Abstract
Anti-mitotic drugs constitute a major class of cytotoxic chemotherapeutics used in the clinic, killing cancer cells by inducing prolonged mitotic arrest that activates intrinsic apoptosis. Anti-mitotics-induced apoptosis is known to involve degradation of anti-apoptotic Bcl-2 proteins during mitotic arrest; however, it remains unclear how this mechanism accounts for significant heterogeneity observed in the cell death responses both within and between cancer cell types. To unravel quantitative determinants underlying variability in anti-mitotic drug response, we constructed a single-cell dynamical Bcl-2 network model describing cell death control during mitotic arrest, and constrained the model using experimental data from four representative cancer cell lines. The modeling analysis revealed that, given a variable, slowly accumulating pro-apoptotic signal arising from anti-apoptotic protein degradation, generation of a switch-like apoptotic response requires formation of pro-apoptotic Bak complexes with hundreds of subunits, suggesting a crucial role for high-order cooperativity. Moreover, we found that cell-type variation in susceptibility to drug-induced mitotic death arises primarily from differential expression of the anti-apoptotic proteins Bcl-xL and Mcl-1 relative to Bak. The dependence of anti-mitotic drug response on Bcl-xL and Mcl-1 that we derived from the modeling analysis provides a quantitative measure to predict sensitivity of distinct cancer cells to anti-mitotic drug treatment.