Characterization of O-acetylation in sialoglycans by MALDI-MS using a combination of methylamidation and permethylation

O-Acetylation of sialic acid in protein N-glycans is an important modification and can occur at either 4-, 7-, 8- or 9-position in various combinations. This modification is usually labile under alkaline reaction conditions. Consequently, a permethylation-based analytical method, which has been widely used in glycomics studies, is not suitable for profiling O-acetylation of sialic acids due to the harsh reaction conditions. Alternatively, methylamidation can be used for N-glycan analysis without affecting the base-labile modification of sialic acid. In this report, we applied both permethylation and methylamidation approaches to the analysis of O-acetylation in sialic acids. It has been demonstrated that methylamidation not only stabilizes sialic acids during MALDI processing but also allow for characterization of their O-acetylation pattern. In addition, LC-MS/MS experiments were carried out to distinguish between the O-acetylated glycans with potential isomeric structures. The repeatability of methylamidation was examined to evaluate the applicability of the approach to profiling of O-acetylation in sialic acids. In conclusion, the combination of methylamidation and permethylation methodology is a powerful MALDI-TOF MS-based tool for profiling O-acetylation in sialic acids applicable to screening of N-glycans.

was most likely to be a product of immune reaction. With the permethylation process, MALDI-MS is the desired technique for profiling glycans [28][29][30][31] . Methylation of all hydroxyl groups of glycans can significantly increase detection sensitivity because the derivatives are considerably more hydrophobic and stable than native glycans. In addition, permethylated glycans produce more information in tandem mass spectra that is very useful for linkage analysis. Unfortunately, permethylation process removes acetyl esters, which precludes its application to the characterization of O-acetylation of sialic acids. So far, detailed O-acetylation profiles and biological roles of these substitutions have been overlooked because of their lability to conventional purification and detection methods. An alternative strategy for stabilizing a sialic acid is the selective modification of its carboxyl group, including methylamidation [32][33][34][35] , dimethylamidation 19 , methyl esterification 36 and ethyl esterification 37,38 . This study aims at the development of MALDI-TOF MS for glycomics, especially the profiles of O-acetylation in sialic acids. To demonstrate the applicability of the proposed method, we analysed sera from seven different species of carp, including crucian carp, grass carp, silver carp, bighead carp, common carp, bream and black carp from wild fisheries 39 . The enzymatically released and purified fish serum N-glycans were subjected to methylamidation and permethylation derivatization and analysed using MALDI-TOF MS and LC-MS/MS. The tandem mass spectrometry experiments were performed to characterize glycan structures.

Results and Discussion
Crucian carp serum. Permethylation reaction removes O-linked modification, thus the resultant MALDI MS spectrum can be significantly simplified and used to derive the composition of N-glycans. A representative MALDI-MS spectrum of permethylated N-glycans from crucian carp (Carassius carassius) serum sample is presented in Fig. 1a. For simplicity, this serum sample was labelled as Crucian-1. The primary structures of N-glycans isolated from crucian carp serum are similar to that of salmon serum 26,27 . Four major glycoforms, m/z 2431.1, 2792.2, 3241.4 and 3602.5, correspond to one, two and three sialic acid-containing oligosaccharides with chemical compositions of Neu5NAc 1 Hex 5 HexNAc 4 , Neu5NAc 2 Hex 5 HexNAc 4 , Neu5NAc 2 Hex 6 HexNAc 5 , and Neu5NAc 3 Hex 6 HexNAc 5 , respectively. Two high mannose structures were detected at m/z 1579.7 (Hex 5 HexNAc 2 ) and m/z 1783.8 (Hex 6 HexNAc 2 ). The non-sialylated biantennary structure was detected at m/z 2069.9 with a composition of Hex 5 HexNAc 4 . The detection of ions at m/z 4051.7 and 4412.9 suggests the presence of tetra-antennary structures with compositions of Neu5NAc 3 Hex 7 HexNAc 6 and Neu5NAc 4 Hex 7 HexNAc 6 , respectively.
The MALDI-TOF MS spectrum of methylamidated N-glycans from crucian carp serum sample (Crucian-1) is shown in Fig. 1b Overall, up to six O-acetyl groups were detected for the N-glycans containing two sialic acid residues. This observation reveals the most extensively O-acetylated N-glycans in any fish species that have been reported so far 26,27 .
