Data-Driven Mechanistic Computational Modeling of Human Cell Cycle Phases for Defining Inhibitor Combinations in Silico

Abstract

The human cell cycle is a highly regulated process that integrates multiple signaling pathways and checkpoints to ensure faithful genome duplication and cell division. Disruptions in these regulatory networks contribute to a wide range of diseases. Here, we present a novel, updateable computational model of the full human cell cycle that shows sustained oscillations over time and reproduces experimental perturbations. We used a hybrid framework combining mass action and Michaelis-Menten kinetics, incorporating the synthesis, degradation, and regulation of key cell cycle proteins and protein complexes. It consists of 63 distinct biochemical species, interacting through 41 major reactions, and functioning through 63 ODEs. The model is built upon a modular framework, structured around the core regulatory networks of the G1, S, G2, and M phases. Due to its complexity, we determined parameter sets that met strict criteria, namely event timing, comparable concentrations, and continuous cycling. We validated the models behavior by reproducing canonical checkpoint responses, including mitogen dependence and the DNA damage response, both of which produced reversible and robust cell cycle arrests. Importantly, the model was trained and calibrated using in vitro data from human U251-MG glioma cells expressing the FastFUCCI cell cycle reporter. We quantitatively aligned the simulated and experimentally determined phase durations and cell doubling times. Next, we experimentally tested and refined model parameters by using abemaciclib-mediated inhibition of CDK4 and volasertib-mediated inhibition of PLK1. In vitro and in silico data show dose-dependent G1 arrest by abemaciclib and dose-dependent mitotic arrest by volasertib. Finally, we demonstrated that the model predicts changes in cell proliferation over a wide range of drug concentrations and combinations. Overall, our work establishes a robust, data-driven computational model for systems-level analysis of the human cell cycle and its disruption by therapeutic perturbations. AUTHOR SUMMARYKnowledge of the protein-protein interaction networks governing the cell cycle is ever-expanding, yet this information is often fragmented across studies focusing on disconnected subsets of the cycle. For decades, researchers have investigated the underlying mechanisms of cell division, but an integrated, quantitative understanding of the entire process remains elusive. This gap is a major hurdle for predicting how targeted therapies affect cell proliferation, especially when used in combination. Our goal is to develop an in silico simulation of the complete human cell cycle by integrating the key mechanistic relationships across all four phases into a single computational model with enough resolution to approximate outcomes upon perturbation. In achieving this, we have developed a novel, comprehensive computational model that provides an integrated quantitative understanding of how cancer drugs alter the human cell cycle. We have rigorously trained, calibrated, and validated the model by suitably estimating its parameter values to produce accurate cell cycle phase timing, using high-resolution, live-cell imaging data and other cardinal features of the cell cycle in a U251-MG glioblastoma line. This work provides an accessible tool for exploring how normal cell cycle control is disrupted in disease, generating new hypotheses, and identifying potential points of therapeutic intervention.