N-terminal regions of Mps1 kinase determine functional bifurcation

The spindle pole body (SPB), the functional equivalent of the mammalian centrosome, provides microtubule (MT)-organizing functions in budding yeast (Jaspersen and Winey, 2004). The SPB is embedded in the nuclear envelope (NE) in a similar way to the nuclear pore complex (Jaspersen and Winey, 2004). This embedding allows the SPB to simultaneously organize both the cytoplasmic and nuclear MTs that execute essential functions in nuclear positioning and chromosome segregation, respectively.

The yeast SPB duplicates once per cell cycle (Adams and Kilmartin, 2000). SPB duplication is a cell cycle–regulated process that starts in G1 phase of the cell cycle with the extension of an SPB substructure named the half bridge. The satellite then develops on the cytoplasmic face of the NE at the distal end of the elongated half bridge (Adams and Kilmartin, 1999; Kilmartin, 2003). After the start of the cell cycle, the satellite expands into a duplication plaque that becomes inserted into the NE followed by the assembly of the nuclear side of the SPB (Byers and Goetsch, 1975; Winey et al., 1991, 1993; Kilmartin and Goh, 1996; Sundberg et al., 1996; Adams and Kilmartin, 1999).

SPB duplication is regulated through phosphorylation. Cdk1 and Mps1 kinase are mainly responsible for this regulation (Lauzé et al., 1995; Ubersax et al., 2003; Loog and Morgan, 2005). Cdk1 either inhibits or promotes SPB duplication, depending on which phase of the cell cycle it is acting in (Haase et al., 2001). In G1 phase, Cdk1 promotes SPB duplication through phosphorylation of Mps1 and the SPB component Spc42 (Jaspersen et al., 2004). Recently, it was shown that Mps1 kinase phosphorylates the SPB component Spc29 at T240. The T240A mutation of Spc29 leads to instability of the protein (Holinger et al., 2009). It is currently unclear whether this is the only phosphorylation event executed by Mps1 to drive SPB duplication.

Jones et al. (2005) constructed the mps1-as1 mutant, which is specifically inhibited in its kinase activity by the ATP analogue 1NM-PP1. Addition of 1NM-PP1 to mps1-as1 cells in G1 disrupted SPB duplication and the spindle assembly checkpoint (SAC). In contrast, poisoning mps1-as1 cells later, when cells have already duplicated their SPBs, leads to failure of sister kinetochore biorientation (Jones et al., 2005; Maure et al., 2007). Targets of Msp1 at kinetochores are Ndc80 and the DASH complex. Mps1-dependent phosphorylation of Ndc80 is important for SAC activation at kinetochores (Kemmler et al., 2009). Phosphorylation of the DASH complex component Dam1 by Mps1 couples efficient binding of kinetochores to MT plus ends (Shimogawa et al., 2006).

In this study, we used a novel combination of deletion analysis and chemical genetics to study the function of the N-terminal nonkinase domain of Mps1. This analysis identified distinct regions within the N terminus of Mps1 with specific functions in SPB duplication and kinetochore biorientation. Using a novel mps1 mutant allele that was defective in SPB duplication but not kinetochore biorientation, we identified the Mps2–Bbp1–Spc29 SPB membrane insertion machinery (Elliott et al., 1999; Muñoz-Centeno et al., 1999; Schramm et al., 2000) as a critical target of Mps1. An additional substrate of Mps1 at the SPB is the yeast centrin Cdc31 (Baum et al., 1986; Spang et al., 1993). Our data indicate that phosphorylation of Cdc31 regulates its binding to the essential half bridge protein Kar1 (Rose and Fink, 1987; Spang et al., 1995).

As a starting model, we assumed that regions in the N-terminal nonkinase domain of Mps1 are instrumental in conferring target specificity to the kinase. To test this idea, we introduced small deletions of 50 codons within the N- and C-terminal nonkinase domains of MPS1 and transformed the mps1 alleles in the MPS1 shuffle strain YAY406 (Fig. 1 A). Transformants were tested for growth on plates in which the inclusion of 5-fluoroorotic acid (5-FOA) selected against the centromere (CEN)-MPS1 URA3-based plasmid of strain YAY406 so that the mps1 allele became the sole source of Mps1 activity. Some deletion mutants were viable on 5-FOA plates (Fig. 1 B, rows 3, 4, 8, and 9), whereas others did not fulfill essential MPS1 functions (Fig. 1 B, rows 2, 5–7, and 10–12). Most of the nonfunctional mps1 alleles were expressed in yeast, were able to dimerize with wild-type Mps1 (Fig. S1 A), and had kinase activity (Fig. 1 C). The mps1^Δ151–200 (mps1-A), mps1^Δ201–250 (mps1-B), and mps1^Δ251–300 (mps1-C) constructs were particularly interesting because they did not support the growth of yeast cells, even though the specific kinase activity of the encoded mps1-A, -B, and -C proteins was comparable with, or even higher than, that of wild-type Mps1 (Fig. 1 C). This suggests that the mps1-A, -B, and -C proteins are unable to execute at least one essential function of Mps1.

The N-terminal Mps1 deletion constructs may fail to function in vivo because the deleted region enables the kinase to perform an essential function in SPB duplication, kinetochore biorientation, or both. The combination of two deletion constructs may restore viability of cells if both mps1 alleles together fulfill all essential functions of MPS1. Combining mps1 deletion constructs in the yeast MPS1 shuffle strain tested this possibility. mps1-A combined with either mps1-B or -C (all chromosomally integrated) supported cell growth on 5-FOA plates (Fig. 2 A, rows 9 and 10). The combined mps1-A mps1-B or mps1-A mps1-C were expressed similarly (Fig. S1 B) and mps1-A–TAP (tandem affinity purification) coimmunoprecipitated with mps1-B–6HA and mps1-C–6HA (Fig. S1 C), suggesting the formation of mps1-A/-B and mps1-A/-C complexes. Importantly, integration of two copies of mps1-A (Fig. 2 A, row 4), -B (row 5), or -C (row 6) did not allow growth on 5-FOA plates. Thus, it is not the increase in Mps1 kinase activity that restores viability of the combined mps1-A, -B, or -C deletion mutants. Rather, mps1-A mps1-B and mps1-A mps1-C cells probably now survive because of complementation of essential functions: one set being executed by the one allele while the other allele performs another set.

