80 tail produced constitutively weak kinetochores but nevertheless remained viable. Its viability indicates that the lethality of ndc80-14D probably depends on phosphorylation of Mps1 sites outside the N-terminal tail whose function will be important to identify in the future.
Phosphorylation at one or more of the eight Mps1-specific target sites within the Ndc80 tail is necessary and sufficient for Mps1-mediated weakening of reconstituted kinetochore-microtubule attachments. While other yeast kinetochore proteins have been identified as substrates of Mps1 (Benzi et al., 2020;London and Biggins, 2014a;London and Biggins, 2014b;Shimogawa et al., 2006), Spc105 phosphorylation does not appear to directly regulate kinetochore attachment strength. The Ndc80 tail is also a major target of regulation by Aurora B kinase, which in yeast phosphorylates four distinct Aurora Bspecific phospho-sites, plus three shared sites phosphorylated by both Mps1 and Aurora B. Phosphorylation of the human Ndc80 (Hec1) tail by Cdk1 kinase was also identified recently and implicated in the correction of erroneous kinetochore-microtubule attachments (Kucharski et al., 2021). Aurora A also phosphorylates a site in the Ndc80 tail during mitosis (DeLuca et al., 2018). Why multiple different kinases phosphorylate a single domain within Ndc80 is not yet clear, but this convergence might allow weakening of kinetochore attachments in response to different kinds of error signals, potentially with different cell cycle timing. At least three other outer kinetochore components are regulated by both Aurora B and Mps1: the yeast Dam1 complex (Cheeseman et al., 2002;Shimogawa et al., 2006), its functional metazoan counterpart the Ska complex (Maciejowski et al., 2017;Redli et al., 2016), and the CENP-E motor protein (Espeut et al., 2008;Kim et al., 2010). Phospho-proteomic analysis also suggests that Ndc80 might be an Mps1 target in human cells (Maciejowski et al., 2017). Thus, outer kinetochore function could be regulated convergently by both kinases in multiple ways and across species.
Mps1-mediated phosphorylation of the Ndc80 tail causes a weakening of kinetochores similar to that caused by phospho-mimetic substitutions at all seven Aurora B target sites within the tail (Sarangapani et al., 2013). This similarity is consistent with the simple view that Ndc80 tail phosphorylation provides rheostat-like control of kinetochore attachment strength (Zaytsev et al., 2015), with each phosphorylation event contributing equally and the resultant strength varying in proportion to the total number of unphosphorylated tail sites. However, a more complex view is suggested by the recent finding that some sites in the Ndc80 tail are dephosphorylated at metaphase while others remain phosphorylated throughout the cell cycle (DeLuca, 2017;Kucharski et al., 2021).
Evidently not all phosphorylation sites are redundant, and therefore some might make differential contributions to chromosome segregation. In the future, it will be important to determine whether Aurora B and Mps1 act sequentially or simultaneously and whether any specific phosphorylation sites are more important for the regulation of kinetochore attachment strength to better understand why cells utilize multiple kinases to regulate the same sites (DeLuca et al., 2018;Kucharski et al., 2021;Zaytsev et al., 2015).
Although both Mps1 and Aurora B kinase can directly weaken kinetochores, mutating either one causes significant biorientation defects in vivo, which indicates that the two cannot fully compensate for one another to support this function. Such a requirement for both kinases in vivo, despite their similar direct effects in vitro, could arise because they each respond to different types of attachment errors or with different timing.
Presumably it also reflects additional indirect effects that occur in cells when these kinases are mutated. If Aurora B is defective, for example, the opposing phosphatase PP1 prematurely localizes to kinetochores (Rosenberg et al., 2011), potentially causing dephosphorylation of Ndc80 and counteracting the role of Mps1 in error correction.
Thus reducing Mps1 activity could indirectly reduce Aurora B activity at kinetochores to a level that is insufficient for error correction. The Bub1-Sgo1 pathway also contributes to kinetochore biorientation (Fernius and Hardwick, 2007;Peplowska et al., 2014), and it was recently reported that Mps1 phosphorylation of Spc105 is a key Mps1 target for kinetochore biorientation in vivo (Benzi et al., 2020). We found that Spc105 phosphorylation does not contribute to the Mps1-dependent weakening of reconstituted kinetochore-microtubule attachments in vitro. This result is consistent with our purified kinetochores lacking the Bub1-Sgo1 pathway and suggests that Mps1 plays multiple roles in achieving proper kinetochore-microtubule attachments in vivo. In the future, it will be important to understand the interplay between these pathways.
In summary, cells appear to rely on multiple kinetochore kinases as well as intrinsic kinase-independent mechanisms to avoid erroneous kinetochore-microtubule attachments during mitosis. While this complexity reflects sophisticated regulation, it also poses a major challenge for distinguishing direct contributions of specific kinases and substrates from indirect effects and, more generally, for developing a complete understanding of mitotic error correction. The combination of in vitro reconstitution with in vivo analysis that we used here to uncover a direct regulation of kinetochore attachment strength by Mps1 should be useful in the future for further dissection of the vital processes by which dividing cells ensure the accuracy of chromosome segregation.

