- Email: [email protected]
Mass Spectrometry and Liquid Chromatography with Tandem Mass Spectrometry
Analytical Biochemistry 269, 297–303 (1999) Article ID abio.1999.4026, available online at http://www.idealibrary.com on
Analysis of Carbohydrate Heterogeneity in a Glycoprotein Using Liquid Chromatography/Mass Spectrometry and Liquid Chromatography with Tandem Mass Spectrometry Nana Kawasaki, 1 Miyako Ohta, Sumiko Hyuga, Osamu Hashimoto, and Takao Hayakawa Division of Biological Chemistry and Biologicals, National Institute of Health Sciences, 1-18-1 Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan
Received October 1, 1998
High-performance liquid chromatography with online electrospray ionization mass spectrometry (ESILC/MS) was investigated for the analysis of carbohydrate heterogeneity using RNase B as a model glycoprotein. Oligosaccharides released from RNase B with endoglycosidase H were reduced and separated on a graphitized carbon column (GCC). GCC-HPLC/MS in the positive-ion mode was successful in the identification of one Man 5GlcNAc, three Man 6GlcNAc, three Man 7GlcNAc, three Man 8GlcNAc, one Man 9GlcNAc, and an oligosaccharide having six hexose units (Hex) and two N-acetylhexosamine units (HexNAc). The branch structures of the three Man 7GlcNAc isomers were determined by liquid chromatography with tandem mass spectrometry (LC/MS/MS). LC/MS/MS analysis was shown to be useful for the detection and identification of a trace amount of Hex 6HexNAc 2 alditol as a hybrid-type oligosaccharide. Its structure was confirmed by the combination of LC/MS with enzymatic digestion using b-galactosidase and N-acetyl-b-glucosaminidase. The relative quantities of high-mannose-type oligosaccharides in RNase B detected by ESI-LC/MS are in reasonable agreement with those by UV, high-pH anion-exchange chromatography with pulsed amperometric detection, fluorophore-assisted carbohydrate electrophoresis. Our results indicate that LC/MS and LC/MS/MS can be utilized to elucidate the distribution of oligosaccharides and their structures, which differ in molecular weight, sugar sequence, and branch structure. © 1999 Academic Press Key Words: ESI-LC/MS; ESI-LC/MS/MS; graphitized carbon column; oligosaccharides.
1 To whom correspondence should be addressed. Fax: 181-3-37076950.
0003-2697/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
Advances in biotechnology have made it possible to provide a variety of recombinant glycoproteins as medical agents (1). It is well known that the structure of carbohydrates in recombinant glycoproteins is influenced by a number of factors, such as the type of host cell, method of establishment of the cell seed, and culture conditions (2). The carbohydrate moieties affect the in vivo activities, metabolic fate, stability, and solubility of the glycoproteins (3), and therefore detailed information on the carbohydrate structure of glycoprotein products is important for their quality control. Mass spectrometry is one of the most useful methods for the structural analysis of carbohydrates. These are many reports on the use of off-line preparative HPLC or high-pH anion-exchange chromatography (HPAEC) 2 with subsequent mass spectrometry for the analysis of carbohydrates (4). Tandem mass spectrometry (MS/MS) techniques can give further structural information. Fast atom bombardment (FAB) was used as an ionization method in early studies of MS/MS for carbohydrates (5–7). Recently, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) (8 –11) and electrospray ionization mass spectrometry (ESI-MS) (12) have become widely used for the analysis of carbohydrates. MALDI is successfully used for distinguishing the branching isomers by a combination of postsource decay (13) or collision2 Abbreviations used: CID, collision-induced dissociation; ESI-MS, electrospray ionization mass spectrometry; Endo H, endoglycosidase H; FAB-MS, fast atom bombardment mass spectrometry; FACE, fluorophore-assisted carbohydrate electrophoresis; GCC, graphitized carbon column; Hex, hexose unit, usually mannose or galactose; HexNAc, N-acetylhexosamine unit, usually N-acetylglucosamine or N-acetylgalactosamine; HPAEC-PAD, high-pH anion-exchange chromatography with pulsed amperometric detection; MALDI-TOF-MS, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry; RNase B, Ribonuclease B; TIC, total ion current.
297
298
KAWASAKI ET AL.
induced dissociation (CID)-MS/MS using a Fourier transform mass spectrometer (14). Reinhold and coworkers reported the characterization of carbohydrate structures using a combination of permethylation and ESI-CID-MS/MS (15). However, these HPLC with offline MS or MS/MS for the structural analysis of carbohydrates in a glycoprotein are inconvenient and complicated when the glycoprotein is highly heterogeneous in the carbohydrate structure. On-line LC/MS and LC/ MS/MS would be more suitable for analyzing the heterogeneity as well as structural characterization of individual oligosaccharides in recombinant glycoprotein products. For the HPLC analysis, reversed-phase columns are well developed and are usually used for separation of carbohydrates (16). LC/MS of oligosaccharides has been carried out using reversed-phase columns (17, 18). However, their use requires a derivatization of carbohydrates with hydrophobic compounds such as 1-phenyl-3-methyl-5-pyrazolone (19) or 2-aminopyridine (20) in order to retain the carbohydrate on the column as well as for improvement of detection. The derivatization and purification of the derivatives are time consuming and sometimes not quantitative. Recently, some groups have reported that graphitized carbon columns (GCC) exhibit properties similar to reversed-phase columns, but are able to retain less hydrophobic oligosaccharides and separate closely related oligosaccharides and alditols without any derivatization (21–26). It is felt that an on-line LC/MS equipped with GCC should be capable of rapid and simple carbohydrate characterization of glycoprotein products, especially those not retained by reversedphase columns. Using ribonuclease B (RNase B) as a model glycoprotein, we report here that the GCC-HPLC with ESI-MS and with ESI-MS/MS are powerful tools for elucidating the distribution and detailed structures of oligosaccharides. MATERIALS AND METHODS
Materials RNase B from bovine pancreas was purchased from Sigma. N-Acetyl-b-D-glucosaminidase and endoglycosidase H (EndoH) were obtained from Boehringer-Mannheim. b-Galactosidase from jackbean was purchased from Seikagaku-Kogyo (Tokyo, Japan). All other chemicals used were of the highest purity available. Preparation of Oligosaccharides and Oligosaccharide Alditols RNase B (6 mg) was dissolved in 150 ml of 100 mM acetate buffer, pH 5.5. The mixture was incubated with 10 mU Endo H for 18 h at 37°C. Protein was precipi-