O-acetylation of sialic acid can occur at four positions, i.e. the 4-, 7-, 8-, or 9-hydroxyl position 7,8 . For any disialylated N-glycan containing six O-acetyl groups, two isomers may exist. Each of two sialic acids has either three O-acetyl groups (3 + 3) or one has four and the other one has two O-acetyl groups (4 + 2). Therefore, it is necessary to perform tandem mass spectrometry to characterize the distribution of O-acetyl groups. We performed LC-MS/MS experiments on the methylamidated sample and the representative MS/MS spectra are shown in Figure S1 Figure S1b in Supporting Information). The intensities of two fragment ions associated with trisaccharide residues, Neu5NAc 1 Hex 1 HexNAc 1 and OAc 1 Neu5NAc 1 Hex 1 HexNAc 1 , were almost identical. However, the intensities of the fragment ions that corresponded to the monosaccharide residues, Neu5NAc and OAc 1 Neu5NAc, exhibited significant difference (m/z 305.1 and m/z 347.1 in Figure S1b Figure S1f in Supporting Information), which implied that the number of O-acetyl modification for a sialic acid was not higher than 3. We also noted that the LC-MS/MS experiments on the glycans containing multi sialic acids did not produce any fragments that could be correlated to the presence of disialic acid residues, i.e. Neu5NAc-Neu5NAc.
To study the biological variation, N-glycans from two additional individual fish serum samples, labelled as Crucian-2 and Crucian-3, were analysed and the obtained mass spectra of the replicate serum samples are presented in Figure S2 in Supporting Information. Although the relative abundance of O-acetylated species varied, the maximum numbers of O-acetyl groups for disialylated N-glycan remained unchanged among the three individual fish. The detected ions and the proposed compositions are summarized in Table S1 in Supporting Information.

Repeatability of methylamidation.
Methylamidation has been previously proven to be an effective derivatization strategy for sialic acid-containing N-glycans [32][33][34][35] . Complete methylamidation of 2,6-sialyllactose could be achieved within 10 min, whereas the complete reaction for 2,3-sialyllactose took about 30 min 32 . So far, the applicability of methylamidation to profiling O-acetylation patterns of sialic acids has not been investigated. Because no O-acetylated N-glycan standards were available, we used the N-glycans from pooled serum samples of multiple crucian carp for repeatability testing. The glycans Neu5NAc 2 Hex 5 HexNAc 4 with different O-acetyl groups were selected to investigate the stability of O-acetyl groups, by comparing ESI-MS analysis of native glycans and MALDI-MS analysis of methylamidated glycans. The intensities of native glycans from 5 technical replicates in ESI-MS experiments were summed and normalized to the sum of intensities of the highest peak (Fig. 2a). To investigate the stability of O-acetyl groups under methylamidation conditions, 5 samples from serum pool were individually PNGase-F digested and derivatized, followed by MALDI-MS analysis. The intensities of each glycan from 3 spots in MALDI target were summed and normalized to the sum of intensities of the highest peak (i.e. the glycan containing three O-acetyl groups). The average relative intensity values for six glycoforms are presented in Fig. 2b, with the CVs ranging from 1.9% to 25.4%. We then derivatized 5 N-glycan samples from the same serum pool collected on each day for two additional days to evaluate the day-to-day repeatability 38 . The relative intensity values of six glycoforms for three different days are illustrated in Fig. 2c, with the CVs ranging from 0.8 to 27.8%. While the profiles between native glycans (ESI-MS) and their methylamine derivatives (MALDI-MS) were similar, relatively lower intensities in ESI-MS spectra were observed for the glycans containing four, five and six O-acetyl groups, respectively. To further demonstrate the applicability of the approach, we analysed N-glycans from serum samples of additional six fish species.
