# Machine learning-based QSAR and molecular modeling of phytocompounds in Barleria buxifolia L. as a potential aldose reductase inhibitor

## TL;DR

This study identifies a plant compound from Barleria buxifolia that strongly inhibits an enzyme linked to diabetes, suggesting it could be a promising treatment.

## Contribution

The study introduces 9CDOT as a novel, non-toxic lead compound from Barleria buxifolia for aldose reductase inhibition using machine learning and molecular modeling.

## Key findings

- 9CDOT showed stronger binding affinity to aldose reductase than sulindac sulfone with a glide score of −9.95 kcal/mol.
- 9CDOT formed multiple stabilizing interactions including π–π, hydrogen bonds, and halogen bonds within the enzyme's active site.
- An AI-assisted QSAR model predicted 9CDOT has a pIC50 of 7.52 (IC50 = 30 nM), indicating strong inhibitory potential.

## Abstract

The traditional medicinal plant Barleria buxifolia L is well-known for its pharmacological properties. This study aims to predict the binding affinity of bioactive compounds obtained from B. buxifolia towards significant molecular targets associated with diabetes mellitus. The polyol pathway enzyme aldose reductase is associated with diabetes, making it a possible therapeutic target.

By redocking the co-crystallised ligand sulindac sulfone into the aldose reductase binding site, the researchers proved the docking approach’s reliability with a 0.117 Å root mean square deviation (RMSD). Virtual screening showed that 9-Carbomethoxy-6,11-dichloroxy-5-oxoxantho [3,2-g] tetralin (9CDOT) as a promising inhibitor, surpassing sulindac sulfone, which had a glide score of −9.95 kcal/mol and binding energy of −55.10 kcal/mol.

The chosen molecule exhibited significant binding affinity through π–π interactions with Trp111 and Trp219, a stabilising hydrogen bond with Leu300, and hydrophobic contacts with Val297, Ala299, and Leu300. Halogen bonding connections were found with one chlorine facing Trp219, suggesting π-halogen interaction, and the other towards Leu300 and Leu301, indicating halogen-hydrophobic stabilisation. Synergistic interactions increase ligand target site affinity and specificity. The compound was non-hepatotoxic, non-neurotoxic, and non-cytotoxic, according to the toxicity prediction. Our investigation shows that 9CDOT, a phytocompound from B. buxifolia, strongly inhibits human aldose reductase.

The molecule is considered to be one of B. buxifolia’s active antidiabetic principles, making it a promising aldose reductase inhibitor lead candidate. Further verifying its potency, an AI-assisted PCA–PLS QSAR model also showed high predictive performance (R2 = 0.692), with 9CDOT showing a projected pIC50 of 7.52 (IC50 = 30 nM). To determine its therapeutic efficacy and investigate its potential as a lead candidate for the management of diabetes complications, more experimental validation is required.