|Home | About | Journals | Submit | Contact Us | Français|
MPS1 kinase is an essential component of the spindle assembly checkpoint (SAC), but its functioning mechanisms are not fully understood. We have shown recently that direct interaction between BUBR1 and MAD2 is critical for assembly and function of the human mitotic checkpoint complex (MCC), the SAC effector. Here we report that inhibition of MPS1 kinase activity by reversine disrupts BUBR1-MAD2 as well as CDC20-MAD2 interactions, causing premature activation of the anaphase-promoting complex/cyclosome. The effect of MPS1 inhibition is likely due to reduction of closed MAD2 (C-MAD2), as expressing a MAD2 mutant (MAD2L13A) that is locked in the C conformation rescued the checkpoint defects. In the presence of reversine, exogenous C-MAD2 does not localize to unattached kinetochores but is still incorporated into the MCC. Contrary to a previous report, we found that sustained MPS1 activity is required for maintaining both the MAD1·C-MAD2 complex and open MAD2 (O-MAD2) at unattached kinetochores to facilitate C-MAD2 production. Additionally, mitotic phosphorylation of BUBR1 is also affected by MPS1 inhibition but seems dispensable for MCC assembly. Our results support the notion that MPS1 kinase promotes C-MAD2 production and subsequent MCC assembly to activate the SAC.
The spindle assembly checkpoint (SAC)2 is an evolutionarily conserved signal transduction mechanism that preserves genomic integrity through regulating proper timing of the metaphase-to-anaphase transition during cell division (1–3). “Wait anaphase” signals originate from kinetochores lacking proper tension or attachment to spindle microtubules, leading to inhibition of the multisubunit E3 ubiquitin ligase anaphase-promoting complex/cyclosome (APC/C) (4). The APC/C utilizes an activator and substrate-specifying subunit, CDC20, to ubiquitylate securin and cyclin B, whose subsequent destruction is essential for anaphase onset (1, 5).
The mitotic checkpoint complex (MCC), composed of BUBR1, BUB3, CDC20, and MAD2 in human cells, is the predominant effector of the SAC and specifically prevents APC/C-mediated degradation of cyclin B and securin during early mitosis (6). The MCC directly binds to the APC/C and blocks CDC20 binding to securin and cyclin B (7). Efficient inhibition of the APC/C could not be achieved by individual MCC subunits or MCC subcomplexes at physiologically relevant concentrations, suggesting that a fully assembled MCC is critical for its potent inhibitory activity (8–11). Despite recent solution of the Schizosaccharomyces pombe MCC crystal structure (naturally lacking BUB3) (12), the molecular mechanisms of human MCC assembly and function remain incomplete. Nevertheless, it is clear that extensive protein-protein interactions exist between human MCC subunits. In addition to a cell cycle-independent BUBR1-BUB3 subcomplex, direct interactions between BUBR1-CDC20, CDC20-MAD2, and BUBR1-MAD2 have also been observed (1, 13–15). Both CDC20 and BUBR1 selectively associate with the closed conformer of MAD2 (C-MAD2), a critical signal transducer for the SAC whose intracellular concentration increases in checkpoint-active mitotic cells (15–17). We were the first to show that direct BUBR1·C-MAD2 interaction is important for MCC integrity, MCC-APC/C association, and APC/C inhibition (15). Our findings have been supported by the S. pombe MCC structure (12) and studies in Saccharomyces cerevisiae (18).
The SAC is also regulated by several mitotic kinases, including MPS1 (1). MPS1 kinase plays essential roles in targeting the MAD1·C-MAD2 complex to kinetochores, allowing the complex to function as a catalyst in converting open MAD2 conformers (O-MAD2) into C-MAD2 (19–24). Hewitt et al. (22) demonstrated that MPS1 kinase activity is also required for recruiting O-MAD2 to the kinetochore-localized MAD1·C-MAD2 catalyst. In addition, MPS1 kinase may also phosphorylate BUBR1 and borealin, although the functional significance of these phosphorylation events in the mitotic checkpoint remains controversial (25–27). Experiments in engineered cell lines, together with novel MPS1-specific small molecule inhibitors, have also shown that MPS1 kinase affects BUBR1-CDC20 and/or CDC20-MAD2 interactions (26, 28–31).
