Protein-Ligand Complex Solvation Thermodynamics: Development, Parameterization, and Testing of GIST-Based Solvent...

Area of Science:

• Computational chemistry and drug design
• Molecular modeling and simulation
• Biophysics and structural biology

Background:

• Molecular solvation and desolvation are critical in drug design, with water molecules actively participating in protein-ligand binding.
• Current computational tools struggle to routinely incorporate solvation features into structure-affinity relationships (SARs).

Purpose of the Study:

• To develop and validate solvent functional-based models for calculating solvation contributions to protein-ligand binding free energy.
• To assess the accuracy of these models using experimental data and explore their applicability across different protein targets.

Main Methods:

• Utilized molecular dynamics simulations and Grid Inhomogeneous Solvation Theory (GIST) processing to develop solvent functionals.
• Calibrated models using experimental binding affinity data from isothermal titration calorimetry (ITC) for 53 thrombin protein-ligand complexes.
• Investigated model performance by considering protein pocket desolvation, ligand desolvation, and combined contributions.

Main Results:

• Models achieved excellent agreement with experimental measurements by considering either protein or ligand desolvation prior to binding.
• Incorporating protein-ligand complex contributions showed good agreement but required spatial cutoff parameter adjustments.
• Transferring trained models to a different protein target yielded deviating results, highlighting the need for target-specific solvation treatments.

Conclusions:

• Solvation effects can be accurately modeled to predict binding free energy, significantly advancing rational drug design.
• The developed GIST-based processing tool (Gips) automates parameter generation for designing compounds with favorable solvation properties.
• The approach offers retrospective insights into the predictive power of established methods like three-dimensional quantitative SAR (3D-QSAR).