Anal. Chem. 2000, 72, 1207-1216
Microsequencing of Glycans Using 2-Aminobenzamide and MALDI-TOF Mass Spectrometry: Occurrence of Unique Linkage-Dependent Fragmentation Yuji Sato,† Minoru Suzuki,‡ Takashi Nirasawa,§ Akemi Suzuki,‡ and Tamao Endo*,†
Department of Glycobiology, Tokyo Metropolitan Institute of Gerontology, Itabashi-ku, Tokyo 173-0015, Japan, Department of Membrane Biochemistry, Tokyo Metropolitan Institute of Medical Science, Bunkyo-ku, Tokyo 113-0021, Japan, and Nihon Bruker Daltonics K.K., Tsukuba-shi, Ibaraki 305-0051, Japan
2-Aminobenzamide-derivatized oligosaccharides were separated by three lectin column chromatographies and then subjected to matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) for structural characterization of the carbohydrates. The combination of sequential exoglycosidase digestion and MALDI-TOF MS greatly facilitates the monosaccharide sequencing and is more feasible than size-exclusion column chromatography in terms of the time consumed and the laboriousness of the procedure. By this strategy, microsequencing of 2-3 pmol of oligosaccharide derivatives could be achieved. Furthermore, spectra obtained by the post source decay (PSD) mode provide excellent sequence information. The relative intensities of metastable ions due to fragmentation at glycosidic linkages were different among linkage isomers of particular oligosaccharides. These results demonstrate that PSD analysis possesses significant potential for the estimation of glycosidic linkage in carbohydrate structures. Many proteins produced by mammalian cells contain glycans, which can be classified into two groups according to their glycanpeptide linkage regions, N-glycans and O-glycans. To elucidate the biological functions of the glycans, the development of sensitive analytical techniques for the structural characterization of glycans is required.1,2 Because of the large number of potential permutations and combinations of glycan types, linkage positions, and anomeric configurations, the methods in use for glycan chain analysis are far more complicated than those for protein or DNA sequence analysis.3 To develop an autosequencer for the determination of glycan structures, similar to protein or DNA sequencers, such essential complexity should be overcome. * Corresponding author: Telephone: +81-3-3964-3241 (ext. 3080). Fax: +81-3-3579-4776. E-mail:
[email protected]. † Tokyo Metropolitan Institute of Gerontology. ‡ Tokyo Metropolitan Institute of Medical Science. § Nihon Bruker Daltonics K.K. (1) Varki, A. Glycobiology 1993, 3, 97-130. (2) Dwek, R. A. Chem. Rev. 1996, 96, 683-720. (3) Fukuda, M.; Kobata, A. Glycobiology: A practical approach; Oxford University Press: Oxford, UK, 1993. 10.1021/ac9908750 CCC: $19.00 Published on Web 02/17/2000
© 2000 American Chemical Society
Recent improvements in mass spectrometry (MS) and soft ionization methods provide successful applications for the determination of the molecular weights of biopolymers, including those of proteins and carbohydrates with high molecular weight, and obtaining detailed structural information.4 One such analytical method is matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS).4-8 Although MALDI-TOF MS does not provide detailed structural information based on fragmentation, analysis by the post source decay (PSD) has been demonstrated to give fine structural information by the detection of metastable ions produced in a flight tube of TOF MS.9,10 On the other hand, various derivatization methods of carbohydrates have been introduced to increase analytical performance and detection sensitivity.3 Reductive amination, involving amination with a primary amine to form a Schiff base and reduction of secondary amines, has been applied for this purpose.11-19 The derivatization of oligosaccharides with a fluorophore compound, 2-aminobenzamide (2AB), has been developed and applied to glycan analysis.16 (4) Watson, J. T. Introduction to Mass Spectrometry, 3rd ed.; Lippincott-Raven Publishers: Philadelphia, PA, 1997. (5) Hillenkamp, F.; Karas, M. Methods Enzymol. 1990, 193, 289-295. (6) Mock, K. K.; Davey, M.; Cottrell, J. S. Biochem. Biophys. Res. Commun. 1991, 177, 644-651. (7) Harvey, D. J.; Ku ¨ ster, B.; Naven, T. J. P. Glycoconjugate J. 1998, 15, 333338. (8) Rouse, J. C.; Strang, A.-M.; Yu, W.; Vath, J. E. Anal. Biochem. 1998, 256, 33-46. (9) Spengler, B.; Kirsch, D.; Kaufman, R.; Jaeger, E. Rapid Commun. Mass Spectrom. 1992, 6, 105-108. (10) Spengler, B.; Kirsch, D.; Kaufman, R. J. Phys. Chem. 1992, 96, 9678-9684. (11) Kuraya, N.; Hase, S. J. Biochem. 1992, 55, 122-126. (12) Wang, W. T.; LeDonne, N. C., Jr.; Ackerman, B.; Sweeley, C. C. Anal. Biochem. 1984, 141, 366-381. (13) Matsuura, F.; Imaoka, A. Glycoconjugate J. 1988, 5, 13-26. (14) Jackson, P. Biochem. J. 1990, 270, 705-713. (15) Jackson, P. Anal. Biochem. 1991, 196, 238-244. (16) Bigge, J. C.; Patel, T. P.; Bruce, J. A.; Goulding, P. N.; Charles, S. M.; Parekh, R. B. Anal. Biochem. 1995, 230, 229-238. (17) Chiba, A.; Matsumura, K.; Yamada, H.; Inazu, T.; Shimizu, T.; Kusunoki, S.; Kanazawa, I.; Kobata, A.; Endo, T. J. Biol. Chem. 1997, 272, 21562162. (18) Sasaki, T.; Yamada, H.; Matsumura, K.; Shimizu, T.; Kobata, A.; Endo, T. Biochim. Biophys. Acta 1998, 1425, 599-606. (19) Anumula, K. R.; Dhume, S. T. Glycobiology 1998, 8, 685-694.