In studying how the BUBR1-MAD2 interaction is regulated, we found that the interaction is impaired when MPS1 kinase activity is inhibited. Importantly, the impairment can be rescued by expressing a C-MAD2 mutant in mitotic cells, supporting that MPS1 contributes to SAC signal transduction mainly through regulating C-MAD2 production.
HeLaM, a subline of HeLa, was maintained in DMEM with 10% fetal bovine serum at 37 °C in 5% CO2 (9). To block cells in prometaphase, HeLaM cells were treated with 2.5 mm thymidine (Sigma-Aldrich) for 24 h and then directly released into medium containing 0.2 μm nocodazole (Sigma-Aldrich) or 10 μm taxol (Biomol International) for 12 h. Alternatively, to treat cells with reversine prior to mitotic entry, cells arrested in G1/S by double thymidine block were released into drug-free medium for 5 h and then treated with nocodazole or taxol in combination with reversine or DMSO for 3 h followed by MG132 addition for another 1.5 h. Some variations of cell synchronization protocols are described in more detail in the figure legends. Reversine (Calbiochem) was used at 500 nm (29). The proteasome inhibitor MG132 (Cayman Chemical) and another MPS1 inhibitor, AZ3146 (Selleckchem) (22), were used at 20 and 2 μm final concentrations, respectively.
The MPS1 shRNA and RNAi resistant pLAP-MPS1WT (wild type) or MPS1KD (kinase-dead) constructs were gifts from Geert Kops (University of Utrecht) (27) and transfected together with pBabe-puromycin at a ratio of 10:5:1. The shRNA-transfected cells were enriched 24 h post-transfection by selection in puromycin (1 μg/ml) for 48 h. The mCherry-Mis12-MAD1WT construct was from Maria Maldonado and Tarun Kapoor (Rockefeller University) (31). pCS2-MAD1-GFP (32), BUBR1 phosphomutants BUBR15A (S543A, S574A, S670A, S720A, and S1043A) (33), and BUBR1QA (S435A, S543A, S670A, and S1043A) (25) were from Ted Salmon (University of North Carolina), Sabin Elowe (Université Laval), and Tim Yen (Fox Chase Cancer Center), respectively. The BUBR1 mutants were subcloned into a mammalian GST fusion vector. GST-BUBR15A was used as a template to create a BUBR111A (adding T54A, S435A, S676A, T792A, T1008A, and T1042A on the basis of known in vivo phosphorylation sites in PubMed Central). Site-directed mutagenesis was performed using the QuikChange multisite kit from Stratagene. All mutants were confirmed by DNA sequencing. Other DNA constructs have been described previously (15). DNA transfection was carried out using TransIT-LT1 reagent (Mirus) following the instructions of the manufacturer.
The extracts were prepared and assayed as described previously (15). Recombinant GST-tagged MPS1 was from Invitrogen.
In vitro ubiquitylation assays were performed essentially as described previously (9), except that anti-CDC16 immunoprecipitates (IPs, as the source of APC/C) were washed four times with 2 bead volumes of TBS-T (20 mm Tris (pH 7.5), 150 mm NaCl, 0.01% Tween 20) and resuspended as a 50% slurry in TBS. Usually, 4 μl of bead slurry was used in a ubiquitylation reaction.
