Tale Sliedrecht

The importance of MPS1 in mitosis has been established for several years. Even though recent years have seen quite some development in the elucidation of the molecular pathways concerning MPS1 signaling, many aspects remained to be uncovered at the onset of the research presented in this thesis. In Chapter 2 we report the development of mutant cell lines that allow specific, highly penetrant and reversible inhibition of MPS1, providing us with the opportunity to study the role of MPS1 in mitosis. One of these functions - the recruitment of MAD1 to unattached kinetochores - is studied in more detail in Chapter 3. We provide evidence that feedback control between MPS1, BUB1 and ZW10 regulates MAD1 kinetochore recruitment and subsequent SAC activity. In this chapter, we furthermore show that the predominant function for RZZ and BUB1 in SAC signaling is ensuring MAD1 localization. Chapter 4 describes the identification of PLK1 as an auxiliary factor in the establishment and maintenance of the SAC. Under conditions of maximal SAC activity, PLK1 is dispensable, but when SAC signaling is suboptimal, PLK1 becomes essential to maintain SAC signaling. Our data suggest that under these conditions PLK1 activity promotes MCC stability. The work presented in Chapter 5 describes the efforts put into identification of direct MPS1 targets and provides useful lessons for future investigation in the search for novel MPS1 targets.

Discussion

The role of MPS1 in the SAC
MPS1 has long been known to be involved in maintenance of the SAC 129,135. In Chapter 2 we used chemical genetics to investigate the role of MPS1 in mitosis. In accordance with previous studies we found MPS1 kinase activity to be essential for SAC activity, the recruitment of checkpoint proteins to kinetochores and MCC formation. In Chapter 3 we investigated in particular the role of MPS1 in the recruitment of MAD1. We reveal that MPS1-dependent recruitment of MAD1 is mediated by BUB1 and ZW10. BUB1 was recently shown to be recruited to KNL1 by MPS1-dependent phosphorylation of conserved MELT-like motifs 118-120. BUB1 therefore may be the crucial and perhaps even the only effector for MPS1 in regulating MAD1 kinetochore recruitment. We were unable to prove this, however, since a kinetochore-tethered BUB1 could not rescue MAD1 kinetochore levels in MPS1-inhibited cells (not shown). This may mean that MPS1 has additional impact on kinetochore MAD1 or, alternatively, that tethered BUB1 cannot recapitulate BUB1’s role in MAD1 localization. With regards to the first option, MPS1 has been shown to phosphorylate MAD1 in yeast 137 but to date a specific phosphosite or functional role for this phosphorylation in the SAC has not been reported. Interestingly in Chapter 5 we identify multiple MAD1 sites that are regulated upon MPS1 inhibition. Mutation of these sites did not, however, influence MAD1 kinetochore localization (data not shown), although this does not exclude the possibility of additional, unidentified MPS1-dependent MAD1 phosphoresidues that impact its localization. Alternatively, the sites we identified might contribute to the recently identified role of MPS1 kinase activity in regulating MAD2 dimerization 90. The recent development of antibodies specifically recognizing O-MAD2 90,237 and C-MAD2 267 might be helpful in investigating this.

MAD2 itself is also an interesting candidate for direct regulation by MPS1, since both the MAD2 conformational change and MCC stability require MPS1 activity 90,91. Fission yeast Mad2 is phosphorylated by Mps1 in vitro and a Mad2 mutant lacking all five sites was SAC deficient 150. Contrary to intuition, phosphorylation of Ser187 of Fission yeast Mad2 was suggested to reduce Mad2-Mad1 binding in fission yeast, suggesting that Mps1 kinase activity inhibits aspects of SAC signaling. In line with this, phosphorylation of the C-terminal region of human MAD2 has also been shown to negatively regulate checkpoint activity by inhibiting the conformational transition of MAD2 268,269. Thus, there is currently no evidence that the MPS1-dependent regulation of MAD2 structural conversion to activate the SAC occurs via direct phosphorylation. Rather, it seems more likely that MPS1 promotes the MAD2 conformational change indirectly, for instance by activating/recruiting a protein phosphatase that counteracts inhibitory phosphorylations. Any of these mechanisms could also account for the observed impact of MPS1 activity on MCC stability.

