Abstract
The biophysical properties of ligand binding heavily influence the ability of receptors to specify cell fates. Understanding the rules by which ligand binding kinetics impact cell phenotype is challenging, however, because of the coupled information transfers that occur from receptors to downstream signaling effectors and from effectors to phenotypes. Here, we address that issue by developing an integrated mechanistic and data-driven computational modeling platform to predict cell responses to different ligands for the epidermal growth factor receptor (EGFR). Experimental data for model training and validation were generated using MCF7 human breast cancer cells treated with the high- and low-affinity ligands epidermal growth factor (EGF) and epiregulin (EREG), respectively. The integrated model captures the unintuitive, concentration-dependent abilities of EGF and EREG to drive signals and phenotypes differently, even at similar levels of receptor occupancy. For example, the model correctly predicts the dominance of EREG over EGF in driving a cell differentiation phenotype through AKT signaling at intermediate and saturating ligand concentrations and the ability of EGF and EREG to drive a broadly concentration-sensitive migration phenotype through cooperative ERK and AKT signaling. Parameter sensitivity analysis identifies EGFR endocytosis, which is differentially regulated by EGF and EREG, as one of the most important determinants of the alternative phenotypes driven by different ligands. The integrated model provides a new platform to predict how phenotypes are controlled by the earliest biophysical rate processes in signal transduction and may eventually be leveraged to understand receptor signaling system performance depends on cell context. ONE-SENTENCE SUMMARY: Integrated kinetic and data-driven EGFR signaling model identifies the specific signaling mechanisms that dictate cell responses to EGFR activation by different ligands.