Centrosomes are frequently amplified in cancer cells, and this amplification has been proposed to contribute to tumorigenesis because cells with several spindles are genomically unstable. But what is the mechanism of amplification? Patrick Meraldi et al. report in The EMBO Journal that centrosome amplification is a consequence, rather than a cause, of defective division.

Overexpression of the mitotic kinase Aurora-A — which occurs in many cancers — causes centrosome amplification. Both wild-type Aurora-A and a kinase-dead (KD) mutant caused an increase in centrosome number when overexpressed in CHO cells, indicating that the kinase activity is not required for centrosome duplication. Addition of hydroxyurea (HU) — a drug that blocks cells in S phase, thereby allowing multiple rounds of centrosome duplication without DNA replication — also increased the percentage of cells with more than two centrosomes, but Aurora-A expression did not enhance this number above that of wild-type CHO cells.

So does Aurora-A induce centrosome amplification by mimicking HU and blocking cells in S phase? Overexpression of Aurora-A in HeLa cells — which are unable to duplicate their centrosomes during S-phase arrest — still increases centrosome number, but only in the absence of HU, indicating that Aurora-A overexpression does not induce S-phase arrest and that cells must pass through mitosis to increase their number of centrosomes.

Closer examination of the Aurora-A-overexpressing HeLa cells revealed that most cells (75%) with multiple centrosomes were also multinucleate, so could the increase in centrosome number be an indirect consequence of aberrant cell division? The number of multinucleate and tetraploid cells increases as Aurora-A is overexpressed, indicative of a cytokinesis defect. So can other defects that lead to an aberrant cytokinesis also cause centrosome amplification? Overexpression of the Aurora-B and PLK-1 mitotic kinases, which also result in multinucleate cells, leads to centrosome amplification, as does addition of the cytokinesis-inhibitory drug cytochalasin D.

Loss of the tumour suppressor p53 is frequently associated with centrosome amplification, so how might this be related to the proposed mechanism of a defect in cell division? Perhaps centrosomes accumulate in p53-null cells because they are unable to respond appropriately — by arrest or apoptosis — to a defective division. Consistent with this, a 66% correlation was observed between the frequency of multinucleation and centrosome amplification in Trp53 −/− mouse embryo fibroblasts (MEFs). Many of the mononucleate cells with more than two centrosomes also had large nuclei, indicating that they might be polyploid. Multinucleate Trp53−/− MEFs were also more likely to proceed through to S phase than multinucleate wild-type cells, and overexpression of the mitotic kinases caused centrosome amplification in 80% of Trp53−/− MEFs, compared with 25% of wild-type MEFs.

So defective cell division can result in tetraploid cells with a concomitant amplification of centrosomes. What is left to discover is whether this is the only mechanism of centrosome amplification in cancer cells, and the impact that centrosome amplification has on genomic instability and tumorigenesis.