Yeast strains and plasmids
Saccharomyces cerevisiae strains used in this study are derivatives of SBY3 (W303) and described in Supplemental Table 1. Standard media and microbial techniques were used and yeast strains were constructed via standard genetic techniques (Rose et al., 1990).
Plasmids were fully sequenced. All plasmids are described in Supplemental Table 2 and   primers in Supplemental Table 3.

Auxin inducible degradation
The auxin inducible degradation system was used as previously described (Miller et al., 2016). Cells expressing a C-terminal fusion of an auxin responsive protein (IAA7) in the presence of TIR1, which is required for auxin induced degradation, were treated with 500μM IAA (indole-3-acetic acid dissolved in DMSO; Sigma) to induce degradation of the desired AID-tagged target protein. For Fig 6G, auxin was added immediately after cells were released from alpha-factor.

Serial dilution assay
The indicated yeast strains were grown overnight in YPD (2% glucose) medium. The next day, the cells were diluted to OD600 ~1.0. A serial dilution (1:5) series was made in a 96well plate and cells were spotted onto YPD or YPD + 500μM auxin (indole-3-acetic acid dissolved in DMSO; Sigma) plates. Plates were incubated for 1-3 days at 30 °C unless otherwise indicated.

Chromosome segregation and time course assays
Cells were grown at 23 °C in YPDA medium (2% glucose, 0.02% adenine). For the spindle assembly checkpoint assays, exponentially growing cells were arrested in G1 with 1 μg/ml alpha factor for 3-4 hours, washed 3 times, and then resuspended in medium lacking pheromone but containing 10ug/ml nocodazole. Samples were collected at the indicated times. 1 μg/ml alpha factor was added to the cultures 40-50 minutes after the G1 release to prevent cells from entering a second cell cycle.
To analyze chromosome segregation, exponentially growing MATa cells containing a LacI-GFP fusion and 256 lacO sequences integrated proximal to CEN8 were arrested in G1 with 1μg/mL alpha-factor. After 2.5 hours, cells were washed and released into medium lacking pheromone, but containing 500μM IAA. Alpha-factor was re-added ~75 minutes after release to prevent entry into the subsequent cell-cycle. 120 minutes after G1 release, aliquots of cells were fixed with 3.7% formaldehyde in 100mM phosphate buffer (pH 6.4) for 10 min. Cells were then washed once with 100mM phosphate buffer (pH6.4), and then permeabilized and stained with DAPI by resuspending in 1.2M Sorbitol/1% Triton X-100/100mM phosphate buffer (pH 7.5) containing 1ug/mL DAPI (Molecular Probes) for 5 min. Cells were then resuspended in the same buffer lacking DAPI and applied to a coverslip treated with 0.5mg/mL concanavalin A. Cells were imaged on a Deltavision Ultra deconvolution high-resolution microscope equipped with a 100x/1.4 PlanApo N oil-immersion objective (Olympus) with a 16-bit sCMOS detector.
Cells were imaged in Z-stacks through the entire cell using 0.2 μM steps. All images were deconvolved using standard settings and analyzed softWorX 7.2.1(GE).
To analyze phosphorylation of Ndc80 during the cell cycle, cells were initially grown at 23°C in YPD medium (2% glucose). Exponentially growing MATa cells were arrested in G1 with 1μg/mL alpha-factor. After 2.5 hours, cells were washed and released into medium lacking pheromone and shifted to 37°C to inactivate mcd1-1. 100mL aliquots of cells were taken at 0, 60, and 90 minutes. Kinetochore purification assays were performed as described below for each timepoint and phosphorylation of Ndc80 threonine-74 was analyzed via immunoblot.

Kinetochore purification assay
Native kinetochore particles were purified essentially as previously described (Akiyoshi et al., 2010;Miller et al., 2016). Kinetochores were purified from asynchronously grown S. cerevisiae cells grown in YPD medium (2% glucose) unless otherwise noted in the text. Beads were washed once with lysis buffer containing 2mM dithiothreitol (DTT) and protease inhibitors, 3 times with lysis buffer with protease inhibitors, 1 time in lysis buffer without inhibitors, and kinetochores particles were eluted by gentle agitation of beads in elution buffer (Buffer H + 0.5mg/mL 3FLAG Peptide (Sigma)) for 30 min at room temperature.