For imaging at a lower magnification (Fig. 3), HeLaM cells grown in 24-well plates were synchronized by double thymidine block and transfected with GFP- or mCherry-tagged MAD2WT, MAD2L13A, or MAD2ΔC10 at the beginning of the second thymidine block. Cells were released 16 h later directly into nocodazole-containing phenol red-free DMEM and allowed to grow for an additional 12 h. HEPES-KOH (pH7.6) and OxyFluor (Oxyrase) were added to 25 mm and 0.3 units/ml, respectively. Then, cells were covered with mineral oil. DMSO or reversine were added to the medium immediately before imaging. Phase-contrast and GFP or mCherry images (using Alexa Fluor 488 or 555 channels) were collected on an automated Olympus IX-81 microscope using a 20 ×, NA 0.50 UPlanFLN objective while cells were maintained at 37 °C in a heated chamber. Single-plane images were acquired for 14 h at 6-min intervals at multiple positions using a CoolSNAP HQ2 camera with 2 × 2 binning.
For imaging at a higher magnification (Fig. 4), HeLaM cells stably expressing mRFP-H2A were grown on No. 1.5 coverslip-bottomed, 35-mm dishes (MatTek); transiently transfected with GFP-MAD2L13A, MAD2ΔC10, or other GFP fusion constructs; and imaged on a Leica TCS SP8 confocal microscope using a 63 ×, NA1.40 objective, usually 24 h after transfection. Z-stacks of 1 μm were collected at each time point with 3-min intervals. When needed, drugs were dissolved in ~1/5 volume of the medium and added underneath the mineral oil.
We have demonstrated previously that BUBR1-MAD2 interaction is crucial for MCC assembly and MCC inhibition toward the APC/C (9, 15). To investigate the potential roles of MPS1 in regulating the BUBR1-MAD2 interaction and MCC assembly, mitotic cells arrested with nocodazole or taxol were treated with the MPS1 inhibitors reversine or AZ3146 (22, 29) (Fig. 1A and supplemental Fig. S1). Inhibition of MPS1 by reversine treatment dramatically reduced MAD2 levels associated with BUBR1 in both nocodazole- and taxol-arrested cells (Fig. 1A, see supplemental Fig. S1, B–D for taxol results). Reciprocal immunoprecipitation using two distinct anti-MAD2 antibodies not only confirmed the inability of MAD2 to associate with BUBR1 but also revealed disruption of the MAD2-CDC20 interaction in reversine-treated cells (Fig. 1A and supplemental Fig. S1, the reproducibly less clear BUB3 signals may be related to MAD2 antibodies). In contrast, the BUBR1-CDC20 interaction showed no apparent changes after reversine treatment (Fig. 1A and supplemental Fig. S1C). Similar results were obtained by treating cells with AZ3146 (supplemental Fig. S1E) or by adding reversine prior to mitotic entry (supplemental Fig. S2). In addition, the level of MAD2 coimmunoprecipitated with BUBR1 was also reduced when MPS1 was silenced by either shRNA or siRNA (Fig. 1B and supplemental Fig. S1F). Furthermore, rescue experiments indicated that wild-type but not kinase-dead MPS1 could restore the level of MAD2 in BUBR1 IP to be comparable with that in control mitotic cells (Fig. 1B). The combined results demonstrate that MPS1 kinase activity is essential for incorporating MAD2 into the MCC.
To correlate defective MCC assembly with SAC override and mitotic exit caused by reversine treatment (29), we examined both APC/C activity and APC/C-MCC association in nocodazole and MG132 arrested mitotic cells (Fig. 2). APC/C activity, measured by percentage of ubiquitylated myc-cyclin B(1–102) after incubation, was higher when the APC/C was isolated from reversine-treated mitotic cells than from DMSO-treated (control) cells (Fig. 2A). Consistently, reduced levels of BUBR1, BUB3, and MAD2 were found to coimmunoprecipitate with APC4 (an APC/C core subunit) following reversine treatment, suggesting loss of MCC association with the APC/C (Fig. 2B). Association of the activator protein CDC20 with APC4 was not altered (Fig. 2B). The results provide molecular evidence to link reversine-mediated mitotic exit with premature APC/C activation.
