Predicting Isoform-Selective Carbonic Anhydrase Inhibitors via Machine Learning and Rationalizing Structural Features...

Area of Science:

• Biochemistry
• Medicinal Chemistry
• Computational Biology

Background:

• Carbonic anhydrases (CAs) are crucial enzymes catalyzing carbon dioxide hydration, with human isoforms (hCAs) targeted for various diseases.
• Achieving selectivity for specific hCA isoforms, like tumor-associated hCA IX, over ubiquitous ones (hCA I, hCA II) is a therapeutic challenge due to conserved active sites.

Purpose of the Study:

• To develop and validate machine learning models for predicting selective inhibitors of human carbonic anhydrase IX (hCA IX) over hCA II.
• To identify key compound features responsible for hCA IX selectivity.

Main Methods:

• Curated dataset of selective and non-selective hCA inhibitors.
• Machine learning algorithms for prediction model development.
• Analysis of X-ray crystal structures of hCA-inhibitor complexes to understand selectivity determinants.

Main Results:

• Machine learning models achieved high accuracy in predicting hCA IX-selective inhibitors.
• Identified specific compound substructures that contribute to selectivity for hCA IX over hCA II.
• Structural analysis confirmed the role of identified features in selective binding.

Conclusions:

• Machine learning is a powerful tool for identifying selective carbonic anhydrase inhibitors.
• The developed models and identified features can guide the discovery of novel hCA IX-targeted therapeutics for cancer treatment.