Calculation of Host-Guest Binding Affinities Using a Quantum-Mechanical Energy Model

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Accurate prediction of protein-ligand binding affinities is crucial for drug discovery.
• Existing methods using empirical force fields often lack the required accuracy.
• Quantum mechanical energy models offer a potential alternative to improve predictions.

Purpose of the Study:

• To evaluate a quantum mechanical energy model for predicting binding affinities using the mining minima (M2) method.
• To compare the performance of quantum mechanics against empirical force fields in binding affinity calculations.
• To analyze the contributions of different energy components to binding free energy.

Main Methods:

• Application of the semi-empirical quantum mechanical energy function PM6-DH+ with the COSMO solvation model.
• Utilizing the mining minima (M2) approach for binding affinity calculations.
• Testing on 29 host-guest systems with diverse binding affinities.

Main Results:

• Computed absolute binding affinities showed good agreement with experimental measurements (mean error 1.6 kcal/mol, R=0.91) after correcting for polar solvation errors.
• The quantum mechanical approach revealed significant differences in energetic and entropic contributions compared to empirical force fields.
• Successful integration of a quantum mechanical Hamiltonian with the M2 affinity method was demonstrated.

Conclusions:

• Quantum mechanical energy models, when combined with the M2 method, provide accurate predictions of protein-ligand binding affinities.
• This approach offers a more reliable alternative to empirical force fields for computer-aided drug discovery.
• Understanding energy component contributions enhances insights into binding mechanisms.