BCRP Inhibition: from Data Collection to Ligand-Based Modeling.

Area of Science:

• Pharmacology
• Biochemistry
• Drug Discovery

Background:

• Breast Cancer Resistance Protein (BCRP), encoded by ABCG2, is an ABC transporter implicated in multidrug resistance and drug-drug interactions.
• Understanding BCRP inhibition is crucial for optimizing cancer therapies and predicting drug behavior.
• Existing datasets for BCRP inhibition lack comprehensive scope, hindering predictive modeling efforts.

Purpose of the Study:

• To present the largest publicly available dataset for BCRP inhibition, comprising 978 unique compounds from 47 studies.
• To develop and validate predictive classification models for BCRP inhibition using this dataset.
• To identify key chemical substructures associated with BCRP inhibition and non-inhibition.

Main Methods:

• Compilation of a comprehensive dataset of BCRP inhibitors and non-inhibitors from diverse scientific literature.
• Application of data cleaning techniques, including duplicate analysis and activity threshold setting, to create a high-quality labeled dataset.
• Utilizing exploratory data analysis and machine learning models to identify structure-activity relationships (SAR) for BCRP inhibition.

Main Results:

• The curated dataset represents the most extensive open resource for BCRP inhibition data to date.
• Predictive modeling successfully identified specific chemical substructures that are characteristic of BCRP inhibitors.
• Novel substructures associated with non-inhibition of BCRP were discovered, expanding the understanding of BCRP SAR.

Conclusions:

• The established dataset and predictive models provide valuable tools for drug discovery and development concerning BCRP.
• The identified substructures offer new insights into the mechanisms of BCRP inhibition and can guide the design of novel therapeutic agents.
• This work facilitates a deeper understanding of BCRP's role in drug resistance and interactions, paving the way for improved drug safety and efficacy.