Structure-based identification of small molecule binding sites using a free energy model

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Predicting drug-target interactions is crucial for drug discovery.
• Existing methods often focus on pocket volume, which may not correlate with druggability.

Purpose of the Study:

• To develop a novel computational approach for identifying druggable binding sites on protein surfaces.
• To reformulate binding site affinity as an additive free energy for efficient surface searching.

Main Methods:

• Utilized a desolvation-based free energy model and physicochemical property ranges for druglike ligands.
• Developed an additive free energy approach to search entire protein surfaces for druggable sites.
• Integrated the method with Statistical Coupling Analysis (SCA) for functional residue prediction.

Main Results:

• The highest-scoring surface patches often correspond to known ligand-binding sites on druggable targets.
• The method successfully identified pockets amenable to druglike small-molecule binding, distinguishing from simple volume-based predictions.
• Successfully predicted potentially druggable allosteric binding sites on p38alpha kinase when combined with SCA.

Conclusions:

• The reformulated additive free energy approach enables efficient searching of protein surfaces for druggable binding sites.
• This method offers a more accurate prediction of druggability compared to volume-based approaches.
• Combining this method with SCA aids in identifying novel allosteric drug targets.