From virtual screening to reality: Identifying potent PI3Kγ inhibitors for inflammation treatment by harnessing...

Research increasingly shows that excessive activation of phosphoinositide 3-kinase gamma (PI3Kγ) is linked to numerous inflammatory factors, highlighting its potential as a drug target for inflammatory disorder treatment. Nonetheless, few inhibitors have advanced to clinical trials for inflammation treatment. One of the major reasons for this is the extreme difficulty in developing selective PI3Kγ inhibitors due to the high homology among kinase structures. Therefore, the development of new PI3Kγ inhibitors is still urgently needed. We developed a machine learning-driven virtual screening (VS) approach that integrates a Naïve Bayesian classification model, molecular descriptors, pharmacophore mapping, and molecular docking to discover novel PI3Kγ inhibitors. The VS method was validated for its exceptional predictive accuracy by successfully identifying marketed and clinically studied PI3K inhibitors. Additionally, this strategy was applied to screen the Bioactive Compounds Library Max, followed by enzyme inhibition assays, leading to the identification of three compounds with definitive PI3Kγ inhibitory activity. Notably, Cpd13 demonstrated an IC₅₀ value of 656 ± 171 nM. In vitro and in vivo studies showed that Cpd13 inhibited the expression of pro-inflammatory cytokines TNF-α, IL-6, and IL-1β by blocking the PI3K/Akt signaling pathway in LPS-stimulated RAW264.7 macrophages and a mouse model of acute lung injury (ALI). Cpd13 significantly inhibited nitric oxide production and decreased inflammatory cell infiltration in lung tissues. Ultimately, the administration of Cpd13 provided substantial protection against LPS-induced ALI. This study provides a validated virtual screening pipeline for the hit discovery of PI3Kγ inhibitors, and lays a preclinical foundation for the application of these inhibitors in the treatment of inflammatory diseases.