Elucidating the Role of Filament Turnover in Cortical Flow using Simulations and Representation Learning

Cell polarization relies on long-range cortical flows, which are driven by active stresses and resisted by the cytoskeletal network. While the general mechanisms that contribute to cortical flows are known, a quantitative understanding of the factors that tune flow speeds has remained lacking. Here, we combine physical simulation, representation learning, and theory to elucidate the role of actin turnover in cortical flows. We show how turnover tunes the actin density and filament curvature and use representation learning to demonstrate that these quantities are sufficient to predict cortical flow speeds. We extend a recent theory for contractility to account for filament curvature in addition to the nonuniform distribution of crosslinkers along actin filaments due to turnover. We obtain formulas that can be used to fit data from simulations and microscopy experiments. Our work provides insights into the mechanisms of contractility that contribute to cortical flows and how they can be controlled quantitatively.