tated with cold ethanol (500 ml). The supernatant was dried, the oligosaccharides were dissolved in 200 ml of H 2O, and aliquots (20 ml) were injected into the LC/MS. Oligosaccharide alditols were prepared by treating the oligosaccharide described above (90 ml) with 0.25 M NaBH 4 (60 ml). The mixture was incubated for 5 h at 25°C and then 30 ml of acetone was added to the solution to decompose the excess NaBH 4. The sample (40 ml) was injected into the LC/MS. Oligomannose 7D1 and 7D3 (10 mg; Oxford GlycoSystems Inc., Rosedale, NY) were incubated with Endo H (5 mU) in 50 ml of 20 mM citrate–phosphate buffer, pH 4.5, at 37°C for 18 h. After the addition of 220 ml of ethanol, the supernatant was dried and dissolved in 50 ml of H 2O. To the oligosaccharide solution, 0.5 M NaBH 4 (50 ml) was added and the mixture was incubated at 25°C for 4 h. The workup procedure was the same as that for the oligomannose samples from RNase B. Aliquots (10 ml) were injected into the LC/MS. Enzyme Digestion of RNase B Oligosaccharides The oligosaccharides from 1 mg of RNase B were dissolved in 40 ml of 20 mM citrate–phosphate buffer, pH 4.5, and incubated with b-galactosidase (50 mU) at 37°C for 4 h or with N-acetyl-b-D-glucosaminidase (50 mU) at 37°C for 18 h. Proteins were removed by precipitation with 80% ethanol. The oligosaccharides recovered from the supernatant were dried and then reduced with 0.25 M NaBH 4 (100 ml) at 25°C for 18 h. HPLC HPLC was carried out using a Finnigan MAT spectra System consisting of a p4000 pump and a UV2000. The GCC used was a Shandon Hypercarb 5 m (4.6 3 100 mm; Shandon Scientific, UK). The eluents were 0.01% NH 4OH (A pump) and 50% CH 3CN containing 0.01% NH 4OH (B pump). For the analysis of oligosaccharides, the column was equilibrated with 7% solvent B. The oligosaccharides were eluted by a linear gradient increasing solvent B to 30% in 50 min at a flow rate of 1.0 ml/min. The oligosaccharide alditols were eluted with a gradient of 8 –15% of solvent B in 50 min at a flow rate of 1.0 ml/min. Effluent was monitored at 206 nm. ESI-MS and ESI-CID-MS/MS Mass spectra were recorded on a Finnigan TSQ 7000 triple-stage quadruple mass spectrometer equipped with an electrospray ion source (Finnigan MAT Instruments Inc., San Jose, CA). The mass spectrometer was operated in the positive-ion mode. The ESI voltage was set to 4500 V and the capillary temperature was 275°C. The pressure of the sheath gas was 70 psi and the auxiliary gas was 10 units.
MASS SPECTROMETRIC ANALYSIS OF OLIGOSACCHARIDES
299
CID-MS/MS was carried out using argon gas as the collision gas at 3.0 m Torr. A collision energy of 230 eV was used for the oligosaccharide alditols. High-pH Anion-Exchange Chromatography with Pulsed Amperometric Detection (HPAEC-PAD) HPAEC-PAD was performed using a Dionex DX-300 (Dionex Corp., Sunnyvale, CA) equipped with a Carbopac PA-1 column. The separation of oligosaccharides released from RNase B with Endo H treatment was achieved using an isocratic elution with 100 mM NaOH containing 35 mM sodium acetate. Fluorophore-Assisted Carbohydrate Electrophoresis (FACE) Oligosaccharides were labeled with 8-aminonaphthalene-1,3,6-trisulfonic acid using the FACE N-linked oligosaccharide profiling kit (Glyko Inc., Novato, CA). The derivatized oligosaccharides were separated on an oligosaccharide profiling gel. Samples were electrophoresed in 0.5-mm-thick gels with 15 mA at 10°C for 90 min. Imaging of the gel was performed using the FACE imaging system. RESULTS AND DISCUSSION
ESI-LC/MS for Oligosaccharides Ribonuclease B from bovine pancreas is reported to have a single N-glycosylation site containing highmannose oligosaccharides ranging from Man 5GlcNAc 2 to Man 9GlcNAc 2 (27–29). The oligosaccharides were released from commercially available RNase B by Endo H treatment. With an aqueous solution as an eluent, each of these reducing oligosaccharides appeared as double peaks on the GCC-HPLC due to their anomeric isomers. Fan et al. used NH 4OH containing eluent to successfully converge the double peaks to single peaks by accelerating the anomerization process (24). Therefore, LC/MS was performed using a solvent containing NH 4OH as an eluent. Figure 1B shows the total ion current (TIC) of the RNase B oligosaccharides in the positive-ion mode. Four peaks were found in TIC, and a similar elution profile was observed by UV detection at 206 nm (Fig. 1A). The positive-ion ESI mass spectra of peaks a, b, c, and d in Fig. 1 are shown in Figs. 2A, 2B, 2C, and 2D, respectively. The dominant ions were the pseudomolecular ions [M 1 H] 1, and peaks corresponding to [M 1 NH 4] 1 and dehydrated peaks were observed as minor peaks. Multiple charged ions were not detected. Figure 2A exhibits the predominant ion at m/z 1680.3, corresponding to [Man 9GlcNAc 1 H] 1. Thus, peak a is identified as Man 9GlcNAc. Intense ions in Fig. 2B had m/z 1518.7 and 1356.4, which correspond to the molecular weights of Man 8GlcNAc and Man 7GlcNAc, respec-
FIG. 1. (A) Chromatogram of RNase B oligosaccharides detected by UV at 206 nm. (B) Total ion current of RNase B oligosaccharides.
tively. Man 8GlcNAc and Man 7GlcNAc could not be separated and were coeluted as peak b under the present conditions. The pseudomolecular ions at m/z 1194.5 in Fig. 2C and at m/z 1032.4 in Fig. 2D indicate that peaks c and d are Man 6GlcNAc and Man 5GlcNAc, respectively. GCC-HPLC with ESI-MS was successful in distinguishing the size of the oligosaccharides obtained from RNase B. ESI-LC/MS for Oligosaccharide Alditols The three positional isomers of Man 7GlcNAc and Man 8GlcNAc were detected in RNase B by capillary gel electrophoresis (28) and 1H nuclear magnetic resonance (NMR) (29). However, GCC-HPLC/MS could not separate these isomers. This is possibly because the rate of anomerization in the presence of NH 4OH is still not sufficiently fast to provide sharp peaks to warrant effective separation. This is borne out by the fact that separation was improved when the oligosaccharides were reduced with NaBH 4. Twelve oligosaccharide alditols were well separated on GCC-HPLC. Figure 3B shows a base peak chromatogram of RNase B oligosaccharide alditols. The structures of the individual peaks identified by ESI-MS spectra are shown in Table 1. The pseudomolecular ions of peaks d, f, and i fit the mass of Man 7GlcNAc; thus, these were considered to be branch isomers of Man 7GlcNAc. Similar branch isomers are found in not only Man 8GlcNAc but also Man 6GlcNAc. GCC-HPLC/MS can separate and identify these isomers simultaneously.
300
KAWASAKI ET AL.
FIG. 3. (A) Chromatogram of RNase B oligosaccharide alditols detected by UV at 206 nm. (B) Base peak chromatogram of RNase B oligosaccharide alditols.
FIG. 2. Fig. 1.
ESI mass spectra of peaks a (A), b (B), c (C), and d (D) in
ESI-LC/MS/MS for Man 7GlcNAc ESI-CID-LC/MS/MS was carried out to identify the isomeric branch structures of peaks d, f, and i. The systematic nomenclature for carbohydrate fragmentation in FAB-MS/MS by Domon and Costello (30) is used for the assignment of fragmentation in ESI-MS/MS. As shown in Fig. 4, LC/MS/MS of Man 7GlcNAc isomers resulted in the cleavage of glycosidic bonds, and b- and y-type fragment ions were produced on the nonreducing and reducing ends, respectively. Scheme 1A–1C shows the structures of Man 7GlcNAc isomers characterized in RNase B by NMR (29). A fragment ion 2Man 3 (y) is likely to arise easily from Man-7D3 in Scheme 1 by cleavage of either the a1-3-linked branch arm or the a1-6-linked branch arm of the core mannose. Likewise, the cleavage of a glycoside bond in the core mannose of Man-7D1 and Man-7D2 will produce predominantly either 2Man 2 (y) or 2Man 4 (y). In the MS/MS spectrum of peak f (Fig. 4B), the 2Man 3 (y) ion at m/z 872 is significantly more intense than 2Man 2 (y)
and 2Man 4 (y), and this fragmentation suggests that peak f could be Man-7D3. In contrast, the intense ion at m/z 710 in Fig. 4C is assigned to the 2Man 4 (y) ion, and the 2Man 2 (y) ion at m/z 1034 is more intense than the 2Man 3 (y) ion. The distribution of the fragment ions in Fig. 4A is more similar to that in Fig. 4C than to that in Fig. 4B. These results suggest that peaks d and i could be Man-7D1/2. The branch structure of peaks d, f, and i were confirmed by comparison of their retention times with TABLE 1
Structural Assignments of the Peaks from ESI-LC/MS of RNase B Oligosaccharide Alditols in Fig. 3 Carbohydrate structure Man 5GlcNAc Man 6GlcNAc
Man 7GlcNAc
Man 8GlcNAc
Man 9GlcNAc Hex 6HexNAc 2
Peak j e g k d f i b c h a l
Calculated mass
Observed mass
1034.4 1196.4
1034.5 1196.5 1196.6 1196.7 1359.1 1358.7 1358.6 1521.4 1520.7 1521.4 1683.4 1400.2
1358.4
1520.5
1682.6 1399.5
MASS SPECTROMETRIC ANALYSIS OF OLIGOSACCHARIDES
301
Man-7D3 were also similar to those of peaks i and f, respectively. These results coincide with the structures expected from the MS/MS spectra. Our results demonstrate that GCC-HPLC with MS/MS can characterize the branch structure of isomers directly without any fractionation of the individual oligosaccharides. ESI-LC/MS/MS of Hex 6HexNAc 2
FIG. 4. Product ion spectra of peaks d (A), f (B), and i (C) in Fig. 3 by ESI-CID-MS/MS.
those of Man-7D1 and Man-7D3 prepared from two commercially available structurally known Man 7GlcNAc 2 isomers. The retention times of Man7D1 and Man-7D3 corresponded to those of peaks i and f, respectively. Product ion spectra of Man-7D1 and
The pseudomolecular ion at m/z 1400 in the mass spectrum of peak l does not correspond to the molecular weight of Man 5–9GlcNAc but does reveal the structure of Hex 6HexNAc 2 aldiol (Table 1). Figure 5 was obtained using LC/MS/MS for peak l in the positive-ion mode. The b-type fragment ions containing HexNAc and y-type fragment ions having a GlcNAc alditol arose from peak l. The y-type fragment ions were observed in a range from GlcNAc alditol to Hex 5GlcNAc alditol but not Hex 6GlcNAc alditol. Fragments containing HexNAc tend to allow intense peaks; however, HexNAc (b) ion was not detected in Fig. 5. From the distribution of the fragment ions, it is suggested that HexNAc is in a position next to a nonreducing end, and therefore peak l could be a hybrid-type oligosaccharide as indicated in Scheme 1D. The product ion at m/z 913 can be a fragment of Gal-GlcNAc-Man-Man-GlcNAc alditol if one can assume that peak l is the hybrid-type oligosaccharide alditol. To confirm the structure of peak l, RNase B oligosaccharides were treated with b-galactosidase or N-acetylb-glucosaminidase, and then the digested oligosaccharide alditols were analyzed by LC/MS. Peak l disappeared after the b-galactosidase treatment but resisted the N-acetyl-b-glucosaminidase digestion. These results support the notion that peak l is a hybrid-type oligosaccharide which appeared in Scheme 1D. There have been no reports on the presence of hybrid-type oligosaccharides in RNase B to the best of our knowledge. We have demonstrated that LC/MS/MS
SCHEME 1. The proposed structures of Man 7GlcNAc 2 isomers (A–C) and a hybrid-type oligosaccharide (D).
302
KAWASAKI ET AL.
FIG. 5. Product ion spectrum of peak l by ESI-CID-MS/MS.
enables us to identify even the structure of a minor oligosaccharide. The Response of Oligosaccharides on ESI-LC/MS To assess the response factors of individual highmannose oligosaccharides, quantitative data of Man 5–9 GlcNAc obtained by ESI-LC/MS was compared with those by UV at 206 nm and HPAEC-PAD and FACE (Table 2). The relative peak areas of the oligosaccharide alditols obtained by ESI-LC/MS are in reasonable agreement with the relative values of the aldoses detected by other analytical methods. This result suggests that the response factor for each high-mannosetype oligosaccharide is nearly the same; therefore, LC/MS can be utilized for analysis of the relative amounts of various high-mannose oligosaccharides in a glycoprotein.
CONCLUSION
GCC-HPLC with on-line ESI-MS and ESI-MS/MS can provide information on the structure of oligosaccharides and their distribution in a glycoprotein. MS or MS/MS alone cannot demonstrate the distribution of oligosaccharides, especially isomers, whereas HPLC cannot identify the carbohydrate structures. Our method is simple, rapid, and applicable for the evaluation of recombinant glycoprotein products which are required for structural and quantitative constancy in their carbohydrates. The power of this technique is demonstrated by characterization of the size, sugar sequences, and branch structures of closely related oligosaccharides without any derivatizations. We are studying LC/MS and LC/MS/MS for complex-type sialylated oligosaccharides from recombinant glycoprotein products. ACKNOWLEDGMENT
TABLE 2
Quantitative Comparison of RNase B Oligosaccharides
Man 5GlcNAc Man 6GlcNAc Man 7GlcNAc Man 8GlcNAc Man 9GlcNAc
ESI-MS a
UV a
PAD b
FACE c
100% 67.2 6 4.9 19.9 6 4.4 20.8 6 4.9 4.5 6 0.7
100% 69.4 6 4.8 ND d 30.6 6 4.8 ND
100% 63.9 6 2.1 20.1 6 2.2 27.2 6 2.6 7.3 6 1.7
100% 65.7 6 2.2 18.6 6 2.3 16.6 6 0.7 3.1 6 1.1
Relative peaks areas (Fig. 3). Values are means 6 SD from triplicate analysis. b Values are means 6 SD from quintuple analysis. c Values are means 6 SD from triplicate analysis. d Peak not detected. a
This work was supported by a grant-in-aid from the Ministry of Health and Welfare Science Research Fund Subsidy granted to the Japan Health Science Foundation.
REFERENCES 1. Werner, R. G., and Noe, W. (1993) Arzneim. Forsch./Drug Res. 43, 1338 –1391. 2. Hayakawa, T. (1991) in Drug Biotechnology Regulation: Scientific Basis and Practices (Chiu, Y.-y. H., and Gueriguian, J. L., Eds.), pp. 468 – 498, Dekker, New York. 3. Varki, A. (1993) Glycobiology 3, 97–130. 4. Dell, A., Khoo, K.-H., Panico, M., McDowell, R. A., Etienne, A. T., Reason, A. J., and Morris, H. R. (1993) in Glycobiology: A Practical Approach (Fukuda, M., and Kobata, A., Eds.), pp. 187–222, Oxford Univ. Press, New York.
MASS SPECTROMETRIC ANALYSIS OF OLIGOSACCHARIDES 5. Laine, R. A., Pamidimukkala, K. M., French, A. D., Hall, R. W., Abbas, S. A., Jain, R. K., and Matta, K. L. (1988) J. Am. Chem. Soc. 110, 6931– 6939. 6. Hofmeister, G. E., Zhou, Z., and Leary, J. A. (1991) J. Am. Chem. Soc. 113, 5964 –5970. 7. Domon, B., and Costello, C. E. (1988) Biochemistry 27, 1534 – 1543. 8. Pitt, J. J., and Gorman, J. J. (1997) Anal. Biochem. 248, 63–75. 9. Ashton, D. S., Beddell, C. R., Cooper, D. J., and Lines, A. C. (1995) Anal. Chim. Acta 306, 43– 48. 10. Whittal, R. M., Palcic, M. M., Hindsgaul, O., and Li, L. (1995) Anal. Chem. 67, 3509 –3514. 11. Stahl, B., Thurl, S., Zeng, J., Karas, M., Hillenkamp, F., Steup, M., and Sawatzki, G. (1994) Anal. Biochem. 223, 218 –226. 12. Gelpi, E. (1995) J. Chromatogr. A 703, 59 – 80. 13. Rouse, J. C., Strang, A.-M., Yu, W., and Vath, J. E. (1998) Anal. Biochem. 256, 33– 46. 14. Penn, S. G., Cancilla, M. T., and Lebrilla, C. B. (1996) Anal. Chem. 68, 2331–2339. 15. Reinhold, V. N., Reinhold, B. B., and Costello, C. E. (1995) Anal. Chem. 67, 1772–1784. 16. Tomiya, N., Awaya, J., Kurono, M., Endo, S., Arata, Y., and Takahashi, N. (1988) Anal. Biochem. 171, 73–90. 17. Suzuki, S., Kakehi, K., and Honda, S. (1996) Anal. Chem. 68, 2073–2083.
303
18. Suzuki-Sawada, J., Umeda, Y., Kondo, A., and Kato, I. (1992) Anal. Biochem. 207, 203–207. 19. Honda, S., Akao, E., Suzuki, S., Okuda, M., Kakehi, K., and Nakamura, J. (1989) Anal. Biochem. 180, 351–357. 20. Hase, S., Ibuki, T., and Ikenaka, T. (1984) J. Biochem. 95, 197–203. 21. Davies, M., Smith, K. D., Harbin, A.-M., and Hounsell, E. F. (1992) J. Chromatogr. 609, 125–131. 22. Davies, M. J., Smith, K. D., Carruthers, R. A., Chai, W., Lawson, A. M., and Hounsell, E. F. (1993) J. Chromatogr. 646, 317–326. 23. Davies, M. J., and Hounsell, E. F. (1996) J. Chromatogr. A 720, 227–233. 24. Fan, J.-Q., Kondo, A., Kato, I., and Lee, Y. C. (1994) Anal. Biochem. 219, 224 –229. 25. Koizumi, K. (1996) J. Chromatogr. A 720, 119 –126. 26. Lipniunas, P. H., Neville, D. C. A., Trimble, R. B., and Townsend, R. R. (1996) Anal. Biochem. 243, 203–209. 27. Liang, C.-J., Yamashita, K., and Kobata, A. (1980) J. Biochem. 88, 51–58. 28. Guttman, A., and Pritchett, T. (1995) Electrophoresis 16, 1906 – 1911. 29. Fu, D., Chen, L., and O’Neill, R. A. (1994) Carbohydr. Res. 261, 173–186. 30. Domon, B., and Costello, C. E. (1988) Glycoconj. J. 5, 397– 409.
Analysis of Carbohydrate Heterogeneity in a Glycoprotein Using Liquid Chromatography/Mass Spectrometry and Liquid Chromatography with Tandem Mass Spectrometry Nana Kawasaki, 1 Miyako Ohta, Sumiko Hyuga, Osamu Hashimoto, and Takao Hayakawa Division of Biological Chemistry and Biologicals, National Institute of Health Sciences, 1-18-1 Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan
Received October 1, 1998
High-performance liquid chromatography with online electrospray ionization mass spectrometry (ESILC/MS) was investigated for the analysis of carbohydrate heterogeneity using RNase B as a model glycoprotein. Oligosaccharides released from RNase B with endoglycosidase H were reduced and separated on a graphitized carbon column (GCC). GCC-HPLC/MS in the positive-ion mode was successful in the identification of one Man 5GlcNAc, three Man 6GlcNAc, three Man 7GlcNAc, three Man 8GlcNAc, one Man 9GlcNAc, and an oligosaccharide having six hexose units (Hex) and two N-acetylhexosamine units (HexNAc). The branch structures of the three Man 7GlcNAc isomers were determined by liquid chromatography with tandem mass spectrometry (LC/MS/MS). LC/MS/MS analysis was shown to be useful for the detection and identification of a trace amount of Hex 6HexNAc 2 alditol as a hybrid-type oligosaccharide. Its structure was confirmed by the combination of LC/MS with enzymatic digestion using b-galactosidase and N-acetyl-b-glucosaminidase. The relative quantities of high-mannose-type oligosaccharides in RNase B detected by ESI-LC/MS are in reasonable agreement with those by UV, high-pH anion-exchange chromatography with pulsed amperometric detection, fluorophore-assisted carbohydrate electrophoresis. Our results indicate that LC/MS and LC/MS/MS can be utilized to elucidate the distribution of oligosaccharides and their structures, which differ in molecular weight, sugar sequence, and branch structure. © 1999 Academic Press Key Words: ESI-LC/MS; ESI-LC/MS/MS; graphitized carbon column; oligosaccharides.
1 To whom correspondence should be addressed. Fax: 181-3-37076950.
0003-2697/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
Advances in biotechnology have made it possible to provide a variety of recombinant glycoproteins as medical agents (1). It is well known that the structure of carbohydrates in recombinant glycoproteins is influenced by a number of factors, such as the type of host cell, method of establishment of the cell seed, and culture conditions (2). The carbohydrate moieties affect the in vivo activities, metabolic fate, stability, and solubility of the glycoproteins (3), and therefore detailed information on the carbohydrate structure of glycoprotein products is important for their quality control. Mass spectrometry is one of the most useful methods for the structural analysis of carbohydrates. These are many reports on the use of off-line preparative HPLC or high-pH anion-exchange chromatography (HPAEC) 2 with subsequent mass spectrometry for the analysis of carbohydrates (4). Tandem mass spectrometry (MS/MS) techniques can give further structural information. Fast atom bombardment (FAB) was used as an ionization method in early studies of MS/MS for carbohydrates (5–7). Recently, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) (8 –11) and electrospray ionization mass spectrometry (ESI-MS) (12) have become widely used for the analysis of carbohydrates. MALDI is successfully used for distinguishing the branching isomers by a combination of postsource decay (13) or collision2 Abbreviations used: CID, collision-induced dissociation; ESI-MS, electrospray ionization mass spectrometry; Endo H, endoglycosidase H; FAB-MS, fast atom bombardment mass spectrometry; FACE, fluorophore-assisted carbohydrate electrophoresis; GCC, graphitized carbon column; Hex, hexose unit, usually mannose or galactose; HexNAc, N-acetylhexosamine unit, usually N-acetylglucosamine or N-acetylgalactosamine; HPAEC-PAD, high-pH anion-exchange chromatography with pulsed amperometric detection; MALDI-TOF-MS, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry; RNase B, Ribonuclease B; TIC, total ion current.
297
298
KAWASAKI ET AL.
induced dissociation (CID)-MS/MS using a Fourier transform mass spectrometer (14). Reinhold and coworkers reported the characterization of carbohydrate structures using a combination of permethylation and ESI-CID-MS/MS (15). However, these HPLC with offline MS or MS/MS for the structural analysis of carbohydrates in a glycoprotein are inconvenient and complicated when the glycoprotein is highly heterogeneous in the carbohydrate structure. On-line LC/MS and LC/ MS/MS would be more suitable for analyzing the heterogeneity as well as structural characterization of individual oligosaccharides in recombinant glycoprotein products. For the HPLC analysis, reversed-phase columns are well developed and are usually used for separation of carbohydrates (16). LC/MS of oligosaccharides has been carried out using reversed-phase columns (17, 18). However, their use requires a derivatization of carbohydrates with hydrophobic compounds such as 1-phenyl-3-methyl-5-pyrazolone (19) or 2-aminopyridine (20) in order to retain the carbohydrate on the column as well as for improvement of detection. The derivatization and purification of the derivatives are time consuming and sometimes not quantitative. Recently, some groups have reported that graphitized carbon columns (GCC) exhibit properties similar to reversed-phase columns, but are able to retain less hydrophobic oligosaccharides and separate closely related oligosaccharides and alditols without any derivatization (21–26). It is felt that an on-line LC/MS equipped with GCC should be capable of rapid and simple carbohydrate characterization of glycoprotein products, especially those not retained by reversedphase columns. Using ribonuclease B (RNase B) as a model glycoprotein, we report here that the GCC-HPLC with ESI-MS and with ESI-MS/MS are powerful tools for elucidating the distribution and detailed structures of oligosaccharides. MATERIALS AND METHODS
Materials RNase B from bovine pancreas was purchased from Sigma. N-Acetyl-b-D-glucosaminidase and endoglycosidase H (EndoH) were obtained from Boehringer-Mannheim. b-Galactosidase from jackbean was purchased from Seikagaku-Kogyo (Tokyo, Japan). All other chemicals used were of the highest purity available. Preparation of Oligosaccharides and Oligosaccharide Alditols RNase B (6 mg) was dissolved in 150 ml of 100 mM acetate buffer, pH 5.5. The mixture was incubated with 10 mU Endo H for 18 h at 37°C. Protein was precipi-
tated with cold ethanol (500 ml). The supernatant was dried, the oligosaccharides were dissolved in 200 ml of H 2O, and aliquots (20 ml) were injected into the LC/MS. Oligosaccharide alditols were prepared by treating the oligosaccharide described above (90 ml) with 0.25 M NaBH 4 (60 ml). The mixture was incubated for 5 h at 25°C and then 30 ml of acetone was added to the solution to decompose the excess NaBH 4. The sample (40 ml) was injected into the LC/MS. Oligomannose 7D1 and 7D3 (10 mg; Oxford GlycoSystems Inc., Rosedale, NY) were incubated with Endo H (5 mU) in 50 ml of 20 mM citrate–phosphate buffer, pH 4.5, at 37°C for 18 h. After the addition of 220 ml of ethanol, the supernatant was dried and dissolved in 50 ml of H 2O. To the oligosaccharide solution, 0.5 M NaBH 4 (50 ml) was added and the mixture was incubated at 25°C for 4 h. The workup procedure was the same as that for the oligomannose samples from RNase B. Aliquots (10 ml) were injected into the LC/MS. Enzyme Digestion of RNase B Oligosaccharides The oligosaccharides from 1 mg of RNase B were dissolved in 40 ml of 20 mM citrate–phosphate buffer, pH 4.5, and incubated with b-galactosidase (50 mU) at 37°C for 4 h or with N-acetyl-b-D-glucosaminidase (50 mU) at 37°C for 18 h. Proteins were removed by precipitation with 80% ethanol. The oligosaccharides recovered from the supernatant were dried and then reduced with 0.25 M NaBH 4 (100 ml) at 25°C for 18 h. HPLC HPLC was carried out using a Finnigan MAT spectra System consisting of a p4000 pump and a UV2000. The GCC used was a Shandon Hypercarb 5 m (4.6 3 100 mm; Shandon Scientific, UK). The eluents were 0.01% NH 4OH (A pump) and 50% CH 3CN containing 0.01% NH 4OH (B pump). For the analysis of oligosaccharides, the column was equilibrated with 7% solvent B. The oligosaccharides were eluted by a linear gradient increasing solvent B to 30% in 50 min at a flow rate of 1.0 ml/min. The oligosaccharide alditols were eluted with a gradient of 8 –15% of solvent B in 50 min at a flow rate of 1.0 ml/min. Effluent was monitored at 206 nm. ESI-MS and ESI-CID-MS/MS Mass spectra were recorded on a Finnigan TSQ 7000 triple-stage quadruple mass spectrometer equipped with an electrospray ion source (Finnigan MAT Instruments Inc., San Jose, CA). The mass spectrometer was operated in the positive-ion mode. The ESI voltage was set to 4500 V and the capillary temperature was 275°C. The pressure of the sheath gas was 70 psi and the auxiliary gas was 10 units.
MASS SPECTROMETRIC ANALYSIS OF OLIGOSACCHARIDES
299
CID-MS/MS was carried out using argon gas as the collision gas at 3.0 m Torr. A collision energy of 230 eV was used for the oligosaccharide alditols. High-pH Anion-Exchange Chromatography with Pulsed Amperometric Detection (HPAEC-PAD) HPAEC-PAD was performed using a Dionex DX-300 (Dionex Corp., Sunnyvale, CA) equipped with a Carbopac PA-1 column. The separation of oligosaccharides released from RNase B with Endo H treatment was achieved using an isocratic elution with 100 mM NaOH containing 35 mM sodium acetate. Fluorophore-Assisted Carbohydrate Electrophoresis (FACE) Oligosaccharides were labeled with 8-aminonaphthalene-1,3,6-trisulfonic acid using the FACE N-linked oligosaccharide profiling kit (Glyko Inc., Novato, CA). The derivatized oligosaccharides were separated on an oligosaccharide profiling gel. Samples were electrophoresed in 0.5-mm-thick gels with 15 mA at 10°C for 90 min. Imaging of the gel was performed using the FACE imaging system. RESULTS AND DISCUSSION
ESI-LC/MS for Oligosaccharides Ribonuclease B from bovine pancreas is reported to have a single N-glycosylation site containing highmannose oligosaccharides ranging from Man 5GlcNAc 2 to Man 9GlcNAc 2 (27–29). The oligosaccharides were released from commercially available RNase B by Endo H treatment. With an aqueous solution as an eluent, each of these reducing oligosaccharides appeared as double peaks on the GCC-HPLC due to their anomeric isomers. Fan et al. used NH 4OH containing eluent to successfully converge the double peaks to single peaks by accelerating the anomerization process (24). Therefore, LC/MS was performed using a solvent containing NH 4OH as an eluent. Figure 1B shows the total ion current (TIC) of the RNase B oligosaccharides in the positive-ion mode. Four peaks were found in TIC, and a similar elution profile was observed by UV detection at 206 nm (Fig. 1A). The positive-ion ESI mass spectra of peaks a, b, c, and d in Fig. 1 are shown in Figs. 2A, 2B, 2C, and 2D, respectively. The dominant ions were the pseudomolecular ions [M 1 H] 1, and peaks corresponding to [M 1 NH 4] 1 and dehydrated peaks were observed as minor peaks. Multiple charged ions were not detected. Figure 2A exhibits the predominant ion at m/z 1680.3, corresponding to [Man 9GlcNAc 1 H] 1. Thus, peak a is identified as Man 9GlcNAc. Intense ions in Fig. 2B had m/z 1518.7 and 1356.4, which correspond to the molecular weights of Man 8GlcNAc and Man 7GlcNAc, respec-
FIG. 1. (A) Chromatogram of RNase B oligosaccharides detected by UV at 206 nm. (B) Total ion current of RNase B oligosaccharides.
tively. Man 8GlcNAc and Man 7GlcNAc could not be separated and were coeluted as peak b under the present conditions. The pseudomolecular ions at m/z 1194.5 in Fig. 2C and at m/z 1032.4 in Fig. 2D indicate that peaks c and d are Man 6GlcNAc and Man 5GlcNAc, respectively. GCC-HPLC with ESI-MS was successful in distinguishing the size of the oligosaccharides obtained from RNase B. ESI-LC/MS for Oligosaccharide Alditols The three positional isomers of Man 7GlcNAc and Man 8GlcNAc were detected in RNase B by capillary gel electrophoresis (28) and 1H nuclear magnetic resonance (NMR) (29). However, GCC-HPLC/MS could not separate these isomers. This is possibly because the rate of anomerization in the presence of NH 4OH is still not sufficiently fast to provide sharp peaks to warrant effective separation. This is borne out by the fact that separation was improved when the oligosaccharides were reduced with NaBH 4. Twelve oligosaccharide alditols were well separated on GCC-HPLC. Figure 3B shows a base peak chromatogram of RNase B oligosaccharide alditols. The structures of the individual peaks identified by ESI-MS spectra are shown in Table 1. The pseudomolecular ions of peaks d, f, and i fit the mass of Man 7GlcNAc; thus, these were considered to be branch isomers of Man 7GlcNAc. Similar branch isomers are found in not only Man 8GlcNAc but also Man 6GlcNAc. GCC-HPLC/MS can separate and identify these isomers simultaneously.
300
KAWASAKI ET AL.
FIG. 3. (A) Chromatogram of RNase B oligosaccharide alditols detected by UV at 206 nm. (B) Base peak chromatogram of RNase B oligosaccharide alditols.
FIG. 2. Fig. 1.
ESI mass spectra of peaks a (A), b (B), c (C), and d (D) in
ESI-LC/MS/MS for Man 7GlcNAc ESI-CID-LC/MS/MS was carried out to identify the isomeric branch structures of peaks d, f, and i. The systematic nomenclature for carbohydrate fragmentation in FAB-MS/MS by Domon and Costello (30) is used for the assignment of fragmentation in ESI-MS/MS. As shown in Fig. 4, LC/MS/MS of Man 7GlcNAc isomers resulted in the cleavage of glycosidic bonds, and b- and y-type fragment ions were produced on the nonreducing and reducing ends, respectively. Scheme 1A–1C shows the structures of Man 7GlcNAc isomers characterized in RNase B by NMR (29). A fragment ion 2Man 3 (y) is likely to arise easily from Man-7D3 in Scheme 1 by cleavage of either the a1-3-linked branch arm or the a1-6-linked branch arm of the core mannose. Likewise, the cleavage of a glycoside bond in the core mannose of Man-7D1 and Man-7D2 will produce predominantly either 2Man 2 (y) or 2Man 4 (y). In the MS/MS spectrum of peak f (Fig. 4B), the 2Man 3 (y) ion at m/z 872 is significantly more intense than 2Man 2 (y)
and 2Man 4 (y), and this fragmentation suggests that peak f could be Man-7D3. In contrast, the intense ion at m/z 710 in Fig. 4C is assigned to the 2Man 4 (y) ion, and the 2Man 2 (y) ion at m/z 1034 is more intense than the 2Man 3 (y) ion. The distribution of the fragment ions in Fig. 4A is more similar to that in Fig. 4C than to that in Fig. 4B. These results suggest that peaks d and i could be Man-7D1/2. The branch structure of peaks d, f, and i were confirmed by comparison of their retention times with TABLE 1
Structural Assignments of the Peaks from ESI-LC/MS of RNase B Oligosaccharide Alditols in Fig. 3 Carbohydrate structure Man 5GlcNAc Man 6GlcNAc
Man 7GlcNAc
Man 8GlcNAc
Man 9GlcNAc Hex 6HexNAc 2
Peak j e g k d f i b c h a l
Calculated mass
Observed mass
1034.4 1196.4
1034.5 1196.5 1196.6 1196.7 1359.1 1358.7 1358.6 1521.4 1520.7 1521.4 1683.4 1400.2
1358.4
1520.5
1682.6 1399.5
MASS SPECTROMETRIC ANALYSIS OF OLIGOSACCHARIDES
301
Man-7D3 were also similar to those of peaks i and f, respectively. These results coincide with the structures expected from the MS/MS spectra. Our results demonstrate that GCC-HPLC with MS/MS can characterize the branch structure of isomers directly without any fractionation of the individual oligosaccharides. ESI-LC/MS/MS of Hex 6HexNAc 2
FIG. 4. Product ion spectra of peaks d (A), f (B), and i (C) in Fig. 3 by ESI-CID-MS/MS.
those of Man-7D1 and Man-7D3 prepared from two commercially available structurally known Man 7GlcNAc 2 isomers. The retention times of Man7D1 and Man-7D3 corresponded to those of peaks i and f, respectively. Product ion spectra of Man-7D1 and
The pseudomolecular ion at m/z 1400 in the mass spectrum of peak l does not correspond to the molecular weight of Man 5–9GlcNAc but does reveal the structure of Hex 6HexNAc 2 aldiol (Table 1). Figure 5 was obtained using LC/MS/MS for peak l in the positive-ion mode. The b-type fragment ions containing HexNAc and y-type fragment ions having a GlcNAc alditol arose from peak l. The y-type fragment ions were observed in a range from GlcNAc alditol to Hex 5GlcNAc alditol but not Hex 6GlcNAc alditol. Fragments containing HexNAc tend to allow intense peaks; however, HexNAc (b) ion was not detected in Fig. 5. From the distribution of the fragment ions, it is suggested that HexNAc is in a position next to a nonreducing end, and therefore peak l could be a hybrid-type oligosaccharide as indicated in Scheme 1D. The product ion at m/z 913 can be a fragment of Gal-GlcNAc-Man-Man-GlcNAc alditol if one can assume that peak l is the hybrid-type oligosaccharide alditol. To confirm the structure of peak l, RNase B oligosaccharides were treated with b-galactosidase or N-acetylb-glucosaminidase, and then the digested oligosaccharide alditols were analyzed by LC/MS. Peak l disappeared after the b-galactosidase treatment but resisted the N-acetyl-b-glucosaminidase digestion. These results support the notion that peak l is a hybrid-type oligosaccharide which appeared in Scheme 1D. There have been no reports on the presence of hybrid-type oligosaccharides in RNase B to the best of our knowledge. We have demonstrated that LC/MS/MS
SCHEME 1. The proposed structures of Man 7GlcNAc 2 isomers (A–C) and a hybrid-type oligosaccharide (D).
302
KAWASAKI ET AL.
FIG. 5. Product ion spectrum of peak l by ESI-CID-MS/MS.
enables us to identify even the structure of a minor oligosaccharide. The Response of Oligosaccharides on ESI-LC/MS To assess the response factors of individual highmannose oligosaccharides, quantitative data of Man 5–9 GlcNAc obtained by ESI-LC/MS was compared with those by UV at 206 nm and HPAEC-PAD and FACE (Table 2). The relative peak areas of the oligosaccharide alditols obtained by ESI-LC/MS are in reasonable agreement with the relative values of the aldoses detected by other analytical methods. This result suggests that the response factor for each high-mannosetype oligosaccharide is nearly the same; therefore, LC/MS can be utilized for analysis of the relative amounts of various high-mannose oligosaccharides in a glycoprotein.
CONCLUSION
GCC-HPLC with on-line ESI-MS and ESI-MS/MS can provide information on the structure of oligosaccharides and their distribution in a glycoprotein. MS or MS/MS alone cannot demonstrate the distribution of oligosaccharides, especially isomers, whereas HPLC cannot identify the carbohydrate structures. Our method is simple, rapid, and applicable for the evaluation of recombinant glycoprotein products which are required for structural and quantitative constancy in their carbohydrates. The power of this technique is demonstrated by characterization of the size, sugar sequences, and branch structures of closely related oligosaccharides without any derivatizations. We are studying LC/MS and LC/MS/MS for complex-type sialylated oligosaccharides from recombinant glycoprotein products. ACKNOWLEDGMENT
TABLE 2
Quantitative Comparison of RNase B Oligosaccharides
Man 5GlcNAc Man 6GlcNAc Man 7GlcNAc Man 8GlcNAc Man 9GlcNAc
ESI-MS a
UV a
PAD b
FACE c
100% 67.2 6 4.9 19.9 6 4.4 20.8 6 4.9 4.5 6 0.7
100% 69.4 6 4.8 ND d 30.6 6 4.8 ND
100% 63.9 6 2.1 20.1 6 2.2 27.2 6 2.6 7.3 6 1.7
100% 65.7 6 2.2 18.6 6 2.3 16.6 6 0.7 3.1 6 1.1
Relative peaks areas (Fig. 3). Values are means 6 SD from triplicate analysis. b Values are means 6 SD from quintuple analysis. c Values are means 6 SD from triplicate analysis. d Peak not detected. a
This work was supported by a grant-in-aid from the Ministry of Health and Welfare Science Research Fund Subsidy granted to the Japan Health Science Foundation.
REFERENCES 1. Werner, R. G., and Noe, W. (1993) Arzneim. Forsch./Drug Res. 43, 1338 –1391. 2. Hayakawa, T. (1991) in Drug Biotechnology Regulation: Scientific Basis and Practices (Chiu, Y.-y. H., and Gueriguian, J. L., Eds.), pp. 468 – 498, Dekker, New York. 3. Varki, A. (1993) Glycobiology 3, 97–130. 4. Dell, A., Khoo, K.-H., Panico, M., McDowell, R. A., Etienne, A. T., Reason, A. J., and Morris, H. R. (1993) in Glycobiology: A Practical Approach (Fukuda, M., and Kobata, A., Eds.), pp. 187–222, Oxford Univ. Press, New York.
MASS SPECTROMETRIC ANALYSIS OF OLIGOSACCHARIDES 5. Laine, R. A., Pamidimukkala, K. M., French, A. D., Hall, R. W., Abbas, S. A., Jain, R. K., and Matta, K. L. (1988) J. Am. Chem. Soc. 110, 6931– 6939. 6. Hofmeister, G. E., Zhou, Z., and Leary, J. A. (1991) J. Am. Chem. Soc. 113, 5964 –5970. 7. Domon, B., and Costello, C. E. (1988) Biochemistry 27, 1534 – 1543. 8. Pitt, J. J., and Gorman, J. J. (1997) Anal. Biochem. 248, 63–75. 9. Ashton, D. S., Beddell, C. R., Cooper, D. J., and Lines, A. C. (1995) Anal. Chim. Acta 306, 43– 48. 10. Whittal, R. M., Palcic, M. M., Hindsgaul, O., and Li, L. (1995) Anal. Chem. 67, 3509 –3514. 11. Stahl, B., Thurl, S., Zeng, J., Karas, M., Hillenkamp, F., Steup, M., and Sawatzki, G. (1994) Anal. Biochem. 223, 218 –226. 12. Gelpi, E. (1995) J. Chromatogr. A 703, 59 – 80. 13. Rouse, J. C., Strang, A.-M., Yu, W., and Vath, J. E. (1998) Anal. Biochem. 256, 33– 46. 14. Penn, S. G., Cancilla, M. T., and Lebrilla, C. B. (1996) Anal. Chem. 68, 2331–2339. 15. Reinhold, V. N., Reinhold, B. B., and Costello, C. E. (1995) Anal. Chem. 67, 1772–1784. 16. Tomiya, N., Awaya, J., Kurono, M., Endo, S., Arata, Y., and Takahashi, N. (1988) Anal. Biochem. 171, 73–90. 17. Suzuki, S., Kakehi, K., and Honda, S. (1996) Anal. Chem. 68, 2073–2083.
303
18. Suzuki-Sawada, J., Umeda, Y., Kondo, A., and Kato, I. (1992) Anal. Biochem. 207, 203–207. 19. Honda, S., Akao, E., Suzuki, S., Okuda, M., Kakehi, K., and Nakamura, J. (1989) Anal. Biochem. 180, 351–357. 20. Hase, S., Ibuki, T., and Ikenaka, T. (1984) J. Biochem. 95, 197–203. 21. Davies, M., Smith, K. D., Harbin, A.-M., and Hounsell, E. F. (1992) J. Chromatogr. 609, 125–131. 22. Davies, M. J., Smith, K. D., Carruthers, R. A., Chai, W., Lawson, A. M., and Hounsell, E. F. (1993) J. Chromatogr. 646, 317–326. 23. Davies, M. J., and Hounsell, E. F. (1996) J. Chromatogr. A 720, 227–233. 24. Fan, J.-Q., Kondo, A., Kato, I., and Lee, Y. C. (1994) Anal. Biochem. 219, 224 –229. 25. Koizumi, K. (1996) J. Chromatogr. A 720, 119 –126. 26. Lipniunas, P. H., Neville, D. C. A., Trimble, R. B., and Townsend, R. R. (1996) Anal. Biochem. 243, 203–209. 27. Liang, C.-J., Yamashita, K., and Kobata, A. (1980) J. Biochem. 88, 51–58. 28. Guttman, A., and Pritchett, T. (1995) Electrophoresis 16, 1906 – 1911. 29. Fu, D., Chen, L., and O’Neill, R. A. (1994) Carbohydr. Res. 261, 173–186. 30. Domon, B., and Costello, C. E. (1988) Glycoconj. J. 5, 397– 409.