We then performed inhibitor washout experiments to confirm that MPS1 kinase regulates MCC assembly (Fig. 2, C and D). Incorporation of MAD2 into the MCC, as evidenced by coimmunoprecipitation of MAD2 with BUBR1, was rapidly restored following reversine washout (within 15 min) (Fig. 2C, BUBR1 IP, 15–45 min lanes). No distinctive changes in BUBR1-CDC20 interaction was observed throughout the course of the experiment. Coinciding with MAD2 loss from the MCC, only APC/C isolated from the 0 min time point displayed ubiquitylation activity (Fig. 2D, lane 2). As MAD2 reassociated with BUBR1, APC/C isolated 15–45 min after reversine washout exhibited no ubiquitylation activity (Fig. 2D, compare lanes 3–5 to lane 2). The above results could also be recapitulated when cells were arrested in taxol (supplemental Fig. S3). The results underscore the importance of MAD2 incorporation into the MCC for efficient APC/C inhibition.
As MPS1 inhibition causes dissociation of MAD2 from both BUBR1 and CDC20, and BUBR1 and CDC20 bind only to C-MAD2 (15, 34), the above results suggest that MPS1 inhibition prevents MCC assembly through impeding C-MAD2 production. This idea is consistent with the roles of MPS1 in recruiting MAD1·C-MAD2 to kinetochores and O-MAD2 to the MAD1·C-MAD2 complex (19–22). To test the idea in a more direct manner, time-lapse live cell imaging was employed to examine whether reversine-compromised SAC could be rescued by expression of a C conformation-locked MAD2 mutant, MAD2L13A (35, 36). In nocodazole-arrested mitotic cells, premature mitotic exit became apparent ~1 h after reversine addition in untransfected cells, as evidenced by cell flattening (GFP-negative cells in Fig. 3, A and B, reversine rows, examples are indicated by arrowheads). Remarkably, GFP-MAD2L13A expression countered the effects of reversine and maintained transfected cells in mitosis for up to 12 h (GFP-positive green cells in Fig. 3A, reversine row, examples are indicated by arrows). In contrast, cells overexpressing GFP-MAD2ΔC10 (a stable O-MAD2 mutant (14, 37)) displayed a weakened checkpoint even in the absence of reversine (Fig. 3B, GFP-positive cells in the DMSO row, an example is indicated by a smaller arrow) and rapidly exited mitosis following reversine addition (reversine row, compare GFP-positive and GFP-negative cells at the 00:32:23 time point, indicated by smaller arrow and arrowhead). These results suggest that C-MAD2 expression can rescue the SAC defects caused by inhibition of MPS1 kinase.
We next show that overexpression of C-MAD2, but not wild-type MAD2, can overcome reversine-induced premature mitotic exit. To this end, mCherry-tagged MAD2 constructs were used because previous observations showed that GFP- or GST-tagged MAD2WT stabilized MAD2 in the C conformation (Ref. 9 and data not shown). Time-lapse movies were taken similarly as those described for GFP-MAD2 constructs, and the results are summarized in Fig. 3, C and D. Although mCherry-MAD2WT-expressing cells rapidly exited from mitosis following reversine treatment, mCherry-MAD2L13A-transfected cells stayed in mitosis for an average of 7.4 h. The variability in mitosis duration in transfected cells can largely be ascribed to the wide range of expression levels of MAD2L13A (data not shown). Cells with lower expression still behaved more like untransfected cells following reversine addition. Together, the above results indicate that premature mitotic exit induced by MPS1 inhibition can be rescued by exogenously expressed C-MAD2.
