A Scalable, Traceable Computational‐to‐Experimental Pipeline for Activity‐Guided Identification of Bioactive Constituents: A Case Study in Salvia miltiorrhiza Bunge

Systematic identification of bioactive constituents from complex natural products remains challenging. Here, a scalable and traceable integrated workflow was developed and validated in Salvia miltiorrhiza Bunge (SMB) for anti–Parkinson's disease (PD) constituent discovery. Network pharmacology first identified four candidate compounds, which were quantified and dynamically tracked during X‐5 macroporous resin enrichment and fractionation using a validated HPLC method. Transfer rates exceeded 80% for all candidates, confirming efficient enrichment and chemical traceability. Bioactivity was then evaluated in MPP + ‐induced PC12 cells and a rotenone‐induced PD mouse model. SMB‐2 showed the strongest neuroprotective activity and was subjected to spectrum–effect relationship analysis, which identified five activity‐associated compounds. Integration of pharmacological validation, molecular docking, and molecular dynamics simulations progressively narrowed the candidate space and identified cryptotanshinone as a major bioactive constituent. Cryptotanshinone exhibited favorable predicted binding affinity and stable interactions with key residues in the monoamine oxidase B binding pocket, supporting its potential MAO‐B inhibitory activity. It also restored intracellular ATP levels and superoxide dismutase activity, indicating improved energy metabolism and antioxidant defense. Overall, this workflow enables stepwise convergence from complex natural matrices to experimentally supported bioactive constituents and provides a broadly applicable strategy for natural‐product‐based drug discovery.