BPS2026 – De novo design of intrinsically disordered desiccation chaperones

Folded proteins can increasingly be designed using deep learning based on protein structures, but it is not clear how this could be extended to intrinsically disordered proteins. Here, we develop a procedure for empirical and computational design of an intrinsically disordered desiccation chaperone, a class of proteins with applications in dry formulation of protein pharmaceuticals. We use a biochemical assay to screen for chaperone activity in a library of synthetic intrinsically disordered proteins. We titrate key sequence parameters, such as charge, hydrophobicity, and aromaticity alone or in combination to identify sequence-function relationships in desiccation chaperones. This reveals that chaperone activity arises readily in disordered proteins but is connected to specific areas of sequences space. To enable rational design of desiccation chaperones, we study selected synthetic chains using in vitro biophysics, coarse-grained simulations and cellular desiccation assays. In combination these suggest a molecular mechanism for how desiccation chaperones work and point toward computational methods for design of intrinsically disordered chaperones.