Mapping Functional Transitions across Protein Homologues from a Single Simulation: Application of gEDES-HM to Adenylate Kinases

Abstract Proteins fluctuate across different conformational states, which constitute a dynamic ensemble that regulates their interaction networks and functions. Predicting the conformational landscape remains a computational challenge, and even more difficult is capturing the link between state distributions and specific functions within protein homologues. Here, we present gEDES-HM (generalized Ensemble Docking with Enhanced sampling of pocket Shape-Homology Modeling), a scalable computational protocol enabling us to remap conformational ensembles obtained from enhanced sampling simulations of a single protein to its homologous family members, removing the need to perform individual simulations for all of them and thus speeding up the whole process by orders of magnitude. Specifically, we exploit our recent gEDES protocol to bias through metadynamics simulations the shape and volume of the binding site(s) of a reference protein. Next, a number of maximally diverse conformations of the protein are extracted from a multistep cluster analysis and used together as a multitemplate library for comparative modeling of the conformational ensemble of each homologue in the pool. This enables the functional conformational ensemble of one protein to be transferred to its homologues and used, for example, in restrained ensemble docking. We validated the protocol on 22 homologues of the E. coli adenylate kinase (hereafter, ADK), corresponding to 26 protein-ligand complexes and spanning 31% to 99% sequence identity to the reference. Our protocol retrieves a bound-like conformation of the protein binding site (corresponding to experimental substrate/inhibitor-bound states) for almost all targets, which allows reproducing native-like binding modes for 92% of the investigated complexes in ensemble docking calculations. Validated so far within a single family, gEDES-HM could provide a practical route to generate binding-competent structural libraries for homologues lacking experimental structures suitable for downstream structure-based modeling.