The MALDI-MS spectra for the N-glycans from serum samples of two additional individual common carp, labelled as Common-2 and Common-3, are presented in Figure S4 in Supporting Information. The O-acetylation pattern in N-glycans from Common-2 ( Figure S4b and Figure S4e) was similar to that shown in Common-1 ( Figure S4a and Figure S4d). However, O-acetylation pattern in N-glycans from Common-3 showed much lower degree of O-acetylation ( Figure S4c and Figure S4f), in which only four and six O-acetyl groups were detected in disialylated glycans and trisialylated glycans, respectively. The detected ions and their corresponding compositions are summarized in Table S2 in Supporting Information.  (Fig. 4a). In the high mass region, the ions at m/z 3737.8, 3941.8, 4098.9 and 4460.1 correspond to tetraantennary oligosaccharides with different numbers of terminal Hex residues and sialic acid residues.
Methylamidation analysis was also carried out to explore the O-acetylation of N-glycans from the same grass carp serum sample. The resultant MALDI-MS spectrum is presented in Fig. 4b. The monosialylated glycan at m/z 2494.7 is a non O-acetylated glycan; whereas the ions at m/z 2536.8 correspond to the addition of one acetyl group with a chemical composition of OAc 1 Neu5NAc 1 Hex 7 HexNAc 5 . The disialylated structures at m/z 2798.8, 2840.8 and 2882.8 correspond to the glycans with addition of zero, one and two acetyl groups, respectively. The detected ions and their corresponding compositions for grass carp serum N-glycans are summarized in Table S3 and the representative MS/MS spectra are shown in Figure S5 in Supporting Information. These results indicate that the sialic acids in grass carp serum glycoproteins can only be modified with one O-acetyl group. The MALDI-MS spectra of serum N-glycans for three individual grass carp (i.e. Grass-1, Grass-2 and Grass-3) are presented in Figure S6 in Supporting Information. Similar O-acetylation patterns were observed in N-glycans from three biological replicates.   Figure S7 in Supporting Information. Two additional biological replicates of silver carp serum samples were analysed and significant variations were observed between the three replicates. For example, the glycan with the composition of Neu5NAc 1 Hex 6 HexNAc 5 (m/z 2880.3) was only detected in the MALDI MS spectra of permethylated glycans from Silver-2 ( Figure S8b and Figure S8e in Supporting Information) or Silver-3 ( Figure S8c and Figure S8f in Supporting Information), and the corresponding methylamidated glycans were detected at m/z 2332.9 (Neu5NAc 1 Hex 6 HexNAc 5 ) and m/z 2374.9 (OAc 1 Neu5NAc 1 Hex 6 HexNAc 5 ). Table S4 in Supporting Information summarizes the detected ions and their corresponding compositions.
Bream carp serum. The N-glycans from a representative serum sample of bream carp, labelled as Bream-1, were initially analysed using permethylation method (Fig. 6a). The most abundant structures are the triantennary glycoforms with different numbers of sialic acids and Hex terminal motifs. For example, the ion detected at m/z 2723.3 is a triantennary structure with the chemical composition of Hex 7 HexNAc 5 . The ions at m/z 3084.4 correspond to the composition of Neu5NAc 1 Hex 7 HexNAc 5 , with the addition of a sialic acid group. More complex structures were observed as the tetraantennary glycans with ions at m/z 3894.8 and 4255.9, respectively. Methylamidation was used to investigate the O-acetyl modification in bream carp serum N-glycans as well (Fig. 6b). The monosialylated glycans at m/z 2212.7, 2374.7 and 2536.7 can be assigned to the addition of an O-acetyl group to the oligosaccharides with m/z 2170.8, 2332.7 and 2494.7, respectively. The disialylated glycans, with the addition of two O-acetyl groups, were detected at m/z 2720.8 (OAc 2 Neu5NAc 2 Hex 6 HexNAc 5 ) and 2882.8 (OAc 2 Neu5NAc 2 Hex 7 HexNAc 5 ). This observation suggests that the O-acetylation patterns of serum N-glycans from bream carp are similar for silver carp and grass carp. The representative MS/MS spectra are presented in Figure S9 in Supporting Information. Similar glycoforms were detected in two additional biological replicates (Bream-2 and Bream-3) as shown in Figure S10b, Figure S10c, Figure S10e and Figure S10f in Supporting Information, although relative intensities varied among the three replicates. The proposed compositions for most detected ions are presented in Table S5 in Supporting Information.
