Molecular Modeling and Virtual Screening: Application to the Design of Benzoisoxazole-Derived Inhibitors for the Treatment of Prostate Cancer, Bromodomain-Containing Protein 4 (BRD4)

Abstract

Bromodomain-containing protein 4 (BRD4) contributes to transcriptional regulation implicated in castration-resistant prostate cancer and is a candidate target for the design of small-molecule inhibitors. This study used molecular modelling and virtual screening to explore benzoisoxazole derivatives with predicted BRD4 inhibitory activity. A three-dimensional quantitative structure–activity relationship (3D-QSAR) pharmacophore was constructed using active conformations of 23 benzoisoxazole derivatives. The selected Hypo1 hypothesis comprised five pharmacophore features and showed a squared correlation coefficient of 0.951, a root mean square deviation of 1.861, and a total cost of 97.682. Cat-Scramble evaluation was performed at a 98% confidence level. Systematic substitution at two positions of the scaffold yielded a virtual combinatorial library of 99,856 analogues. After drug-likeness filtering and pharmacophore screening, 120 analogues matched at least four of the five features and were assessed using a previously developed complexation QSAR model. Predicted absorption, distribution, metabolism and excretion properties were examined with QikProp. Molecular dynamics simulations of selected BRD4–ligand complexes were performed over 200 ns to assess conformational behaviour and ligand interactions. Six prioritised analogues—57-42, 7-39, 47-11, 7-42, 45-316, and 44-14—showed comparatively favourable predicted oral absorption and molecular dynamics profiles. The results support their prioritisation for synthesis and experimental evaluation, while remaining computational predictions rather than evidence of biological efficacy.