Computational Analysis of LOX1 Inhibition Identifies Descriptors Responsible for Binding Selectivity

Area of Science:

• Biochemistry
• Enzymology
• Pharmacology

Background:

• Lipoxygenases (LOX) are enzymes involved in polyunsaturated fatty acid metabolism.
• LOX enzymes play a role in the pathogenesis of several human diseases.
• Developing selective LOX inhibitors is challenging due to enzyme subtype similarity.

Purpose of the Study:

• To identify structural descriptors that confer binding selectivity for 15S-LOX1.
• To develop Quantitative Structure-Activity Relationship (QSAR) models for LOX1 activity and selectivity.
• To create predictive models capable of discriminating between inhibitors of 15S-LOX1 and 12S-LOX1.

Main Methods:

• Assembled a dataset of 317 diverse molecules active against 15S-LOX1, 12S-LOX1, and 15S-LOX2.
• Employed supervised machine learning to identify selectivity-driving structural descriptors.
• Trained ensemble QSAR models using Support Vector Machine, Random Forest, and k-Nearest Neighbor algorithms.

Main Results:

• Identified specific structural descriptors crucial for 15S-LOX1 binding selectivity.
• Developed highly accurate QSAR models for predicting LOX1 activity and inactivity.
• Achieved high accuracy in discriminating between molecules selective for 15S-LOX1 versus 12S-LOX1.

Conclusions:

• Machine learning can identify key structural features for selective enzyme inhibition.
• Ensemble QSAR models offer a powerful approach for drug design targeting lipoxygenase isozymes.
• These models facilitate the development of selective inhibitors for therapeutic applications.