Bighead carp serum. The N-glycan profiles of a representative serum sample from bighead carp (Hypophthalmichthys nobilis) were also investigated using both permethylation and methylamidation methods (Fig. 7). For permethylated glycans (Fig. 7a), the most abundant ions corresponded to triantennary glycoforms at The MS spectrum of methylamidated N-glycans illustrates the presence of O-acetyl modification in sialic acids (Fig. 7b). For example, the ion at m/z 2374.7 is the triantennary structure with one O-acetyl group (OAc 1 Neu5NAc 1 Hex 6 HexNAc 5 ). The disialylated glycans at m/z 2636.8, 2678.8 and 2720.8 are with zero, one and two O-acetyl groups, e.g. Neu5NAc 2 Hex 6 HexNAc 5 , OAc 1 Neu5NAc 2 Hex 6 HexNAc 5 and OAc 2 Neu5NAc 2 Hex 6 HexNAc 5 , respectively. It is worth noting that sialic acids in bighead carp serum sample have only one O-acetyl group, similarly to the O-acetylation found in grass carp, silver carp and bream carp serum glycans. Tandem mass spectrometry experiment confirmed the O-acetylation of sialic acids ( Figure S11 in Supporting Information). The permethylated and methylamidated N-glycans of two additional replicate serum samples (i.e. Bighead-2 and Bighead-3) are shown in Figure S12b, Figure S12c, Figure S12e and Figure S12f in Supporting Information, respectively. The ion at m/z 3037.8 in the spectrum of permethylated glycans corresponded to a composition of Neu5NAc 2 Hex 5 HexNAc 5 . The methylamidated glycans with zero, one or two O-acetyl groups were detected at m/z 2474.9, 2516.9, and 2558.9, respectively. The detected ions and their compositions are summarized in Table S6 in Supporting Information.
Black carp serum. MALDI-MS profiling of the permethylated glycans released from a representative serum sample of black carp (Mylopharyngodon piceus), labelled as Black-1, indicated the presence of multi-antennary structures (Fig. 8a), which have been previously found in several different species of fresh-water fish and both unfertilized 40,41 and fertilized fish eggs [42][43][44] . The most abundant glycan structures at m/z 2723.3, 3084.5 and 3445.7, correspond to the oligosaccharides with chemical compositions of Hex 7 HexNAc 5 , Neu5NAc 1 Hex 7 HexNAc 5 and Neu5NAc 2 Hex 7 HexNAc 5 , respectively. The glycans from black carp serum sample were mainly found to exhibit complex tri-and tetra-antennary structures. Comparatively, the glycan compositions in crucian carp and common carp sera mainly exhibit biantennary structures and extensive O-acetyl modification. The N-glycans of black carp serum samples were also investigated using the methylamidation method (Fig. 8b). Surprisingly, no O-acetyl modification was detected for this species. The representative MS/MS spectra are shown in Figure S13 in Supporting Information. Two additional biological replicates (i.e. Black-2 and Black-3) confirmed the absence of O-acetylation in sialic acids of N-glycans from black carp serum samples ( Figure S14 in Supporting Information). The relative abundance and type of each glycan in black carp serum samples are consistent between the permethylation and methylamidation spectra. The detected ions and their corresponding compositions are summarized in Table S7 in Supporting Information.

Conclusion
In this study, we applied a new derivatization strategy, the combination of permethylation and methylamidation, for the characterization of O-acetylation of sialic acid in N-glycans. Methylamidation allows the stabilization of O-acetylated sialic acid residues; whereas permethylation provides a sensitive technique with simplified MALDI MS spectra that contain unambiguous information on glycan compositions. The serum samples from seven freshwater carp species were analysed, including crucian carp, grass carp, silver carp, common carp, bream and black carp. The results revealed that the N-glycans from sera of different fish species presented significant difference in composition, sialylation pattern and degree of O-acetylation. For example, the crucian carp serum exhibited up to three O-acetyl groups on a single sialic acid residue. The results also suggested that no more than three O-acetyl groups were attached to one sialic acid residue in all glycans studied, even in the ones with the most extensive O-acetyl substitution. While the method was developed for application in aquaculture, it is applicable for the analysis of sialoglycans derived from sera of any other animal species.  were purchased from Aladdin (Shanghai, China). All solutions were prepared using deionized water purified by a Milli-Q purification system (Millipore, MA, USA).

