Effect of Water\nNetworks On Ligand Binding: Computational\nPredictions vs Experiments

Rational drug design focuses on the explanation and prediction\nof complex formation between therapeutic targets and small-molecule\nligands. As a third and often overlooked interacting partner, water\nmolecules play a critical role in the thermodynamics of protein–ligand\nbinding, impacting both the entropy and enthalpy components of the\nbinding free energy and by extension, on-target affinity and bioactivity.\nThe community has realized the importance of binding site waters,\nas evidenced by the number of computational tools to predict the structure\nand thermodynamics of their networks. However, quantitative experimental\ncharacterization of relevant protein–ligand–water systems,\nand consequently the validation of these modeling methods, remains\nchallenging. Here, we investigated the impact of solvent exchange\nfrom light (H<sub>2</sub>O) to heavy water (D<sub>2</sub>O) to provide\ncomplete thermodynamic profiling of these ternary systems. Utilizing\nthe solvent isotope effects, we gain a deeper understanding of the\nenergetic contributions of various components. Specifically, we conducted\nisothermal titration calorimetry experiments on trypsin with a series\nof <i>p</i>-substituted benzamidines, as well as carbonic\nanhydrase II (CAII) with a series of aromatic sulfonamides. Significant\ndifferences in binding enthalpies found between light vs heavy water\nindicate a substantial role of the binding site water network in protein–ligand\nbinding. Next, we challenged two conceptually distinct modeling methods,\nthe grid-based WaterFLAP and the molecular dynamics-based MobyWat,\nby predicting and scoring relevant water networks. The predicted water\npositions accurately reproduce those in available high-resolution\nX-ray and neutron diffraction structures of the relevant protein–ligand\ncomplexes. Estimated energetic contributions of the identified water\nnetworks were corroborated by the experimental thermodynamics data.\nBesides providing a direct validation for the predictive power of\nthese methods, our findings confirmed the importance of considering\nbinding site water networks in computational ligand design.