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000 1207
In this study, we examined whether the introduction of 2AB to oligosaccharides affects their binding to lectin columns, because lectin columns are well-known as an effective tool not only for the separation of glycans on the basis of binding specificity but also for elucidating partial structures.20 After sequential exoglycosidase digestion of oligosaccharides, thus fractionated by lectin column chromatography, each product was determined by MALDITOF MS very simply and with high sensitivity. Moreover, we obtained linkage-dependent fragmentation using the PSD mode. EXPERIMENTAL SECTION Enzymes and Lectins. Diplococcal β-galactosidase and β-Nacetylhexosaminidase were purchased from Boehringer Mannheim (Mannheim, Germany). Jack bean β-N-acetylhexosaminidase was purified from jack bean meal as described in a previous paper.21 Recombinant Aleuria aurantia lectin (AAL)-Sepharose was prepared according to the cited reference.22 Erythroagglutinin (E-PHA)-Sepharose and concanavalin A (Con A)-Sepharose were purchased from Seikagaku Corp. (Tokyo, Japan) and Amersham Pharmacia Biotech (Uppsala, Sweden), respectively. 2-Aminobenzamide (2AB) was purchased from Nacalai Tesque (Kyoto, Japan). Sodium cyanoborohydride (NaBH3CN) was obtained from Sigma Aldrich Japan K. K. (Tokyo, Japan). Oligosaccharides. All the oligosaccharides used in this study are listed in Table 1. The derivatization of oligosaccharides with 2AB was performed as described previously.16-18 Briefly, the oligosaccharides obtained as described below were lyophilized in a microcentrifuge tube. An aliquot (5 µL) of a reagent mixture [0.35 M 2AB/30% acetic acid in dimethyl sulfoxide (v/v)] was added to the oligosaccharide sample tube, and the mixture was incubated at 65 °C for 2 h. The derivatized oligosaccharide was purified by paper chromatography using Whatman 3MM paper in a solvent system of 1-butanol/ethanol/water (4:1:1, v/v). Analytical Methods. Lectin column chromatography using immobilized AAL, Con A, or E-PHA was performed as described previously.22,23,24 In brief, a mixture of 2AB-labeled oligosaccharides was applied to a Con A-Sepharose column equilibrated with 10 mM Tris-HCl buffer, pH 7.4, containing 0.15 M NaCl, 1 mM CaCl2, 1 mM MgCl2, and 1 mM MnCl2. The column was washed with 15 bed volumes of the buffer, and the bound oligosaccharides were eluted with a buffer containing 100 mM methyl R-D-mannoside. The AAL-Sepharose column was equilibrated with 10 mM TrisHCl buffer, pH 7.4, and washed with 10 bed volumes of the buffer after application of the sample, and the bound oligosaccharides were eluted with a buffer containing 1 mM L-fucose. The E-PHASepharose column was equilibrated with 10 mM Tris-HCl buffer, pH 7.4, and washed with 20 bed volumes of the buffer. For the 2AB-labeled oligosaccharides, fluorescence was monitored at 420 nm (excitation, 330 nm). Glycosidase Digestion. Oligosaccharides were incubated with one of the following mixtures for 18 h at 37 °C: (i) diplococcal β-galactosidase (5 mU) in 40 µL of 0.3 M citrate phosphate buffer, pH 6.0; (ii) diplococcal β-N-acetylhexosaminidase (0.25 mU) in (20) Endo, T. J. Chromatogr. 1996, A720, 251-261. (21) Li, Y.-T.; Li, S.-C. Methods Enzymol. 1972, 28, 702-713. (22) Fukumori, F.; Takeuchi, N.; Hagiwara, T.; Ohbayashi, H.; Endo, T.; Kochibe, N.; Nagata, Y.; Kobata, A. J. Biochem. 1990, 107, 190-196. (23) Ogata, S.; Muramatsu, T.; Kobata, A. J. Biochem. 1975, 78, 687-696. (24) Yamashita, K.; Hitoi, A.; Kobata, A. J. Biol. Chem. 1983, 258, 14753-14755.