We have noticed that, when observed at higher magnification, the majority of mitosis-arrested cells expressing GFP-MAD2L13A exhibited well defined metaphase plates but did not retain GFP signals at kinetochores (data not shown). In current models, the kinetochore-localized MAD1·C-MAD2 complex is widely assumed to be essential for amplifying C-MAD2 and activating the mitotic checkpoint. In addition, Hewitt et al. (22) have shown previously that, when targeted to kinetochores, the MAD1·C-MAD2 catalyst complex became insensitive to MPS1 inhibition. To better understand the C-MAD2 production process, we decided to further study kinetochore localization of the MAD2 conformers, MAD1 and MPS1, in the presence of reversine. When GFP-MAD2L13A-transfected cells were first treated with MG132 for 1 h, cells with a well aligned metaphase plate showed no GFP signals at kinetochores, as predicted (Fig. 4A, left panel). When the same cells were exposed to nocodazole (3.3 μm), the metaphase plates were disrupted within minutes as GFP dots developed on chromosomes, indicating C-MAD2 recruitment as kinetochores lost microtubule attachment (Fig. 4A, center panel, and supplemental Movies S1 and S2). Interestingly, when the same cells were then challenged with reversine, C-MAD2 disappeared from kinetochores very rapidly (< 6 min under our experimental conditions) (Fig. 4A, right panel, and supplemental Movies S3 and S4). This is in contrast to the results reported previously by Hewitt et al. (22). There is a slight possibility that GFP-MAD2L13A behaves differently from endogenous C-MAD2. However, MAD1-GFP also disappeared from unattached kinetochores within 15 min after reversine treatment, albeit at somewhat slower kinetics, strongly suggesting that the MAD1·C-MAD2 complex was delocalized from kinetochores after reversine treatment (Fig. 4B and supplemental Movie S5). We did confirm that GFP-MAD2ΔC10 could not be recruited to unattached kinetochores in the presence of reversine (Fig. 4C) (22). In contrast, GFP-MPS1 signal intensities at kinetochores increased following exposure to reversine (Fig. 4D), consistent with previous reports that MPS1 targeting to kinetochores does not depend upon its kinase activity but MPS1 dynamic exchange at kinetochores does (21, 22, 38). The MPS1 result also demonstrated that loss of MAD1 and C-MAD2 from kinetochores was not due to photobleaching during live cell imaging. The combined results suggest that maintenance of the MAD1·C-MAD2 complex at kinetochores requires MPS1 kinase activity. Therefore, when MPS1 kinase is inhibited, failure to recruit O-MAD2 to kinetochores may simply reflect loss of the O-MAD2 receptor (the MAD1·C-MAD2 complex) from kinetochores.
To further validate this interpretation, we examined the localization of GFP-MAD2L13A and GFP-MAD2ΔC10 in cells cotransfected with constitutively kinetochore-localized mCherry-Mis12-MAD1 (31). The prediction is that MAD2 conformers will stay at kinetochores as long as MAD1 is at kinetochores, regardless of MPS1 inhibition or not. Reversine treatment slightly reduced mCherry-Mis12-MAD1 levels at kinetochores under our experimental conditions, possibly because of a number of reasons (for example, loss of mCherry-Mis12-MAD1 heterodimerized with endogenous MAD1). Nevertheless, it was clear that both GFP-MAD2L13A and GFP-MAD2ΔC10 stayed at kinetochores at least 1 h after reversine addition at levels roughly proportional to the remaining MAD1 at kinetochores (Fig. 4, E and F).
We conclude that exogenous C-MAD2 expression arrests cells in metaphase without having to localize at kinetochores, whether in the presence or absence of reversine. On the other hand, MPS1 kinase activity is essential for maintaining the MAD1·C-MAD2 catalyst at unattached kinetochores, which, in turn, functions as the O-MAD2 receptor.