Immunoblot and silver stain analysis
Cell lysates were prepared as described above (kinetochore purification assay) or by bead-beat pulverization (Biospec Products) with glass beads in sodium dodecyl sulfate (SDS) buffer. Standard procedures for immunoblot and SDS-polyacrylamide gel electrophoresis (SDS-PAGE) were followed (Biggins et al., 1999). SDS-PAGE gels were transferred to 0.2μm nitrocellulose membrane (Bio-Rad) using either the semi-dry (Bio-Rad) or wet method (Hoefer). The anti-Mps1 antibodies were generated in rabbits against a recombinant Mps1 protein fragment (residues 440-764) of the protein by Genscript. The company provided affinity purified antibodies that we validated by purifying kinetochores from yeast strains with Mps1 or Mps1-13Myc and confirming that the antibody recognized a protein of the correct molecular weight that migrated more slowly with the 13Myc epitope tags. We subsequently used the antibody at a dilution of 1:10,000. The phospho-specific T74 Ndc80 antibody was generated by Pacific Immunology against the phospho-peptide spanning resides 69-80 of Ndc80 (KRTRS-pT-VAGGTN-Cys) and subsequently affinity purified. Immunoblotting with this antibody was performed at a dilution of 1:1,000 with 1μg/mL of non-phosphorylated competitor peptide (KRTRS-T-VAGGTN-Cys). The Ask1 antibody was affinity purified with recombinant GST-Ask1 from serum generated in rabbits against recombinant Dam1 complex and was used at a dilution of 1:5,000 (Gutierrez et al., 2020). The following commercial antibodies were used for immunoblotting: α-PGK1 (Invitrogen) 1:10,000, α-FLAG (M2; Sigma) 1:3,000, α-Myc (7D10; Cell Signaling) 1:1000, and α-HA (12CA5; Roche) 1:10,000 dilution. The α-Ndc80 antibody (OD4), used at 1:10,000, was a kind gift from Arshad Desai. Secondary antibodies used were donkey anti-rabbit antibody conjugated to horseradish peroxidase (HRP) (GE Biosciences) at 1:10,000 or a sheep anti-mouse antibody conjugated to HRP (Ge Biosciences) at 1:10,000. Antibodies were detected using SuperSignal West Dura Chemiluminescent Substrate (Thermo Scientific). Immunoblots were imaged with a ChemiDock MP system (Bio-Rad) or film. For silver stain analysis, protein samples were separated using pre-cast 4%-12% Bis-Tris Gels (Thermo-Fisher) and stained using the Silver Quest Staining Kit (Invitrogen).

Kinase assays
For radioactive kinase assays containing kinetochore particles (Figure 2A), kinetochores bound to beads were washed into kinase buffer (50 mM Tris-HCl pH 7.5, 75 mM NaCl, 5% glycerol, 10 mM MgCl 2, 1 mM DTT, 10 μM ATP) containing 66 nM γ-32 P-ATP with either 5 µM reversine (Sigma R3904) or vehicle (DMSO) and incubated at 30 °C for 30 min. Kinetochores were eluted by boiling in sample buffer containing SDS and analyzed via silver stain and Phosphorimager. In addition to the previously detected phosphorylation on Spc105, Ndc80, and Dsn1 (London et al., 2012), we also detected phosphorylation on an additional unidentified protein (marked by an asterisk in Figure 2A), possibly due to altered SDS-PAGE conditions that allowed better resolution of kinetochore proteins. For kinase assays containing purified Ndc80 complex ( Figure 5A), an anti-HA immunoprecipitation was performed and kinase assays were completed identically. Quantification of kinase assays was performed in Fiji (ImageJ) across three independent biological replicates. Intensity values from the Phosphorimager were normalized to Dsn1-His-Flag levels ( Figure 2A) or Ndc80-3HA levels ( Figure 5A) and the mean of the three replicates was used to calculate the change in phosphorylation between conditions.

Phosphatase assay
Native kinetochore particles were purified (see kinetochore purification assay) and kept bound to Dynabeads (Invitrogen

Laser trap instrument
The laser trap has been described previously (Franck et al., 2010). Position sensor response was mapped using the piezo stage to raster-scan a stuck bead through the beam, and trap stiffness was calibrated along the two principle axes using the drag force, equipartition, and power spectrum methods. Force feedback was implemented with custom LabView software. During force measurements, bead-trap separation was sampled at 40 kHz while stage position was updated at 50 Hz to maintain the desired tension (force-clamp assay) or ramp-rate (force-ramp assay). Bead and stage position data were decimated to 200 Hz before storing to disk.

