Thrombin toggles between procoagulant fibrinogen cleavage and anticoagulant protein C activation when complexed with thrombomodulin (TM). The W215A/E217A (WE) double mutant displays a ∼19,000-fold reduction in fibrinogen cleavage yet only ∼7-fold loss in TM-dependent protein C activation, but the structural basis for this substrate bias has been unclear. Using 32 μs of all-atom molecular dynamics across four states (WT, WT+TM456, WE, WE+TM456) combined with RMSF profiling, unsupervised clustering (HDBSCAN; Amorim-Hennig), PCA-derived free-energy surfaces, and logistic-regression analysis of hydrogen-bond networks, we map the allosteric determinants governing thrombin’s specificity switch. Analyses reveal two recurrent catalytic-triad conformational ensembles: one associated with fibrinogen cleavage and another favoring protein C activation. WE and TM binding independently shift occupancy toward the protein C-competent state; in WE+TM, additional minor triad states emerge that rationalize the modest (≈7-fold) impairment of protein C activation. TM binding stabilizes exosite I and, in WE+TM, constrains the 60s loop to a predominantly “closed” ensemble, disfavoring fibrinogen recognition. The 170s and 220s loops undergo state-dependent rearrangements, and the γ-loop free-energy landscape is uniquely remodeled only when WE and TM act together, indicating synergistic allostery rather than a simple superposition of effects. A logistic-regression classifier trained on 509 hydrogen bonds (accuracy ≥97%) identifies THR277-ASP135 and LEU132-ARG216 as key conduits linking distal sites to the catalytic machinery and highlights a persistent ARG125-centered interaction across all systems as a robust allosteric node. These results provide a mechanistic explanation for the anticoagulant bias of WE thrombin: mutation- and TM-induced rewiring of long-range hydrogen-bond networks stabilizes a protein C-activating catalytic-triad ensemble while suppressing fibrinogen-competent states.
Salsbury et al. (Sun,) studied this question.