Uncoupling of the spindle-checkpoint and chromosome-congression functions of BubR1

Equal distribution of duplicated chromosomes during cell division is essential for maintaining genome integrity. During mitosis, accurate chromosome segregation is dependent on productive interactions between microtubules (MTs) and kinetochores (KTs), large proteinaceous structures that assemble onto centromeric DNA. Most errors in chromosome attachment activate the spindle assembly checkpoint (SAC), a surveillance mechanism that delays the onset of anaphase until chromosomes have attached to the mitotic spindle in a bipolar fashion. A major target of the SAC is Cdc20, a substrate-binding subunit of the APC/C (anaphase promoting complex/cyclosome). The APC/C is a large E3 ubiquitin ligase that initiates the ubiquitylation and subsequent degradation of key mitotic proteins, particularly cyclin B and securin (Musacchio and Salmon, 2007). Several critical components of this checkpoint were originally identified in budding yeast and include Mad1, Mad2, Mad3 (BubR1 in animals), Bub1, Bub3 and Mps1 (Musacchio and Salmon, 2007). These checkpoint proteins localize dynamically to unattached KTs. Elegant studies from several groups have collectively resulted in a model, according to which the KT provides an interface for the catalytic conversion of inactive, open-Mad2 to a closed-Mad2 capable of binding to Cdc20 and triggering the recruitment of BubR1-Bub3 into an APC/C inhibitory complex (Kulukian et al., 2009; Musacchio and Salmon, 2007; Yu, 2002). Although the exact composition of the resulting mitotic checkpoint complex (MCC) remains controversial, this complex is thought to shuttle off the KT as a diffusible `stop anaphase' signal. This model is supported by the observation that the association between BubR1 and Cdc20 is dependent on Mad2 (Hardwick et al., 2000; Hwang et al., 1998; Nilsson et al., 2008), and it explains why both Mad2 and BubR1 are required for APC/C^Cdc20 inhibition in vivo, although either protein is sufficient in vitro (Fang, 2002).

SAC signaling appears to be more complex in animals than in yeast. Notably, BubR1 harbors a C-terminal Bub1-like kinase domain, and a second C-terminal Cdc20 binding region, both of which are lacking in the putative yeast ortholog Mad3 (Davenport et al., 2006; Harris et al., 2005; Tang et al., 2001). In addition to direct inhibition of APC/C^Cdc20, BubR1 contributes to the control of mitotic timing and the accurate chromosome capture by spindle microtubules (Ditchfield et al., 2003; Lampson and Kapoor, 2005). BubR1 is highly phosphorylated in mitosis, and we and others have recently shown that phosphorylation by Plk1 is important for maintaining proper kinetochore-microtubule (KT-MT) attachments and for timely mitotic progression (Elowe et al., 2007; Matsumura et al., 2007). More recently, additional (non-Plk1) phosphorylation sites have been identified, mutation of which also delayed mitotic exit (Huang et al., 2008). However, the identity of the kinase(s) responsible for phosphorylation of these sites remains unclear, although Mps1 might be involved.

Insight into how BubR1 mediates APC/C^Cdc20 inhibition came initially from studies in yeast, where it was demonstrated that two KEN boxes in Saccharomyces cerevisiae Mad3p are required for the SAC. These KEN motifs, which often mediate substrate recognition, are thought to inhibit the APC/C^Cdc20 by competing with bona fide Cdc20 substrates (Burton and Solomon, 2007; King et al., 2007). The conservation of Mad3/BubR1 KEN motifs across evolution suggests that a similar mechanism may function in higher eukaryotes. Indeed, recent evidence suggested that N-terminal fragments of mouse BubR1 carrying both KEN boxes are sufficient for cell survival, whereas mutation of either KEN box led to loss of Cdc20 binding and cell viability (Malureanu et al., 2009).

Here, we examine the contribution of conserved motifs to the mitotic functions of full-length human BubR1. We identify several conserved phosphorylation sites in BubR1 and demonstrate that Cdk1 (CDK1) rather than Mps1 (also known as TTK) phosphorylation contributes significantly to KT-MT attachments, but not to MCC formation or other SAC functions. Conversely, cells expressing BubR1 KEN-box mutants exit mitosis prematurely and are unable to arrest in response to nocodazole treatment. Such cells are nevertheless capable of efficiently aligning their chromosome complement, provided that they are maintained in mitosis by MG132 treatment. Furthermore, we demonstrate that efficient Cdc20 and APC/C binding requires full-length BubR1, as N-terminal fragments of BubR1 bound poorly to both Cdc20 and the APC/C. These data suggest that the chromosome congression and checkpoint functions of BubR1 can be uncoupled from each other. Interestingly, a BubR1 GLEBS domain mutant, which is unable to bind Bub3 and shows reduced accumulation at the KT, no longer becomes detectably phosphorylated. Strikingly, this mutant exhibits both congression and checkpoint defects, highlighting the importance of the BubR1 GLEBS domain for its mitotic functions.

