Benchmarking Alchemical Relative Binding Free Energy Calculations for Nucleotide Binding to Multimeric ATPases.

Abstract

Multimeric ATPases generally bind nucleotides at intersubunit interfaces, where cooperative or allosteric interactions complicate experimental measurement of binding affinities. Here, we present a large-scale benchmarking study of alchemical relative binding free energy (RBFE) calculations using fixed-charge force fields across six classes of oligomeric ATPases: F1-ATPase, MalK, MCM, Rho, FtsK, and gp16. While previous absolute or relative binding free energy studies have largely focused on monomeric or single-site ATPases, this work extends RBFE calculations to 55 interfacial binding sites in multimeric ATPases, providing insight into the successes and limitations of alchemical free energy methods in complex, cooperative systems. RBFE simulations were conducted both in the presence and absence of the central substrate (DNA or RNA) to assess its impact on nucleotide-binding free energies. The highly charged and conformationally flexible nature of nucleotide ligands necessitated extensive sampling (>20 ns per alchemical window) to account for slow relaxation associated with long-range electrostatic interactions. Our RBFE results reproduced experimentally observed binding preferences for 91% of the sites in F1-ATPase, MalK, and MCM ATPases, which exhibited low global and local structural deviations during simulations across alchemical windows. In contrast, only 60% agreement was observed for Rho, FtsK, and gp16─systems with greater structural variability. This study not only highlights the predictive potential of alchemical free energy methods for nucleotide binding in protein complexes but also systematically identifies key sources of RBFE error, including structural fidelity, protein flexibility, ligand pose instability, and disruption of critical binding interactions during alchemical transformations. Furthermore, AlphaFold3 (AF3) was used to model a gp16 structure with higher structural stability than the available cryo-EM structure, and RBFE results indicate that the two models may correspond to distinct functional states (nucleotide-binding preferences) during the substrate translocation cycle.