Raman imaging uncovers Ebracteolatain A's subtype-selective metabolic reprogramming in luminal versus triple-negative...

Ebracteolatain A (EA), a potential anti-cancer agent, has demonstrated efficacy against breast cancer through protein kinase D1 inhibition. However, its effects on cellular metabolic reprogramming and activity against aggressive cancer subtypes remain poorly understood. Using confocal Raman spectroscopy coupled with machine learning, we investigated spatial and quantitative metabolic alterations induced by EA (0, 5, 10 μM) in luminal (MCF-7) and triple-negative (SUM149) breast cancer cells at single-cell resolution. Our integrated analysis revealed distinct, concentration-dependent metabolic responses between subtypes. While both cell lines exhibited conserved EA-induced increases in glucose (1123 cm^-1) and lipid (1440 cm^-1) levels with protein depletion (2928 cm^-1), MCF-7 cells displayed significantly greater metabolic vulnerability. These proliferative cells showed progressive phenylalanine reduction (1003/1171 cm^-1) and developed characteristic perinuclear lipid-protein droplets at 5 μM EA, progressing to complete biomolecular disorganization at cytotoxic concentrations (10 μM). In contrast, SUM149 cells maintained metabolic stability, demonstrating sustained nucleic acid production (1223 cm^-1) and preserved subcellular integrity throughout treatment. Multivariate analysis confirmed this differential metabolic remodeling sensitivity, with linear discriminant analysis showing pronounced dose-dependent clustering in MCF-7 cells (Mahalanobis distance = 5.90) compared to SUM149 populations (distance = 3.72). These findings not only elucidate EA's subtype-specific mechanisms of action but also validate the combined Raman spectroscopy-machine learning platform as a powerful tool for pharmacometabolomic assessment. This approach provides spatial and quantitative insights into cancer cell metabolic adaptation, offering new opportunities for targeted therapeutic development.