N-Glycan Modifications with Negative Charge in a Natural Polymer Mucin from Bovine Submaxillary Glands, and Their Structural Role

Bovine submaxillary mucin (BSM) is a natural polymer used in biomaterial applications for its viscoelasticity, lubricity, biocompatibility, and biodegradability. N-glycans are important for mucin stability and function, but their structures have not been fully characterized, unlike that of O-glycans. In this study, BSM N-glycans were investigated using liquid chromatography-tandem mass spectrometry. The microheterogeneous structures of 32 N-glycans were identified, and the quantities (%) of each N-glycan relative to total N-glycans (100%) were obtained. The terminal N-acetylgalactosamines in 12 N-glycans (sum of relative quantities; 27.9%) were modified with mono- (10 glycans) and disulfations (2 glycans). Total concentration of all sulfated N-glycans was 6.1 pmol in BSM (20 µg), corresponding to 25.3% of all negatively charged glycans (sum of present N-glycans and reported O-glycans). No N-glycans with sialylated or phosphorylated forms were identified, and sulfate modification ions were the only negative charges in BSM N-glycans. Mucin structures, including sulfated N-glycans located in the hydrophobic terminal regions, were indicated. This is the first study to identify the structures and quantities of 12 sulfated N-glycans in natural mucins. These sulfations play important structural roles in hydration, viscoelasticity control, protection from bacterial sialidases, and polymer stabilization to support the functionality of BSM via electrostatic interactions.


Introduction
Asparagine-linked (N-) glycans containing negatively charged moieties, such as sialic acid, phosphate, or sulfate groups, frequently have crucial roles in the function of glycoproteins [1]. The presence of sialylated N-glycans is related to the stability and half-life of glycoproteins [2]. Phosphorylated N-glycans, including the mannose-6-phosphate group, are responsible for the efficacy of enzyme replacement therapy [3], and sulfated N-glycans are essential for electrostatic interactions and signal transduction pathways [4]. However, these N-glycans are difficult to analyze because of the labile nature of their charged moiety, their heterogeneous composition and structure, and their small proportions relative to total glycan quantities [5,6].

N-Glycan Analysis Using UPLC
ProA-labeled N-glycans were analyzed using the Waters ACQUITY UPLC H-class system equipped with a Waters ACQUITY UPLC fluorescence detector. Hydrophilic interaction LC-based separation was performed using an ACQUITY UPLC Glycan BEH amide column (1.7 µm, 2.1 × 150 mm; Waters, Milford, MA, USA). Solvents A and B were 50 mM ammonium formate (pH 4.4) and 100% acetonitrile, respectively. The gradient conditions allowed the linear changes of solvent B from 75% to 50% at a flow rate of 0.5 mL/min for 46.5 min. The excitation and emission wavelengths were set at 310 nm and 370 nm, respectively.

N-Glycan Analysis Using LC-ESI-HCD-MS/MS
LC-ESI-HCD-MS/MS was performed using an Ultimate 3000 LC system (Thermo Scientific, Waltham, MA, USA) coupled with a mass spectrometer (Q Exactive Hybrid Quadrupole-Orbitrap; Thermo Scientific). The same column and gradient were used for LC-ESI-HCD-MS/MS as for UPLC. The MS measurement was performed in positive ion mode over a range of m/z 500, and the HCD-MS/MS spectra were acquired over m/z 100 to observe N-glycan-generated fragment ions. The detailed mass spectrometer settings for MS analyses are discussed below. The full scan (m/z 500-3000) used 70,000× resolution with an automatic gain control target of 1 × 10 6 ions and a maximum injection time of 100 ms. Data-dependent MS/MS was performed in a "Top 5" data-dependent mode using the following parameters: 17,500× resolution, 2 × 10 5 automatic gain control target, and 50 ms maximum injection time. The source ionization was conducted with 3.5 kV spray voltage, 320 • C capillary temperature, 300 • C probe heater, and an S-Lens level of 50. The data acquisition was performed with an Xcalibur 2.1 (Thermo Fisher Scientific). Mass error was calculated as follows: ((observed mass − calculated mass)/(calculated mass)) × 10 6 [3]. All analytical experiments were repeated three times with identical LC chromatograms and MS ion chromatograms.

