MODELING OF CELL CYCLE CHECKPOINTS WITH APPLICATIONS TO THE ANALYSIS OF INTERMITOTIC TIME DATA

Abstract

The mammalian cell can be thought of as an information processing unit, much like a computer. External stimuli initiate very complex networks of interacting proteins. When activated, these different signaling networks result in varying cell fate decisions such as cell growth, division, differentiation, and cell death. These signaling networks are subject to randomness, so that cell fate decisions are variable. For example, in a population of homogeneous cells, the time between two successive mitotic events (cell divisons), or intermitotic time (IMT), is subject to considerable, seemingly random, variation. To determine the potential sources of variability in IMT, we first use an existing model of the Rb-E2F network, which controls cell cycle entry at the restriction point. A network perturbation analysis is performed on the model to determine which part or parts of the network contribute to temporal variability in cell cycle entry. Network perturbation reveals that regulation of the $Rb$ node contributes to variability in IMT. This analysis is complemented by the development and application of numerical methods for the analysis of IMT data. Specifically, we apply multi-part stochastic models to the study of IMT data, develop and test procedures for performing maximum likelihood estimation of model parameters, explain how the random variables associated with the model can be linked to intracellular protein concentrations, and analyze IMT variability within our model framework. Model selection theory is then used to determine which model performs best from a set of candidate models. Our results show that the cell cycle is best conceptualized as a two-part stochastic process. Collectively, the presented approaches provide a greater understanding of how mammalian cells process information and the noise sources involved in temporal variability in cell cycle entry.