Cell state plasticity emerging from co-regulated, competitive, and configurable interactions within the AP-1 network

Cell state plasticity drives metastasis and therapy resistance in cancers. In melanoma, these behaviors map onto a melanocytic-to-mesenchymal-like continuum regulated by AP-1 transcription factors. However, how the AP-1 network encodes a limited set of discrete states, why their distributions vary across tumors, and what drives phenotypically consequential AP-1 state transitions remain unclear. We develop a mechanistic ODE model of the AP-1 network capturing their dimerization-controlled, co-regulated, competitive interactions. Calibrated to heterogeneous single-cell data across genetically diverse melanoma populations and combined with statistical learning, the model reveals network features explaining population-specific AP-1 state distributions. These features correlate with MAPK activity across tumor lines and with variability within clones, linking MAPK signaling to AP-1 states. The model predicts and experiments validate adaptive AP-1 reconfiguration under MAPK inhibition, inducing a dedifferentiated, therapy-resistant state that can be blocked by model-guided AP-1 perturbations. These results establish AP-1 as a configurable network and provide a computational framework for predicting and modulating AP-1 driven cell state plasticity.