Rational Design of Broad-Spectrum Non-Nucleoside Reverse Transcriptase Inhibitors via Pharmacophore-Oriented Generative Artificial Intelligence

Guided by the pharmacophore-oriented molecular generation platform PhoreGen, we employed rilpivirine (RPV) as a lead compound to generate 300 structurally diverse analogs that preserve key pharmacophoric features. Subsequent drug-likeness evaluation, molecular docking score, and synthetic feasibility led to the identification of compound <b>No.102</b> (<b>A19</b>), which displayed potent inhibition activity against WT HIV-1 (EC₅₀ = 3.15 nM) and low cytotoxicity (CC₅₀ > 335 μM). Subsequent structure-activity relationship optimization identified <b>A24</b>, which demonstrated robust activity against clinically relevant drug-resistant HIV-1 mutants (EC₅₀ = 2.07-28.8 nM). Notably, compared to RPV, <b>A24</b> exhibited significantly reduced cytotoxicity (CC₅₀ = 82.3 μM vs. 3.98 μM) and enhanced potency against the Y188L (EC₅₀ = 28.8 nM vs. 79.4 nM) and F227L + V106A (EC₅₀ = 19.0 nM vs. 81.6 nM) mutants. Pharmacokinetic evaluations revealed that <b>A24</b> exhibited attenuated CYP enzyme inhibition (IC₅₀ ≥ 1.88 μM), reduced hERG-related toxicity (IC₅₀ = 2.633 μM), and improved metabolic stability (human <i>t</i>_1/2 = 39.1 min). Collectively, these favorable properties position <b>A24</b> as a promising NNRTI for further development.