1208 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
Table 1. Structures of the Oligosaccharides Used in This Studya
a Oligosaccharides I and II were prepared from fetuin,30 and oligosaccharides III, IV, V, VI, and VII were from human IgM myeloma protein31 by hydrazinolysis.32 LNT, LNnT, 3′SL, and 6′SL were from human milk.33 The derivatization of oligosaccharides with 2AB was performed as described in the Experimental Section.
40 µL of 0.3 M citrate phosphate buffer, pH 6.0; (iii) jack bean β-N-acetylhexosaminidase (0.5 U) in 55 µL of 0.3 M citrate phosphate buffer, pH 5.0. One drop of toluene was added to all the reaction mixtures to inhibit bacterial growth during the incubation period. The digestions were terminated by heating the reaction mixture in a boiling water bath for 3 min. The digested samples were desalted by using the Cosmosil 5C18-AR column (Nacalai Tesque, Kyoto, Japan), and one-tenth of each (1 µL) was subjected to mass spectrometry analysis as described later. Measurement of 2AB-Labeled Oligosaccharides by MatrixAssisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS). The experiments were carried out on the Reflex (Bruker Daltonics GmbH, Bremen, Germany) TOF mass spectrometer, which was equipped with a reflector. Ions generated by a pulsed UV laser beam (nitrogen laser, λ ) 337 nm) were accelerated to a kinetic energy of 30 keV. All spectra were measured in the reflector mode with time-lag focusing. Analyte Preparation for Mass Spectrometric Investigation. 2,5-Dihydroxybenzoic acid (DHBA), which was purchased from
Sigma Aldrich, was used as the matrix. It was dissolved to a concentration of 10 mg/mL of 30% aqueous ethanol. Dried analytes were dissolved in the matrix solution. Aliquots of the resulting mixtures (1 µL or less) were placed onto probe tips. The solvent was removed in a gentle stream of air, and the solid sample/matrix mixture was then transferred into the mass spectrometer. Post Source Decay (PSD)-MALDI Mass Spectrometry Measurement. The total acceleration voltage was 28.5 kV, the reflection voltage was decreased in successive 20% steps, 14 segments were obtained, and 560 shot spectra were summed up. One picomole of each sample was subjected to PSD analysis. Mass Calibration. The measurements were externally calibrated with the following compounds depending on the mass range: R-cyano-4-hydroxycinnamic acid (m/z 379.093 03), angiotensin II (m/z 1046.542), substance P (m/z 1347.736), neurotensin (m/z 1672.918), and ACTH clip 18-39 (m/z 2465.199). Aliquots of calibrants were added to the existing matrix. RESULTS AND DISCUSSION Fractionation of 2AB-Labeled Oligosaccharides by Lectin Column Chromatography. First, to develop sensitive and simple methods by using 2AB-derivatized oligosaccharides, it is indispensable to determine that 2AB-derivatization of oligosaccharides did not affect on the binding specificity of each lectin. Therefore, we examined whether 2AB-labeled oligosaccharides could be fractionated by lectin columns on the basis of carbohydrate binding specificity. For this purpose we selected three lectins which recognized different portions of N-glycan. Our current knowledge of the binding specificities of the three lectins is as follows. All complex-type N-glycans containing an R-fucosyl residue linked at the C-6 position of the proximal N-acetylglucosamine residue of their trimannosyl core bind to an AAL-Sepharose column, but those without the fucose residue do not.22 The minimal requirement for binding to a Con A-Sepharose column is the presence of at least two R-mannosyl residues unsubstituted at the C-3, 4, and 6 positions.23 The structural unit required for binding to E-PHA should contain the following, in which R represents either hydrogen or sugars.24
Among the N-glycans listed in Table 1, all of the high-mannose type (oligosaccharide VI and oligosaccharide VII) and biantennary complex-type chains (oligosaccharide IV and oligosaccharide V) are retained in the Con A column (Figure 1A). The presence or absence of the R-fucosyl residue linked at the C-6 position of the proximal N-acetylglucosamine residue (oligosaccharide IV and oligosaccharide V) does not affect the binding of oligosaccharides to the column. As the presence of the β-N-acetylglucosamine residue linked at the C-4 position of the β-mannosyl residue of the trimannosyl core (bisecting GlcNAc) considerably decreases the affinity of oligosaccharides to the column, oligosaccharide III is not retained in the column. Additionally, oligosaccharide I and oligosaccharide II did not bind to the column at all. Only
Figure 1. Separation of 2AB-derivatized oligosaccharides by various lectin column chromatographies. (A) Concanavalin A column chromatography. Oligosaccharides I, II, and III were recovered in fraction a and oligosaccharides IV, V, VI, and VII were recovered in fraction b. (B) Aleuria aurantia lectin column chromatography. Oligosaccharides I, II, V, VI, and VII were recovered in fraction c and oligosaccharides III and IV in fraction d. (C) Erythroagglutinin column chromatography. Oligosaccharides I, II, IV, V, VI, and VII were recovered in fraction e and only oligosaccharide III in fraction f. Arrows in (A) and (B) indicate the positions where the elution buffers were switched to buffer containing R-methyl mannoside (100 mM) and fucose (1 mM), respectively.
