PerspectiveMolecular Biology

Surfing Chromosomes (and Survivin)

See allHide authors and affiliations

Science  08 Oct 2010:
Vol. 330, Issue 6001, pp. 183-184
DOI: 10.1126/science.1197261

Mothers are generally inclined to provide each of their offspring with similar opportunities. Even a mother cell abides by providing her daughters with an equal dowry of chromosomes. Diversions from fairness create imbalances in chromosome numbers that affect cell viability and are associated with tumor formation, infertility, and birth defects. Three papers in this issue—by Yamagishi et al. on page 239 (1), Wang et al. on page 231 (2), and Kelly et al. on page 235 (3)—describe how an enzyme (Aurora B protein kinase) that is essential for accurate chromosome inheritance is recruited to chromosomes that are poised to segregate into daughter cells.

Chromosomes attach to the microtubules of a spindle apparatus that separates them during cell division (mitosis), a process that is error prone. By reversibly modifying target proteins with a phosphate group on serine or threonine residues, Aurora B acts as a fidelity factor (so-called checkpoint) for this process. Inhibiting Aurora B leads to errors in chromosome attachment to the spindle, and to premature exit from mitosis, which is indicative of checkpoint failure (4).

Current models posit that a centromeric pool of Aurora B corrects attachment errors. A centromere is a chromosomal region that specifies the formation of a kinetochore, a complex protein assembly devoted to spindle capture (see the figure). Phosphorylation of substrates at the centromere and kinetochore by Aurora B may control the “fluidity” of the kinetochore-spindle interface, preventing premature stabilization of the interaction. This would allow resolution of improperly formed connections.

Kinase recruitment.

A cascade of molecular interactions and phosphorylation (P) events occur at the centromere and inner kinetochore of sister chromatids that are attached to the mitotic spindle (microtubules), poised to segregate. These include recruitment of the kinases Haspin and Bub1, as well as the CPC complex (which harbors the Aurora B kinase). These events control the normal segregation of chromosomes during cell division.


The signals that attract Aurora B to the centromere had been unknown. Now, Yamagishi et al., Wang et al., and Kelly et al. report that a conserved protein kinase, Haspin, is crucial for this process in three distinct model organisms, including humans. Haspin is an atypical protein kinase that phosphorylates threonine 3 of histone H3 (P-Thr3-H3). Kelly et al. show that Survivin, a protein that binds to Aurora B in the chromosome passenger complex (CPC), contains a bipartite binding site that recognizes the phosphate moiety of P-Thr3-H3 and also an alanine residue at the amino terminus of histone H3. Thus, the CPC may be recruited to mitotic centromeres by interacting with P-Thr3-H3. But what determines the localization of this modification mark specifically at centromeres? Yamagishi et al. observed that Haspin interacts with Cohesin, a protein complex required for sister chromatid cohesion. Because the bulk of Cohesin resides at centromeres during mitosis, P-Thr3-H3 may be generated at these chromosomal regions through the interaction of Haspin with Cohesin.

An unresolved puzzle concerns the mechanisms through which Aurora B gains access to its substrates. Specifically, Aurora B targets proteins at the interface of the centromere and the inner kinetochore, including Cenp-A, mitotic centromere-associated kinesin (MCAK), Ndc80, and Knl1 (4). The ability of Aurora B to target kinetochore substrates is inversely proportional to the degree of tension impinging on kinetochores (5, 6). Kinetochore tension arises from the binding of microtubules. Tubulin polymers are the main constituent of the mitotic spindle and are important for discriminating correct from incorrect kinetochore-spindle attachments (7). Tension stretches kinetochores, so that certain kinetochore constituents become separated by ∼30 to 50 nm (8). As the kinetochore stretches, kinetochore substrates may be put beyond the reach of Aurora B that is localized at the centromere, leading to the stabilization of attachment and to checkpoint satisfaction. It is unclear how kinetochore tension causes a progressive separation of centromeric Aurora B from its kinetochore substrates (8).

Yet, the effects of tampering with Haspin-mediated localization of Aurora B to the centromere are not as pervasive as one might expect. Wang et al. show that at the interface between the centromere and inner kinetochore, the extent of phosphorylation of Cenp-A was not appreciably affected by depletion of P-Thr3-H3 in mammalian cells. By contrast,the recruitment of MCAK to centromeres and kinetochores, which depends on phosphorylation by Aurora B, was blocked. Cenp-A and MCAK are considered bona fide substrates of Aurora B. The strong difference in Cenp-A and MCAK phosphorylation upon Aurora B mislocalization is surprising. Perhaps the degree of substrate phosphorylation is determined by selective depletion or retention of specific pools of Aurora B. For example, there may be a diffusible pool of the kinase that does not require centromere recruitment to phosphorylate Cenp-A. Alternatively, there may be specific centromeric pools of Aurora B, each recruited through a distinct modification to distinct targets.

Through studies in human and fission yeast cells, Yamagishi et al. have implicated a second mark, phosphorylated threonine 120 of histone H2A (P-Thr120-H2), in the recruitment of the CPC to centromeres (9, 10). This mark, created by the protein kinase Bub1, is proposed to recruit Shugoshin proteins (hSgo1 and hSgo2 in humans). These, in turn, may recruit components of the CPC (Survivin or Borealin), provided that they have been previously phosphorylated by cyclin-dependent kinase 1 (CDK1) (10). Bub1 and Sgo2, like Aurora B, are required for MCAK recruitment to the centromere and inner kinetochore in human cells (11), in agreement with the hypothesis that these proteins control Aurora B localization and thus MCAK recruitment. However, there are contradictory reports in the role of Bub1 in Aurora B localization in human cells (12, 13). Furthermore, previous studies showed that Aurora B acts upstream, not downstream, of Shugoshin proteins in human cells (11, 14). Thus, further clarity is needed on the role of Bub1 and Shugoshin in Aurora B recruitment.

There likely are additional needles in the CPC protein modification haystack. Dephosphorylation of Incenp (Sli15p in Saccharomyces cerevisiae), another component of the CPC, is crucial for removing the CPC from the centromere and relocalizing it to the spindle after chromosomes have segregated to opposite poles of the dividing cell (15), suggesting yet additional mechanisms for centromere recruitment or retention of the CPC. Experiments centered on reconstituting the CPC and its interactions may be ultimately necessary to illuminate the path and make sense of this complexity.


View Abstract

Navigate This Article