Comparing experimental and computational alanine scanning techniques for probing a prototypical protein-protein interaction

Abstract

The central role of protein-protein interactions in a wide range of cellular processes makes them a target for research and drug discovery. A variety of methods, both experimental and theoretical, exist for probing protein interfaces for residues that affect activity and binding affinity. Using as an example a protein-protein complex between trypsin and a nine-residue synthetic peptide, we experimentally assay-binding affinities for a variety of mutants and determine their relative free energy of binding, ΔΔG, to rank the importance of interface residues to binding. We then compare how accurately, precisely and reliably computational methods for calculating ΔΔG can replicate these results. We find that a 'post-process alanine scanning' protocol of a single native complex trajectory gives results with better accuracy than running separate molecular dynamics (MD) trajectories for individual mutants. Compared across 10 independent simulations, we find that results from the post-process alanine scanning are also more precise and are obtained over five times faster than their equivalent with the 'full MD' protocol. These results suggest that, although not suitable in every case, post-process alanine scanning is a useful and reliable tool in predicting important residues at protein interfaces with potential for modulation.