Ligand-based pharmacophore exploration and QSAR analysis of transition state analogues followed by in silico screening guide the discovery of new sub-micromolar β-secreatase inhibitors

Transition state analogues (TSA)-based inhibitors have particularly potent inhibitory profiles. The fact that beta-secretase (BACE) inhibitors have potential as anti-Alzheimer's treatments prompted us to explore the pharmacophoric space of 68 known TSA BACE inhibitors. Subsequently, quantitative structure-activity relationship (QSAR) analysis was employed to select optimal combination of binding model(s) and 2D physicochemical descriptors capable of explaining bioactivity variation. One pharmacophoric model emerged in the successful QSAR equation. However, to closely mimic the sterically demanding transition state of BACE we were obliged to complement the successful pharmacophore with strict shape-based query decorated with carefully positioned hydroxyl fragment that simulates the statin hydroxyl of known TSA BACE inhibitors. Both models, i.e., the hybrid shape-OH and optimal pharmacophore, were validated via receiver-operating characteristic curve analysis and were found to exhibit excellent abilities in discerning active compounds from decoys. Subsequent in silico screening against the National Cancer Institute structural database using the two models yielded two novel inhibitors of nanomolar and low micromolar IC50 values.