To further understand the time-lapse results (Figs. 3 and and4),4), we examined the ability of MAD2L13A and MAD2ΔC10 to rescue MCC assembly in reversine-treated cells. Cells transfected with GST-tagged MAD2L13A or MAD2ΔC10 or mock-treated cells (control) were arrested in mitosis with nocodazole and then challenged with reversine in the presence of MG132 (Fig. 5). In agreement with Fig. 1B, little endogenous MAD2 was found in BUBR1 IPs in the presence of reversine (Fig. 5B, BUBR1 IP, bottom rows). In contrast, GST-MAD2L13A was clearly detected in BUBR1 IPs, together with CDC20 and BUB3 (Fig. 5B, BUBR1 IP, L13A lane), indicating an assembled MCC. Reciprocal pull-downs of GST-MAD2L13A confirmed the BUBR1 immunoprecipitation results (Fig. 5B, GST pull-down, L13A lane). The MAD2ΔC10 mutant failed to coimmunoprecipitate with BUBR1 or pull down any MCC subunit (Fig. 5B, ΔC10 lanes). Consistent results were obtained when cells were arrested in taxol (supplemental Fig. S4). The results suggest that, in reversine-treated cells, exogenously expressed MAD2L13A is still incorporated into the MCC, despite delocalization from kinetochores.
We next investigated two other possible mechanisms through which MPS1 might affect the mitotic checkpoint. The first possibility concerns BUBR1 phosphorylation by MPS1 kinase. MPS1 was suggested to regulate the phosphorylation status of BUBR1, but the impact of the modification on the mitotic checkpoint remains controversial (25, 26, 33, 39). We have noticed the increase of a faster-migrating BUBR1 species in cell lysates prepared from reversine-treated mitotic HeLa cells, as reported previously (26). A similar mobility shift was obtained using alternative MPS1 inhibitors SP600125 or AZ3146 (Fig. 6A). The results indicate that BUBR1 phosphorylation may indeed be affected by MPS1 directly or indirectly (Fig. 6A). Additionally, faster-migrating BUBR1 species were also observed following treatment with the Plk1 inhibitor BI2536 or the Aurora kinase inhibitor ZM447439 (Fig. 6A), confirming earlier observations that more than one kinase contribute to the slower-migrating species of BUBR1 in mitotic cell lysates (39–41). To investigate whether BUBR1 phosphorylation affects MCC assembly, nocodazole-arrested lysates were subjected to mock or λ phosphatase treatment, followed by BUBR1 immunoprecipitation to examine coimmunoprecipitation of MCC subunits (Fig. 6B). No distinctive changes in the association of CDC20, BUB3, or MAD2 with BUBR1 were observed after phosphatase treatment, suggesting that BUBR1 phosphorylation is dispensable for MCC integrity. To test this idea further, we examined two previously characterized BUBR1 phosphomutants, BUBR15A (S543A, S574A, S670A, S720A, and S1043A) (33) and BUBR1QA (S435A, S543A, S670A, and S1043A) (25). Both mutants showed relatively intact checkpoint responses (25, 33). Consistently, cells transfected with GST-tagged BUBR15A and BUBR1QA were able to pull down CDC20, BUB3, and MAD2 as efficiently as BUBR1WT (Fig. 6C). In addition, a BUBR111A (see “Experimental Procedures” for details) also pulled down other MCC subunits at comparable levels as BUBR1WT, despite a clear mobility shift (Fig. 6D). This is in contrast to a BUBR1KEN26AAA mutant, where alanine substitutions at the critical lysine-glutamic acid-asparagine (KEN) box (26–28 residues) disrupted BUBR1 interactions with both CDC20 and MAD2 (Fig. 6C). A conservative conclusion can at least be drawn that our efforts so far have not revealed a significant contribution of BUBR1 phosphorylation to MCC assembly or stability.
