Comparative dynamics of EcDHFR-NADPH-ligand complexes

Abstract#

Escherichia coli dihydrofolate reductase (EcDHFR) is a validated antibacterial target whose ligand recognition cannot be adequately described by docking scores alone, a limitation shared by any protein-ligand system of comparable complexity.

We present a multi-scale computational protocol designed to characterise EcDHFR-ligand complexes with physicochemical and electronic detail, while generating structured data suitable for downstream machine-learning parameterisation. Classical molecular dynamics assesses binding-mode stability, residue-level flexibility and the dominant conformational states of the binding pocket, including its response to structurally diverse ligands. PLUMED-driven metadynamics further probes pocket conformational transitions. MM-GBSA provides comparative binding energetics across selected systems. For representative trajectory frames, CP2K is used to examine the local electronic structure of the binding region, including atomic charge distribution, charge transfer and polarisation effects, phenomena systematically neglected by classical force fields.

Together, these analyses build a domain-informed dataset that links computed descriptors to experimentally known values, enabling iterative refinement of earlier-stage approximations in the pipeline. The methodology is developed for EcDHFR but is designed to generalise to homologous targets.

Protocol#

1. EcDHFR-NADPH-ligand systems.Representative trimethoprim analogues modelled in the cofactor-bound pocket for comparable trajectory analysis.
2. Classical dynamics.Trajectory descriptors capture binding-mode persistence, residue-level flexibility and recurring pocket states.
3. Enhanced sampling.Biased sampling probes conformational transitions that short unbiased simulations under-sample.
4. Energetics and electronics.MM-GBSA and CP2K expose effects omitted by docking scores and classical force fields.

Figures#