Abstract
Membraneless organelles (MLOs) are assemblies of biomolecules that function without a dividing lipid membrane in a cellular environment. These MLOs, termed biomolecular condensates, are commonly formed by the thermodynamic process of liquid-liquid phase separation and assembly of large numbers of proteins, nucleic acids, and co-solvent molecules. Within MLOs, certain biomolecule types are particularly causative of phase separation and are termed "scaffolds" as they provide the major driving forces for self-assembly. Other molecules that are present in a condensate but are less causative than the scaffold molecules are termed "clients." Much effort has recently decoded many of the molecular interactions underlying liquid-liquid phase separation in search of predicting equilibrium concentrations and materials properties of condensates. In this work, we provide a simple computational approach that may predict the partitioning of globular protein clients into condensates primarily composed of disordered protein scaffolds. Specifically, we use multiple methods to calculate hydration free energy of a series of globular green fluorescent protein variants and find that hydration free energy is relatively well-correlated with the partition coefficient of these proteins into FG nucleoporin condensates. We then provide a comparison of different hydration free energy predictors and discuss why some may provide a more accurate prediction of partitioning. Finally, we discuss the shortcomings of hydration free energy as a predictor by identifying other possible confounding factors such as specific interactions, charge matching, and differential solvation inside a condensate, which will aid in making more robust predictions in future studies trained on more diverse data sets.