We also examined the possibility that MPS1 kinase increases the steady level of MCC by slowing its disassembly. In cells, MCC assembly and disassembly may exist in a dynamic equilibrium (42). The net loss of MAD2 from BUBR1 IPs following MPS1 inhibition may arise from either failure to incorporate MAD2 into the MCC, expedited disassembly of the pre-existent MAD2-containing MCC, or both. When MCC (in the form of BUBR1 IPs) isolated from mitotic cells was incubated in vitro for various times, the complex was stable, and no apparent loss of MCC subunits was observed, even after 24 h (Fig. 7A). This is consistent with previous reports that MCC disassembly requires energy input and p31comet (43–46). Indeed, MCC disassembly and APC/C activation were observed in a chromosome-free mitotic cell extract (15) (Fig. 7, B and C, left), most likely because of the combined effects of added energy input, the presence of disassembly factors, and the absence of wait anaphase signals and C-MAD2 amplification thereof (because of lack of chromosomes/kinetochores) in the extract (15, 47). When reversine was added to the mitotic extract, no acceleration of MCC disassembly was observed (Fig. 7B, right). Correspondingly, the rate of cyclin B degradation in the extract remained comparable in the absence or presence of reversine (Fig. 7C). Moreover, addition of active recombinant MPS1 kinase to the mitotic extract did not delay the kinetics of cyclin B degradation either (Fig. 7, D and E). Together, the above results are consistent with the notion that MPS1 mainly promotes MCC assembly rather than inhibits MCC disassembly.
We have demonstrated previously that BUBR1 directly interacts with C-MAD2 at the MAD2 dimerization domain (αC helix) and that incorporation of C-MAD2 is critical for the MCC to become an effective APC/C inhibitor (9, 15). These results have been largely confirmed by the crystal structure of the S. pombe MCC, in which MAD3 (the yeast BUBR1), MAD2, and CDC20 closely interact to prevent CDC20 from accessing substrates containing degrons such as the KEN box or Destruction (D) box (12). Although the S. pombe study focused on MAD3 segments (including the N-terminal KEN box and the Lys-92 residue) in blocking substrate binding to CDC20, MAD2 in the fully assembled MCC structure may be essential for maintaining key MAD3 residues at proper spatial positions or directly shifting CDC20 away from the Doc10 coactivator in the APC/C (12). Lau and Murray (48) even showed that artificially tethering Mad2 to Cdc20 arrests S. cerevisiae cells in metaphase independently of other checkpoint components. Consistently, when MPS1 was inhibited by reversine and MCC lost MAD2, the APC/C could not be effectively inhibited, and the checkpoint was compromised, even though CDC20 still bound to BUBR1 (Figs. 1, B–D, and and5).5). These results also agreed with our earlier discovery when comparing the APC/C inhibitory activity of recombinant MCC and recombinant BUBR1-BUB3-CDC20 ternary complexes (9). C-MAD2 has been regarded as a major wait anaphase signaling molecule for the SAC, and its amplification is temporally restricted to when the SAC is active (16, 17). In contrast, neither protein level nor posttranslational modification changes of BUBR1/BUB3 and CDC20 can be reliably correlated to the active SAC state. In fact, the interaction between BUBR1/BUB3 and CDC20 has been observed both in vitro and in interphase cells independent of an active checkpoint (11, 15). Therefore, we support an Occam's razor type of SAC model, suggesting that the signal transducer C-MAD2 is directly incorporated into the MCC through interaction with BUBR1 and CDC20, producing a highly potent APC/C inhibitor. We did notice a very recent report that suggests a two-step catalysis model for the mitotic checkpoint (49). The model proposes that only BUBR1-BUB3 constitutes the ultimate inhibitor of APC/CCDC20 and that the interaction between BUBR1 and CDC20 is catalyzed by a C-MAD2·CDC20 complex. Obviously more work is needed to further clarify the pathway and the effector(s) that lead to APC/CCDC20 inhibition.
