Cell division is a dangerous business — every chromosome kinetochore must attach to the mitotic spindle for the two chromatids to be segregated correctly to opposite poles of the cell. The integrity of this process is carefully monitored by the spindle-assembly checkpoint, and it has been suggested that a defective checkpoint could lead to aneuploidy, as found frequently in human cancer. However, mutations in the checkpoint genes BUB1 , BUBR1 and MAD2 are rarely found. Ashok Venkitaraman and colleagues now suggest that cancer cells might use an alternative mechanism for deregulating the checkpoint — amplification of AURORA-A .

Spindle formation is defective when AURORA-A is overexpressed. Top panel is the the control image, the bottom panel is with overepxression of AURORA-A. DNA is stained red and microtubiles are stained green.

The AURORA-A serine-threonine kinase is overexpressed in a significant number of epithelial cancers, often as a result of gene amplification. A similar protein in budding yeast — Ipl1 — regulates the attachment of chromosomes to the mitotic spindle. Reasoning that this process might become dysregulated if AURORA-A were over expressed, the authors created mouse embryo fibroblasts or human HeLa cells in which AURORA-A was overexpressed to similar levels as found in human cancers. They found that this causes several abnormalities in mitosis that are consistent with defects in chromosome-spindle attachment. Chromosomes fail to align at the metaphase plate in 59% of mitoses, and 11% of cells have abnormal, often multipolar, spindles to which chromosomes do not properly attach (see figure).

Usually, problems with spindle attachment would be expected to activate the spindle checkpoint, leading to arrest at the metaphase–anaphase transition. The authors tested checkpoint activation by examining the localization of the MAD2 protein. Normally, this protein remains at the kinetochores as long as the checkpoint is active in metaphase, but is re-distributed when the checkpoint is switched off to let cells enter anaphase. However, the authors found that MAD2 behaves abnormally when AURORA-A is overexpressed in HeLa cells. It remains at the kinetochores in 80% of cells in anaphase, when it should have disappeared, indicating that anaphase occurs despite the continued activation of the spindle checkpoint.

As noted previously by several laboratories, one consequence of AURORA-A over expression is the accumulation of polyploid cells. Could this be related to the spindle checkpoint problems that are induced by increased AURORA-A expression? The authors show that after passing through anaphase, cells that over express AURORA-A complete nuclear division, but not cytokinesis, leading to multi-nucleation and polyploidy. But co-expressing a mutant BUB1, which inactivates the spindle checkpoint, alleviated these defects, indicating that they are, indeed, linked to the ability of AURORA-A to 'override' the checkpoint at the metaphase-anaphase transition.

Cancer chemotherapeutics, such as paclitaxel (Taxol), induce metaphase arrest and apoptosis because they interfere with microtubule dynamics and hence the mitotic spindle. So, does AURORA-A amplification affect the response to this drug? Interestingly, overexpression of AURORA-A causes an increase in the resistance to paclitaxel-induced apoptosis, indicating that it might contribute to drug resistance in cancer patients who have this amplification.

So, AURORA-A amplification — a frequent occurrence in human tumours — can contribute to genomic instability by interfering with correct kinetochore attachment to the spindle and activation of the spindle checkpoint. The mechanism by which this occurs is unclear at present, but the implication of this result for cancer treatment by paclitaxel certainly warrants further investigation.