It was recently shown that phosphorylation of BubR1 by Plk1 is required for stable KT-MT attachments (Elowe et al., 2007; Matsumura et al., 2007). Nevertheless, Plk1 phosphorylation cannot alone account for the congression defects observed in BubR1-depleted cells (Lampson and Kapoor, 2005). Indeed, in addition to Plk1, Cdk1 is able to efficiently phosphorylate BubR1 in vitro (Elowe et al., 2007). We therefore sought to identify novel phosphorylation sites on BubR1 that are specifically upregulated in mitosis. Using mass spectrometry, we identified mitotic-specific phosphorylation at the serine-proline (SP) motifs around S543, S574, S670, S720 and S1043 (Fig. 1A and supplementary material Fig. S1A), as well as 12 other residues. Using the stable isotope labeling with amino acids in cell culture (SILAC) approach, we were able to quantify the extent of mitotic-specific phosphorylation at four of these sites (supplementary material Fig. S1B), and show that phosphorylation is upregulated during nocodazole arrest (Fig. 1A). S574 was identified in a non-tryptic peptide; as it contained neither arginine nor lysine it could not be directly quantified by the SILAC procedure used here. Independently, Huang et al. recently reported the identification of S543, S670, and S1043 as mitotic-specific phosphorylation sites on BubR1 (Huang et al., 2008), but S720 and S574 are novel sites.

To test whether the phosphorylation sites identified here contribute to the characteristic electrophoretic upshift of mitotic BubR1, MYC-tagged constructs for BubR1-WT, BubR1-620A (Polo-box domain binding mutant), BubR1-5A, and the phosphomimetic 5D (where all five phosphorylation sites were mutated to aspartate) were expressed in HeLa cells. The S620A mutant was included in this analysis, as we have previously shown that Plk1 phosphorylation regulates the characteristic mitotic upshift in BubR1 (Elowe et al., 2007). Cells were then harvested after a 16-hour release from thymidine arrest into nocodazole, before proteins were analyzed by western blotting. BubR1-WT exhibited the characteristic double band pattern, whereas BubR1-620A, as predicted, migrated as a single faster migrating species (Fig. 1B). Both BubR1-5A and BubR1-5D also exhibited a double band pattern, suggesting that mutation of these five residues is not sufficient to eliminate the mitotic BubR1 upshift (Fig. 1B).

Interestingly, the five sites described here consist of a serine followed by a proline residue, suggesting that a proline-directed kinase such as Cdk1 may phosphorylate these sites (Fig. 1C), although phosphorylation at several of these has been suggested to be Mps1-dependent (Huang et al., 2008). To compare BubR1 phosphorylation by Cdk1 and Mps1, we performed in vitro kinase assays using recombinant full-length BubR1 as a substrate (Fig. 1D). Whereas Cdk1 was able to efficiently phosphorylate BubR1, there was no detectable phosphorylation of BubR1 by Mps1 kinase at the same specific activity (Fig. 1D, left panels), although both Cdk1 and Mps1 efficiently phosphorylated Borealin under the same conditions (Fig. 1D, right panels). To clarify whether the SP sites identified here are indeed Cdk1 target sites, we generated 12-mer peptides centered on each of the five serines and used these peptides as in vitro substrates for Cdk1. Corresponding control peptides were synthesized with the serine phospho-acceptor positions changed to alanine to determine signal specificity. The peptides that included S543, S670 and S1043 became phosphorylated in this assay, whereas the alanine versions of the same peptides were either not detectably phosphorylated or phosphorylated to a significantly reduced extent (Fig. 1E). Collectively, these observations suggest that Cdk1 may directly phosphorylate BubR1 in vivo, and that the loss of BubR1 phosphorylation upon Mps1 depletion may reflect an indirect mechanism.

Initially we sought to determine whether loss of BubR1 phosphorylation at the SP sites identified here results in changes to BubR1 localization or in gross structural defects in the protein. Endogenous BubR1 was depleted using siRNA oligos that target the 3′-UTR region of BubR1 (supplementary material Fig. S2) (Elowe et al., 2007), and cells were simultaneously transfected with MYC-tagged BubR1-WT or phosphorylation site mutants. MYC–BubR1-WT localized as expected to the outer-KT, as demonstrated by colocalization with Hec1 (supplementary material Fig. S3A), as did MYC–BubR1-620A, -5A and -5D. In addition BubR1-5A and -5D retained the ability to coimmunoprecipitate Cdc20 and Bub3 (supplementary material Fig. S3B), confirming that MCC assembly is maintained. Finally, mutation of the five SP sites did not affect phosphorylation of BubR1 at the previously reported S676 site of Plk1 (supplementary material Fig. S3C). These results demonstrate that mutation of the five phosphorylation sites on BubR1 did not result in gross conformational or structural defects in the BubR1 protein that would preclude KT localization and MCC formation.

To explore whether the phosphorylation of the five identified SP sites on BubR1 contributes to the congression or checkpoint function of BubR1 (or both), we took several independent approaches. Initially, we examined mitotic progression by staining of fixed cells (Fig. 2A,B). After MG132 treatment, BubR1-620A, -5A, and -5D-expressing cells were all unable to align proper metaphases to the same extent as BubR1-WT-expressing cells (supplementary material Fig. S4A). This indicates that lack of phosphorylation on one or several of the five sites identified in this study causes congression defects, and that the BubR1-5D phosphomimic mutant cannot rescue the BubR1 depletion phenotype. A BubR1-4A/D mutant (which retained the previously described S670) also resulted in aberrant metaphases after MG132 treatment, indicating a significant contribution from the remaining four serine residues to the chromosome alignment function of BubR1 (supplementary material Fig. S4B).

