br Experimental br Acknowledgements This work

Experimental
Acknowledgements This work was financially supported by Ministry of Knowledge and Economy (10032113) and Ministry of Education, Science and Technology (20100000297).
Introduction To ensure accurate chromosome segregation, cohesion between sister chromatids must be maintained until its release in a concerted fashion at anaphase. During mitosis in vertebrates, cohesin removal occurs in two steps (Losada et al., 1998, Waizenegger et al., 2000). In prophase and prometaphase, cleavage-independent release of cohesin from chromosome arms takes place under the control of Polo-like kinase 1 (Plk1) and Aurora B, while cohesion at centromeres and a few loci on the arms is maintained (Gimenez-Abian et al., 2004, Losada et al., 2002, Sumara et al., 2002). The protease Separase can cleave the Rad21/Scc1 cohesin subunit (Hauf et al., 2001), but it is kept inactive in early mitosis by two inhibitory factors: association with Securin and phosphorylation by Cyclin B/Cdk1. A ubiquitin ligase, the Ifenprodil Tartrate promoting complex/cyclosome (APC/C), promotes degradation of Securin and Cyclin B at the metaphase-anaphase transition, and Separase is activated. The resulting release of cohesin triggers the separation of sister chromatids (Nasmyth, 2002). The mechanisms that act to sustain centromeric cohesion in the face of the prophase removal pathway remain poorly defined. Recently, Shugoshin (Sgo1) has been proposed to be a protector of centromeric cohesin. Repression of Sgo1 by RNA interference (RNAi) leads to premature loss of cohesion between chromatids and arrest at prometaphase with misaligned chromosomes (Kitajima et al., 2005, McGuinness et al., 2005, Salic et al., 2004, Tang et al., 2004). The centromeric localization of Sgo1 relies on Bub1 (Kitajima et al., 2004, Kitajima et al., 2005, Riedel et al., 2006, Tang et al., 2004). After Bub1 repression, Sgo1 is relocated along the length of the chromosomes and appears then to prevent the dissociation of cohesin from the arms (Kitajima et al., 2005). Sgo1 may act in part by promoting the activity of the phosphatase PP2A at centromeres, counteracting the effects of mitotic kinases by preventing phosphorylation of the cohesin subunit Scc3/SA2 (Kitajima et al., 2006, Riedel et al., 2006, Tang et al., 2006). We previously showed that the kinase Haspin associates with chromosomes during mitosis. Phosphorylation of Haspin itself and its substrate histone H3 threonine-3 (H3T3ph) occurs specifically during mitosis. Haspin RNAi causes an accumulation of cells in which partial metaphase plates are present, but many chromosomes appear stranded near the spindle poles (Dai et al., 2005). In this study, we reveal that an important defect underlying this phenotype is a failure of mitotic chromosome cohesion. Our results indicate that Haspin is a positive regulator of centromeric cohesion. We also suggest that the complex influence of Aurora B on cohesion may be explained by its effect on the localization of Sgo1.
Results
Discussion Our studies reveal that Haspin is vital to maintain chromosome cohesion during mitosis. Other proteins required for cohesion in mitosis have been identified, including Sgo1, Bub1, CENP-F, and Soronin (Holt et al., 2005, Kitajima et al., 2004, Kitajima et al., 2005, Losada et al., 2002, McGuinness et al., 2005, Rankin et al., 2005, Salic et al., 2004, Tang et al., 2004), and the kinases Plk1 and Aurora B are known to be involved in cohesin removal (Gimenez-Abian et al., 2004, Losada et al., 2002, McGuinness et al., 2005, Sumara et al., 2002). The kinase Haspin, however, appears to be critical for protecting centromeric cohesion during mitosis. As with other treatments that disrupt cohesion, such as Sgo1 or Scc1 deletion or depletion (Hoque and Ishikawa, 2002, Kitajima et al., 2004, Kitajima et al., 2005, McGuinness et al., 2005, Salic et al., 2004, Sonoda et al., 2001, Tang et al., 2004, Vagnarelli et al., 2004, Vass et al., 2003), depletion of Haspin leads to loss of cohesin association, premature chromatid separation, activation of the spindle assembly checkpoint, and a block in mitosis in a prometaphase-like state. Our results indicate that both Haspin and Sgo1 are required to maintain centromeric cohesion prior to anaphase (Figure 4E). While the Sgo1 RNAi phenotype is more penetrant than that of Haspin RNAi, it is difficult to determine whether the two proteins are equally depleted in such experiments. The overall similarity of the effects of depleting Haspin and Sgo1 suggests that they might act in a common pathway. Haspin depletion, however, does not have a dramatic effect on Sgo1 localization, and Sgo1 depletion does not appear to prevent Haspin action. To date, we have also been unable to detect phosphorylation of Sgo1 by Haspin in vitro (unpublished data). Indeed, our results suggest that in some experimental circumstances, such as when Haspin is overexpressed or Aurora B is depleted, Haspin and Sgo1 can act independently to stabilize cohesion on chromosome arms. It remains to be seen if this reflects the normal situation in vivo. One can envision a variety of other mechanisms by which Haspin might regulate cohesion. For example, Haspin may directly phosphorylate a subunit of cohesin, making it resistant to cleavage-independent removal. Alternatively, the overlapping distributions of H3T3ph and cohesin suggest that Haspin could act via histone modification to influence cohesion. We have been unable to demonstrate any effect of threonine-3 phosphorylation on cohesin binding to H3 peptides in vitro (unpublished data), but this simplified experimental system does not account for the potential contribution of other proteins and histone modifications that are present in cells. It is also possible that Haspin has an indirect role in regulating cohesion. Cohesion loss occurring after Haspin RNAi could be a consequence of disrupted centromere structure, for example, although the ability of Haspin overexpression to stabilize cohesion is suggestive of a more direct effect. Whatever the case, it seems unlikely that protection of centromeric cohesion can be described simply by the opposing action of kinases and phosphatases on cohesin at the centromere.