Multidimensional single-cell modeling of cellular signaling

Abstract

Cell-to-cell differences in signaling components can lead to qualitatively different responses to stimuli. Understanding this heterogeneity in signaling response is limited by the inability of time-lapse methods to measure multiple pathway components simultaneously in situ. Here, we present Distribution-Independent Single-Cell ODE modeling (DISCO), a computational method for inference of continuous single-cell signaling dynamics from multiplexed snapshot data. We used DISCO to analyze signaling in the MAPK/ERK pathway of HEK293T cells stimulated with the growth factor EGF. Our model recapitulates known features of the ERK signaling response and enables the detection of hidden cell-to-cell variation in seemingly homogeneous samples. Further, DISCO analysis suggested that the MAPK/ERK pathway transmits signal duration rather than amplitude, and that cell-to-cell variation in MAPK/ERK signaling response depends primarily on initial cell states. Finally, we applied an extended version of DISCO to explain changes in signaling kinetics due to overexpression of a disease-relevant protein. Overall, DISCO enables a deeper understanding of how single-cell variation affects cellular responses in complex signaling systems.