Close examination of cells expressing the various mutants revealed differences with regard to their precise congression defects. As expected, cells depleted of endogenous BubR1 and rescued with the empty plasmid failed to congress chromosomes efficiently, so that KTs were spread along the length of the pole-to-pole axis (Fig. 2B). By contrast, cells expressing BubR1-WT generally formed tight metaphase plates, with KTs efficiently aligned at the spindle equator. Cells expressing BubR1-620A were largely able to align their KTs, although resulting metaphases were considerably broader, whereas both BubR1-5A- and -5D-expressing cells resembled those depleted of endogenous BubR1, with many KTs remaining unattached and spread along the entire pole-to-pole axis (Fig. 2B, supplementary material Fig. S4C for MYC construct expression). Quantification of KT misalignment with the various BubR1 mutants revealed that the degree of misalignment in cells expressing either BubR1-5A or BubR1-5D was almost as severe as in cells entirely depleted of BubR1 (25% in BubR1-5A and 22% in BubR1-5D compared to 30% of cells transfected with control vector; Fig. 2C). These observations suggest that these five phosphorylation sites probably make the most significant contribution to the congression function of BubR1.

As a second, complementary approach, we examined the function of BubR1 phosphorylation site mutants in real-time by time-lapse video microscopy performed on HeLa cells stably expressing histone H2B-GFP. To track individual cells, we used BubR1 constructs tagged with mCherry, as previously described (Elowe et al., 2007). Representative stills from each movie, taken at the indicated time points after the onset of chromosome condensation, are shown in Fig. 2D, and quantification of time in mitosis is summarized in Fig. 2E. Cells depleted of endogenous BubR1 and expressing the mCherry tag alone exited mitosis very rapidly (90% of the cells within 60 minutes, supplementary material Movie 1), without metaphase alignment and often with lagging chromosomes, as expected. This phenotype was rescued by BubR1-WT expression (supplementary material Movie 2). As described previously, BubR1-620A-expressing cells exited mitosis more slowly, with approximately 40% of the cells entering anaphase more than 140 minutes after chromosome condensation, compared with about 25% of BubR1-WT expressing cells (supplementary material Movie 3). Cells expressing BubR1-5A or BubR1-5D displayed severe congression defects (supplementary material Movies 4 and 5, respectively); they were often unable to reach metaphase and many cells were unable to exit mitosis within the imaging period (16 hours). As a result, only 25% of BubR1-5A- and 42% of BubR1-5D-expressing cells reached anaphase within 140 minutes. Interestingly, amongst the cells expressing BubR1-5A and BubR1-5D that were able to exit mitosis, we were unable to detect any lagging chromosomes in anaphase. This, together with the observation that the many cells unable to align at metaphase did not exit mitosis, suggests that the checkpoint function of BubR1 is maintained in the BubR1-5A mutant.

To more rigorously test this conclusion, we assessed the ability of cells expressing either MYC–BubR1-WT or BubR1 phosphosite mutants to remain mitotically arrested upon microtubule depolymerization. Non-rescued cells were unable to arrest in the presence of nocodazole, whereas cells rescued with BubR1-WT arrested efficiently (Fig. 2F). Similarly, cells expressing MYC–BubR1-620A, -5A or -5D all arrested to the same extent as those expressing BubR1-WT, indicating that the SAC function of BubR1 is functional in these cells. Collectively, our observations indicate that phosphorylation at the five sites described here is critical for BubR1 function in KT-MT attachment and chromosome congression, but is not required for mediating the SAC response in either an unperturbed mitosis or after a nocodazole challenge.

S670 and S676 are the best-conserved of the BubR1 phosphorylation sites identified to date. Indeed many residues in this region of the protein are conserved in higher eukaryotes (Fig. 3A). To study S670 phosphorylation, we generated an anti-pS670 antibody. Phosphospecificity of the antibody was demonstrated by loss of reactivity after phosphatase treatment as seen by immunofluorescence and western blotting (supplementary material Fig. S5A,B). Furthermore, antibody reactivity in cells depleted of endogenous BubR1 could be restored by MYC–BubR1-WT but not the non-phosphorylatable MYC–BubR1-670A (supplementary material Fig. S5C). Analysis of an asynchronous cell population by immunofluorescence microscopy revealed that BubR1 S670 was phosphorylated predominantly at unaligned KTs during prometaphase and in cells nearing metaphase (supplementary material Fig. S5D). Anti-pS670 was also observed to decorate the aligned KTs of metaphase plates (supplementary material Fig. S5D), which contrasts with the results obtained with the anti-pS676 antibody (Elowe et al., 2007). This suggests that phosphorylation at S670 may be important for KT attachments throughout mitosis.