In addition, the other MCC subunits BUBR1 and CDC20 are also subjected to phosphorylation and other post translation modifications 41,85,233,270,271. MPS1-mediated MCC stability might thus also be promoted via CDC20 or BUBR1. Another intriguing option is that MPS1 does not actively promote MCC stability but rather counters its disassembly. There are several, possibly related, mechanisms described that promote MCC disassembly. These include p31 comet-mediated extraction of MAD2 from the MCC, aided by CDK1-dependent phosphorylation of CDC20 and possibly by CUEDC2 110,112,270,272, and auto-ubiquitination of CDC20, aided by APC15 and perhaps even by p31 comet 102-105,111,112. Any of these events could be impacted by MPS1 activity. Due to the complexity of the pathways, I propose that the best way to identify the mechanisms by which MPS1 regulates the various aspects of SAC signaling is quantitative proteomics studies of protein complexes. For a more elaborate discussion, see below.

Budding yeast NDC80 can be phosphorylated by MPS1 in vitro, and some evidence in budding yeast supports the idea that MPS1-dependent phosphorylation of NDC80 contributes to SAC activity 263. Although both human and budding yeast MPS1 require HEC1/NDC80 for its recruitment to kinetochores 142,145,146,263,273, there is no evidence that the phosphoregulation of HEC1/NDC80 by MPS1 is conserved in humans. Indeed, despite extensive phosphoproteomic examinations of the KMN network (8,14,202, and our data in Chapter 5), no MPS1-like sites in the tail of HEC1/NDC80 have been found. We furthermore showed previously that the sole function of HEC1/NDC80 in SAC signaling in human cells is recruitment of MPS1 to kinetochores 146, so it seems unlikely that MPS1 itself could initiate that event. It is of interest to note that the mitotic kinase NEK2 was reported to phosphorylate HEC1 in its CH domain on Ser165, a phosphorylation that was shown to modulate SAC activity in human cell 12,264. Although the fact that NEK2 is degraded in early mitosis 274 argues against a prominent role for this kinase in the SAC, these findings could indicate that SAC-modulating phosphorylation of HEC1 is present in human cells as it is in yeast, even if the kinase may not be conserved. Interestingly, in Chapter 5 we found phosphorylation of SPC24, a member of the NDC80 complex, to be regulated. Future experiments need to elucidate if SPC24 phosphorylation in humans serves a similar purpose as NDC80 phosphorylation in yeast. Relatively straightforward RNAi rescue experiments of SPC24 phosphosite mutants should give an initial clue as to whether the SAC, weakened or not, depends to some extent on SPC24 phosphorylation. More sophisticated subsequent analysis will be needed to determine the mechanism by which such phosphoresidues might act.

Identification of novel MPS1 targets
Despite the multiple critical functions of MPS1 in mitosis, only a limited number of targets proteins have been identified and studied in detail 118-120,164. It has proven to be quite difficult to identify bona fide MPS1 substrates, which might stem from low abundance of MPS1 substrates, low phosphorylation site occupancy, lack of a well-defined consensus motif and lack of stringent binding between MPS1 and its substrates. The latter finds support in the above-mentioned studies that, although successful at identifying KNL1 and Borealin as direct MPS1 substrates, fail to identify a direct interaction between MPS1 and its substrates. The recent developments in quantitative MS-based proteomics have allowed the identification of a large number of in vivo phosphorylation sites from complex samples and has proven a successful method for the identification of PLK1, Aurora B and CDK1 substrates 235,251-253. Chapter 5 describes the experimental optimization of quantitative phosphoproteomics methods aimed at identifying new MPS1 substrates.

Although our search has yielded several interesting candidates which will be subjected to future research, we conclude that the different experimental set-ups used were not ideal for the identification of MPS1 substrates. Although we have been successful in identifying large quantities of phosphopeptides, the relative abundance of kinetochore- and spindle-associated proteins was low. Furthermore, the amount of significantly regulated, MPS1-dependent phosphopeptides also seems relatively low compared to, for instance, studies that used similar approaches to identify PLK1 substrates 235,251. The latter might imply that MPS1 regulates the SAC and error-correction through only a small number of target proteins. However, since we also failed to identify the presumably heavily phosphorylated bona-fide MPS1 substrate KNL1 67,118-120, we suspect that poor recovery of kinetochore and spindle proteins has limited successful identification of novel MPS1 targets.

