Survivin is having an identity crisis. Is it an inhibitor of apoptosis ( IAP), as suggested by its baculoviral IAP repeat ( BIR) domain? Or is it necessary for cell division, like its closest relations in yeast and worms? Reporting in Proceedings of the National Academy of Sciences, Daniel O'Connor and colleagues present evidence that these two functions need not be mutually exclusive, whereas Anthony Uren and co-workers come down firmly on the side of a mitotic function for survivin.

Localization of survivin (green; appears pale blue when colocalized with DNA and yellow when colocalized with tubulin), tubulin (red) and DNA (blue) at different stages of the cell cycle. Courtesy of Lee Wong and K. H. Andy Choo, The Murdoch Children's Research Institute, Parkville, Victoria, Australia.

O'Connor et al. found that survivin is unique among IAPs in that it has a consensus sequence for phosphorylation by the cyclin-dependent kinase CDC2, and is phosphorylated by CDC2 in vitro and in vivo. They could immunoprecipitate survivin phosphorylated on threonine 34 (T34) only from cells undergoing mitosis. Survivin could also be co-immunoprecipitated with CDC2, and this interaction doesn't depend on phosphorylation at T34 because it worked just as well when T34 was mutated to alanine (T34A). This suggests that the T34A mutant might act as a dominant-negative inhibitor. Sure enough, when T34A was overexpressed, mitotic cells died by apoptosis; but how does survivin prevent death during mitosis? The apoptotic protease caspase-9 could also be found in survivin immunoprecipitates, but the T34A mutant didn't associate with caspase-9, suggesting that this interaction, whether direct or indirect, requires phosphorylation of T34.

Uren and colleagues used antibodies to track survivin's behaviour thoughout the cell cycle, and found that survivin's movements closely mimicked those of a group of proteins known as chromosome passenger proteins. These hitch a ride on the centromeres to the spindle equator, where they remain until sister chromatids separate (see picture). Survivin bound to centromeres along the same axis as the inner centromere protein INCENP, which is needed to localize Aurora1 kinase to centromeres. To get a handle on what survivin might be doing at centromeres, the authors knocked it out in mice. At first glance, knockout embryos looked normal until embryonic day 4.5, but then looked irregular, with cells that failed to separate and disorganized mitotic spindles. By day 5.5, the knockout embryos contained an average of only 13 nuclei, compared with around 200 in wild-type embryos.

Is survivin a death defier, an orchestrator of division or something in between? We're still far from an answer but, whatever the final verdict, one thing is clear: dividing cells can't manage without it.