Drug-tolerant persisters to TRAIL emerge from a dose-dependent surface in a cell-state continuum of sensitivity.

Abstract

Clonal cancer cells show heterogeneous responses to cytotoxic drugs, raising the question of whether this variability reflects discrete phenotypes or a continuum of underlying cell states. We address this by quantifying single-cell caspase-8 activation dynamics after TRAIL treatment and developing an extended mechanistic model of the extrinsic apoptosis pathway that incorporates c-FLIP-mediated control of initiator caspase activation. Fitting this model to individual trajectories across multiple doses recovers cell-specific procaspase-8 and c-FLIP abundances, together with three kinetic parameters, and reproduces the full diversity of observed responses. Embedding these inferred parameters into a shared state space reveals that sensitive and tolerant outcomes do not correspond to discrete subpopulations. Instead, a single biochemical pathway generates a continuous distribution of cell states whose position at treatment determines fate. Linking each trajectory to its early activation rate identifies a dose-dependent hyperplane that partitions this landscape into apoptotic and tolerant regions. Increasing drug dose translates this decision surface predictably, altering the outcome only for cells positioned near the boundary. This geometric perspective explains fractional killing in clonal populations and shows how drug-tolerant persister cells can arise from reversible variation in cell state. It further suggests that shifting state-space distributions relative to the decision surface may offer new strategies to limit persistence.