Bead functionalization and slide preparation for laser trap experiments
Native kinetochore particles were linked to beads as previously described (Franck et al., 2010). First, streptavidin-coated polystyrene beads (0.56 μm in diameter, Spherotech Inc., Libertyville IL) were functionalized with biotinylated anti-5His antibodies (Qiagen, Valencia CA) and stored with continuous rotation at 4°C in BRB80 (80 mM PIPES, 1 mM MgCl2, and 1 mM EGTA, pH 6.9) supplemented with 8 mg·mL -1 BSA for up to 3 months.
Immediately prior to each experiment, beads were decorated with kinetochore particles by incubating 6 pM anti-5His beads for 60 min at 4°C with purified kinetochore material, corresponding to Dsn1-His-Flag concentrations ranging between from 2 to 4 nM. The resulting fraction of active beads capable of binding microtubules remained below 50%, thus ensuring single particle conditions (Akiyoshi et al., 2010;Sarangapani et al., 2013;Sarangapani et al., 2014).
Flow chambers (~10 µL volume) were made using glass slides, double-stick tape, and KOH-cleaned coverslips, and then functionalized in the following manner. First, 15 to 20 µL of 10 mg·mL −1 biotinylated BSA (Vector Laboratories, Burlingame CA) was introduced and allowed to bind to the glass surface for 15 min at room temperature. The chamber was then washed with 100 µL of BRB80. Next, 25 µL of 1 mg·mL −1 avidin DN (Vector Laboratories, Burlingame CA) was introduced, incubated for 3 min, and washed out with 100 µL of BRB80. GMPCPP-stabilized biotinylated microtubule seeds were introduced in BRB80, and allowed to bind to the functionalized glass surface for 3 min.
The chamber was then washed with 100 µL of growth buffer (BRB80 containing 1 mM GTP and 1 mg·mL −1 κ-casein). Finally, kinetochore particle-decorated beads were introduced at eight to ten-fold dilution from the incubation mix (see above) in a solution of growth buffer containing 1.5 mg·mL −1 purified bovine brain tubulin and an oxygen scavenging system (1 mM DTT, 500 µg·mL −1 glucose oxidase, 60 µg·mL −1 catalase, and 25 mM glucose).
For the ATP (or mock) exposure experiments (Figures 1, 2, 3  The edges of the flow chamber were sealed to prevent evaporation, and the time was set to 0 min. All laser trap experiments were performed in temperature-controlled rooms, maintained at 23 °C, for up to 90 min after the chamber was sealed.

Rupture force measurements
For efficiency of data collection, beads that were already bound to microtubules (on the lattice, away from the dynamic tip) were usually chosen for measurements of rupture strength. Initially, the attachments were preloaded with a constant tensile force of 1 to 3 pN, which caused the lattice-bound beads to slide until reaching the microtubule plus end. Once at the end, we verified that the beads moved under the preload force at a rate consistent with that of microtubule growth or shortening. The laser trap was subsequently programmed to ramp the force at a constant rate (0.25 pN·s -1 ) until the linkage ruptured, or until the load limit of the trap was reached (~23 pN under the conditions used here).
Fewer than 5% of all trials ended in detachment during the preload period before force ramping began, while 0 to 15% reached the load limit (depending on the conditions tested). These out-of-range events were included in the median force calculations and the survival probability distributions. Under selected conditions, we also tested beads that were floating freely in solution in order to estimate the fraction of active beads, capable of binding microtubules. When beads decorated with wild type kinetochore particles (from SBY8253) were tested in the absence of adenosine, six beads out of thirty tested (20%) bound to microtubules. Similarly, when identically prepared beads were tested in the presence of ADP, five beads out of twenty two (23%) bound microtubules.
However, when the beads were tested in the presence of ATP, only three out of twenty seven tested (11%) bound to microtubules, suggesting that the binding activity of the kinetochore particles (like their rupture strength) is reduced upon exposure to ATP. We found no statistically significant difference in the rupture strengths of pre-bound versus free beads. Most of the measurements (> 98%) were obtained using pre-bound beadmicrotubule pairs.
Supplemental Table S4 summarizes the median rupture force values for each kinetochore type or condition. All the individual rupture force values, the Kaplan-Meier survival probability estimates (with 95% confidence intervals), the numbers of trials that reached the load limit of the trap without rupture, and other statistics (e.g., p-values from log-rank tests), are included in Supplemental Table S5. 49 Supplemental