Prediction of phase separation propensities of disordered proteins from sequence

Abstract

AbstractPhase separation is thought to be one possible mechanism governing the selective cellular enrichment of biomolecular constituents for processes such as transcriptional activation, mRNA regulation, and immune signaling. Phase separation is mediated by multivalent interactions of biological macromolecules including intrinsically disordered proteins and regions (IDRs). Despite considerable advances in experiments, theory and simulations, the prediction of the thermodynamics of IDR phase behaviour remains challenging. We combined coarse-grained molecular dynamics simulations and active learning to develop a fast and accurate machine learning model to predict the free energy and saturation concentration for phase separation directly from sequence. We validate the model using both experimental and computational data. We apply our model to all 27,663 IDRs of chain length up to 800 residues in the human proteome and find that 1,420 of these (5%) are predicted to undergo homotypic phase separation with transfer free energies<−2kBT. We use our model to understand the relationship between single-chain compaction and phase separation, and find that changes from charge-to hydrophobicity-mediated interactions can break the symmetry between intra-and inter-molecular interactions. We also analyse the structural preferences at condensate interfaces and find substantial heterogeneity that is determined by the same sequence properties as phase separation. Our work refines the established rules governing the relationships between sequence features and phase separation propensities, and our prediction models will be useful for interpreting and designing cellular experiments on the role of phase separation, and for the design of IDRs with specific phase separation propensities.