Unveiling potent xanthine oxidase inhibitors in two Balanophora spp. using machine learning-based virtual screening and molecular docking approach.

Pharmacological studies revealed that the Balanophora species contains diverse phytochemicals which enable interesting biological activities and emphasize their pharmaceutical relevance. Previously, we identified significant xanthine oxidase (XO) inhibitory activity from extracts of the two Balanophora spp. (Balanophora subcupularis P.C. Tam and Balanophora tobiracola Makino). However, the specific compounds responsible for this activity remain unidentified so far. Thus, in the present study, we focused on elucidating the compounds inducing the XO inhibitory effect of extracts from Balanophora species. Therefore, a combination of advanced liquid chromatography and mass spectrometry (LC-QToF-HRMS), virtual screening using machine learning (ML) models, and molecular docking simulation was applied. Using LC-QToF-HRMS, 23 and 21 compounds were identified in the ethyl acetate fractions of B. subcupularis and B. tobiracola, respectively. Next, a curated dataset of natural and synthetic compounds with known XO inhibitory activity was employed to train several ML models. Adducing five selected ML models, the virtual screening process identified the potentially active compounds 1-(3,4-dihydroxyphenyl)-6,7-dihydroxy-1,2-dihydro-2,3-naphthalenedicarboxylic acid, taxifolin, and 1-O-caffeoyl-6-O-(S)-brevifolincarboxyl-β-D-glucopyranose. All the compounds found in the two Balanophora spp. underwent docking simulations, in which MTE, FES, and AFH were retained in the active site of XO, ensuring reliable re-docking results. Finally, taxifolin emerged as the most promising novel XO inhibitor, demonstrating greater potential than the established drug allopurinol, as supported by both the virtual screening nomination and docking simuation. These findings contribute to the development of natural XO inhibitors and may open new opportunities for gout treatment and uric acid level control.