Integrated RNA Sequencing and Multi-Bioinformatics Analyses Reveal the Molecular Mechanisms Underlying the Anti-Breast Cancer Effects of the Fritillariae Thunbergii Bulbus-Bolbostemmatis Rhizoma Herb Pair.

Breast cancer (BC) is a highly heterogeneous malignancy with complex molecular mechanisms, highlighting the need for effective biomarkers and therapeutic targets. The Fritillariae Thunbergii Bulbus-Bolbostemmatis Rhizoma (ZBM-TBM) herb pair has been used in the treatment of BC, although its potential molecular mechanisms remain incompletely understood. In this study, an integrated strategy combining transcriptomic analysis, machine learning, liquid chromatography-tandem mass spectrometry (LC-MS/MS), network pharmacology, molecular docking, molecular dynamics (MD) simulations, and single-cell virtual knockout analysis was employed to identify BC-associated molecular biomarkers and explore the potential mechanisms underlying the effects of ZBM-TBM. Integration of weighted gene co-expression network analysis with least absolute shrinkage and selection operator regression, random forest, and support vector machine-recursive feature elimination identified EPCAM as the only common core gene. EPCAM was significantly upregulated in BC tissues and showed favorable diagnostic performance (AUC = 0.903), while high EPCAM expression was associated with shorter overall survival (log-rank P = 0.032). Functional analyses indicated that EPCAM was associated with tumor-related and immune-related biological processes, including cell cycle regulation, DNA replication, and immune regulation. LC-MS/MS analysis identified 32 major active constituents of ZBM-TBM, and subsequent network pharmacology analysis identified 33 overlapping targets and 10 hub genes enriched in multiple tumor- and immune-related signaling pathways, including the Notch, IL-17, HIF-1, and TGF-β signaling pathways. Although ZBM-TBM did not directly target EPCAM in the predicted target network, computational analyses indicated functional associations between EPCAM and hub gene-associated regulatory networks. Molecular docking and MD simulations further suggested that peiminine may represent a potential active constituent, exhibiting favorable predicted binding characteristics with multiple candidate targets, particularly AURKA. Single-cell virtual knockout analysis showed that computational perturbation of AURKA was associated with alterations in cell cycle-, extracellular matrix-, and cell adhesion-related processes, supporting its potential functional relevance at the cellular level.