Affinity Enrichment and Characterization of Mucin Core-1 Type Glycopeptides from Bovine Serum

The lack of consensus sequence, common core structure, and universal endoglycosidase for the release of O-linked oligosaccharides makes O-glycosylation more difficult to tackle than N-glycosylation. Structural elucidation by mass spectrometry is usually inconclusive as the CID spectra of most glycopeptides are dominated by carbohydrate-related fragments, preventing peptide identification. In addition, O-linked structures also undergo a gas-phase rearrangement reaction, which eliminates the sugar without leaving a telltale sign at its former attachment site. In the present study we report the enrichment and mass spectrometric analysis of proteins from bovine serum bearing Galβ1–3GalNAcα (mucin core-1 type) structures and the analysis of O-linked glycopeptides utilizing electron transfer dissociation and high resolution, high mass accuracy precursor ion measurements. Electron transfer dissociation (ETD) analysis of intact glycopeptides provided sufficient information for the identification of several glycosylation sites. However, glycopeptides frequently feature precursor ions of low charge density (m/z > ∼850) that will not undergo efficient ETD fragmentation. Exoglycosidase digestion was utilized to reduce the mass of the molecules while retaining their charge. ETD analysis of species modified by a single GalNAc at each site was significantly more successful in the characterization of multiply modified molecules. We report the unambiguous identification of 21 novel glycosylation sites. We also detail the limitations of the enrichment method as well as the ETD analysis.Glycosylation is among the most prevalent post-translational modifications of proteins; it is estimated that over half of all proteins undergo glycosylation during their lifespan (1). Glycosylation of secreted proteins and the extracellular part of membrane proteins occurs in the endoplasmic reticulum and the contiguous Golgi complex. The side chains of Trp, Asn, and Thr/Ser residues can be modified, termed as C-, N-, and O-glycosylation, respectively (2, 3). In addition, O-glycosylation also occurs within the nucleus and the cytosol: a single GlcNAc residue modifies Ser and Thr residues. O-GlcNAc glycosylation fulfills a regulatory/signaling function similar to phosphorylation (4).From an analytical point of view, C-glycosylation is the simplest. A consensus sequence has been defined: WXXW where the first Trp is modified, and the modification, a Man moiety, readily survives sample preparation and mass spectrometric analysis, including collisional activation (5). Investigation of N-glycosylation is also facilitated by several factors. First, N-glycosylation again has a well defined consensus sequence: NX(S/T/C) where the middle amino acid cannot be Pro (6). Second, there is a universal core glycan structure: GlcNAc₂Man₃; and this core is conserved across species. Third, a specific endoglycosidase, peptide N-glycosidase F, has been identified. This enzyme cleaves the carbohydrate structure from the peptide, leaving behind a diagnostic sign: the Asn residue is hydrolyzed to Asp, inducing a mass shift of +1 Da. By contrast, analysis of O-glycosylation is hampered by a lack of (i) a consensus sequence, (ii) a universal core structure, and (iii) a universal endoglycosidase or gentle chemical hydrolysis method to facilitate analysis.Glycosylation shows a high degree of species and tissue specificity; the same site may be modified by a wide variety of different glycan structures, and unmodified variants of the protein may occur simultaneously (7–9). Disease and physiological changes also may alter the glycosylation pattern (10–12). The biological role(s) of glycosylation has been studied extensively (13–15), although such studies are seriously hampered by the difficulties of glycosylation analysis.Most secreted proteins are glycosylated; and thus, mammalian serum is rich in glycoproteins. On the other hand, O-linked glycoproteins represent a small percentage of the serum protein content. Glycoproteins may display a befuddling heterogeneity both in site specificity and site occupancy. Thus, the enrichment of modified proteins or peptides is necessary for their characterization, and different techniques have been tested for this purpose. Lectin affinity chromatography is a popular method for selective isolation of glycoproteins and glycopeptides. Concanavalin A can be used to isolate oligomannose type glycopeptides (16), wheat germ agglutinin is applied for GlcNAc-containing compounds (16, 17), and jacalin is selective for core-1 type O-glycopeptides (18, 19). Lectins with preferential affinity for fucosylated and sialylated structures can also be utilized (12). Non-selective capture of glycopeptides can be performed using hydrophilic interaction chromatography (20, 21) or size exclusion chromatography (22). A recent approach applies porous graphite columns for semiselective enrichment (23), whereas the acidic character of sialylated glycopeptides has also been exploited via titanium dioxide-mediated enrichment (24). Finally vicinal cis-diols can be selectively captured using boronic acid derivatives (25–27). All methods described here provide some glycopeptide enrichment from non-glycosylated peptide background, but all also suffer from significant non-selective binding. N-Linked glycoproteins may also be selectively captured on hydrazide resin following periodate oxidation (28). This approach requires enzymatic deglycosylation to release the captured peptides for analysis, therefore excluding the determination of the carbohydrate structure.Intact glycopeptide characterization still represents a significant challenge. Edman degradation, either alone or in combination with mass spectrometry, has been utilized for such tasks (29, 30). CID analysis of O-linked glycopeptides has limited utility. (i) MS/MS analysis cannot differentiate between the isomeric carbohydrate units and usually does not reveal the linkage positions and the configuration of the glycosidic bonds. (ii) Such spectra are typically dominated by abundant product ions associated with carbohydrate fragmentation, namely non-reducing end oxonium ions and product ions formed via sequential neutral losses of sugar residues from the precursor ions. (iii) The glycan is cleaved from the peptide via a gas-phase rearrangement reaction, and as a result the peptide itself and most peptide fragments (if any) are detected partially or completely deglycosylated (31–33). Recently a different approach, the combination of positive and negative ion mode infrared multiphoton dissociation, was found to provide conclusive structural assignment for some O-linked glycopeptides (34). However, two novel MS/MS techniques, electron capture dissociation (ECD),¹ which is performed in FT-ICR mass spectrometers (35), and electron transfer dissociation (ETD), which is performed in various ion trapping devices (36), may represent the real breakthrough. In both cases an electron is transferred to multiply protonated peptide cations, triggering peptide fragmentation at the covalent bond between the amino group and the α-carbon, producing mostly c and radical z· product ions while leaving the side chains intact. ETD is typically more efficient than ECD and thus leads to more comprehensive fragmentation. In addition, ETD can be performed in ion traps and thus, at a higher sensitivity level, especially in a linear ion trap. Because it has been observed that there are instances when the electron transfer is efficient and still no significant fragmentation occurs, ETD is usually combined with supplementary (and gentle) CID activation (37). O-Glycosylation analysis using these new dissociative techniques has been investigated (38, 39). However, because of the complexity of extracellular O-glycosylation, analysis of complex mixtures is rarely attempted (18), and the above techniques are usually restricted to the analysis of purified proteins.In this study we present the analysis of secreted O-linked glycopeptides. Lectin (jacalin) affinity chromatography was used to achieve some enrichment of core-1 O-GalNAcα type carbohydrate-carrying glycopeptides from bovine serum. The glycopeptide fractions were subjected to CID and ETD analysis. These experiments were performed on a linear ion trap-Orbitrap hybrid mass spectrometer (40). The Orbitrap delivered high resolution, high mass accuracy for the precursor ions, whereas the linear trap provided high sensitivity MS/MS analyses. Some fractions were also subjected to sequential exoglycosidase digestions, and glycopeptides retaining only the proximal GalNAc residues were analyzed. ProteinProspector v5.2.1, developed to accommodate ETD product ion spectra, aided data interpretation (41). We identified 26 glycosylation sites from bovine serum unambiguously; 21 of these sites have never been reported by any other study. No other single study to date has yielded so much information about O-linked glycosylation sites.