A Computational Investigation of Small-Molecule Engagement of Hot Spots at Protein–Protein Interaction Interfaces

Abstract

The binding affinity of a protein–protein interaction is concentrated at amino acids known as hot spots. It has been suggested that small molecules disrupt protein–protein interactions by either (i) engaging receptor protein hot spots or (ii) mimicking hot spots of the protein ligand. Yet, no systematic studies have been done to explore how effectively existing small-molecule protein–protein interaction inhibitors mimic or engage hot spots at protein interfaces. Here, we employ explicit-solvent molecular dynamics simulations and end-point MM-GBSA free energy calculations to explore this question. We select 36 compounds for which high-quality binding affinity and cocrystal structures are available. Five complexes that belong to three classes of protein–protein interactions (primary, secondary, and tertiary) were considered, namely, BRD4•H4, XIAP•Smac, MDM2•p53, Bcl-xL•Bak, and IL-2•IL-2Rα. Computational alanine scanning using MM-GBSA identified hot-spot residues at the interface of these protein interactions. Decomposition energies compared the interaction of small molecules with individual receptor hot spots to those of the native protein ligand. Pharmacophore analysis was used to investigate how effectively small molecules mimic the position of hot spots of the protein ligand. Finally, we study whether small molecules mimic the effects of the native protein ligand on the receptor dynamics. Our results show that, in general, existing small-molecule inhibitors of protein–protein interactions do not optimally mimic protein–ligand hot spots, nor do they effectively engage protein receptor hot spots. The more effective use of hot spots in future drug design efforts may result in smaller compounds with higher ligand efficiencies that may lead to greater success in clinical trials.