In contrast to mps1-A, -B, and -C, the gene products of mps1^Δ4–50 and mps1^Δ716–764 had reduced specific kinase activity (Fig. 1 C), and two copies of each allele restored viability (Fig. 2 A, rows 3 and 8). This result is consistent with the idea that aa 4–50 and 716–764 are important for full kinase activity. However, our analysis does not exclude the possibility that aa 4–50 and 716–764 of Mps1 may fulfill additional functions such as in localization of the protein.

The defects of the mps1-A, -B, and -C cell alleles were analyzed using a novel strategy. We combined two complementing mps1 deletion alleles, one of which carried an as1 mutation in its kinase domain (Bishop et al., 2000; Jones et al., 2005). 1NM-PP1 competes with ATP for binding to the active site of mps1-as1, whereas the unmodified Mps1 kinase remains unaffected. mps1-A–as1 mps1-B cells were viable under normal growth conditions but failed to grow when the kinase activity of mps1-A–as1 was inhibited by 1NM-PP1. Similar observations were made with other mps1-as1 mps1 combinations (Fig. 2 B).

If our hypothesis of the complementation of essential functions were correct, we would expect distinct phenotypes depending on which of the two mps1 alleles was inhibited. Addition of 1NM-PP1 should inactivate mps1-A–as1 and therefore reveal the phenotype of the mps1-B allele. To test this, we first compared the flow cytometry profiles of α-factor–synchronized mps1-A–as1 mps1-B cells with mps1-A mps1-B–as1 cells after incubation with 1NM-PP1 (Fig. 2 C). In the presence of 1NM-PP1, mps1-A–as1 mps1-B cells replicated the DNA (2C peak after ∼90 min) and then showed strong genetic instability giving rise to >2C and <1C DNA peaks. This phenotype is similar to that of mps1-as1 cells, which have defects in both SPB duplication and SAC function (Winey et al., 1991; Dorer et al., 2005). In contrast, the presence of 1NM-PP1 did not grossly affect the flow cytometry profile of mps1-A mps1-B–as1 cells (Fig. 2 C).

A comparison between synchronized mps1-A–as1 mps1-C cells and mps1-A mps1-C–as1 cells incubated with 1NM-PP1 revealed that the first cell type replicated the DNA and became aneuploid soon after this (120 min; Fig. 2 C), whereas mps1-A mps1-C–as1 cells duplicated the chromosomes (t = 60) followed by only a modest degree of genetic instability (Fig. 2 C). This analysis validates our approach and indicates that deletions within essential regions in the N terminus of MPS1 cause different phenotypes.

We asked whether mps1-as1 mps1 cells are defective in SPB duplication. The addition of 1NM-PP1 to mps1-A–as1 mps1-B or mps1-A–as1 mps1-C cells led to defects in mitotic spindle assembly (Fig. 3, A and B). About 50% of cells had monopolar spindles that were organized by only one SPB (Fig. 3, A and B [i]), whereas the remainder showed a monopolar spindle alongside a dead SPB, as characterized by a red Spc42-eqFP dot (Spc42 marks the central SPB layer; Donaldson and Kilmartin, 1996) that lacked a green nuclear GFP-tubulin signal (Fig. 3, A and B [ii, arrowhead]). The appearance of a dead SPB is characteristic for mutants with defects in duplication plaque insertion. Therefore, in these mutants, the newly formed SPB is unable to organize nuclear MTs (Winey et al., 1991). Importantly, mps1-A–as1 mps1-B–as1 and mps1-A–as1 mps1-C–as1 cells showed the same SPB defect as mps1-as1 cells (Fig. S2 A), suggesting that 1NM-PP1 efficiently blocks kinase activity of combined mps1-as1 mps1 alleles. In contrast, most mps1-A mps1-B–as1 and mps1-A mps1-C–as1 cells duplicated their SPBs normally, as indicated by the formation of a bipolar spindle (Fig. 3, A and B [iii]). This analysis demonstrates that aa 201–300 of Mps1 are required for the kinase to promote SPB duplication, whereas aa 151–200 of Mps1 are not important for SPB duplication.

Next, we analyzed mps1-A mps1-B–as1 and mps1-A mps1-C–as1 cells for kinetochore segregation defects with an assay in which the URA3 locus on chromosome V was marked by binding of Tet repressor–GFP to multiple tandem Tet operators (Michaelis et al., 1997). Whereas >95% of MPS1 anaphase cells achieved successful biorientation of GFP-marked chromosome V (URA3-GFP) in the presence of 1NM-PP1, this number dropped to 78% in mps1-A mps1-B–as1 cells and to 25% in mps1-A mps1-C–as1 cells (Fig. 3 C). In the 22%/75% of monooriented mps1-A mps1-B–as1 or mps1-A mps1-C–as1 cells, respectively, about half of the sister kinetochores were associated with the SPB in the mother cell and the other half with the SPB in the daughter (Fig. 3 D, ii and iii). This random distribution contrasts with mutants of the aurora B complex and mps1-as1 that show preferential association of sister kinetochores with the SPB in the daughter (Tanaka et al., 2002; Maure et al., 2007). One way to explain this difference is the disturbance of SPB inheritance in mps1-A mps1-C–as1 cells (Pereira et al., 2001). However, the newly assembled SPB of mps1-A mps1-C–as1 cells, as indicated by the weaker Spc42-eqFP signal, was still inherited preferentially to the mother cell (Fig. 3 E). Thus, aa 151–200 of Mps1 function in kinetochore biorientation.

mps1-as1 cells that have already duplicated the SPB show defects in kinetochore biorientation upon the subsequent addition of the inhibitor 1NM-PP1 (Maure et al., 2007). Thus, mps1-A–as1 mps1-C cells could be defective in kinetochore biorientation, but the phenotype is being masked by the failure in SPB duplication. To bypass the SPB duplication defect, we incubated α-factor–synchronized mps1-as1, mps1-A–as1 mps1-C, and mps1-A mps1-C–as1 cells with 1NM-PP1 at different time points after the G1 release. The addition of 1NM-PP1 45, 60, or 75 min after release from the G1 block to mps1-A–as1 mps1-C cells allowed formation of bipolar spindles (addition at t = 0 caused mainly monopolar spindles and a dead pole). In the t = 45–75 min cells, most sister chromatids segregated in a bipolar fashion. Cells with anaphase spindles and monopolar orientation of URA3-GFP were rarely observed (Fig. 3 F). In contrast, mps1-as1 and mps1-A mps1-C–as1 cells grown under identical conditions exhibited defects in sister kinetochore biorientation (Fig. 3, F and G). Thus, mps1-A–as1 mps1-C cells are defective in SPB duplication but are fully competent to execute Mps1’s function in kinetochore biorientation.

