Online Energy-Resolved Mass Spectrometry Enables Differentiation of Glycosylation Sites and Sugar Moieties Supporting More Accurate Identification of Ginsenoside Isomers.

Structural characterization of natural saponins is highly challenging by tandem mass spectrometry, lacking evidence for differentiating multiple glycosylation sites and diverse sugar moieties. Optimal collision energy (OCE) in collision-induced dissociation has demonstrated potential in identifying multiple-site isomers of saponins; however, solid evidence supporting differentiation of isomeric saponins remains insufficient. We aimed to exploit the OCE characteristics associated with the diversity of saponin substructures (involving sapogenins, glycosylation sites, and sugar moieties) by analyzing 86 ginsenoside compounds. Through comparative analysis of absolute OCE (|OCE|) characteristics for ion pairs, we discovered that (i) monodesmosidic ginsenosides exhibited higher |OCE| than bidesmosidic counterparts for protopanaxadiol (PPD)/protopanaxatriol (PPT)-types, with an opposite trend for oleanolic acid (OA)-type; (ii) |OCE| correlated with glycosylation sites, with C-20 glycosylation requiring lower energy for glycosidic bond cleavage than C-3 (PPD-type) or C-6 (PPT-type) glycosylation, and (iii) for PPT/PPD-type ginsenosides with up to three sugars, |OCE| ranked as Ara(p)- > Xyl- > Ara(f)-containing chains. On the basis of these findings, we formulated a strategy integrating ion mobility separation and OCE characteristics and validated it by ginsenoside characterization in leaves of Panax ginseng and Panax quinquefolius (PGL and PQL). Among 395 identified ginsenosides, 27 were assigned with enhanced confidence. Furthermore, five markers differentiating PGL and PQL were identified via pseudo-targeted metabolomics and machine learning. Comparative analysis across the mainstream MS platforms established dimeric ions in full-scan spectra as diagnostic markers for distinguishing mono- and bidesmosidic ginsenosides. This study demonstrates the applicability of online ER-MS-derived OCE as a complementary structural descriptor in improved annotation of ginsenoside isomers.