Using multiple microenvironments to find similar ligand-binding sites: application to kinase inhibitor binding

Area of Science:

• Biochemistry
• Structural Biology
• Computational Chemistry

Background:

• Identifying small-molecule binding sites is crucial for drug discovery and understanding drug side effects.
• Existing methods often rely on strict geometric and chemical complementarity, potentially missing binding sites with dynamic or altered interaction modes.

Purpose of the Study:

• Introduce PocketFEATURE, a novel algorithm for assessing binding site similarity.
• Overcome limitations of geometric-based approaches by considering microenvironment similarities.
• Enable recognition of distant evolutionary relationships and unexpected shared ligand binding.

Main Methods:

• PocketFEATURE algorithm analyzes shared microenvironments between binding sites.
• Employs weak geometric constraints to accommodate ligand and protein dynamics.
• Utilizes biophysical and biochemical measures to characterize diverse binding modes.

Main Results:

• PocketFEATURE effectively discriminates sites binding similar ligands.
• Achieved 92% Area Under the Curve (AUC) for recognizing FAD-binding sites across proteomes.
• Identified 40 kinase pairs predicted to share ligands, including evolutionarily distant kinases.

Conclusions:

• PocketFEATURE offers a robust method for identifying shared ligand binding across diverse protein structures.
• The algorithm can uncover novel drug repurposing opportunities and potential therapeutic strategies.
• Predicts new avenues for inhibiting multiple proteins within a single pathway through shared ligand interactions.