We have presented evidence demonstrating that MPS1 kinase mainly impacts MCC assembly and subsequent APC/C inhibition through promoting C-MAD2 formation. MPS1 may work at a minimum of two discrete steps during C-MAD2 amplification. First, MPS1 is required for targeting the catalytic platform MAD1·C-MAD2 to unattached kinetochores (19, 20). Second, MPS1 may modify the MAD1·C-MAD2 platform to empower or enhance its catalytic efficiency. The second mechanism is consistent with the result that MPS1 activity is still required for SAC signaling even when MAD1 is constitutively expressed at kinetochores through fusion with Mis12 (31). Theoretically, MPS1 may work through increasing substrate (O-MAD2) binding, stimulating O → C-MAD2 conversion efficiency or promoting product (C-MAD2) release. Hewitt et al. (22) suggested previously that MPS1 activity is required for O-MAD2 recruitment. In particular, they observed that MPS1 inhibition after mitotic entry only affects O-MAD2 recruitment but not localization of the MAD1·C-MAD2 complex. We have drawn a different conclusion on the basis of live cell imaging experiments. We have shown that MPS1 kinase activity is not only required for initial establishment but also for maintenance of the MAD1·C-MAD2 complex at unattached kinetochores. This is consistent with our earlier result that MPS1 is required for both establishment and maintenance of the mitotic checkpoint when activated by CENP-E perturbation (20). Our results also suggest that defects in O-MAD2 recruitment following MPS1 inhibition could simply be caused by loss of the MAD1·C-MAD2 complex from unattached kinetochores (Fig. 4). The reasons behind differences in results are not yet clear, although we did notice some differences in methodology. For example, Hewitt et al. used stable cell lines and immunofluorescence (22), whereas our results were derived from live cell imaging after transfection of GFP- or mCherry-tagged proteins.
MAD1 forms a cell cycle-independent complex with C-MAD2 (34, 50, 51), but current SAC models all assume that only kinetochore-localized MAD1-MAD2 complexes are capable of catalyzing O → C-MAD2 conversion (1). The reason underlying the differences between interphase and prometaphase MAD1-MAD2 complexes cannot be simply ascribed to the “capping” of p31comet, as p31comet is also found at prometaphase kinetochores (46, 50, 52). Westhorpe et al. (45) have suggested that p31comet at unattached kinetochores does not bind to the MAD1-MAD2 complex (possibly binding to CDC20-MAD2 instead). The interpretation still begs for an explanation of why the MAD1-MAD2 complex at unattached kinetochores displays different behavior from the same complex in interphase cells. Posttranslational modifications of the MAD1·C-MAD2 complex at unattached kinetochores is likely to play a major role in activating its catalytic power. Consistently, the mitotic arrest induced by constitutively kinetochore-localized mCherry-Mis12-MAD1 expression was abrogated by inhibition/depletion of MPS1, Aurora B, or BUBR1 (31). BUBR1 depletion may directly disrupt MCC assembly without any impact on C-MAD2 production, but the result strongly indicates that MPS1 and Aurora B kinases regulate the catalysis step per se during O → C-MAD2 conversion. Along this line, it is interesting to note that a fraction of O-MAD2 remains at unattached kinetochores in mCherry-Mis12-MAD1-expressing cells even when MPS1 is inhibited by reversine (Fig. 4). In fact, GFP-O-MAD2 can also be retained at centromeres harboring mCherry-Mis12-MAD1 in interphase cells when MPS1 kinase activity is supposedly low (data not shown). These observations further suggest that increasing substrate (O-MAD2) binding to the MAD1·C-MAD2 catalyst may only be part of the reasons for the requirement of MPS1 activity during SAC activation. MPS1 kinase may directly affect the O → C conversion itself or C-MAD2 product release from the catalyst. Future studies will focus on dissecting how MPS1 and Aurora B kinases facilitate the catalysis of O → C-MAD2 conversion by the MAD1·C-MAD2 complex.
We conclude that the major role of MPS1 kinase during SAC activation is to facilitate C-MAD2 production and subsequent MCC assembly through engaging the MAD1·C-MAD2 catalytic core at unattached kinetochores.
We thank Drs. Sabine Elowe, Tarun Kapoor, Geert Kops, Ted Salmon, and Tim Yen for DNA constructs and Tim Yen for multiple antibodies.
*This work was supported by the deArce Memorial Fund, by Ohio Cancer Research Associates, and by National Science Foundation Grant MCB-1052413 (to S. T. L.).
This article contains supplemental Figs. S1–S5, Table 1, and Movies S1–S5.
2The abbreviations used are: