Chinese Chemical Letters  2019, Vol. 30 Issue (3): 676-680   PDF    
Two-step derivatization and mass spectral distinction of α2, 3 and α2, 6 sialic acid linkages on N-glycans by MALDI-TOF
Xiaoman Zhoua, Shang Yangb, Ganglong Yanga, Zengqi Tanb, Feng Guana,b,*     
a The Key Laboratory of Carbohydrate Chemistry and Biotechnology, Ministry of Education, School of Biotechnology, Jiangnan University, Wuxi 214122, China;
b Joint International Research Laboratory of Glycobiology and Medicinal Chemistry, College of Life Science, Northwest University, Xi'an 710069, China
Abstract: Sialylation is one of important glycosylation in human beings and plays an important role in cancer development. α2, 3-Linked and α2, 6-linked sialic acids are normally observed on the end of N-glycans and have different functions. Derivatization on sialic acid was designed to detect the different linkages by MALDI-TOF MS. In this study, a two-step derivatization by dimethylamine and ammonium hydroxide was improved to modify the sialic acid and made it easier to detect the different linkages of sialic acids on MALDI-TOF MS. Using this derivatization method, specific sialic acids linkages on N-glycans of protein samples such as fetuin and lactoferrinwere detected. For complex cell samples, increased α2, 3-linked and α2, 6-linked sialic acids on bi-antennary and tri-antennary N-glycans were observed in A549 cells induced by hypoxia environment. Taken together, our two-step derivatization of sialic acids offers a simple and accurate way to detect specific linkages on N-glycans with MALDI-TOF mass spectrometer.
Keywords: Sialic acid     N-Glycans     Derivatization     MALDI-TOF     Hypoxia    

Sialic acids, a monosaccharide family sharing a common 9- carbon bone structure, binding to C3 or C6 of the former sugar with its C2 under most conditions, are usually present at the end of glycochains on the surface of cells [1, 2]. The amine group on the C5 of sialic acid is mainly modified as N-acetyl or Nhydroxyacetyl, forming Neu5Ac and Neu5Gc, respectively. Neu5Ac is the most common structure of sialic acids found in human body [3]. And due to the terminal expression and wide distribution, sialic acids play a key role in various biological and pathological processes [4]. For example, some glycoproteins on the surface of influenza virus specifically bind to host epithelial cells which enriched with α2, 3-linked sialylated glycans, but not α2, 6-linked sialylated glycans [5-7].

Increased sialylation is a common phenomenon in cancer cells, and further affects human immunity and assist tumor progression [8, 9]. α-2, 6-Sialyltransferase I (ST6Gal-I) has been demonstrated as a characteristic biomarker in the migration of many cancers, such as breast cancer, colorectal cancer, cervical cancer, and others [10, 11]. Ac53FaxNeu5Ac, a sialic acid mimetic, blocks tumor sialic acid expression in vivo and suppresses tumor growth in multiple tumor models by enhancing cytotoxic CD8+ T cell-mediated killing [12]. The analysis of sialic acids has been addressed using different methods. The sialic acids which released and labelled by DMB (1, 2- diamino-4, 5-methylenedioxybenzene·2HCl) [13] can be detected by high-performance liquid chromatography (HPLC). This method offers qualitative analysis of sialic acids but without linkage information. Lectins MAL-II (Maackia amurensis lectin II) and SNA (Sambucus nigra lectin) are used to enrich α2, 3 or α2, 6-linked sialylated glycans or glycopeptides, respectively [14, 15]. However, the unspecific and low binding affinity of lectins limited the wide application. Matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) can rapidly provide profiles with information on glycan composition and is widely used for high-throughput glycomics [16]. However, negative charge and unstable glycosidic bond on sialic acids make it hard to be detected and analyzed in vacuum MALDI source [17, 18]. Thus modifications, such as amidation [19] and esterification [20] are often used to stabilize the sialic acids. In our previous research, aniline was used to modify sialic acid and enhance ionizing [21]. However, those derivatizations cannot provide sialic acid linkage information since the α2, 3-linked and α2, 6-linked sialic acids have same mass value and can hardly be distinguished by mass spectrum. Reiding et al. innovated the method of ethanol derivatization successfully differentiated α2, 3-linked and α2, 6- linked sialylated glycans [22] and revealed glycan changes on IgG happening during pregnancy [23]. Stephanie Holst et al. modified this process by adding amidation to stabilize sialic acids during MALDI-TOF MS measurements [24].

In this study, sialoglycans were derivated using a simple twostep reaction by dimethylamine and ammonium hydroxide, and the different linkages of sialic acids on N-glycans in different protein samples were determined by MALDI-TOF MS. Furthermore, altered sialylation of N-glycans was detected in hypoxia induced cancer cells using current strategy. Detailed materials and methods were provided in Supporting information.

Firstly, the structure of terminal sialic acids makes it possible to distinguish the linkage by derivatization. The α-glycosidic bond between sialic acids and the adjacent sugar residue make sialic acids unstable and readily lost both within the ion source and during the flying in the detector [25]. Derivatization approaches such as amidation [19] and esterification [20] have been introduced to stabilize sialic acids during MALDI-TOF MS based glycan analysis. α2, 3-Linked sialic acid can form a stable lactone to the neighbor galactose whereas α2, 6-linked sialic acid cannot. This character makes it possible to differ specific linkages of sialic acid on glycans. In this study, derivatization reaction was divided into two steps (Scheme 1). In step 1, glycans of glycosylated proteins were treated with 0.25 mol/L EDC, 0.5 mol/L HOBt and 0.25 mol/L dimethylamine in DMSO solution at 60 ℃ for 2 h. EDC was used as carboxylic acid activator and HOBt as catalyst. In step 2, equivalent volume of ammonium hydroxide (pH 10) was added into following reaction solution and kept at 60 ℃ for 2 h. α2, 3-Linked sialic acid formed a stable amidation structure [26], which showed a mass shift of -0.984 Da (Scheme 1A). Whereas α2, 6-linked sialic acid formed a stable dimethyl amide and remained stable in ammonium hydroxide [26]. The dimethylamidation of α2, 6-linked sialic changed glycans mass by + 27.047 Da (Scheme 1B). The distinct shifts of mass caused by derivatization on sialic acid made it possible to detect specific linkages of sialic acid on glycans through MALDI-TOF MS.

Scheme 1. Reaction scheme for the derivatization of sialic acid. (A) α2, 3-Linked sialic acid forms a stable lactone to the neighbor galactose when treated by dimethylamine under EDC + HOBt condition. (B) α2, 6-Linked sialic acid forms a stable dimethyl amide, and remains stable in ammonium hydroxide.

Two simple sialyl oligosaccharides, sialyl lactose with α2, 3- linked sialic acid and α2, 6-linked sialic acid, which share the same mass value at 656.162 Da were used to verify the above derivatization reaction (Figs. 1A and B). After derivatization, α2, 3-linked sialic acid formed lactone and hydrolyzed to amidation, but α2, 6-linked sialic acid formed dimethylamidation. As the result, the derived α2, 3-linked sialyl lactose was detected at 655.184 Da (Fig. 1C), while the derived α2, 6-linked sialyl lactose was observed at 683.150 Da (Fig. 1D). The mass derived -0.987 Da for α2, 3-linked sialyl lactose and + 26.988 Da for α2, 6-linked sialyl lactose were closely to theoretic value, and can easily be distinguished by MALDI-TOF MS.

Fig. 1. Verify the derivatization reaction by simple sialyl oligosaccharides. Standard α2, 3-linked and α2, 6-linked sialyl lactose were treated by dimethylamine under EDC + HOBt condition as described above, and detected with MALDI-TOF MS at positive ion mode. MALDI-TOF/TOF-MS spectra of standard α2, 3-linked sialyl lactose (A), standard α2, 6-linked sialyl lactose (B), derived α2, 3-linked sialyl lactose (C) and derived α2, 6-linked sialyl lactose (D).

Then, we detected sialic acids on simple protein N-glycans. Fetuin, a glycoprotein containing sialylated N-linked and O-linked glycans, was usually used for the glycomic methodologies [27]. Fetuin samples were treated by dimethylamine and N-glycans were released by PNGase F enzyme through FASP method [21]. Twelve kinds of sialyl N-glycans were detected (Fig. S1A in Supporting information) by GlycoWorkbench search engine according to MS spectra and MS2 fragments spectra. Most sialyl N-glycans were found ranging from 2000 Da to 3000 Da, and two kinds of sialic acid linkages were observed. For example, peak at 2927.448 Da was a tri-antennary N-glycans containing two α2, 3 and one α2, 6-linked sialic acids, shifting a mass of 25.450 Da which exactly obtaining a dimethylamine and losing two H2O molecules (Fig. S1A). Peak at 2955.498 Da had one α2, 3 and two α2, 6-linked sialic acids. The typical MS2 peaks of main glycans showed the accuracy derivation of the sialic acids (Fig. S1B in Supporting information). The fragmentation of selected glycans matched the predicted glycan structures, confirming the derivatization reaction is effective.

Further, we tested whether this method can be applied in complex N-glycans analysis. Lactoferrin (LF), an iron-binding glycoprotein, is one of the most important bio-activators in milk [28, 29]. N-Glycans on lactoferrin are important in resisting proteolysis, keeping solubility and inhibiting bacteria [30]. Recombinant human lactoferrin (rhLF) is also rich in N-glycosylation. The whole N-glycans on rhLF were derived and analyzed as described above. As illustrated in Fig. 2A, fifteen kinds of N-glycans were detected and five of them were α2, 6-sialylated, consistent with other reports [31]. The sialic acid linkage was further confirmed with MS2 fragmentspectra(Fig. 2B).ComparedtooriginalN-glycans or N-glycans derivated by aniline, dimethylamine treated glycans not only exhibited a relatively higher ionization efficiency of sialyl N-glycans, but also kept perfect accuracy on non-sialyl N-glycans (Fig. 2C).

Fig. 2. N-Glycans analysis of recombinant human lactoferrin. (A) Recombinant human lactoferrin was treated by dimethylamine and ammonium hydroxide for derivatization (as mentioned in section "Materials and methods" in Supporting information). N-Linked glycans were digested by PNGase F and collected through 10 kD centrifugal filter. (B) MALDI-TOF/TOF-MS/MS analysis of N-glycan precursor ions 2168.111 Da in MS spectra. Representative N-glycan spectra are shown. Detailed structures were analyzed using GlycoWorkbench software. Proposed structures are indicated by m/z and fragments mass value. (C) The comparison of N-linked glycans detection results derivated by dimethylamine (red), aniline (blue) and without derivation (green). Sialylated N-glycans were mainly distributed at mass range colored in gray.

Since the method worked well on complex protein samples, Nglycan changes of hypoxia induced tumor cells were evaluated. Glycosylation changes during tumor progression and plays important parts in tumor cell survival and metastasis [11, 32]. The metabolism of cancer cells is altered in hypoxia microenvironment [33-35]. Adenocarcinoma human alveolar basal epithelial cells A549 was used to investigate the sialylation change effected by hypoxia treatment. The N-glycans analysis procedure was defined to detected N-glycans and sialic acid linkages (Fig. 3A). HIF- 1α (hypoxia-inducible factor 1-alpha), a subunit of HIF1 which responses to systemic oxygen levels in mammals, can be used as the indicator of hypoxia [36]. The increased expression of HIF-1α confirmed that the effective hypoxia condition was used in this study (Fig. 3B). Biotinylated lectin SNA (recognizes α2, 6-linked sialic acid) and MAL-II (recognizes α2, 3-linked sialic acid) were used to distinguish the sialic acid expression. Lectin blotting results showed that both types of sialic linkages were elevated (Fig. 3B). Thirty-three kinds of N-glycans were observed in MALDI-TOF pattern (Fig. 3C). Further analysis revealed that complete sialylation modification of bi-antennary glycans without core fucose (peaks at m/z 2244.158, 2272.197, 2300.232) and triantennary glycans (peaks at m/z 2927.584, 2955.620, 2983.632) were upregulated compared to partial sialylation modification (Table S1 in Supporting information). In form of complete sialylation, every terminus of glycan branch is modified with sialic acid. Specifically, a serious of tri-antennary glycans (peaks at m/z 2899.529, 2927.584, 2955.620, 2983.632) were observed. Those glycans shared the same sequence and had the same mass value at m/z 2901.999 in non-derivated experiment. Using our approach, these glycans could be sorted to four groups (Fig. 3D), based on different sialic acid linkage information, making it possible to quantify detailed N-glycan groups with specific sialic linkages. Since both types of sialyl N-glycans were upregulated, they may perform important functions during cancer cells survival when faced with hypoxia environment.