Several recent studies have circumvented this by selective enrichment of phosphorylated substrates of a kinase of interest using chemical genetics 275-277. Using the analog-sensitive MPS1 mutants described in Chapter 2, we attempted similar approaches. Unfortunately, poor activity of recombinant gatekeeper MPS1 mutants prevented us from using such analyses to identify MPS1 substrates. Since I feel that defining the MPS1 substrates and pinpointing the relevant phosphoresidues is key in understanding the mechanisms of SAC control, this begs the question how future research towards the identification of novel MPS1 substrates should be directed? Although quantitative phosphoproteomics provides us with the opportunity to identify phosphorylation sites that occur at least in cell culture conditions, future research might benefit from selective enrichment of kinetochore and spindle-associated proteins and protein complexes. It was recently shown that native kinetochore particles can be isolated from budding yeast by affinity capture of the MIS12 complex component Dsn1 278. A quantitative MS-based approach combined with selective enrichment of kinetochore particles or subcomplexes of the more elaborate human kinetochore might aid in the identification of novel MPS1 substrates, as was previously shown for KNL1 in budding yeast 118. Similar strategies can be envisioned for the APC/C and MCC complex as well as the KMN network, CCAN proteins and, for instance, the SKA-complex. Elucidating which of the undoubtedly many candidate phosphorylations identified in this way are truly done by MPS1 will be greatly aided by the recent identifications of the MPS1 consensus phosphorylation motif 239,279 that closely resembles the sequence surrounding the MPS1 targets in the bona fide substrate KNL1 67,118-120.

Auxiliary checkpoint proteins
In Chapter 4 we investigated if the activity of the SAC depends on the presence or activity of so called auxiliary, or modulating, proteins. We previously identified Aurora B as such a modulating protein. The role of Aurora B in SAC regulation is not an essential one since SAC maintenance is not significantly affected by depletion or inactivation of Aurora B. Aurora B is only required for timely establishment of the SAC by promoting efficient recruitment of MPS1 to unattached kinetochores at the onset of mitosis 146,153,212. The role of Aurora B in the SAC is therefore auxiliary and only revealed in a sensitized SAC assay 146,212. Like Aurora B, PLK1 activity is not essential for SAC maintenance when it is fully activated, but becomes essential to maintain a mitotic arrest when SAC activity is low. A role for PLK1 in the SAC has been tentatively proposed 226 but has remained unclear. This is perhaps due to the fact that PLK1 acts as an auxiliary factor combined with its essential role in a multitude of other mitotic processes, inhibition of which prominently activate the SAC 224,225,227,228. In contrast to Aurora B, PLK1 does not influence MPS1 kinetochore localization in human cells, as was recently suggested in Drosophila cells 228. Instead, we show that PLK1 regulates checkpoint activity downstream of MAD1 and serves to promote MCC stability when SAC activity is low. Although future experiments need to elucidate the exact mechanism by which PLK1 promotes MCC stability, our data suggests that phosphorylation within the BUBR1 GLEBS motif might be an important factor, by promoting the BUBR1-BUB3 interaction. If so, BUBR1 is a focal point for PLK1 signaling at the kinetochore: It was recently shown that PLK1 phosphorylates BUBR1 in another motif termed the KARD, promoting a BUBR1-PP2A interaction that is required for stabilizing kinetochore-microtubule interactions 41,42.

