Integrated PASS Prediction, Molecular Networking-Guided Isolation, and Quantum Mechanics-Assisted Structure Elucidation Enable the Discovery of Cytotoxic Dimeric Sesquiterpenoids from Inulae Flos

The process of discovering new drugs from natural products is laborious and requires abundant resources and effort. By integration of data science into natural product research, process efficiency can be enhanced, as demonstrated herein by our discovery of cytotoxic constituents from Inulae Flos (IF), assisted by a data-science-integrated approach. Prediction of Activity Spectra for Substances (PASS), a database and neural-network tool for structure–activity relation prediction, suggested dimeric sesquiterpenoids as promising candidates with antineoplastic effects. Seven dimeric sesquiterpenoids (1–7) were obtained by molecular-networking-guided isolation from the methanol extract of IF, including two new compounds (1 and 2). The structures of these compounds were elucidated using nuclear magnetic resonance (NMR) spectroscopy, mass spectrometry (MS), and electronic circular dichroism (ECD) experiments, supported by quantum mechanics-based analyses, including 1H iterative Full-Spin Analysis (HiFSA), Gauge-Independent Atomic Orbital (GIAO) NMR, and ECD calculations. The in silico predicted biological activity was validated by in vitro experiments in which the isolated dimeric sesquiterpenoids showed significant cytotoxicity against Hep G2 and DU 145 cancer cell lines with inhibitory rates ranging from 23.9–98.9% at a dose of 10 μM. These findings validate the utility of the data-science-integrated approach in streamlining the discovery of novel bioactive compounds within the natural product drug discovery pipeline.