Inferring nonlinear dynamics of cell migration.

The motility of eukaryotic cells is strongly influenced by their environment, with confined cells often developing qualitatively different motility patterns from those migrating on simple two-dimensional substrates. Recent experiments, coupled with data-driven methods to extract a cell's equation of motion, showed that cancerous MDA-MB-231 cells persistently hop in a limit cycle when placed on two-state adhesive micropatterns (two large squares connected by a narrow bridge), while they remain stationary on average in rectangular confinements. In contrast, healthy MCF10A cells migrating on the two-state micropattern are bistable, i.e., they settle into either basin on average with only noise-induced hops between the two states. We can capture all these behaviors with a single computational phase field model of a crawling cell, under the assumption that contact with non-adhesive substrate inhibits the cell front. Our model predicts that larger and softer cells are more likely to persistently hop, while smaller and stiffer cells are more likely to be bistable. Other key factors controlling cell migration are the frequency of protrusions and their magnitude of noise. Our results show that relatively simple assumptions about how cells sense their geometry can explain a wide variety of different cell behaviors, and show the power of data-driven approaches to characterize both experiment and simulation.