Accelerating Factor Xa inhibitor discovery with a de novo drug design pipeline

Small-molecule drugs are essential for maintaining human health. The objective of this study is to identify a molecule that can inhibit the Factor Xa protein and be easily procured. An optimization-based de novo drug design framework, DrugCAMD, that integrates a deep learning model with a mixed-integer nonlinear programming model is used for designing drug candidates. Within this framework, a virtual chemical library is specifically tailored to inhibit Factor Xa. To further filter and narrow down the lead compounds from the designed compounds, comprehensive approaches involving molecular docking, binding pose metadynamics (BPMD), binding free energy calculations, and enzyme activity inhibition analysis are utilized. To maximize efficiency in terms of time and resources, molecules for in vitro activity testing are initially selected from commercially available portions of customized virtual chemical libraries. In vitro studies assessing inhibitor activities have confirmed that the compound EN300-331859 shows potential Factor Xa inhibition, with an IC₅₀ value of 34.57 μmol·L⁻¹. Through in silico molecular docking and BPMD, the most plausible binding pose for the EN300-331859-Factor Xa complex are identified. The estimated binding free energy values correlate well with the results obtained from biological assays. Consequently, EN300-331859 is identified as a novel and effective sub-micromolar inhibitor of Factor Xa.