Exploring Natural Compounds for HIV Treatment: A Virtual Screening Study

Background: Bioprospecting for natural compounds has led to the discovery of diverse molecules that have become effective drugs for treating various diseases. However, exploring these compounds remains challenging because of the vast, uncatalogued global biodiversity and the structural complexity of the identified molecules. Virtual screening (VS) has emerged as an effective strategy to support molecular discovery, significantly reducing drug development time and costs. Methods: We virtually screened a natural compound library using quantitative structure-activity relationship (QSAR) models and molecular docking assays, targeting the human immunodeficiency virus (HIV) and its relevant clinical targets. We docked against 11 structures using a single protocol, tested its ability to reproduce a known binding mode by redocking every co-crystallized ligand, and docked each compound alongside 21 established inhibitors of the same sites so predictions could be read against a clinical baseline rather than in isolation. We evaluated the most promising molecules for pharmacokinetic properties and the toxicity of both the parent compounds and their predicted secondary metabolites. Results: In terms of docking score, the compounds identified here do not outperform the approved drugs of the pockets they occupy; what they show is interaction fidelity, the clearest case being a prenylated flavonoid that reproduces the complete interaction fingerprint of nevirapine in the non-nucleoside pocket of reverse transcriptase at a score 1.0 kcal/mol weaker. Conclusions: We present these compounds as scaffolds for optimization rather than candidates for direct testing and report the pockets where our protocol could not be validated so readers do not overinterpret the corresponding predictions.