Development of Simple and Accurate in Silico Ligand-Based Models for Predicting ABCG2 Inhibition.

Area of Science:

• Pharmacology
• Computational Chemistry
• Drug Discovery

Background:

• The ATP binding cassette transporter ABCG2 significantly impacts drug ADMET profiles and multidrug resistance.
• Predictive in silico models for ABCG2 inhibitors are crucial for early-stage drug discovery.

Purpose of the Study:

• To develop and validate robust in silico classification models for identifying ABCG2 inhibitors.
• To identify key molecular properties and fragments that determine ABCG2 inhibition.
• To create a simple, accurate model for evaluating drug-transporter interactions.

Main Methods:

• Ligand-based classification models using partial least squares-discriminant analysis (PLS-DA).
• Molecular interaction field and fingerprint-based structural descriptions.
• Validation using public and in-house experimental datasets.
• Docking simulations to explore inhibitor binding.

Main Results:

• Developed PLS-DA models incorporating physicochemical and fragmental properties.
• Identified a key chemical property as the principal determinant of ABCG2 inhibition.
• Derived a simple rule for differentiating inhibitors from non-inhibitors.
• Integrated rule significantly improved model performance and prediction accuracy.

Conclusions:

• An accurate and simple integrative model for evaluating ABCG2 inhibitor potential was developed.
• The model aids in predicting drug-transporter interactions during drug development.
• Identified molecular features can guide the design of novel inhibitors to overcome ABCG2-mediated resistance.