Water network perturbation in ligand binding: adenosine A(2A) antagonists as a case study

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• In silico methods for evaluating water's role in ligand binding are advancing.
• Existing techniques like WaterMap and Grand Canonical Monte Carlo simulations predict water molecule position and free energy in protein active sites.

Purpose of the Study:

• To extend computational water network (WNP) analysis to predict kinetics and relative binding free energy for small molecules.
• To apply these methods to large, diverse datasets and assess their predictive power for ligand binding affinity.

Main Methods:

• Utilized WaterMap, SZMAP, GRID/CRY probe, and Grand Canonical Monte Carlo simulations.
• Developed a linear combination method (WNP-MMSA) integrating WNP analysis with molecular mechanics force fields.
• Employed machine learning with probabilistic classification trees to predict binding affinity.

Main Results:

• Qualitative correlation observed between residence time of triazine adenosine A(2A) receptor antagonists and trapped solvent molecules.
• WNP-MMSA successfully predicted relative ligand binding free energy for a subset of compounds.
• Machine learning approach achieved 90% accuracy on the training set and 67% on the test set for binding affinity classification.

Conclusions:

• Computational WNP analysis can inform ligand binding kinetics and thermodynamics.
• A linear WNP-MMSA approach is effective for specific ligand sets but has limitations for diverse datasets.
• Machine learning offers a promising, fast, and effective alternative for predicting ligand binding affinity.