Predicting the Impact of Drug Resistance Mutations on Inhibitor Potency with Molecular Dynamics and Machine Learning

Area of Science:

• Biochemistry and Structural Biology
• Computational Biology and Cheminformatics
• Drug Discovery and Development

Background:

• Enzymes are critical drug targets, but mutations can lead to drug resistance, diminishing treatment efficacy.
• Resistance mutations, particularly distal ones, can significantly reduce inhibitor binding affinity, posing challenges for drug design.
• Existing structure-based analyses of enzyme-inhibitor complexes often fail to fully explain resistance mechanisms.

Purpose of the Study:

• To develop an accurate computational method for predicting drug resistance in enzyme variants.
• To identify key molecular features that predict the loss of inhibitor binding potency due to mutations.
• To create a robust machine learning model for forecasting resistance to potent inhibitors like darunavir against HIV-1 protease variants.

Main Methods:

• Rerefining cocrystal structures of 28 HIV-1 protease variants with darunavir to ensure accurate inhibitor geometries.
• Employing parallel molecular dynamics simulations to capture enzyme dynamics and interactions.
• Utilizing feature selection techniques combined with machine learning to build predictive models for binding affinity and resistance.

Main Results:

• Machine learning models incorporating physics-based features of intramolecular interactions accurately predicted binding affinity.
• Four specific features, located distally from the active site, were sufficient to predict binding affinity within 1 kcal/mol of experimental values.
• The developed models significantly outperformed those based solely on enzyme structures or sequences in predicting resistance.

Conclusions:

• A robust strategy for predicting drug resistance due to unseen mutations has been demonstrated.
• Physics-based features, particularly intramolecular interactions distal to the active site, are crucial for predicting inhibitor binding affinity loss.
• This approach offers a powerful tool for anticipating and overcoming drug resistance in enzyme-targeted therapies.