Glycoproteomics: a challenge (balance) to mass spectrometry acquisition and interpretation
Résumé
Glycopeptide-based tandem-mass-spectrometry enables site-specific glycosylation analysis, allowing us to explore the glycomes and potential functions. While performing glycopeptides MS/MS, fragmentation can be complicated by the heterogeneous glycans and amino-acid combination, which is the first bottleneck. Several fragmentation modes, such as stepped-collision-energy (sce), and a hybrid of both electron transfer and collision-induced dissociation, were reported to have more potential to generate an informative spectrum(1, 2). Furthermore, the superimposition of fragments from peptides and glycans on a single spectrum is a major limitation for annotation and identification accuracy(3, 4). Hence, obtaining unambiguous interrogation depends on the quality of both acquisition and interpretation.
We have performed MS/MS of tryptic peptides separated prior by reversed-phase chromatography on a Tribrid system(ThermoFisher Eclipse) with HCD followed by the aforementioned two fragmentation modes triggered by product-dependent (pd) ion of glycan. MS/MS spectra were searched against the UniProt database using Proteome Discoverer (PD) with the Byonic searching node, and the results were manually verified. Further validation was conducted by releasing glycans with a homemade glycosidase and mapping with a MALDI-TOF/TOF(Sciex 5800) system. Our analysis of N-glycosylated samples demonstrated that pd-EthcD didn’t improve the identifications compared to other methods.
Moreover, several aspects require improvements during annotation and interrogation, regardless of the fragmentation types. Firstly, glycopeptide MS1-peaking can fail when performing a match-between-run chromatogram and spectrum alignment. Secondly, MS/MS peaks-fragments annotation may be misleading when one isotopic-cluster is simultaneously assigned to two fragments. Thirdly, filtering results by scores as a common proteomics workflow might not lead to confident identifications. For example, scores> 300 and DeltaMod scores> 10 are Byonic-defined thresholds for confident peptide-sequence-match and modification localization. However, manual reviewing of the identifications showed insufficient justifications.
Eventually, we proposed a workflow combining glycan MALDI-MS-MS/MS analyses to improve the robustness of glycopeptide nanoLC-MS/MS spectrum annotation and identification. The complementary profiling was applied to the porcine Luteinizing hormone study.