A Computational Framework for the Determination of the Binding Affinity of Peripheral Proteins to Homogeneous Membranes

Abstract Peripheral membrane-binding proteins (PMPs) are essential to immunological and neurological processes, and disruptions in their membrane interactions are implicated in cancer, neurodegeneration, and autoimmune diseases. Their functional diversity arises from the wide range of mechanisms by which they associate with lipid surfaces, motivating atomistic computational frameworks capable of quantifying binding energetics. We present such a methodology and demonstrate it using Bacillus thuringiensis phosphatidylinositol-specific phospholipase C (BtPI–PLC), a 296-residue virulence factor that preferentially associates with phosphatidylcholine (PC) membranes. Experimental studies report dissociation constants near 24 μM for wild-type BtPI–PLC binding to PC vesicles, corresponding to a standard free energy of −6.6 kcal/mol. BtPI–PLC associates with membranes through cation−π interactions involving Y88 and Y246, hydrophobic contacts from helix B, and other noncovalent interactions, making comprehensive sampling challenging. Here, we present a statistical mechanical geometrical route formulation based on the potential of mean force for the quantitative characterization of dilute PMP binding to homogeneous membranes. Using enhanced sampling via well-tempered metadynamics and extended adaptive biasing force, we obtain a computational estimate of the standard binding free energy, −6.2 ± 0.3 kcal/mol, consistent with experimental estimates for BtPI–PLC membrane association.