Our sensitized SAC assay is a powerful tool to examine SAC modulating factors, as shown by our identification of Aurora B and PLK1 as auxiliary components of the SAC. Besides allowing functional analysis of these two kinases in the SAC, we show in Chapter 3 that the sensitized SAC assay enhances the checkpoint phenotypes of ZW10 and BUB1 RNAi, proteins that are difficult to deplete efficiently. This may also hold true for NDC80 depletion, and the sensitized SAC assay has been instrumental in determining its contribution and that of its CH domain to SAC signaling 142. Future investigations using this sensitized set up might yield more auxiliary SAC regulators and might aid the study of essential regulators that have proven difficult to functionally inactivate. For instance, depletion of MAD1 has proven difficult and has obstructed the investigation of MAD1 phosphorylations that we identified in Chapter 5. Combining the described MAD1 phosphomutants with the sensitized SAC assay might allow functional studies, and this will be a subject of future studies.

Feedback in the SAC
Correct regulation of many cellular processes is controlled by feedback networks that enable cellular mechanism to rapidly respond to changing stimuli. The decision to enter mitosis for instance, is mediated by a network of several feedbacks loops that correctly regulate the activation of the Cyclin B-CDK1 complex to ensure that once the commitment to cell division is made, it is executed rapidly and robustly 3. A snowballing amount of evidence suggests that SAC regulation also depends on multiple feedback mechanisms. We previously showed that timely establishment of MPS1 kinetochore localization depends on Aurora B 146. Activity and localization of Aurora B in turn is regulated by MPS1 through direct phosphorylation of Borealin and promoting BUB1-dependent histone H2A phosphorylation respectively 132,151,153,164,191,192. MPS1 and Aurora B are thus engaged in a positive feedback loop ensuring rapid establishment of both the SAC and the error correction mechanisms as cells enter mitosis (Figure 1).

In Chapter 3 we identify feedback control in the regulation of MAD1 kinetochore binding. MPS1 recruitment of MAD1 is mediated by BUB1 and ZW10. BUB1 regulates ZW10 localization, and ZW10 in turn recruits MAD1 while simultaneously facilitating KMN network stability consequently regulating MPS1 localization. Feedback regulation thus ensures rapid activation and subsequent maintenance of the SAC by promoting MPS1-dependent MAD1 kinetochore binding (Figure 6.1). Although not fully recognized, other feedback mechanisms involving MPS1 are easily envisioned. By promoting MCC formation and stability 90,91,151, MPS1 guarantees high Cyclin B-CDK1 activity, which in turn promotes MPS1 activity by direct phosphorylation 280. The involvement of MPS1 in MCC stability might also be subjected to feedback regulation, for instance by regulation of a phosphatase, p31 comet or the APC/C as mentioned above. Lastly, it has been shown that MPS1 promotes PLK1 kinetochore binding 91. Since in Chapter 4 we show that MCC stability requires PLK1 activity, it might thus be possible that an additional feedback loop regulates MCC stability. We suspect that the multitude of feedback loops in SAC regulation provides timely establishment of SAC signaling as cells enter mitosis, while also enabling the SAC to rapidly respond to kinetochore attachment status and ensuring rapid inactivation of the SAC once all kinetochores have attached. Although technically challenging, it will therefore be interesting to address whether feedback loops, like the Aurora B - MPS1 feedback, is essential to maintain SAC signaling when only few unattached kinetochores remain, while the same feedback might simultaneously function in the rapid extinguishing of SAC signaling once the last kinetochore has attached. Since MPS1 plays a central role in various SAC feedbacks, small alterations in its activity or expression might have drastic effects on genome stability and result in aneuploidy. Future studies investigating the consequence of minor alterations in MPS1 activity or expression on genome stability in cells and mouse models can provide more insight in the contribution of the SAC to chromosomal instability and cancer.

MPS1 Aurora B MAD1 ZW10 BUB1 KMN network PLK1 MCC

Figure 6.1. MPS1-centered feedback loops control the SAC
Schematic representation of various feedbacks that regulate SAC signaling and error correction. The first feedback loop between Aurora B and MPS1 (green) ensures timely establishment of both SAC signaling and error correction. The second feedback control between MPS1, BUB1 and ZW10 (blue) promotes MAD1 localization. The third, and more elaborate, feedback regulates MCC production and stability (red). In this feedback, MPS1 activity is required for MCC production through MAD1 localization while simultaneously ensuring PLK1 kinetochore localization, which is required for MCC stability when SAC signaling is low.