A dynamical-nonequilibrium molecular dynamics (D-NEMD) alanine scanning approach for identifying allosteric positions in proteins

Allosteric effects are widespread in proteins, but predicting the impact of sequence substitutions at positions distant from ligand-binding or enzyme active sites remains challenging, as their effects are mediated through complex dynamical networks. Pinpointing such positions experimentally is labour-intensive and low throughput. Equilibrium molecular dynamics (MD) simulations are useful: analysis methods can identify functional distal sites and networks, while free energy simulations can predict effects on stability or binding, given sufficient sampling; such simulations are typically time-consuming, requiring extensive simulation of each individual mutant. Here, we introduce a dynamical-nonequilibrium MD (D-NEMD) protocol using alanine substitutions as targeted perturbations to identify distal positions that influence enzyme function. Using the well characterised class A β-lactamase KPC-2 as a test system, we show that D-NEMD distinguishes functional from non-functional sites based on whether substitutions generate structured, long-range responses that reach the active site. KPC-2 is clinically important due to its broad substrate spectrum and the emergence of resistance-associated point mutations, many of which act through non-local effects on catalytic residues. Alanine substitutions at positions 179 and 164, which disrupt the Ω-loop salt bridge, trigger persistent structural responses propagating through the protein and reach both catalytic residues and the oxyanion-hole backbone. Likewise, alanine substitution at position 220 elicits pronounced responses extending to the active-site β-sheet and key loops, consistent with its known role in substrate specificity. In contrast, substitution at position 276—mutations of which have negligible kinetic impact—produces only local displacements with minimal propagation and no effect on catalytically relevant regions. These differential response patterns align with experimentally observed resistance phenotypes. D-NEMD therefore provides a fast, generalisable, and predictive approach for identifying allosterically connected distal positions. The protocol complements equilibrium MD, is straightforward to implement, and offers a tractable route for prioritising candidate sites for mechanistic study or future mutational scanning efforts.