In this study we showed that hMps1 is a kinetochore-associated protein kinase. Until recently, no information was available on which checkpoint proteins are essential for targeting Mps1-like protein kinases to kinetochores. We found that Hec1 was required for kinetochore recruitment of hMps1. In Hec1-depleted cells, kinetochore-association of hMps1 was completely abolished, but not vice versa. The budding yeast homolog of Hec1, Ndc80p, is found in a complex together with Nuf2p, Spc24p, and Spc25p, and mutational inactivation of these components causes defects in chromosome segregation (Wigge and Kilmartin, 2001; He et al., 2001; Janke et al., 2001). The Ndc80p-complex is thought to mediate kinetochore-microtubule attachments, as does the Nuf2 homolog in C. elegans (Him-10) and Ndc80/Hec1 in human cells (Chen et al., 1997; Wigge et al., 1998; Howe et al., 2001; Janke et al., 2001; Nabetani et al., 2001). Homologs of Ndc80 and Nuf2 exist from fission yeast to humans, yet Spc24 and Spc25 homologs have not been identified in metazoans. Therefore, we have recently tested in mammalian cells whether Nuf2-depletion also leads to displacement of hMps1 from kinetochores and found that to be the case (data not shown).
Importantly, yeast cells lacking either Ndc80p or Nuf2p are checkpoint-proficient;
however, cells lacking both proteins are checkpoint-deficient (McCleland et al., 2003), suggesting that both proteins have a redundant checkpoint role.
Interestingly, two independent reports illustrated that Ndc80p/Hec1 is a phosphoprotein. In budding yeast, the Ipl1p kinase can phosphorylate Ndc80p in vitro, and in human cells, Ndc80/Hec1 is phosphorylated by the centrosomal kinase Nek2 (Chen et al., 2002a). The significance of these phosphorylation events remains to be determined, but it is possible that hMps1 is also able to phosphorylate Hec1.
However, until now we have been unsuccessful in demonstrating a biochemical interaction between Hec1 and hMps1 (data not shown) and Hec1 is not phosphorylated by hMps1 in in vitro kinase assays (data not shown).
As recently shown by Martin-Lluesma et al. (2002), Hec1 elimination also interfered with kinetochore localization of hMad1 and hMad2. Thus, Hec1 is required for localizing hMps1 to kinetochores, and both Hec1 and hMps1 are required for recruiting the hMad1/hMad2 complex. Importantly, depletion of Hec1 caused
lacked significant amounts of hMad2. In addition, none of the chromosomes were aligned at the metaphase plate, although they were attached to kinetochore microtubules, which points to an underlying defect in chromosome congression (Martin-Lluesma et al., 2002). It is possible that persistent checkpoint activity is caused by lack of tension in Hec1-depleted cells. In this regard it is important to note that the checkpoint proteins hBub1, CENP-E and CENP-F are not displaced from the kinetochores in Hec1-depleted cells. Moreover, several reports suggested that the checkpoint proteins hBub1 and hBubR1 might be involved in a signaling pathway that senses tension (Musacchio and Hardwick, 2002; Skoufias et al., 2001; Zhou et al., 2002a; Taylor et al., 2001). Therefore, it is tempting to speculate that kinetochore-microtubule interactions and checkpoint signaling may involve two distinct pathways in vertebrates: one centered on Hec1, hMps1 and hMad1/hMad2 (for sensing microtubule attachment) and the other on CENP-E interacting with CENP-F and, most likely, hBub1 and hBubR1 (for sensing tension). Both pathways will converge onto the APC/C, which gets inhibited and cells arrest in mitosis (Chan et al., 1999;
McEwen et al., 2001). To strengthen this model from the perspective of hMps1, we found that neither siRNA-mediated depletion of CENP-E, CENP-F or hBubR1 produced any effect on kinetochore localization of hMps1 nor vice versa.
