Recovering hidden information from live-cell trajectories to predict the biological pathway activity underlying cell state change decisions

Abstract

Extracellular signals induce changes in cellular phenotype, activating diverse cellular programs and decisions at the single-cell level-- proliferation or cell-cycle arrest, multicellular clustering or contact avoidance, and enhanced collective or individual motility. High-throughput, extended duration, and single-cell resolved trajectories can be obtained via live-cell imaging assays. Trajectory analysis indicates that the single-cell responses to perturbation from extracellular signals can be heterogenous dynamically in the transition pathways and transition times between cell states, as well as in the extent of overall response at the final timepoint. It is unknown the extent to which the dynamics of cell state transitions are intrinsically stochastic and random, or deterministic and thus predictable. We present a live-cell morphodynamical trajectory-based analysis which extracts “hidden” information present in extended single-cell trajectory windows, and indicates much of the observed stochasticity in cell state transition dynamics can be attributed to incomplete single-timepoint cell characterization, and not intrinsically random cell state change dynamics. Our analysis enables the extraction of information rich single-cell trajectories of cell state changes, enabling data-driven modeling and translation from live-cell imaging morphodynamical cell states to gene transcript levels. We map the biological pathway activity along transition intermediates as the cellular commitment to extracellular signal-induced cell state change occurs, investigating specific pathway activity gating the development of multicellular clusters, and the switch from individualized to collective motility.