Relative Quantity of N-Glycans
The quantity (%) of each N-glycan relative to total N-glycans (as 100%) was determined from the relative extracted-ion chromatogram (EIC) areas for all observed charge states of each N-glycan relative to the total EIC area for all N-glycans generated by LC-ESI-HCD-MS/MS.

Absolute Concentration of N-Glycans
The absolute concentration (mol) of each N-glycan in BSM (20 µg) was determined from the fluorescence intensity of two distinct N-glycan peaks (negatively charged and neutral glycans) in the UPLC chromatogram using a linear calibration curve generated from the ProA concentrations (0.1-12.5 pmol) with two-fold serial dilutions. The other concentrations were calculated using the relative quantities from the concentrations of the distinct peaks.

UPLC and LC-ESI-HCD-MS/MS Analysis of N-Glycans
BSM protein purity was confirmed using SDS-PAGE (data not shown), as described in our previous reports [19,29]. The N-glycans released from BSM were derivatized with ProA and analyzed using UPLC and LC-ESI-HCD-MS/MS. All of the analytical experiments were repeated three times and they yielded identical chromatograms and MS data.

Structural Analysis of N-Glycans
N-glycan structures were identified using LC-ESI-HCD-MS/MS in the following order: (1) a full-scan mass spectrum of the precursor ion with its isotopes; (2) MS/MS spectrum for the fragment ions generated from the precursor ion, as HCD fragmentation cleaves the glycosidic bonds of glycan; and (3) a full-scan mass spectrum for the multiple-charged precursor ion. The fragmentation spectra for the N-glycan structural analysis are presented in Figure 2 and Figure 16), and (C) core-fucosylated (peak 31) N-glycans with ProA-labeling detected as doubly charged precursor ion [M+2H] 2+ . Peak numbers correspond to the respective peak numbers in Table 1 and  Figure 2A shows the MS/MS spectrum of the sulfated N-glycan, which was determined by identifying a precursor ion (observed m/z = 991.3793; mass error = −2.6 ppm; peak 18 in Table 1 Table 1), and their fragment ion peaks ( Figure 2C). The other MS/MS spectra of the identified N-glycans are indicated in the Supplementary Materials ( Figure S2).             Sum of relative quantities of each glycan in the group of total N-glycans. b Symbols: S, sulfate; Fuc; , GlcNAc; , alNAc; , Gal; , Man. c Precursor ion mass obtained from full scan mass spectra. d Mass error was calculated as [(obrved mass − calculated mass)/(calculated mass)] × 10 6 . e Relative quantity (%) was determined from EIC areas for all bserved charge states of each glycan relative to total EIC area for all N-glycans. f The numbers correspond to the peak umbers in Figure 1 Figure 2A shows the MS/MS spectrum of the sulfated N-glycan, which was determined by identifying a precursor ion (observed m/z = 991.3793; mass error = −2.6 ppm; peak 18 in Table 1) and the characteristic oxonium ion-including sulfate [S1HexNAc1 + H] + (observed m/z = 284.0430; mass error = −1.8 ppm). We observed fragment ion peaks that monosaccharides consecutively separated from  Figure 2B shows the structure of oxonium ions, including sulfate [S1HexNAc1 + H] + (observed m/z = 284.0441; mass error = −2.1 ppm) and core-Fuc [ProA-Fuc1HexNAc1 + H] + (observed m/z = 587.3280; mass error = −1.2 ppm). Core-Fuc in N-glycan generally migrated to the end of the glycan during the analyses [31]. However, the MS/MS spectra indicated that the fragment ion [ProA-Fuc1HexNAc1 + H] + was suggestive of a core-fuco- Sum -100.0 m of relative quantities of each glycan in the group of total N-glycans. b Symbols: S, sulfate; Fuc; , GlcNAc; , lNAc; , Gal; , Man. c Precursor ion mass obtained from full scan mass spectra. d Mass error was calculated as [(obved mass − calculated mass)/(calculated mass)] × 10 6 . e Relative quantity (%) was determined from EIC areas for all erved charge states of each glycan relative to total EIC area for all N-glycans. f The numbers correspond to the peak mbers in Figure 1 Figure 2A shows the MS/MS spectrum of the sulfated N-glycan, which was determined by identifying a precursor ion (observed m/z = 991.3793; mass error = −2.6 ppm; peak 18 in Table 1) and the characteristic oxonium ion-including sulfate [S1HexNAc1 + H] + (observed m/z = 284.0430; mass error = −1.8 ppm). We observed fragment ion peaks that monosaccharides consecutively separated from  Figure 2A shows the MS/MS spectrum of the sulfated N-glycan, which was determined by identifying a precursor ion (observed m/z = 991.3793; mass error = −2.6 ppm; peak 18 in Table 1 um of relative quantities of each glycan in the group of total N-glycans. b Symbols: S, sulfate; Fuc; , GlcNAc; , alNAc; , Gal; , Man. c Precursor ion mass obtained from full scan mass spectra. d Mass error was calculated as [(obrved mass − calculated mass)/(calculated mass)] × 10 6 . e Relative quantity (%) was determined from EIC areas for all served charge states of each glycan relative to total EIC area for all N-glycans. f The numbers correspond to the peak mbers in Figure 1 Figure 2A shows the MS/MS spectrum of the sulfated N-glycan, which was determined by identifying a precursor ion (observed m/z = 991.3793; mass error = −2.6 ppm; peak 18 in Table 1) Table 1 summarizes the structure and relative quantity of 32 BSM N-glycans confirmed by the fragmentation patterns in the MS/MS spectra, which revealed that oxonium ions depend on their constituents. The N-glycans were detected as 6 singly and 26 doubly charged ion forms, all observed with a high accuracy (<5.0 ppm).