Beside the dependency on Hec1, we also found that kinetochore-association of hMps1 is partially dependent on hMad1, but not on hMad2. In hMad1-depleted cells, hMps1 could still be detected on kinetochores, albeit to a lower extent. Therefore, both proteins mutually depend on each other for kinetochore recruitment. It is interesting to note that in Xenopus egg extracts immunodepleted for XMad1 a reduction of BubR1 at kinetochores (Chen, 2002), but not of Bub1 (Sharp-Baker and Chen, 2001), was reported. Furthermore, on unattached chromosomes, both Bub1 and BubR1 showed hyperphosphorylation, as visualized by a mobility shift that is strongly dependent on Mad1 (Chen, 2002). These observations lead to the proposal that Mad1 may act as a scaffold protein, which allows the recruitment of many checkpoint proteins to kinetochores. As a consequence, these molecules would be brought into close proximity, which would allow the reformation of checkpoint complexes and/or stimulate posttranslational modifications through the checkpoint kinases Mps1, Bub1 and BubR1.
Model for hMps1 function in the mitotic spindle checkpoint
Based on data obtained primarily from this study we can propose a model of how hMps1 assembles onto kinetochores and how it might function in checkpoint signaling. During early mitosis, unattached kinetochores recruit most of the spindle checkpoint proteins such as hMps1, hMad1/hMad2, hBub1/hBub3 and hBubR1/hBub3 (Yu, 2002; Musacchio and Hardwick, 2002; Zhou et al., 2002b;
Millband et al., 2002). Particularly, Hec1 is required in order to localize hMps1 efficiently to kinetochores. Upon checkpoint activation, rearrangements of the Mad/Bub checkpoint complexes occur, and it seems likely that such rearrangements take place at kinetochores. Certain checkpoint components, such as hMad2, hBubR1, hBub3, are extremely dynamic and have a high turnover rate at kinetochores (Howell et al., 2000; Yu, 2002). It is possible that kinases like hMps1 promote complex rearrangements, probably by phosphorylating critical substrates like hMad1, which releases free hMad2. It is important to note that in budding yeast the formation of a Mad1/Bub1/Bub3 complex has been reported upon checkpoint activation, which is dependent on the activity of Mps1p (Brady and Hardwick, 2000). In general, reorganization of checkpoint complexes might create platforms, which allow the formation of checkpoint complexes that can subsequently inhibit the activity of the APC/C, like the MCC consisting of hBubR1/hBub3/hMad2/Cdc20 (Sudakin et al., 2001) or the hBubR1/hBub3/Cdc20 complex (Tang et al., 2001).
However, an unresolved issue concerns the contribution of the kinetochores to checkpoint complex formation. The MCC was shown to be already present in interphase and is therefore unlikely to be generated exclusively by kinetochores (Sudakin et al., 2001). Similarly, budding yeast MCC can still be formed in ndc10-1 mutants, which are thought to lack kinetochore structures entirely (Fraschini et al., 2001b). Furthermore, overexpression of Mps1p in ndc10-1 mutant cells leads to cell cycle arrest, suggesting that Mps1p does not absolutely require kinetochores for its signaling function in the checkpoint pathway (Fraschini et al., 2001a). Thus, it is apparent that large cytoplasmic pools of checkpoint complexes exist beyond the complexes generated at the kinetochores. Interestingly, the human MCC is only inhibiting the activity of the APC/C that has been isolated from mitotic cells (Sudakin et al., 2001). It is therefore important to determine the modifications that make the APC/C susceptible to inhibition by the MCC. It is an intriguing possibility that
kinases like hMps1 might directly modify APC/C subunits. In fact, Liu et al. (2003) have recently proposed that hMps1 associates with the APC/C, however the consequences of this interaction remain unclear. It is possible that basal levels of the MCC can form in interphase cells. Upon checkpoint activation, unattached kinetochores might then increase the efficiency of MCC assembly. In parallel, modifications of the APC/C in mitosis could increase its affinity toward the MCC, thus establishing a stable APC/C-MCC complex. Future studies are needed to resolve the contribution of hMps1 to this important issue.
Figure 29: A model proposing a kinetochore recruitment pathway of mitotic spindle checkpoint components in human cells. Upon mitotic spindle checkpoint activation, checkpoint proteins are thought to be recruited to kinetochores in distinct pathways and complexes, and most likely in a specific order. The formation of the inhibitory MCC might be generated by unattached kinetochores, but could also come from cytosolic pools. Mitotic APC/C might be phosphorylated by kinases like hBub1/hBubR1 or hMps1 making it more susceptible for MCC inhibition. Pi: phosphorylation.