Fig. 3. Sialyl N-glycans variation of A549 cells in hypoxia and normoxia culture condition. (A) The process of linkage-specific sialic acid derivatization on A549 cells in hypoxia and normoxia condition. A549 cells were cultured in hypoxia champer filled with 1% O2, 5% CO2 and 94% N2 mixture gas, or in normoxia condition (air containing 5% CO2) at 37 °C humidified incubator for 48 h. Cells were harvested and total protein was extracted by T-PER lysis buffer. (B) Total protein sialylation of hypoxia and normoxia treated A549 cells were detected by lectin blot. Biotinylated SNA (recognize α2,6-linked sialic acid) and MAL-II (recognize α2,3-linked sialic acid) were used to identify sialic acid linkages. Coomassie blue staining as loading control. HIF-1α blot was used as the hypoxia marker. (C) Glycans were collected through FASP method after sialic acid derivatization. Total N-glycomics were detected by MALDI-TOF mass spectrum and peaks were analyzed with GlycoWorkbench software (hypoxia in red and normoxia in gray). Proposed structures were indicated by m/z value and MS2 fragments. (D) Four triantennary glycans (peaks at m/z 2899.529, 2927.584, 2955.620, 2983.632) were specifically indicated in hypoxia and normoxia conditions.

In summary, we established a two-step derivatization method based on the previous ethanol derivatization. We applied this strategy to the analysis of N-glycans with α2, 3 and α2, 6-linked sialic acids from different samples, including standard sialyl lactose, fetuin and recombinant human lactoferrin, and the A549 cancer cells which cultured in hypoxia and normoxia conditions. Compared to aniline derivatization and non-derivatization, our strategy not only identified the α2, 3 and α2, 6-linked sialic acids on the N-glycans from the protein samples with a relative higher signal, but also kept the same accuracy in detecting non-sialyl glycans. In conclusion, this procedure provides the simple and accurate way to distinguish specific sialic linkages, and can be used in high-throughput N-glycomics study and tissue sialic acid distribution imaging in future study [37].


This study was supported by the National Natural Science Foundation of China (Nos. 81672537 and 81470294), the National Science and Technology Major Project of China (No. 2018ZX10302205) and Hundred-Talent Program of Shaanxi Province.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi :

C. Traving, R. Schauer, Cell. Mol. Life Sci. 54 (1998) 1330-1349. DOI:10.1007/s000180050258
G.J. Liu, C.Y. Li, X.T. Zhang, et al., Chin. Chem. Lett. 29 (2018) 1-10. DOI:10.1016/j.cclet.2017.09.034
R. Schauer, Adv. Carbohydr. Chem. Biochem. 40 (1982) 131-234. DOI:10.1016/S0065-2318(08)60109-2
N.M. Varki, A. Varki, Lab. Invest. 87 (2007) 851-857. DOI:10.1038/labinvest.3700656
A. Chandrasekaran, A. Srinivasan, R. Raman, et al., Nat. Biotechnol. 26 (2008) 107-113. DOI:10.1038/nbt1375
A. Varki, Trends Mol. Med. 14 (2008) 351-360. DOI:10.1016/j.molmed.2008.06.002
X. Meng, M.B. Yang, Y. Li, et al., Chin. Chem. Lett. 29 (2018) 76-80. DOI:10.1016/j.cclet.2017.10.032
O.M. Pearce, H. Laubli, Glycobiology 26 (2016) 111-128. DOI:10.1093/glycob/cwv097
H. Xiao, E.C. Woods, P. Vukojicic, et al., Proc. Natl. Acad. Sci. U. S. A. 113 (2016) 10304-10309. DOI:10.1073/pnas.1608069113
Dall'Olio F., M. Chiricolo, Glycoconjugate J. 18 (2001) 841-850. DOI:10.1023/A:1022288022969
B.N. Vajaria, P.S. Patel, Glycoconjugate J. 34 (2017) 147-156. DOI:10.1007/s10719-016-9755-2
C. Bull, T.J. Boltje, N. Balneger, et al., Cancer Res. 78 (2018) 3574-3588. DOI:10.1158/1538-7445.AM2018-3574
N.D. Pham, P.C. Pang, S. Krishnamurthy, et al., J. Biol. Chem. 292 (2017) 9637-9651. DOI:10.1074/jbc.M116.764597
C.F. Verkoelen, van der Boom B.G., D.J. Kok, et al., Kidney Int. 57 (2000) 1072-1082. DOI:10.1046/j.1523-1755.2000.00934.x
R. Thompson, A. Creavin, O'Connell M., et al., Anal. Biochem. 413 (2011) 114-122.
Z.N. Wu, N. Xu, W.W. Li, J.M. Lin, Chin. Chem. Lett. 30 (2019) 95-98. DOI:10.1016/j.cclet.2018.01.048
K. Jiang, H. Zhu, L. Li, et al., Anal. Chim. Acta 981 (2017) 53-61. DOI:10.1016/j.aca.2017.05.029
M. Toyoda, H. Ito, Y.K. Matsuno, et al., Anal. Chem. 80 (2008) 5211-5218. DOI:10.1021/ac800457a
X. Liu, H. Qiu, R.K. Lee, et al., Anal. Chem. 82 (2010) 8300-8306. DOI:10.1021/ac101831t
G.L. Yang, J.X. Zhang, C.M. Liu, et al., Prog. Biochem. Biophys. 42 (2015) 1144-1154.
K.R. Reiding, D. Blank, D.M. Kuijper, et al., Anal. Chem. 86 (2014) 5784-5793. DOI:10.1021/ac500335t
A. Bondt, Y. Rombouts, M.H. Selman, et al., Mol. Cell. Proteomics 13 (2014) 3029-3039. DOI:10.1074/mcp.M114.039537
S. Holst, B. Heijs, de Haan N., et al., Anal. Chem. 88 (2016) 5904-5913. DOI:10.1021/acs.analchem.6b00819
S.I. Snovida, Rak-Banville J.M., H. Perreault, J. Am. Soc. Mass. Spectrom. 19 (2008) 1138-1146. DOI:10.1016/j.jasms.2008.04.033
Gomez-Bombarelli R., E. Calle, J. Casado, J. Org. Chem. 78 (2013) 6868-6879. DOI:10.1021/jo400258w
X. Dong, H. Qin, J. Mao, et al., Anal. Chem. 89 (2017) 3966-3972. DOI:10.1021/acs.analchem.6b04394
L. Hao, Q. Shan, J. Wei, et al., Curr. Protein Pept. Sci. 19 (2018) 1389-2037.
M. Barboza, J. Pinzon, S. Wickramasinghe, et al., Mol. Cell. Proteomics 11 (2012) 22261723.
S. Karav, J.B. German, C. Rouquie, et al., Int. J. Mol. Sci. 18 (2017) 870.
A.L. Parc, S. Karav, C. Rouquie, et al., PLoS One 12 (2017) 28170415.
C. Bull, M.A. Stoel, den Brok M.H., et al., Cancer Res. 74 (2014) 3199-3204. DOI:10.1158/0008-5472.CAN-14-0728
D.M. Gilkes, G.L. Semenza, D. Wirtz, Nat. Rev. Cancer 14 (2014) 430-439. DOI:10.1038/nrc3726
D. Hanahan, R.A. Weinberg, Cell 144 (2011) 646-674. DOI:10.1016/j.cell.2011.02.013
Z.Q. Tan, C.X. Wang, X. Li, et al., Front. Physiol. 9 (2018) 29593568.
J.W. Kim, I. Tchernyshyov, G.L. Semenza, et al., Cell Metab. 3 (2006) 177-185. DOI:10.1016/j.cmet.2006.02.002
T. Wu, X.H. Yang, C.J. Zhang, et al., Chin. Chem. Lett. 27 (2016) 901-904. DOI:10.1016/j.cclet.2016.02.020