Structure-Based Virtual Screening and De Novo Design Guided by 3D-QSAR Triage for the Identification of Nanomolar STK3 Inhibitors

Abstract

Serine/threonine kinase 3 (STK3) is a core kinase in the Hippo signaling pathway that regulates cell proliferation, differentiation, and apoptosis. Dysregulation of this pathway is linked to various cancers and immune disorders, emphasizing STK3 as a promising therapeutic target for diseases that involve abnormal cell growth. Despite their therapeutic relevance, only a limited number of potent STK3 inhibitors have been reported, likely due to the difficulty in designing molecules that can bind tightly to the ATP-binding site of STK3. Using the 2-(2-amino-5-phenylpyrimidin-4-yl)-5-methoxyphenol (APPP) scaffold identified through virtual screening, we discovered a series of novel, potent STK3 inhibitors with half-maximal inhibitory concentration (IC50) values as low as 3.52 nM. This was made possible through an integrated molecular design strategy that combined structure-based de novo design with a predictive 3D-QSAR model to estimate IC50 values and prioritize compounds for synthesis. This approach improved the prioritization of compounds for synthesis by enhancing binding energy discrimination beyond the limitations of scoring functions in de novo design. Strategic derivatization at four defined positions on the APPP core enhanced interactions in the ATP-binding site of STK3, resulting in 18 of the 25 synthesized derivatives exhibiting IC50 values below 100 nM. These findings are particularly significant given the scarcity of known STK3 inhibitors with comparable potency, while also indicating that 3D-QSAR modeling contributed as a complementary triage tool rather than a standalone driver of lead discovery. This advancement stems from the use of quantum-mechanically derived 3D structural alignments and the incorporation of electrostatic potential distributions as quantitative molecular descriptors. Taken together, this study establishes a robust design paradigm for the rapid development of highly potent kinase inhibitors through structure-based computational strategies.