Finally, we analyzed the role of the N terminus of Mps1 in the SAC. mps1 deletion mutants were incubated with the MT-depolymerizing drug nocodazole to trigger a SAC response. Wild-type MPS1 and SAC-deficient mad2Δ cells were used as controls for cells with a functional and defective SAC, respectively. mps1-A–as1 mps1-B, mps1-A mps1-B–as1, mps1-A–as1 mps1-C, and mps1-A mps1-C–as1 cells were all SAC deficient in the presence of the inhibitor 1NM-PP1 because they behaved as mad2Δ cells in the flow cytometry analysis (Fig. S1 D). A rather large portion of the N terminus of Mps1 is essential for SAC function.

In our strain background, all mps1-as1 alleles were sensitive for growth at elevated temperatures (Fig. S2 B). mps1-as1, mps1-A–as1 mps1-B, and mps1-A–as1 mps1-C cells failed to duplicate their SPBs at 37°C because of the malfunction of the mps1-as1 allele (Fig. S2 C). In contrast, mps1-A mps1-B–as1 or mps1-A mps1-C–as1 cells grown at 37°C without 1NM-PP1 or at 23°C with 1NM-PP1 showed defects in kinetochore biorientation but not SPB duplication (Fig. 3 C, Fig. S2 C, and not depicted). Thus, the phenotypes of mps1-as1 mps1 cells at 23°C with 1NM-PP1 and at 37°C without 1NM-PP1 are similar.

We used the failure of mps1-A–as1 mps1-B and mps1-A–as1 mps1-C cells to grow at 37°C for suppression analysis. mps1-A–as1 mps1-B and mps1-A–as1 mps1-C cells with high gene dosage of MPS2, NBP1, and SPC29 were able to grow efficiently at 37°C (Fig. 4 A and not depicted). SPC110 and BBP1 were weak suppressors of the growth defect. In contrast, MPS2, NBP1, and SPC29 did not suppress the growth defect of mps1-as1 and mps1-A mps1-C–as1 cells at 37°C (unpublished data). These data suggest that Mps1 regulates SPB duplication at the level of the Mps2–Bbp1–Nbp1 complex, which interacts with Spc29. MPS2, BBP1, and NBP1 code for interacting proteins that are essential for the insertion of the newly assembled SPB into the NE (Winey et al., 1991; Elliott et al., 1999; Muñoz-Centeno et al., 1999; Schramm et al., 2000; Araki et al., 2006).

Because SPC29 was the most efficient suppressor of the mps1-A–as1 mps1-C cells (Fig. 4 A), we focused our analysis on the regulation of Spc29 by Mps1. Analysis of α-factor–synchronized SPC29-Flag cells grown at 23°C showed a moderate but reproducible upshift of Spc29 in the SDS-PAGE gel when cells were in G1/S 60 min after the release of the cell cycle block (Fig. 4 B). Incubation of immunoprecipitated Spc29-Flag from synchronized cells of the 60-min time point (Fig. 4 B) with λ phosphatase led to the collapse of the slower-migrating Spc29 species into a single, faster-migrating protein band (Fig. 4 C). Thus, the mobility shift was caused by phosphorylation of the Spc29 protein. Phosphorylation was dependent on Mps1 activity because the reduced mobility of Spc29 was abolished when mps1-as1 cells were incubated with 1NM-PP1 (Fig. 4 D). Together, these data support the notion that Mps1 phosphorylation of Spc29 increases in G1/S during the time of SPB duplication (Lim et al., 1996).

Liquid chromatography (LC)–tandem mass spectrometry (MS; LC-MS/MS) analysis of Spc29 phosphorylated by recombinant Mps1 in vitro identified residues T18, T159, S187, and T240 of Spc29 as Mps1 phosphorylation sites (Fig. 4 E). In vitro phosphorylation of purified Spc29 mutant proteins with altered phosphorylation sites indicated that T18, S187, and T240 are the major amino acids in Spc29 that are modified by Mps1 (Fig. S2 D). Inactivation of all four phosphorylation sites in Spc29 reduced the level of Mps1 phosphorylation to 9% of that conferred upon wild-type Spc29 (spc29-4A; Fig. S2 D).

In vivo phosphorylation of Spc29 on T18, S187, and T240 was recently reported (Chi et al., 2007; Holinger et al., 2009; Holt et al., 2009). Our LC-MS/MS analysis of purified Spc29-TAP from yeast cells confirmed phosphorylation of T18 and S240 in vivo (Fig. 4 E). The spc29-4A mutant protein, which lacks all four Mps1 phosphorylation sites of Spc29, migrated at the same position on the SDS-PAGE gel as Spc29 from mps1-as1 cells incubated with 1NM-PP1 (Fig. 4 D). Thus, Spc29 is phosphorylated by Mps1 in vivo.

We next analyzed whether the phosphorylation cycle of Spc29 is important for its function. If dynamic phosphorylation of Spc29 on these residues were to be crucial for it to execute its essential role, we would expect both the phosphorylation-deficient spc29-4A and the phosphorylation-mimicking spc29-4D mutants to fail to provide SPC29 function. Indeed, the spc29-4A and -4D alleles were unable to support the growth of spc29Δ cells (Fig. S2 E) despite the fact that both proteins were expressed to similar levels as the wild-type protein (Fig. S2 F). These data and Fig. 4 D suggest that Mps1 has, in our strain background, no major role in stabilizing the Spc29 protein. Analysis of spc29-4A in an SPC29 degron strain (spc29-td; Dohmen et al., 1994), which allows efficient degradation of wild-type Spc29 upon shifting cells to 37°C, revealed that spc29-4A cells have the same phenotype as Spc29-depleted cells (Fig. 4 F). Thus, the nonphosphorylated spc29-4A does not show any SPB duplication function.