To test the effect of alterations in microtubule dynamics on S670 phosphorylation, we treated mitotic cells with DMSO, nocodazole, or Taxol for 30 minutes before fixation. Cells were then probed for S670 and S676 phosphorylation by immunofluorescence using the corresponding phosphospecific antibodies. Focusing on prometaphase cells we observed only minor differences in S676 phosphorylation between conditions tested (data not shown, see quantification in Fig. 3C), consistent with the lack of inter-KT tension in prometaphase. By contrast, S670 phosphorylation was significantly elevated in prometaphase cells treated with 300 nM nocodazole, as compared to the DMSO control, whereas no increase in phosphorylation was detected after treatment with 100 nM Taxol (Fig. 3B, and quantification in C), even though this dose of Taxol efficiently induces rephosphorylation of S676 when cells are aligned at metaphase (Elowe et al., 2007). In agreement with these findings, a marked increase in pS670 reactivity after nocodazole but not Taxol or DMSO treatment was observed by western blotting (Fig. 3D). These results indicate that S670 phosphorylation is elevated at unattached rather than tensionless KTs. A similar conclusion was recently reached by Yen and co-workers, using an independently generated antibody against pS670 (Huang et al., 2008).

As S676 phosphorylation is lost at KTs upon metaphase alignment and establishment of tension, we considered the possibility that the phosphorylation status of S670 may regulate S676 phosphorylation by Plk1 at this stage. Indeed, when cells were maintained at metaphase by MG132 treatment, S676 phosphorylation was still detectable in BubR1-S670A or BubR1-S670D expressing cells, although, as expected, it was lost in BubR1-WT expressing cells (Fig. 3E). This indicates that the KT-MT attachments in cells expressing BubR1-S670A/D were unable to generate sufficient tension at metaphase to turn off S676 phosphorylation by Plk1, although we cannot directly exclude microenvironment alterations that effect anti-pS676 reactivity in response to changes in S670 phosphorylation.

In addition to phosphorylation sites, BubR1 and Mad3 proteins contain several evolutionarily conserved motifs. These include two KEN boxes and a GLEBS-like motif which is required for Bub3 binding and Mad3 KT localization. In S. cerevisiae and Schizosaccharomyces pombe, the direct interaction between Mad3p and Cdc20p is mediated through the N-terminal, but not C-terminal KEN box (Burton and Solomon, 2007; King et al., 2007; Sczaniecka et al., 2008), whereas a recent study has suggested that both N- and C-terminal KEN motifs of murine BubR1 are required for association with Cdc20 (Malureanu et al., 2009). To resolve this discrepancy, we initially generated immobilized peptides encompassing either N-terminal (KEN26) or C-terminal (KEN304) KEN-box motifs and tested their ability to directly bind recombinant GST-Cdc20 or GST alone. Whereas the KEN26 motif of human BubR1 bound efficiently to recombinant Cdc20, a peptide encompassing the KEN304 box showed no binding when assayed under the same conditions (Fig. 4A). Neither peptide was able to associate with GST alone, demonstrating the specificity of the interaction. This is in contrast to a recent report on mBubR1 (Malureanu et al., 2009), but in full agreement with data from S. cerevisiae and S. pombe (King et al., 2007; Sczaniecka et al., 2008).

How the KEN motifs function in the context of the full-length BubR1 molecule is not clear, as BubR1 binding to Cdc20 involves multiple sites, and previous reports only focused on truncated N-terminal BubR1 (Davenport et al., 2006; Malureanu et al., 2009). We therefore asked whether various full-length human BubR1 proteins carrying point mutations in these critical motifs are able to form a complex with other MCC components and the APC/C in mitosis. Endogenous Cdc20, Bub3 and Cdc27 could readily be co-immunoprecipitated with MYC–BubR1-WT, as expected (Fig. 4B). Similar results were observed after immunoprecipitation of MYC-tagged BubR1-620A and BubR1-5A, indicating that MCC formation is intact in the BubR1 phosphorylation site mutants. Surprisingly, we found that full-length BubR1-KEN26AAA as well as BubR1-KEN304AAA associated with appreciable amounts of Cdc20, Cdc27 and Bub3 (Fig. 4B), and even mutation of both BubR1 KEN boxes together (BubR1-ΔKEN) did not abrogate the interaction with these proteins (Fig. 4C, supplementary material Fig. S6A). Therefore, in the context of full-length BubR1, mutation of KEN26 alone or in conjunction with KEN304 is not sufficient to eliminate the interaction with Cdc20 and the APC/C, probably because of the presence of a second C-terminal Cdc20 binding region in BubR1 (Davenport et al., 2006; Tang et al., 2001). For comparison, BubR1-E413K, carrying a mutation in the Bub3 binding (GLEBS) region also co-immunoprecipitated Cdc20 and Cdc27, but, as expected, not Bub3 (Fig. 4B).

To directly compare the efficiency of Cdc20 binding between full-length and N-terminal BubR1, we generated an N-terminal fragment of human BubR1 (BubR1-N, residues 1-370) in its WT form or with one or both KEN motifs mutated, and tested the ability of these fragments to bind Cdc20 and the APC/C. Full-length BubR1-WT bound efficiently to Cdc20, Cdc27, and Bub3, as expected, whereas BubR1-N-WT recruited Cdc20 at much lower levels (Fig. 4D, supplementary material Fig. S6B). This interaction was completely eliminated in BubR1-N-KEN26AAA, and BubR1-N-ΔKEN, whereas BubR1-N-KEN304AAA bound Cdc20 at levels comparable to BubR1-N-WT. The full complement of Cdc20 binding therefore requires the BubR1 C-terminus; the N-terminus of BubR1 can only recruit low levels of Cdc20 through KEN26 (but not KEN304), in agreement with our peptide binding studies and observations in yeast (King et al., 2007; Sczaniecka et al., 2008). Moreover, in the absence of sufficient Cdc20 binding, BubR1 is unable to efficiently associate with the APC/C. Indeed, upon siRNA-mediated depletion of Cdc20, BubR1 immunoprecipitates contained reduced levels of the APC/C subunits Cdc27, Apc7 and Apc4 compared with control cells (Fig. 4E).

