Binding free energy estimation for protein-ligand complex based on MM-PBSA with various partial charge models

Area of Science:

• Computational chemistry
• Drug discovery
• Molecular modeling

Background:

• Accurate estimation of protein-ligand binding free energy is vital for drug discovery.
• Challenges include accurately modeling long-range electrostatics and solvation effects.
• Optimization of atomic net charges is a key factor in binding free energy calculations.

Purpose of the Study:

• To investigate the impact of various atomic net charge models on binding free energy profiles.
• To evaluate the performance of different charge models for diverse protein-ligand systems.
• To identify optimal charge models for specific protein targets like PKB and CDK2.

Main Methods:

• Tested multiple atomic charge models: AM1-BCC, MNDO, PM5, Mulliken, CM2, CM3, RESP, and QM/MM.
• Evaluated models on diverse protein complexes, protein kinase B (PKB), and cyclin-dependent kinases 2 (CDK2) datasets.
• Compared calculated binding free energies with experimental affinities using MM-PBSA.

Main Results:

• The MNDO charge model was optimal for the PKB system.
• The QM/MM model performed best for the CDK2 system.
• No single charge model was universally suitable for all diverse protein complexes tested.
• MM-PBSA trends generally aligned with experimental data for CDK2, but less so for PKB.

Conclusions:

• The choice of atomic charge model significantly impacts binding free energy estimations.
• Specific charge models show promise for particular protein targets (e.g., MNDO for PKB, QM/MM for CDK2).
• Careful consideration of ligand and protein charges is essential for accurate binding free energy calculations in drug design.