We also analyzed mutations in single phosphorylation sites of Spc29 (Fig. S2 E). spc29-S187D supported growth of cells as wild-type SPC29, whereas spc29-S187A cells were inviable. In contrast, spc29-T18A, -T18D, -T240A, and -T240D failed to support the growth of cells despite stable expression of mutant proteins (Fig. S2 F). This suggests that the individual Mps1 phosphorylation sites of Spc29 may have distinct functions. In conclusion, the phosphorylation cycle of Spc29 is important for the viability of cells.

To understand the molecular role of Mps1 in SPB duplication, we characterized SPB duplication of mps1-A–as1 mps1-C cells in greater detail. This experiment was performed with G1/S-arrested cells (SIC1 overexpression; Schwob et al., 1994) to avoid diffusion of the difficult to detect dead pole away from the preexisting SPB (Araki et al., 2006). Thin serial section electron microscopic analysis of mps1-A–as1 mps1-C cells without the inhibitor showed two functional, NE-inserted SPBs that were still connected by the bridge structure (Fig. 5 A). This is the SPB phenotype of cells arrested in G1/S (Byers and Goetsch, 1975). mps1-A–as1 mps1-C cells incubated with 1NM-PP1 showed a distorted SPB that had a similar defect to that seen in spc29-3 cells (Fig. 5, B and C; Elliott et al., 1999). Frequently, an SPB duplication intermediate, which was not inserted in the NE, was associated with the larger, NE-embedded SPB structure (Fig. 5 C).

To obtain a molecular understanding of Mps1 in SPB duplication, we analyzed the binding of SPB proteins to the dead pole of mps1-A–as1 mps1-C cells in the presence of 1NM-PP1. This analysis showed that Spc29, Ndc1, and Nbp1 were associated with the defective SPB, whereas the Mps2–Bbp1 complex and Spc110 were mostly not (Fig. 6, A and B). Similar results were obtained when mps1-A–as1 mps1-C cells were shifted to 37°C (Fig. S2 G). The failure of Spc110 to bind to the dead pole is a phenotype exhibited by all SPB insertion mutants because Spc110 only binds to the SPB after the insertion of the duplication plaque into the NE (Kilmartin and Goh, 1996). The absence of the Mps2–Bbp1 complex from the dead pole was unexpected and may indicate that Mps1 facilitates the recruitment of the Mps2–Bbp1 complex to the new SPB.

Previously, we have shown that Bbp1 interacts with Spc29 (Schramm et al., 2000). To confirm a role for Mps1 in the regulation of the interaction between Spc29 and Bbp1, we asked whether the spc29-4D allele was able to recruit Bbp1 to the newly formed SPB of mps1-A–as1 mps1-C cells incubated at 37°C. High gene dosage of SPC29 or spc29-4D was partly able to recruit Bbp1 to the dead pole (Fig. 6 C), although they were not able to suppress the growth defect of mps1-A–as1 mps1-C cells (not depicted). In this assay, spc29-4D was clearly more efficient than SPC29 in bringing Bbp1 to the dead pole, suggesting that the phosphorylation-mimicking spc29-4D allele is able to bypass the requirement for Mps1 in the recruitment of Bbp1 to the SPB during the duplication process. In contrast, spc29-4A did not significantly increase binding of Bbp1-GFP to the newly formed SPB of mps1-A–as1 mps1-C cells. These data suggest that phosphorylation of Spc29 by Mps1 facilitates recruitment of the Mps2–Bbp1 complex to the newly formed SPB.

In vitro experiments tested whether Mps1 regulates the binding of Spc29 to Bbp1. We incubated purified GST-Spc29 with either 6His-Mps1WT or the kinase-dead 6His-msp1KD in the presence of ATP, after which the kinase molecules were efficiently removed from the SPB component by the addition of Ni–nitrilotriacetic acid beads (Fig. 6 D, bottom). Phosphorylated GST-Spc29 showed a moderate gel shift compared with nonphosphorylated GST-Spc29 (Fig. 6 D, top; compare lane 2 with lane 3). In the next step, equal quantities of GST-Spc29 (Fig. 6 D, lanes 2 and 3) and GST (lane 1) were incubated with Bbp1-TAP (lanes 4, 6, and 8) or TAP (lanes 5 and 7). The levels of GST-Spc29 (Fig. 6 D, lanes 5–8) and GST (lane 4) bound to the Bbp1-TAP or TAP beads were determined by quantitative immunoblotting. Neither GST nor GST-Spc29 bound to Bbp1-TAP (Fig. 6 D, lane 4) or TAP beads, respectively (lanes 5 and 7). Importantly, phosphorylation of Spc29 by Mps1 enhanced binding of Spc29 to Bbp1 fivefold (Fig. 6, D [compare lane 6 with lane 8] and E). Thus, binding of Spc29 to Bbp1 is stimulated by Mps1 kinase activity.

Analysis of mps1-as1 cells revealed distinct SPB phenotypes depending on when in the cell cycle the inhibitor 1NM-PP1 was added. The addition of 1NM-PP1 to synchronized G1-phase cells (α-factor release; satellite is already formed; Byers and Goetsch, 1975) caused SPB duplication defects that were associated with the occurrence of a dead pole (Fig. 3 A). This phenotype is consistent with the function of Mps1 in the insertion of the duplication plaque into the NE (Schutz and Winey, 1998). However, when 1NM-PP1 was added to synchronized mps1-as1 cells in S/M phase (cells have already duplicated the SPB), nearly 100% of cells formed a proper anaphase spindle (Fig. 7 A, 120 min) and then exited mitosis into the second cell cycle. In the second cell cycle, the SPB failed to duplicate, as indicated by a single Spc42-eqFP SPB signal; however, a dead pole was rarely observed (Fig. 7 A, 240 min). This suggests that cells progressing from mitosis in G1 without Mps1 activity fail SPB duplication before duplication plaque formation probably on the level of satellite assembly (Schutz and Winey, 1998).

