A pharmacodynamic model of the spindle assembly checkpoint and the action of the aurora kinase inhibitors

A defining characteristic of cancer is loss of control of accurate chromosome segregation resulting in aneuploidy, genetic instability and tumour progression, ultimately leading to invasive and metastatic phenotypes. A quantitative description of the underlying biological processes would be a valuable tool for developing chemotherapy that is selective against cells in which chromosome segregation is defective or damaged. We describe a mechanistic model of the mitotic spindle assembly checkpoint that incorporates the function of aurora kinase A, aurora kinase B and other drug target molecules. Correct cell division requires that the kinetochore on each sister chromatid is attached to and only to microtubules from one spindle pole. Incorrect attachments result when both kinetochores are attached to microtubules all from the same pole (syntelic attachment) or if at least one kinetochore is attached to both poles (merotelic attachment). Aurora-B can detect and dissociate these aberrant attachments. Our model also describes the process by which unattached kinetochores activate Mad2 and BubR1 into an inhibitor of cdc20, thereby generating a wait signal that prevents the cell from proceeding to anaphase before all kinetochores have been captured correctly. Aurora-B is also a key kinase in phosphorylating histone-H3 which appears during mitosis. Inhibition of aurora-B results in abrogation of the checkpoint and progression into premature mitosis, resulting in aneuploidy or polyploidy. These abnormalities are likely to lead to cell death. We have modelled the effects of selective inhibitors of aurora-A kinase and aurora-B kinase as well as non-selective inhibitors of aurora-A and B kinases including ZM447439. The model predicts that aurora-A inhibitors delay anaphase and that aurora-B inhibitors decrease time to anaphase. The model also predicts that inhibition of aurora-B will result in a large increase in the percentage of incorrect kinetochore-microtubule attachments, this is accompanied by a decreased level of phosphorylated histone H3. The model predicts that non-specific inhibitors will be more potent than a combination of selective inhibitors at the same dose. The biomarker phosphorylated histone H3, and the proportion of polyploid cells, have been used to provide experimental validation of the model. This model can be used to devise selective treatment strategies against tumors with specific mitotic defects, to suggest synergistic drug combinations and as part of a PK/PD modeling approach to optimise chemotherapy.