The glycosylation of proteins is pivotal for their structural integrity and significantly impacts a variety of biological processes. The detailed characterization of protein glycosylation is essential for optimizing biological functions, understanding pharmacokinetics, managing immunogenicity, ensuring quality control, and complying with regulatory standards. However, the microheterogeneity of potential N/O-glycosylation sites and the clustering of multiple adjacent O-glycans present a considerable challenge in the analysis of glycosylation.

In our study, we have developed a novel, highly efficient parallel enrichment strategy, named DeepGP, capable of simultaneously enriching both N and O-glycopeptides. We conducted a systematic comparison of five commonly used enrichment columns for their effectiveness in enriching intact N-glycopeptides from human IgG and mouse plasma. Our results demonstrated that DeepGP not only identified the highest number of glycopeptides but also achieved the highest efficiency. Moreover, we showcased DeepGP's ability to effectively enrich minimal sample quantities; for instance, it successfully enriched as little as 20ug of HeLa peptides, leading to the identification of 829 glycopeptides. Impressively, with 200ug of HeLa protein, DeepGP identified 4582 glycopeptides, with a consistent coefficient of variation (CV) of less than 7.8% across three replicates.

We also employed DeepGP in conjunction with GlycanFinder to characterize N and O-glycosylation qualitatively and quantitatively in Etanercept. Our innovative approach offers several distinct advantages over existing methods: (1) A significant enhancement in the detection of O-glycosylation sites, with the identification of up to 11 previously unreported sites; (2) An increased diversity of glycans at various O-glycosylation sites; (3) Improved accuracy through the incorporation of additional corroborating evidence, such as b/y or c/z ions generated via Electron Transfer/Higher-energy Collision Dissociation (EtHCD) mode, which bolsters the confidence level of our findings.