In contrast to mps1-as1 cells, when 1NM-PP1 was added to mps1-A–as1 mps1-C cells in S/M phase, a dead pole was observed in the second cell cycle in at least 30% of cells (Fig. 7 A). Thus, mps1-A–as1 mps1-C cells are specifically defective in duplication plaque insertion and not in the initial events of SPB duplication. The multiple defects of mps1-as1 cells in SPB duplication suggest that there are further Mps1 targets in addition to Spc29.

Purification of a stabilized Mps1 in a proteasome mutant was used to identify additional Mps1 substrates. LC-MS/MS analysis of the proteins in complex with Mps1-TAP revealed components of the Ndc80 complex, the kinetochore proteins Spc105 and Mtw1 (Westermann et al., 2007), and the yeast centrin Cdc31 (Fig. 7 B), which functions early in SPB duplication in the formation of the satellite (Baum et al., 1986).

Cdc31 is an in vitro substrate of Mps1 (Fig. S3 A, second lane). LC-MS/MS analysis of in vitro phosphorylated Cdc31 identified Mps1 phosphorylation sites in Cdc31 at T110 and T128 (Fig. 7 C and Fig. S3 B). T110 is the major phosphorylation site because mutation to alanine radically reduced the ability of Mps1 to phosphorylate Cdc31 in vitro (Fig. S3 A). LC-MS/MS analysis of the functional Z-tagged Cdc31 purified from yeast cells (Kilmartin, 2003) confirmed phosphorylation of T110 in vivo (Fig. S3 C).

To understand the relevance of the Cdc31 phosphorylation, we assessed the phenotypes of cdc31-T110 and -T128 phosphorylation mutants. cdc31-T110A, -T110D, -T128A, and -T128D supported the essential function of CDC31 (Figs. S3 D). However, cdc31-T110A cells were sensitive for growth at evaluated temperatures (Fig. 7 D). In fact, cdc31-T110A cells failed to duplicate their SPB at 37°C, as indicated by the arrest as large-budded cells with a single SPB and a monopolar spindle (Fig. 7 E). As expected from the early function of Cdc31 in SPB duplication (Baum et al., 1986), no dead pole was observed in cdc31-T110A cells. These data suggest that the Mps1 phosphorylation cycle on T110 of Cdc31 is important for SPB duplication.

To understand the nature of the defect conferred by blocking T110 phosphorylation, we tested genes coding for SPB components for their ability to suppress the growth defect cdc31-T110A cells at 33/35°C. KAR1 on a CEN-based plasmid was able to confer partial suppression of the growth defect of cdc31-T110A cells (Fig. 7 F). Note that high gene dosage of KAR1 is lethal for cells (Rose and Fink, 1987). Other genes involved in SPB duplication (e.g., BBP1, NBP1, SPC29, SFI1, and MPS3) were unable to suppress the cdc31-T110 growth defect even when present in high gene dosage (unpublished data). Thus, T110 of Cdc31 is important for the functional interaction with the Kar1 protein.

As T110 is the major Mps1 phosphorylation site in Cdc31 (Fig. S3 A), it may be important for the association of Cdc31 with the SPB component Kar1 (Fig. 7 F). We addressed this directly by comparing the ability of nonphosphorylated Cdc31 and in vitro phosphorylated Cdc31 (Fig. 8 A) to bind a peptide that constitutes the Cdc31-binding site in Kar1 (Geier et al., 1996). The affinity of phosphorylated Cdc31 toward Kar1 peptide was fivefold lower than the ability of nonphosphorylated Cdc31 (Fig. 8 B). Consistently, the ability of Cdc31-T110D to bind the Kar1 peptide was ∼10-fold lower than that of wild-type Cdc31 (unpublished data).

Mps1 is a conserved protein kinase that functions from yeast to human cells in centrosome/SPB duplication, kinetochore biorientation, and the SAC (Winey et al., 1991; Hardwick et al., 1996; Fisk and Winey, 2001; Espeut et al., 2008; Jelluma et al., 2008; Tighe et al., 2008). In this study, we have used a novel combination of deletion analysis and chemical genetics to dissect the role of the noncatalytic regions of Mps1 kinase. We found that distinct regions within the noncatalytic N-terminal domain of Mps1 have specific functions either in SPB duplication or kinetochore biorientation. Our defined N-terminal deletion mutants differ from previously characterized mps1(ts) conditional-lethal alleles, which either carried single amino acid substitutions in the kinase domain (e.g., mps1–737) or 10 aa substitutions dispersed along the entire N-terminal nonkinase domain (mps1–8; Schutz and Winey, 1998; Castillo et al., 2002).

The most interesting deletion alleles were mps1-A, -B, and -C (Fig. 1 A). They did not support growth of yeast cells despite having specific kinase activity that was similar to wild-type Mps1. However, combinations of mps1-A mps1-B and mps1-A mps1-C restored viability, whereas cells with two copies of either mutant could not. These data suggest intragenic complementation of essential functions by the combined mps1-A mps1-B or mps1-A mps1-C alleles. To confirm this notion, we inactivated one of the two alleles through the as1 mutation, which allows specific inhibition of the kinase by the addition of the ATP analogue 1NM-PP1 (Bishop et al., 2000). mps1-A–as1 mps1-C and mps1-A mps1-C–as1 cells showed strikingly different phenotypes. The addition of 1NM-PP1 to mps1-A–as1 mps1-C cells caused lethality with defects in SPB duplication, although kinetochore biorientation was not affected. In contrast, mps1-A mps1-C–as1 cells duplicated their SPB as wild-type cells but then had defects in kinetochore biorientation.