Chemicals and
Preparation of N-glycans. Fish serum (10 μ L) was dissolved in 50 μ L of sodium phosphate (20 mM, pH 7.5) containing 0.2% SDS and 0.1 M DTT and denatured at 100 °C for 10 min. After cooling, 12 μ L of 10% NP-40 and 38 μ L of water were added. The reaction mixture was incubated with PNGase F (10 units) for 24 h at 37 °C. The sample was then boiled for 5 min to stop the reaction and the released glycans were purified using PGC cartridges. The first serum sample from crucian carp was labelled as Crucian-1 and similar definition was used for the first serum sample of other fish species, i.e. Common-1, Grass-1, Silver-1, Bream-1, Bighead-1 and Black-1.
Biological variation was established by analysing two additional replicates of serum samples for each fish species. Accordingly, the additional serum replicates for all species examined were abbreviated as Crucian-2 and Crucian-3, Common-2 and Common-3, Grass-2 and Grass-3, Silver-2 and Silver-3, Bream-2 and Bream-3, Bighead-2 and Bighead-3, Black-2 and Black-3. For comparison, the MALDI-TOF MS results for all three replicates of each species were included in Supporting Information, i.e. Figure S2, Figure S4, Figure S6, Figure S8, Figure S10, Figure S12 and Figure S14.
All experiments were performed in accordance with the protocols approved by Ethical Committee of Jianghan University.
Purification of glycans. PGC cartridges were washed with 3.0 mL of 80% (v/v) ACN containing 0.1% TFA followed by 3.0 mL distilled water. The glycans released by PNGase F were loaded on PGC cartridges and then washed with distilled water (3.0 mL) to remove buffer and salts. Glycans were eluted with 25% ACN in 0.1% TFA (3 × 1 mL). The fraction was collected and dried for further processing. The liquid phase was recovered and dried using a SpeedVac concentrator.

Permethylation of glycans.
Dried oligosaccharide sample was dissolved in an Eppendorf tube using 50 μ L of DMSO. 100 μ L of DMSO-NaOH slurry and 50 μ L of methyl iodide were then added. Tubes were capped tightly and stirred for about 10 min at room temperature. The reaction was stopped by the addition of distilled water (0.5 mL). The permethylated oligosaccharides were extracted into chloroform (0.25 mL) by vortex mixing. The lower organic layer containing the permethylated oligosaccharides was washed with water (3 × 0.5 mL). The reaction mixture was vortexed and allowed to proceed at room temperature for 30 min. The glycan derivatives were purified according to a hydrophilic method described previously 32 . Briefly, cellulose particles were first washed with distilled water and then with solvent mixture of 1-butanol/ethanol/H 2 O (4:4:1, v/v/v). The reaction mixture solution was mixed with 5 mg cellulose particles in 1 mL of the organic solvent described above. After gentle shaking for 45 min, the cellulose particles were washed three times with 0.5 mL of the organic solvent. The cellulose particles were incubated with 0.3 mL of a mixture of ethanol/H 2 O (1:2 v/v) for 30 min. The liquid phase was recovered and dried using a SpeedVac concentrator. All samples were redissolved in distilled water prior to MS analysis.

MALDI-MS analysis.
The MALDI-MS spectra were acquired using 4800 MALDI-TOF/TOF (SCIEX, Concord, Canada) equipped with an Nd:YAG laser with 355 nm wavelength of < 500 ps pulse and 200 Hz repetition rate. The spectrometer was operated in the positive reflectron mode. The spectra were accumulated by 1000 laser shots. The MS data were further processed using Dataexplorer 4.0. The samples were loaded onto MALDI target in 0.5 μ L of water and mixed with 0.5 μ L of freshly prepared DHB solution (10 mg/mL in 50% ACN) and allowed to dry in a gentle stream of air. Repeatability analysis. Repeatability of methylamidation conditions was investigated by multiple analyses of the same serum samples, including glycan releasing and purification. Using similar experiment design as for testing the repeatability of ethyl esterification conditions 38 , we repeated the entire procedure twice on consecutive days with freshly prepared reagents to establish day-to-day variability. For comparison, the native glycans from the same sample were also analysed using a TripleTOF 5600 System.