Structure and Quantification of N-Glycans
The area of each peak of the UPLC chromatogram usually indicates the glycan quantity. However, in this study, the quantity (%) of each N-glycan relative to the total N-glycan amount (as 100%) was characterized according to the EIC rather than the UPLC chro-ProA to enhance mass ionization efficiency, minimize ionization deviation of charged glycans, and inhibit core-fucose migration to the terminal site [10,31]. Of the 32 N-glycans, 15 unreported N-glycans were not identified without ProA-labeling approach. These included 5 sulfated (6.2%), 5 sulfated and core-fucosylated (9.6%), 3 core-fucosylated (4.6%), and 2 nonsulfated and non-fucosylated (3.5%) N-glycans. The relative quantity of each unreported N-glycan ranged from 0.7 to 2.6% (average 1.6% each), and the sum of their relative quantities was 23.9% of the total N-glycans.
Sulfation reportedly modifies the terminal GalNAc of N-glycans [33]. The importance of sulfation in mucin has been demonstrated in previous studies focusing on water-holding capacity [32] and the protection of degradation from bacterial sialidases [12]. Additionally, glycan length and the degree of sulfation in glycans might control mucin viscoelasticity [34]. Moreover, the removal of mucin sulfate groups by treatment with sulfatase type H-1 from Helix pomatia reduced its adsorption and lubricity, suggesting that the functionality of mucins is closely related to their glycosylation pattern [14].
Core-fucosylated N-glycans in glycoproteins are important in the regulation of protein functions involving receptor-binding affinity, malignant tumors, immunity, and disease biomarkers [35]. Core-fucosylated N-glycans are also reportedly crucial for the efficacy and safety of biotherapeutic glycoproteins [31]. However, this core-fucosylation is difficult to analyze because of the migration of the core-Fuc at the innermost GlcNAc in the N-glycans to the terminal site when using MS/MS, but ProA reportedly inhibits this migration [31].
Schematic illustrations of the mucin structures based on the present results and previous reports [12,14] are shown in Figure 4. The N-glycans, including the negatively charged sulfated N-glycans identified in this study, are located in the hydrophobic terminal regions ( Figure 4A) [12,19].  [12,14].
The negatively charged glycans or positively charged amino acids (containing Lys, Arg, and His) in mucin might enhance its adsorption onto positive or negative surfaces in an aqueous environment, respectively. Sulfated N-glycans might contribute to the adsorption (or desorption) onto positive (or negative) surfaces due to attractive (or repulsive) The negatively charged glycans or positively charged amino acids (containing Lys, Arg, and His) in mucin might enhance its adsorption onto positive or negative surfaces in an aqueous environment, respectively. Sulfated N-glycans might contribute to the adsorption (or desorption) onto positive (or negative) surfaces due to attractive (or repulsive) forces, as well as hydration with water-holding and protection of degradation from foreign bacterial sialidases in mucin [12].
The C-terminal Cys-rich part (including the Cys knot) of mucin forms disulfidebonded mucin dimers and multimers [36]. The present results suggest that the physicochemical characterization (e.g., pH sensitivity and stability) of BSM is derived from the sulfate (pKa = 1) groups in N-glycans as well as the reported sialic acid (pKa = 2.6) with some sulfate groups in O-glycans and charged amino acids [37]. Finally, sulfated N-glycans may structurally affect mucin mobility due to inter-and intramolecular electrostatic interactions, and associative polymer networks through self-association could form and stabilize together with hydrogen bonding ( Figure 4B).