In the presence of the inhibitor 1NM-PP1, a large percentage of mps1-as1 cells show monoorientation of sister kinetochores. In these mps1-as1 cells, sister kinetochores bind preferentially to the old spindle pole that moves into the daughter cell (Maure et al., 2007). In contrast, in mps1-A mps1-C–as1 cells, we did not observe preferential binding of sister kinetochores to one of the two SPBs. This was not because of a defect in SPB inheritance (Pereira et al., 2001). The reason for this difference is unclear, but it is possible that Mps1 has multiple functions during kinetochore biorientation and that only one is disrupted in mps1-A mps1-C–as1 cells. Similar data were obtained for mps1-A–as1 mps1-B and mps1-A mps1-B–as1 cells. Thus, aa 151–200 of Mps1 are important for kinetochore biorientation but not for SPB duplication. In contrast, aa 201–300 of Mps1 are specifically important for SPB duplication but not biorientation of kinetochores. Our findings emphasize the importance of the N-terminal domain of Mps1 for functional selectivity.

We identified SPC29, NBP1, and MPS2 as efficient high dosage suppressors of the lethality of mps1-A–as1 mps1-C cells. The growth defect of the pleiotropic mps1-as1 cells or the kinetochore-defective mps1-A mps1-C–as1 cells were not suppressed by SPC29, NBP1, and MPS2, indicating allele specificity for this suppression. SPC29 encodes an essential component of the SPB that functions together with the Mps2–Bbp1–Nbp1 complex in the insertion of the newly formed SPB into the NE. Spc29 interacts via the Bbp1 protein with the integral membrane protein Mps2 (Elliott et al., 1999; Muñoz-Centeno et al., 1999; Schramm et al., 2000).

Several lines of evidences indicate that phosphorylation of Spc29 by Mps1 controls an essential step in SPB duplication and is not just important for Spc29 stabilization (Holinger et al., 2009). First, spc29-4D or -4A was expressed similarly as SPC29 in yeast cells, and inhibition of mps1-as1 in our strain background did not decrease the levels of Spc29 (Fig. 4 D and Fig. S2 F). Second, conditional-lethal spc29-3 cells have a similar SPB duplication defect to that found in mps1-A–as1 mps1-C cells (Elliott et al., 1999). Third, Mps1 phosphorylates Spc29 in G1/S when the newly formed SPB becomes inserted into the NE. Fourth, the newly formed SPB of mps1-A–as1 mps1-C cells that is not inserted into the NE contains Spc29, Nbp1, and the integral membrane protein Ndc1 but is devoid of the Bbp1–Mps2 complex. This may indicate that phosphorylation of Spc29 by Mps1 on four sites recruits the Bbp1–Mps2 complex to the newly formed SPB to facilitate its insertion into the NE. Fifth, expression of the phosphorylation-mimicking spc29-4D in mps1-A–as1 mps1-C cells restored the recruitment of Bbp1 to the newly formed SPB. Finally, in vitro experiments with purified GST-Spc29 and Bbp1-TAP showed that the interaction between these proteins relied on the phosphorylation of Spc29 by Mps1 (Fig. 6 D).

Thus, what is the specific function of aa 201–300 of Mps1 in SPB duplication? Purified mps1-C still bound to Spc29 (Fig. S4 A), and its specific kinase activity toward Spc29 was as wild-type Mps1 or mps1-A (Fig. S4 B). It is challenging to determine the subcellular localization of Mps1 proteins because of low expression and the close vicinity of the SPB to clustered kinetochores (Jin et al., 2000). This makes it difficult to exclude that mps1-C is no longer associated with SPBs. However, considering that the function of Cdc31 in SPB duplication and proteins in kinetochore biorientation is maintained in mps1-C cells, we suggest that aa 201–300 of Mps1 are specifically required to regulate Spc29 in the context of the SPB.

Mps1 has multiple functions in SPB duplication. The phenotype of the mps1–8 allele suggests that Mps1 is also important early in SPB duplication (Castillo et al., 2002) along with the half bridge proteins Cdc31, Mps3, Kar1, and Sfi1 (Rose and Fink, 1987; Spang et al., 1993, 1995; Biggins and Rose, 1994; Adams and Kilmartin, 1999; Jaspersen et al., 2002; Kilmartin, 2003). Consistently, the addition of 1NM-PP1 to mps1-as1 cells in metaphase leads to an SPB duplication defect that is distinct from that seen when the inhibitor is added in late G1 (Fig. 7 A). Purification of Mps1 from yeast cells identified yeast centrin Cdc31 as an interactor of Mps1. Mps1 phosphorylates Cdc31 at T128 and the conserved T110 residue in the third EF hand. Blocking phosphorylation of the major Cdc31 phosphorylation site of Mps1 by mutating it to alanine (cdc31-T110A) caused a conditional-lethal growth defect in SPB duplication that was suppressed by a single additional gene copy of KAR1. Kar1 is an essential half bridge component that interacts with Cdc31 (Biggins and Rose, 1994; Vallen et al., 1994; Spang et al., 1995; Geier et al., 1996). Our results are consistent with a previous study (Ivanovska and Rose, 2001), in which the cdc31-57, which has a T110I exchange, displayed synthetic lethality with the kar1-Δ17 allele.

Phosphorylation of Cdc31 by Mps1 caused a fivefold decrease in the affinity of Cdc31 for a peptide consisting of the Cdc31 docking site on Kar1. Thus, phosphorylation of Cdc31 by Mps1 may regulate the cell cycle–specific physical association of Cdc31 with Kar1. It will be interesting to see whether mammalian Mps1 regulates centrin-2 through phosphorylation and how this impacts upon centrosome duplication (Fisk and Winey, 2001; Salisbury et al., 2002).

Duplication of the mammalian centrioles and the yeast SPB can be envisioned as self-duplication events that are controlled at different steps alongside the duplication process (Fig. 8 C). One of the first steps in SPB duplication is the formation of the satellite in early G1 phase before the start of the cell cycle. Satellite formation requires the activities of Mps1 kinase and, in addition, the half bridge components Cdc31, Kar1, Mps3, and Sfi1. Our data now suggest that the regulation of the Cdc31–Kar1 interaction by Mps1 is critical in satellite formation. Whether phosphorylation of Cdc31 by Mps1 also regulates the interaction with the half bridge components Sfi1 and Mps3 needs to be investigated (Jaspersen et al., 2002; Kilmartin, 2003; Li et al., 2006).