BubR1 mutated in either the N-terminal or C-terminal KEN box efficiently localized to the KT in cells depleted of endogenous BubR1, whereas BubR1-E413K was unable to accumulate at KTs, as predicted (Fig. 5A, for quantification see supplementary material Fig. S7). Recently, it was shown that BubR1 interacts with Blinkin via N-terminal TPR motifs and independent of Bub3 (Kiyomitsu et al., 2007). The relevant region is conserved in Bub1 and mutations in this region abrogate Bub1 KT recruitment (Kiyomitsu et al., 2007). By contrast, we find that the Blinkin-binding region of BubR1 (residues 1-203) was largely cytoplasmic, indicating that Blinkin binding alone is not sufficient for KT localization of BubR1 (supplementary material Fig. S7).

We next sought to determine whether SAC function was perturbed in full-length BubR1 lacking functional KEN and KT recruitment motifs. Having shown that full-length BubR1 mutated at either KEN box was still able to associate with significant amounts of Cdc20, we asked whether cells expressing these mutants would be able to mount a SAC response. However, expression of BubR1-KEN26AAA or -KEN304AAA in cells depleted of endogenous BubR1 was not sufficient to sustain a mitotic arrest in the presence of nocodazole (Fig. 5B), in full agreement with observations in yeast and mouse. These data underscore the notion that Cdc20 binding alone is not sufficient for APC/C^Cdc20 inhibition by BubR1. Interestingly, HeLa cells expressing BubR1-E413K were equally defective in arresting in response to a nocodazole (Fig. 5B), suggesting that Bub3 binding and/or BubR1 localization to KTs is requisite for SAC function.

To study the role of the KEN and GLEBS motifs in regulating mitotic timing, we turned to time-lapse videomicroscopy of HeLa cells stably expressing histone H2B-GFP. Expression of BubR1-KEN26AAA, -KEN304AAA, or -E413K failed to restore the timely and accurate mitotic progression that occurred with expression of BubR1-WT, but interesting differences in phenotypes were observed. BubR1-KEN26AAA- and -E413K-expressing cells exited mitosis with kinetics very similar to cells depleted of endogenous BubR1 (with a median mitotic time of 24±1 and 30±4 minutes, respectively, see Fig. 5D and supplementary material Movies 6 and 7) and often exhibited misaligned chromosomes upon anaphase onset, indicative of a defective SAC (see arrowhead). By contrast, although cells expressing BubR1-KEN304AAA also exited mitosis faster than BubR1-WT-expressing cells (median 45±2 minutes, see Fig. 5D and supplementary material Movie 8), in each of three independent experiments they were consistently slower than either BubR1-KEN26AAA- or BubR1-E413K-expressing cells, or cells depleted of BubR1 (P=0.03 for KEN304AAA-expressing vs BubR1-depleted cells). Moreover, the chromosomes of these cells were often observed to align at metaphase, although alignment was not maintained for the same duration as in BubR1-WT-expressing cells (Fig. 5C). Similar results were obtained upon analysis of fixed cells by immunofluorescence microscopy. Cells expressing the control empty vector or expressing BubR1-KEN26AAA, -KEN304AAA or -E413K exited mitosis prematurely compared with BubR1-WT-expressing cells, as demonstrated by the larger proportion of transfected cells in anaphase (Fig. 5E). Together, these results demonstrate that, in the context of full-length human BubR1, KEN26, KEN304 and the GLEBS motif are all essential for SAC function and, to varying extents, for timely mitotic progression.

The difference in mitotic progression time between cells expressing BubR1-KEN304AAA and other checkpoint mutants tested here (on average 15 minutes), prompted us to question whether this extra time contributed to the correction of chromosome attachment errors that occur frequently in early mitosis. We therefore explored the extent to which chromosomes were found lagging in anaphase cells expressing the various BubR1 SAC mutants in depletion-rescue experiments. Less than 20% of cells depleted of endogenous BubR1 exhibited normal anaphases, with the vast majority showing more than two lagging KTs, whereas the majority of cells expressing BubR1-WT exited mitosis accurately (Fig. 6A,B). Cells expressing BubR1-KEN26AAA or BubR1-E413K exhibited lagging KTs in anaphase at levels significantly higher than those expressing BubR1-WT, in line with their rapid and premature exit from mitosis. By contrast, lagging KTs were observed in BubR1-KEN304AAA anaphases at a much lower frequency than in the other SAC mutants (Fig. 6A,B). This suggests that the longer duration of mitoses in BubR1-KEN304AAA-expressing cells may allow sufficient time for establishing accurate KT-MT attachments.

