Alignment between MS/MS and MS3 identifications. Glycans were identified after analyzing the HCD/CID-MS/MS spectrum pairs, but the sequences of peptide backbones were still unknown. In pGlyco, MS3 was used for the identification of peptide sequences of Y1 ions. HCD-pd-MS3 was performed in one-hour MS analysis of the standard glycoprotein mixture, and HCD-pd-CID-MS/MS was performed in another one-hour MS run of the same sample. The identified peptide backbones from MS3 analyses and glycans from HCD/CID-MS/MS analyses generated adequate information for the identification of glycopeptides. For MS3 data analyses, 414 PSMs with a HexNAc on “J” (identities of Y1) were identified at 1% PSM FDR, and were then compiled into a PSM list for alignment. Alignment between these two MS runs was performed based on the peptide backbone mass and the retention time, as shown in Fig. 3a, where each dot on the map represented a qualified glycopeptide precursor with its peptide backbone mass and retention time as coordinates. Extensive microheterogeneity could be directly perceived from the map: a cluster of horizontally distributed dots represents the microheterogeneity of glycopeptides (identical peptide backbone with different glycans attached). As shown in the map in Fig. 3a, the typical retention time window was 1–6 minutes for a cluster of glycopeptides with the same peptide backbone. Therefore, the peptide backbones identified by MS3 could be aligned with glycans identified by HCD- and CID-MS/MS within a certain retention time window. In the right table of Fig. 3a, a cluster of horizontally distributed dots, corresponding to 34 distinct glycopeptide precursors with the same peptide backbone “LVPVPITJATLDR”, was shown with the information of spectral scan number, the precursor mass and the glycan composition. The match of the peptide backbone “LVPVPITJATLDR” in MS3 analysis was illustrated in Fig. 3b, the peptide backbone was interpreted confidently with nearly twenty matched b/y ions. Combining the results from Figs 2 and 3b, the glycopeptide with the peptide backbone “LVPVPITJATLDR” and the glycan (6,5,1,0,1) was confidently identified.

A PSM list was obtained after MS3 identification at 1% PSM FDR. All identified HCD/CID-MS/MS spectrum pairs before glycan FDR cutoff were then aligned with this PSM list. After alignment, 765 glycopeptide-spectrum matches (GPSMs) were obtained, each GPSM had a glycan identified by a HCD/CID-MS/MS spectrum pair and a peptide backbone identified by a MS3 spectrum. And at 1% glycan FDR, we got 556 confidently identified GPSMs, corresponding to 309 non-redundant glycopeptides. Information including all glycosylation sites and the microheterogeneity was shown in Table S-2. There were 46 potential glycosylation sites (with the sequon N-X-S/T/C, X ≠ P) in our standard protein database, and 25 sites were identified by using pGlyco, but no peptide backbones with the sequon N-X-C were identified, although there were really 3 N-X-C sites in our protein database.

Figure 3. Alignment between the glycans identified by HCD- and CID-MS/MS and the peptide backbones identified by MS3. (a) The map of alignment results of glycan identifications and peptide backbone identifications. Different colors of the dots represent the different peptide backbones. The right table shows the microheterogeneity of “LVPVPITJATLDR” characterized by pGlyco. (b) The peptide “LVPVPITJ[+ HexNAc] ATLDR” identified by MS3. The mass of the peptide “LVPVPITJ[+ HexNAc]ATLDR” is 1408.8158 Da, the deduced peptide backbone mass in Fig. 2 is 1408.8194 Da, the precursor mass deviation is 0.0036 Da (2.56 ppm). The tolerance of the fragment ions for MS3 is ± 0.5 Da.