Phosphorylation of the satellite component Spc42 by Cdk1 after the start of the cell cycle may allow extension of the satellite into a duplication plaque, which is layered on top of the cytoplasmic face of the NE (Donaldson and Kilmartin, 1996; Adams and Kilmartin, 1999; Jaspersen et al., 2004). Spc29 that interacts with Spc42 is already associated with the SPB satellite in early G1 phase of cell cycle, and it probably copolymerizes with Spc42 during satellite extension (Adams and Kilmartin, 1999; Elliott et al., 1999). In this study, we showed that the duplication plaque first associates with a subset of the insertion machinery: Ndc1, Nbp1, and Spc29. Next, Mps1 kinase phosphorylates Spc29 at several sites, leading to the recruitment of Bbp1 probably in complex with Mps2 (Fig. 8 C). This step triggers the insertion of the SPB into the NE with the concomitant binding of the SPB component Spc110 from within the nucleus (Kilmartin and Goh, 1996; Sundberg et al., 1996; Elliott et al., 1999). When taken together, our data suggest that Mps1 kinase regulates critical steps in SPB duplication through controlling the affinity of structural components of the duplication machinery. We propose a similar function for kinases such as Cdk2, mMps1, and Plk4 in mammalian centrosome duplication (Fisk and Winey, 2001; Bettencourt-Dias and Glover, 2007).

Strains are listed in Table S1. Yeast strains were derivatives of YPH499 (Sikorski and Hieter, 1989) unless otherwise indicated. Gene deletions and epitope tagging of genes at their endogenous loci were constructed by PCR-based methods (Janke et al., 2004). Mutations or internal deletions in MPS1, SPC29, or CDC31 genes were introduced by the QuikChange site-directed mutagenesis kit (Agilent Technologies). For synchronization, yeast cells were grown in YPDA (yeast extract, peptone, dextrose, and adenine) medium and arrested in G1 by treatment with 10 µg/ml α-factor at 23°C until >95% of cells showed a mating projection. Cells were washed twice with growth medium to remove α-factor. Cells were resuspended in YPDA medium at the indicated temperatures. Synthetic complete medium was used for live cell imaging experiments. 1NM-PP1 was purchased from Merck Biosciences. The mps1-as1 allele was as described previously (Jones et al., 2005).

Cells with chromosomal gene fusions with the fluorophores eqFP611 (Wiedenmann et al., 2002) and GFP (Janke et al., 2004) were analyzed by fluorescence microscopy without fixation. Series of z-focal planes images were collected on a microscope equipped with a 100× NA 1.45 Plan-Fluar oil immersion objective (Carl Zeiss, Inc.), a camera (Cascade; 1K; Photometrics), and MetaMorph software (Universal Imaging Corp.). Images in different z planes were projected and processed in Photoshop (Adobe). No manipulations other than contrast and brightness adjustments were used.

The DNA content of 20,000 mid–log phase cells stained with propidium iodide was determined by flow cytometry (FACScan flow cytometer; BD) using CELL QUEST software (BD; Schramm et al., 2000).

Yeast cells processed for EM were high-pressure frozen as described previously (Ding et al., 1993). In brief, cells from mid–log phase cycling culture were harvested by vacuum filtration on 0.45-mm filters (Millipore). The yeast paste was frozen using a high-pressure freezer (HPM-010; ABRA Fluid). Freeze substitution of frozen cells was performed in a freeze substitution device (EM-AFS1; Leica) for 2 d in 0.1% glutaraldehyde, 0.25% uranyl acetate, and 0.01% OsO₄ in acetone at −90°C (Müller-Reichert et al., 2003). The freeze-substituted cells were further embedded in Lowicryl resin (HM20) at −45°C. Thin sections (50–70 nm) were cut by ultramicrotomy, contrasted with 2% uranyl acetate and Reynold’s lead citrate, and viewed in an electron microscope (CM120-Biotwin; Philips) operating at 100 kV. Digital acquisitions were made with a charge-coupled device camera (Keen View; Soft Imaging System).

GST- and 6His-fused Mps1 proteins were expressed in Escherichia coli Rosetta (DE3) by adding 0.25 mM IPTG for 3 h at 25°C. The recombinant proteins were eluted either with PreScission Protease (GE Healthcare) for GST-Mps1 or with imidazole for 6His-Mps1. Mps1-TAP proteins were prepared from yeast cells. In brief, MPS1-TAP cells were resuspended in lysis buffer (50 mM Tris-Cl, pH 8, 400 mM NaCl, 10% glycerol, 1 mM DTT, 1% Triton X-100, 1 mM PMSF, and complete protease inhibitor cocktail, EDTA free [Roche]) and were lysed with a glass bead homogenizer. The cleared cell extract was then incubated with IgG-Dynabeads (Dynal; Invitrogen) for 1 h at 4°C. After intensive washing steps of the IgG Dynabeads, the bound Mps1 was used for the in vitro phosphorylation experiment. GST-Spc29 and GST recombinant proteins were expressed and purified as described previously (Elliott et al., 1999). Purified GST-tagged recombinant proteins were phosphorylated in vitro by the use of Mps1 kinase purified from E. coli. The phosphorylation reactions contained 50 mM Tris-Cl, pH 7.5, 75 mM NaCl, 5% glycerol, 10 mM MgCl₂, 1 mM DTT, 10 µM ATP, and 10 µCi γ-[³²P]ATP. Reactions were incubated at 30°C for 30 min. Coomassie brilliant blue (CBB)–stained bands were quantified by the Odyssey system (LI-COR Biosciences). ³²P-labeled proteins were detected with a PhosphorImager (FLA-300; Fujifilm) and quantified with Image Gauge software (Fujifilm). For the analysis of phosphorylation sites by MS, GST-Spc29 and -Cdc31 were mixed with recombinant Mps1 from E. coli as described above in this section but in the presence of 10 mM ATP and incubated at 30°C for 1.5 h. In the experiment in Fig. 1 C, Mps1-TAP and mps1KD-TAP proteins bound to IgG-Dynabeads were incubated at 30°C for 30 min with myelin basic protein (MBP; Invitrogen) as described above in this section.