As with the BubR1 phosphorylation site mutants, we sought to investigate whether the motifs important for the SAC function of BubR1 are also important for the KT-MT attachment function. To this end we treated cells again as shown in Fig. 2A and then examined metaphase alignment in the presence of MG132. Congression defects caused by endogenous BubR1 depletion were largely rescued by BubR1-WT, -KEN26AAA, or -KEN304AAA expression, but not by expression of BubR1-E413K (Fig. 6C). Together, these results indicate that whereas BubR1 KEN motifs contribute to the SAC, these motifs are dispensable for chromosome attachments. By contrast, the GLEBS motif of BubR1 is required for both SAC and KT-MT attachment functions.

The Plk1-dependent BubR1 mitotic upshift correlates with its function in maintaining stable KT-MT attachments (Elowe et al., 2007; Matsumura et al., 2007). We therefore tested whether BubR1 KEN and GLEBS mutants also become upshifted in mitosis in response to nocodazole treatment. Whereas MYC–BubR1-WT, - KEN26AAA and -KEN304AAA mutants exhibited the double-band pattern characteristic of BubR1, the mitotic upshift was no longer seen in MYC–BubR1-E413K (Fig. 6D). However, because the BubR1 upshift during mitosis is a result of Plk1 activity, it is not indicative of the phosphorylation sites described here. Thus, to study possible effects of BubR1 mutations within the KEN and GLEBS motifs on S670 phosphorylation we used the anti-pS670 antibody for immunofluorescence microscopy (Fig. 6E) and western blotting (Fig. 6F). As demonstrated by anti-pS670 antibody staining, BubR1-WT expression in HeLa cells depleted of endogenous BubR1 resulted in phosphorylation of S670 and, similarly, robust pS670 staining was detected at KTs in both BubR1-KEN26AAA- and BubR1-KEN304AAA-expressing cells (Fig. 6E). By contrast, BubR1-E413K expressing cells showed no anti-pS670 staining (Fig. 6E). We considered the possibility that S670 phosphorylation was not detected in these cells because of a loss of BubR1-E413K from the KT and a corresponding dilution of the epitope into the cytoplasm. However, when the phosphorylation state of the BubR1 mutants was examined by western blotting, we again observed that BubR1-WT, BubR1-KEN26AAA and BubR1-KEN304AAA readily became phosphorylated at both S670 and S676, but BubR1-E413K was not phosphorylated at either residue (Fig. 6F). Taken together, these observations suggest that BubR1 phosphorylation, and thus BubR1 functionality during KT attachment, depends on the BubR1-Bub3 interaction. This is lost in the BubR1-E413K mutant, but not in either KEN-box mutant.

Here, we demonstrate that the mitotic functions of BubR1 are regulated by distinct motifs and thus can be uncoupled from each other (Fig. 7). Loss of BubR1 phosphorylation results in faulty KT-MT attachments and poor chromosome congression (Elowe et al., 2007; Huang et al., 2008; Matsumura et al., 2007). Furthermore, phosphorylation, at least at the sites described here, appears to be largely dispensable for the SAC functions of BubR1. By contrast, although KEN motifs are essential for SAC functionality of BubR1, no direct role was observed in either chromosome congression or BubR1 phosphorylation. Finally, the GLEBS domain is essential for both SAC activity and BubR1 phosphorylation, and thus for accurate and timely chromosome congression and separation.

We have identified five proline-directed phosphorylation sites on BubR1. We demonstrate that at least three of the five sites identified are direct Cdk1 targets in vitro. Huang et al. recently suggested that Mps1 might be a major kinase for BubR1 in vivo (Huang et al., 2008). Although we also observe loss of BubR1 phosphorylation upon Mps1 inhibition (data not shown), we have been unable to directly phosphorylate BubR1 by Mps1 and thus favor the view that the reported result may represent an indirect effect, perhaps reflecting mislocalization of BubR1. Indeed, a mislocalization of endogenous BubR1 upon Mps1 inhibition (our unpublished results), and the lack of any detectable phosphorylation on the mislocalized BubR1-E413K mutant support the conclusion that KT localization of BubR1 is required for its phosphorylation. Importantly, phosphorylation on BubR1 at one of these SP sites, S670, is significantly enhanced at unattached KTs but not at KTs lacking tension, in excellent agreement with previous work (Huang et al., 2008).

Recent studies indicate that Cdk1-cyclin B1 localizes to KTs during prometaphase, where it contributes to the correct attachment of MTs to KTs and efficient chromosome alignment through phosphorylation of local substrates (Bentley et al., 2007; Chen et al., 2008). Moreover, KT recruitment of Cdk1 is increased in nocodazole- but not Taxol-treated cells (Bentley et al., 2007), in agreement with the increase in S670 phosphorylation observed here upon nocodazole but not Taxol treatment.

Studies of yeast Mad3 and N-terminal fragments of murine BubR1, as well as our own observations reported here, indicate that both KEN motifs are required for SAC function (Burton and Solomon, 2007; King et al., 2007; Malureanu et al., 2009; Sczaniecka et al., 2008). Importantly, however, our binding studies reveal that the interaction between BubR1 and Cdc20 is in itself not sufficient for APC/C^Cdc20 inhibition, as full-length BubR1 mutated at its KEN boxes bound Cdc20 efficiently, and yet cells expressing these mutants exited mitosis prematurely and could not arrest upon microtubule depolymerization. Nevertheless, we also demonstrate that efficient recruitment of Cdc20 is necessary for binding and inhibition of the APC/C and that this strictly requires the C-terminal region of BubR1. It is possible that APC/C inhibition provided by the BubR1 KEN-box interaction with Cdc20 is sufficient for cell survival during normal passage through mitosis, but that the C-terminal BubR1 Cdc20 binding region becomes essential for full SAC activity upon microtubule depolymerization. In support of this view, a mutant of murine BubR1 [mBubR1Δ(525-700)], which lacks only the C-terminal Cdc20 docking site, supports cell survival but is unable to fully sustain the SAC (Malureanu et al., 2009).

