Molecular Mechanics-Based Symmetric Two-State Energy-Gap Estimation of Binding Free Energies

We examine a symmetric two-state energy-gap (STEG) approximation for binding free energies using a molecular mechanics/Poisson–Boltzmann surface area (MM/PBSA) energetic baseline. An endpoint linear-response approximation, together with a negligible OFF-state mean gap, yields an estimate equal to one-half of the MM/PBSA energetic baseline. The numerical protocol requires one 50 ns bound trajectory per complex and no simulations at intermediate coupling states. For eight neutral cucurbit[7]uril host–guest complexes, MM-STEG gives mean absolute/root-mean-square errors of 1.41/1.64 kcal mol−1, compared with 5.36/6.52 kcal mol−1 for MM/PBSA with interaction entropy (IE), and reproduces the experimental rank order, as does raw MM/PBSA. Across fifteen T4 lysozyme complexes with experimental affinities, MM-STEG reduces every absolute error relative to MM/PBSA+IE, giving errors of 3.04/3.21 versus 7.02/7.58 kcal mol−1. The T4 predictions show persistent overbinding and a pooled Spearman coefficient of 0.375, with the narrow experimental affinity range making accurate ranking particularly challenging. Raw MM/PBSA and MM/PBSA+WEIE give T4 errors of 11.15/11.40 and 7.13/7.59 kcal mol−1, respectively. Within MM-STEG, WEIE affects only the optional enthalpy inference; the inferred 1-adamantanol binding enthalpy differs from experiment by 1.48 kcal mol−1. For eight L99A ligands, inferred enthalpies give MAE/RMSE of 3.58/4.35 kcal mol−1 against calorimetry at a 4 K higher temperature, with no consistent advantage over ordinary IE at the same free-energy baseline. These results support further evaluation of