Cells were resuspended in lysis buffer (50 mM Tris-Cl, pH 8, 400 mM NaCl, 10% glycerol, 1 mM DTT, 1% Triton X-100, 1 mM PMSF, complete protease inhibitor cocktail, EDTA free, 5 mM EDTA, 50 mM NaF, 2 mM Na₃VO₄, 10 mM Na₄P₂O₇, and 60 mM β-glycerophosphate) and were lysed as described in the previous section. Spc29-Flag was immunoprecipitated using anti-Flag antibody M2 (Roche) conjugated to protein G–Dynabeads. Reactions both with and without λ phosphatase (New England Biolabs, Inc.) with and without phosphatase inhibitors (5 mM EDTA, 50 mM NaF, 2 mM Na₃VO₄, 10 mM Na₄P₂O₇, and 60 mM β-glycerophosphate) were set up and incubated at 30°C for 30 min. Before removal from the beads, the beads were washed once with lysis buffer. SDS loading buffer was then added to each sample, and these were incubated at 95°C to elute the bound proteins.

Proteins present in the gel lane were visualized with colloidal Coomassie staining. Bands were cut out with a scalpel. Gel slices were reduced, alkylated, and digested with trypsin using a Digest pro MS liquid handling system (Intavis AG). The sample was analyzed by a nano–high-performance LC system (Dionex) coupled to a mass spectrometer (ESI LTQ Orbitrap; Thermo Fisher Scientific). The sample was loaded on a C18 trapping column (Inertsil; LC Packings), and peptides were eluted and separated on an analytical column (75 µm × 150 mm) packed with Inertsil 3-µm C18 material (LC Packings). The column was connected to a nano–electrospray ionization emitter (New Objectives). 1,500 V were applied via liquid junction. One survey scan (resolution, 60,000) was followed by five information-dependent product ion scans in the LTQ. Only doubly and triply charged ions were selected for fragmentation. Tandem mass spectra were extracted by Mascot Distiller and grouped within a precursor m/z tolerance of 0.03 amu and with five intermediate scans at maximum. All MS/MS samples were analyzed using Mascot (Matrix Science). Mascot was set up to search the SwissProt database assuming the digestion enzyme trypsin. Mascot was searched with a fragment ion mass tolerance of 0.20 D and a parent ion tolerance of 4.0 ppm. Scaffold (Proteome Software) was used to validate MS/MS-based peptide identifications. Peptide identifications were accepted if they could be established at >95.0% probability as specified by the Peptide Prophet algorithm (Keller et al., 2002). The peptide sequences of the phosphorylated peptides were additionally confirmed by manual evaluation of the fragment spectra.

Purified GST-Spc29 proteins were phosphorylated in vitro as described in In vitro kinase assays. In the experiment in Fig. 6 D, purified 6His-Mps1WT wild type or the kinase-dead mutant (6His-mps1KD) was incubated with Ni-agarose to remove 6His-Mps1. The treated proteins were mixed with TAP or Bbp1-TAP–bound beads for 1 h at 4°C. The beads were washed three times with washing buffer (50 mM Tris-Cl, pH 8, 1 mM DTT, 150 mM NaCl, 10% glycerol, and 0.2% Triton X-100). 10% of bound fraction was used for an immunoblot with a GST antibody (GE Healthcare) to detected GST-Spc29. GST proteins were quantified with ImageJ (National Institutes of Health). The efficiency of Spc29 binding to Bbp1 was quantified and calibrated as follows: the intensity of the band in Fig. 6 D (lane 4) was subtracted from GST-Spc29 intensities, and the outcome was divided by the intensities of the Bbp1-TAP and GST-Spc29 inputs. To compare results from multiple experiments, data were normalized to the wild-type Mps1 kinase reaction.

Binding of nonphosphorylated and phosphorylated Cdc31 to Kar1 peptide was measured as follows (Geier et al., 1996). Mps1-TAP proteins were prepared from yeast cells with 2 µm–based plasmids carrying pGal1-MPS1-TAP. Purified Cdc31 at a concentration of 15 µg/µl was treated with either wild-type Mps1WT-TAP or kinase-dead mps1KD-TAP bound to IgG-Dynabeads for 2 h at 30°C under the same buffer conditions as in the in vitro kinase assay described in In vitro kinase assays. The efficiency of phosphorylation was checked by running 2 µg Cdc31 on a 15% native polyacrylamide gel using the Laemmli buffer system without SDS and staining with SimplyBlue SafeStain (Invitrogen). Fluorescence measurements were performed at 20°C on a spectrofluorometer (FP-6500; Jasco). The wavelength for excitation was 295 nm. Emission spectra were recorded over the range of 300–400 nm. 0.5-µl aliquots of appropriate dilutions of nonphosphorylated or phosphorylated Cdc31 were added to 300 µl of a 1-µM solution of the Kar1 peptide (KKRELIESKWHRLLFHDKK) in Ca²⁺ buffers (Invitrogen) with 225, 602, or 1350 nM free [Ca²⁺], 30 mM MOPS, and 100 mM KCl, pH 7.3. The emission spectrum of the Kar1 peptide showed a maximum at 350 nm. Binding of Cdc31 resulted in a blue shift of the fluorescence emission maximum to 320 nm accompanied by an increase in fluorescence intensity. Dissociation constants (K_d) were determined by fitting the fluorescence intensity at 320 nm to a single site-specific binding model using the program Prism (GraphPad Software, Inc.). Mean values of three titrations were taken.

Fig. S1 shows the function of subdomains of Mps1. Fig. S2 shows growth phenotypes of mps1-as1 cells and phosphorylation of Spc29 by Mps1. Fig. S3 shows identification of the Mps1 phosphorylation sites in Cdc31 and functionality of cdc31 phosphorylation mutants. Fig. S4 shows biochemical analyses of Mps1 mutant proteins. Table S1 lists the yeast strains used in this study.

We thank G. Pereira and I. Hagan for comments on the manuscript, T. Ruppert for the MS analysis, and A. Neuner for the EM. Z-tagged Cdc31 was a gift from J. Kilmartin.

The project was supported by Sonderforschungsbereich 638.