Our data also suggest that the two KEN motifs make distinct contributions to BubR1 functionality during mitosis. We found that an N-terminal but not a C-terminal KEN-box peptide interacts directly with Cdc20, in agreement with data from both S. cerevisiae and S. pombe (Burton and Solomon, 2007; King et al., 2007; Sczaniecka et al., 2008). This data supports the idea that BubR1 can act as a competitive inhibitor of substrate binding through a direct interaction between the N-terminal KEN box and Cdc20. In addition, the KEN26-Cdc20 interaction may be required for the recently reported Cdc20 turnover in early mitosis, which is thought to be required for maintaining the SAC (Nilsson et al., 2008; King et al., 2007). It is attractive to speculate that the N-terminal KEN box of BubR1 plays a dual role in the SAC: first, by establishing tight binding to Cdc20 it may exclude bona fide mitotic substrates, and second, by functioning as a destruction box in trans, it may facilitate Cdc20 ubiquitylation and degradation.

The role of the C-terminal KEN box is less clear. Although the KEN304 motif is required for SAC function, cells expressing BubR1-KEN304AAA were consistently and significantly slower in completing mitosis and exited with a significantly lower incidence of lagging KTs than cells expressing other SAC mutants. This subtle difference would not have been readily detected in end point survival assays such as those used in the studies on yeast Mad3 and murine BubR1 (Burton and Solomon, 2007; King et al., 2007; Malureanu et al., 2009; Sczaniecka et al., 2008). One possible function of the C-terminal KEN box could relate to the proper docking and orientation of BubR1 on the APC/C, similar to what has been proposed for the D-box of Emi1 (Miller et al., 2006). In this context, it is interesting that Drosophila BubR1 has only one KEN box, corresponding to the N-terminal KEN motif of the vertebrate enzyme. How Drosophila compensates for the absence of a C-terminal KEN box is not presently clear.

Several lines of evidence indicate that BubR1-Bub3 association is important for SAC function and accurate chromosome congression. First, Bub3 is one of the original SAC components identified in S. cerevisiae (Hoyt et al., 1991), and loss of Bub3 expression or the Mad3p-Bub3p interaction confers benomyl sensitivity (Hardwick et al., 2000). Our own results corroborate the data from S. cerevisiae and demonstrate that cells expressing BubR1-E413K, a mutant that cannot bind Bub3 or localize to KTs, have a defective SAC (Fig. 5B). Second, BubR1 dynamically associates with KTs, and FRAP studies show that BubR1 and Cdc20 display similar biphasic kinetics at unattached KTs, arguing that they shuttle off the KT in a complex (Howell et al., 2004). Third, overexpression of a peptide encompassing the BubR1 GLEBS motif was shown to disrupt the nocodazole-activated SAC in HeLa cells (Harris et al., 2005). Taken together, these data strongly support the view that the GLEBS motif in BubR1 contributes to the SAC function of BubR1. In addition, we show here that BubR1-E413K does not exhibit the characteristic upshift in mitosis and does not become phosphorylated at either S670 or S676, two sites shown to be critical for productive KT-MT attachments. Whether the loss of functionality of the BubR1 GLEBS mutant reflects an impaired recruitment of BubR1 to KTs or a disruption of the BubR1-Bub3 interaction (or both) remains to be clarified.

Recent studies were interpreted to suggest that cytosolic BubR1 is sufficient for efficient mitosis. First, Drosophila cid (CENP-A) mutants retain an intact SAC response to spindle disruption despite the inability of SAC components, including BubR1, to target to KTs (Blower et al., 2006). Second, a cytosolic murine BubR1 fragment was recently reported to support cell survival (Malureanu et al., 2009). However, although unattached KTs are apparently not strictly required for MCC formation in vitro, they do accelerate this process (Kulukian et al., 2009). In vivo, the rate at which MCC formation occurs may not be sufficient to support a fully functional checkpoint. Anchoring of BubR1 to the KT through Bub3 would provide an elevated concentration of BubR1 readily available for association with the primed Cdc20-Mad2 complexes, and this may well be important for timely MCC formation. Although it is possible that the requirement for the GLEBS domain reflects an essential function of the BubR1-Bub3 complex that is independent of KT localization, it remains difficult to rationalize how a cytoplasmic BubR1 fragment that lacks not only the GLEBS domain but also phosphorylation sites known to be important for chromosome capture and alignment (Elowe et al., 2007; Huang et al., 2008; Matsumura et al., 2007) would be able to confer stable KT-MT interactions (Malureanu et al., 2009).

HeLa S3, HEK293T and HeLa S3 cells expressing histone H2B-GFP were routinely maintained in DMEM (Invitrogen) supplemented with 10% FBS and penicillin-streptomycin (100 IU/ml and 100 mg/ml, respectively, Gibco). MG132 (Calbiochem) was used at 20 μM for 3 hours unless otherwise stated. For synchronization studies, nocodazole and thymidine were used at 300 nM and 2 mM respectively for 16 hours unless otherwise stated. All BubR1 constructs were generated in the pcDNA3.1 plasmid (Invitrogen), driven by the CMV promoter, and modified to carry an N-terminal triple-MYC tag.

