Causal Deconvolution of a Mechanistic Model of EGFR and ERK Signaling Explains Adaptive and Genetic Resistance in Melanoma
TimeFriday, 13 November 202012:30pm - 12:45pm EDT
DescriptionAllosteric interactions are at the core of many signal transduction processes and provide robustness and enable context dependency for the underlying molecular mechanisms. This is prominently captured by paradoxical activation, a clinically observed phenomenon where RAF inhibitors inhibit tumor growth in BRAF mutant cancers, but promote tumor growth in BRAF wild-type cancers. Energy based formalisms to describe such allosteric effects in kinetic models have been developed, but approaches to enable intelligibility of and address the computational complexity associated with such large, multi-scale models are currently missing.
Here we demonstrate the use of a programmatic, thermodynamic, energy-balanced rule-based formalism in PySB to describe allosteric interactions. We tackle the numerical challenges of large kinetic models by using and extending state of the art high performance computing simulation and calibration tools. To address the conceptual challenge of rendering large kinetic models intelligible, we introduce a novel approach to causally separate intertwined signaling channels.
We apply these methods to an ordinary differential equation model of adaptive resistance in melanoma (EGFR and ERK pathways, >1k state variables, >10k reactions), accounting for paradoxical activation. We trained the model on absolute proteomic and phospho-proteomic as well as time-resolved immunofluorescence data, both in dose-response to small molecule inhibitors. We deconvolve oncogenic and physiological causal paths to derive simple explanations for complex dose-response relationships, explain how synergy and antagonism can arise without direct drug interaction and establish a link between adaptive and genetic resistance in melanoma.