Cell division, sometimes referred to as the cell-cycle, is one of the most fundamental and highly regulated processes in biology. The decision of a cell to divide is tightly controlled, integrating both extra- and intracellular signals. Despite the existence of complex control and surveillance systems guiding cell-cycle progression, malfunctions do occur as evidenced by the uncontrolled proliferation underlying many cancers. Not every malfunction in cell-cycle leads to cancer, thus, cell-cycle is robust to some malfunctions and sensitive or fragile to others. Stelling et al., have proposed that points of fragility in a network can be computationally identified by performing sensitivity analysis upon mathematical models. Our working hypothesis is that a single-cell mathematical model of cell-cycle, which mechanistically describes the molecular events controlling cell division, could be used to computationally determine points of network fragility that correspond to clinically observed cell-cycle malfunctions in solid and hematological malignancies. Knowledge of fragility could aid in the basic understanding of cancer progression and could serve as the basis of therapeutic strategies. We test our working hypothesis by analyzing published models of cell-cycle progression and compare our computationally identified fragile points with literature. Three different published cell-cycle models of varying complexity were analyzed; the skeleton mammalian cell-cycle model of Novak and Tyson, the G1/S model of Qu et al., and the G2/M DNA-damage model of Aguda. The cell-cycle models were implemented in Matlab v7.01 (The Mathworks, Natick MA) and solved using the ODE15s routine. Three different numerical schemes were used to solve the sensitivity equations; forward finite difference and two backward difference techniques. To capture the effect of parametric uncertainty and its impact upon the predicted points of fragility, N = 500 independent parameter sets were generated by randomly perturbing the published parameter sets by up to 2 orders of magnitude. Overall State Sensitivity Coefficients (OSSC) were calculated for each parameter set. Of the top 15 fragile points (top five from each model), 14/15 are known control or clinically observed points where malfunctions have occurred. These include global elements like translational efficiency and specific interactions like the programmed proteolysis of cyclins by the Ubiquitin Proteasome System (UPS), phosphorylation and dephosphorylation of CDC25A in G1/S or the synthesis of 14-3-3σ proteins. In addition, overlap between models was found indicating the prediction of fragility was not model specific. When taken together, the fragility results indicate that it may be possible to computationally screen for sites of probable malfunction, despite model uncertainty.

Extended Abstract Status: Not Uploaded