Systematic Modeling of Covalent Inhibitors of SARS-CoV-2 Main Protease and Its Mutants

The SARS-CoV-2 main protease (Mpro) is a key antiviral target due to its essential role in viral replication and the absence of close human homologues. Covalent inhibitors, which form durable bonds with critical active-site residues have shown particular promise for viral proteases, combining high potency with prolonged suppression of enzymatic activity. Among these, the aldehyde-based inhibitor H102 exhibits nanomolar inhibition of Mpro (IC50 = 8.8 nM), outperforming other reported inhibitors such as GC376 and the clinically approved nitrile-based inhibitor PF-07321332. Despite its potency, the threat of viral mutations raises concerns about the robustness of covalent inhibitors against drug resistance. In this study, we employ ab initio and empirical valence bond (EVB) simulations to elucidate the covalent binding mechanism of H102, revealing that a water-assisted concerted nucleophilic attack and proton transfer (PT-NA) dominates the reaction pathway via a single transition state, with a calculated overall activation free energy of ∼20.5 kcal/mol, in good agreement with experimental kinetics (ΔG⧧ = 21.3 kcal/mol). Although the crystal structure of H102 bound to Mpro displays an unusual distortion of the catalytic dyad, in which the P2 benzyl group forms a “sandwich-like” arrangement between Cys145 and His41, our simulations indicate that this distortion arises after covalent bond formation. The catalytic dyad remains structurally intact during the reaction, suggesting that the crystallographic pose reflects post-reaction conformational relaxation rather than the reaction mechanism itself. Furthermore, we successfully predict the absolute binding free energies (ABFEs) of H102 and its analogues across wild-type and mutant Mpro variants, accurately reproducing experimental affinities and providing quantitative insight into how mutations impact inhibitor potency. Collectively, these findings not only elucidate the mechanistic basis of aldehyde-based covalent inhibition but also highlight the utility of ABFE-guided computational strategies in the rational design of potent, mutation-resilient antiviral therapeutics.