The structural modification of glycans has been identified and characterized using radioactive-isotope labeling, nuclear magnetic resonance spectroscopy, and LC-MS/MS [27,38,39]. A comprehensive assessment of the use of AB, ProA, aminoxyTMT (Thermo Scientific), RapiFluor-MS (Waters), and permethylation for N-glycan analysis was recently reported [7]. RapiFluor-MS labeling was suggested to provide the highest MS signal enhancement for neutral glycans; however, these increased ratios were not significantly different (<10 times). The aminoxyTMT and RapiFluor-MS kits involve a short sample preparation time but are more expensive than the others. The analysis of sulfated N-glycans using permethylation to effectively remove negative charges has also elucidated increases in hydrophobicity and ionization. However, permethylation generates heterogeneous N-glycans as a result of non-specific processing, and large methylated N-glycans are difficult to elute from a column [7].
The stoichiometric ratio of ProA to glycan is reportedly 1:1 [38]. Therefore, glycan concentrations were estimated based on the fluorescence intensities of the negatively charged (peak 13 in Figure 1) and neutral (peak 2 in Figure 1) glycan peaks using a linear calibration curve (r 2 = 0.99) generated from the ProA concentrations (0.1-12.5 pmol) (data not shown). The concentration range of each unreported N-glycans was 0.2-0.6 pmol. The total concentration, which was the sum of 12 negatively charged (6.1 pmol) and 20 neutral (15.4 pmol) N-glycans, was 21.5 pmol in 20 µg of BSM.
Recently, O-glycans of BSM have been quantitatively reported to be in total concentrations of 1.1 pmol per 1 µg of BSM, and the negatively charged sialylated O-glycans occupy 81.9% of total O-glycans [29]. From the present and reported results, it is reasonable to suggest that sulfated N-glycans account for 14.0% of total glycans (sum of the present total N-glycans and reported total O-glycans) [6.1 pmol/(21.5 pmol + 22.0 pmol)], and these were 25.3% of negatively charged N-glycans per total negatively charged glycans (sum of the present negatively charged N-glycans and reported negatively charged O-glycans) [6.1 pmol/(6.1 pmol + 18.0 pmol)].
There have been some reports of N-glycans in human and rat mucins. According to MALDI-TOF MS analysis, human corneal mucin MUC16 has N-glycans containing core-Fuc and sialic acid [17], whereas rat sublingual mucin contains only hybrid N-glycans [40]. The human mucin MUC1 also contained neutral N-glycans based on MALDI-TOF MS, gas chromatography-MS, and collisionally activated dissociation-ESI-MS/MS [41]. BSM was recently reported as the most appropriate complement for an extremely complex biocolloid human salivary mucin for use in biomaterial coatings [42] and is an optimal commercially available mucin source because of its similarity to human saliva in terms of surface adsorption and lubrication [33]. However, none of these reports provide information on the presence of sulfation in N-glycans, and further studies are needed to identify sulfated N-glycans.

Conclusions
The present study identified 12 sulfated N-glycans with negatively charged moieties in BSM. The sulfate modifications were observed in 27.9% (6.1 pmol total concentration) of total N-glycans in BSM (20 µg), and these occupy 14.0% of total glycans (sum of the present N-glycans and reported O-glycans [29]) and 25.3% of all negatively charged Nand reported O-glycans [29]. Together with the reported sialylation of O-glycans, the negatively charged sulfations in N-glycans play important structural roles in hydration [32], viscoelasticity control [32], and polymer network stabilization for maintaining gel-forming mucin functionality because of their electrostatic interactions. These N-glycans were located in the hydrophobic terminal regions of mucin, and we indicated the mucin structures, which will be useful for understanding the functionality of mucins and their application as biomaterials. This is the first study to structurally characterize and quantify 12 microheterogeneous sulfate modifications in N-glycans of natural mucins.