In vitro phosphorylation of recombinant BubR1 was carried out in 30 μl of kinase reaction buffer as previously described (Elowe et al., 2007). Recombinant active Cdk1 was purchased from Upstate Biotechnology, and GST-Mps1 from Invitrogen. For Cdk1 assays on peptides immobilized on cellulose membranes, dried membranes were first washed in ethanol and then hydrated in kinase buffer [50 mM Tris-HCl pH 7.5, 10 mM MgCl₂, 1 mM DTT (dithiothreitol), 100 μM NaF, 10 μM sodium vanadate] for 1 hour, followed by overnight blocking in kinase buffer with 100 mM NaCl and 0.5 mg/ml BSA. The next day, the membrane was blocked again with kinase buffer containing 1 mg/ml BSA, 100 mM NaCl and 50 μM non-radioactive ATP at 30°C for 45 minutes. The blocking solution was subsequently replaced with kinase reaction buffer containing 0.2 mg/ml BSA, 50 μCi/ml [γ-³²P]ATP (3000 Ci/mmol, 10 mCi/ml), 2 μg/ml recombinant Cdk1, and 50 μM ATP for 3 hours on a shaker at 30°C. The membranes were then washed extensively: 10×15 minutes in 1 M NaCl, 3×5 minutes in H₂O, 3×15 minutes 5% H₃PO_4, 3×5 minutes in H₂O, and then sonicated overnight in 8 M urea, 1% SDS (w/v) and 0.5% (v/v) β-mercaptoethanol to remove residual nonspecific radioactivity. Phosphorylation was visualized by autoradiography.

Cells grown on coverslips were fixed and permeabilized simultaneously in PTEMF buffer (0.2% Triton X-100, 20 mM PIPES pH 6.8, 1 mM MgCl₂, 10 mM EGTA and 4% formaldehyde). Processing for immunofluorescence and image acquisition on a Deltavision microscope (Applied Precision) were performed as previously described. For time-lapse microscopy, all treatments within a single experiment were performed simultaneously. During imaging, the atmosphere was maintained at a temperature of 37°C, humidity 60% and 5% CO₂. Imaging was performed using a Zeiss Axio Observer Z1 microscope equipped with a Plan Neofluar 40× objective. Metamorph 7.1 software (Molecular Devices) was used to collect and process data. Images were captured at 3-minute intervals for 16 hours.

The monoclonal BubR1 antibody was previously described (Elowe et al., 2007). Anti-p55Cdc20 (Santa Cruz Biotechnology), anti-Bub3 (BD Transduction labs), anti-MYC (9E10, ATCC), anti-α-tubulin (DM1A, Sigma), anti-Cdc27 (BD Transduction labs), anti-APC7 (Biolegend), anti-APC4 (Santa Cruz Biotechnology), anti-Bub1 (Chemicon), as well as CREST anti-human auto-immune serum (Immunovision), were obtained commercially. Anti-pS670 polyclonal antibody was generated by immunizing rabbits with KLH-conjugated phosphopeptide (H-CSIKKLS(P)PIIED-OH), and then isolated from a protein-A-purified IgG fraction using the same peptide. For immunofluorescence experiments, all primary antibodies were detected with Cy2/Cy3-conjugated donkey antibodies (Dianova) and Alexa-Fluor-647-conjugated goat antibodies (Invitrogen). For western blots, signals were detected using HRP-conjugated anti-mouse or anti-rabbit antibodies (Pierce).

Peptide arrays were constructed according to the Spots-synthesis method according the manufacturer's directions (Intavis). For Cdc20 binding experiments, purified GST-Cdc20 fusion protein generated in SF9 insect cells was added at 5 μg/ml in TBST and incubated together with the membrane overnight at 4°C. Membranes were washed three times in TBST and bound protein was visualized with anti-Cdc20 antibodies. The sequences of synthesized peptides are as follows: KEN30: DEWELSKENVQPLRQ; KEN304: PPMPRAKENELQAGP; S543: SEKKNKSPPADP; S574: TSNEDVSPDVCD; S670: LSIKKLSPIIED; S720: SENPTQSPWCSQ; S1043: KVGKLTSPGALL.

HeLa S3 cells were cultured in DMEM formulated with either unlabeled L-lysine or L-arginine or labeled with L-[6-¹³C, 4-¹⁵N]arginine and L-[6-¹³C, 2-¹⁵N]lysine (Cambridge Isotope Laboratories) at a concentration of 44 and 86 μg/ml respectively, and supplemented with 10% dialyzed fetal bovine serum. Extracts prepared from 4×10⁷ unlabelled nocodazole-arrested cells and of 4×10⁷ isotopically labeled thymidine-blocked cells were mixed at a ratio of 1:1. This mixture was divided into equal parts, and immunoprecipitation was performed with anti-BubR1 monoclonal antibody or 9E10 anti-MYC monoclonal antibody as a negative control. Samples were separated by SDS-PAGE and excised gel fragments were processed for mass spectrometry.