fucosylated N-glycans (oligosaccharide III and oligosaccharide IV) bound to the column and were eluted with 1 mM fucose (Figure 1B). Binding was not affected either by the structure of the outer chain moieties or the presence of bisecting GlcNAc residues (oligosaccharide III). As shown in Figure 1C, only oligosaccharide III was retarded in the E-PHA column. Based on these results, we could conclude that the 2AB derivatization of oligosaccharides did not affect their binding specificity to each lectin column and the method can be used for the fractionation of various N-glycans, similar to oligosaccharides and glycopeptides as reported previously.3,20 Identification of Products of Sequential Exoglycosidase Digestion of N-Glycans with Matrix-Assisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry (MALDITOF MS). Neutral oligosaccharides were fractionated by gel filtration as shown previously.25 Although gel filtration column (25) Yamashita, K.; Mizuochi, T.; Kobata, A. Methods Enzymol. 1982, 83, 105126.
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1209
Figure 2. MALDI mass spectra of a 2,4-branched triantennary complex-type oligosaccharide and its sequential exoglycosidase digestion products. (A) The 2,4-branched triantennary oligosaccharide (oligosaccharide I) obtained from fetuin and derivatized by 2AB. (B) The component in (A) after digestion with diplococcal β-galactosidase. (C) The component in (B) after digestion with diplococcal β-Nacetylhexosaminidase. (D) The component in (C) after digestion with jack bean β-N-acetylhexosaminidase. Mass spectra of the digested products desalted by using the Cosmosil 5C18-AR column were recorded in the positive ion reflectron mode using 2,5-dihydroxybenzoic acid as the matrix. An aliquot of the sample, 43 (A), 72 (B), 29 (C), or 56 fmol (D), was analyzed.
chromatography separates oligosaccharides very effectively, it is time-consuming; it actually takes several hours to separate each product after exoglycosidase digestion.3,17,25 To shorten this analytical time, we identified each product by MALDI-TOF MS. The application of MS for structure elucidation is attractive because the method is less time-consuming and more sensitive than other methods such as gel filtration column chromatography or gel electrophoresis. The results are represented in Figure 2. The measured massto-charge ratio of the 2AB-derivatized triantennary sugar chain (oligosaccharide I) was m/z 2148 and this was assigned as a sodium-adduct ion, [M + Na]+ (Figure 2A). Diplococcal β-galactosidase digestion reduced the mass-to-charge ratio by 486u (i.e., 162 × 3u), indicating that three galactose residues were removed and all the linkages were β1f4 linkages (Figure 2B), because of its substrate specificity.26 Treatment of m/z 1662 oligosaccharide 1210 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
derivatives with diplococcal β-N-acetylhexosaminidase, which cleaved only the GlcNAcβ1f2Man linkage,27 resulted in the detection of the glycoform at m/z 1256 with the loss of 406u (i.e., 203 × 2u) (Figure 2C). Further digestion of this m/z 1256 oligosaccharide with jack bean β-N-acetylhexosaminidase converted it to m/z 1053 oligosaccharide with an additional loss of 203u (Figure 2D). These results and the specificity of the enzymes reveal that the oligosaccharide is one of the 2,4-branched triantennary complex-type sugar chains. Analyses of the digested samples by MALDI-TOF MS can be completed in 30 min and the mass can be determined accurately. Further detailed information on the oligosaccharide structure is obtained by the combined use of sequential exoglycosidase digestion and lectin-affinity column chromatography as described above. Sequential digestion is an important nonspectroscopic technique for carbohydrate analysis, but it is tedious and timeconsuming. Therefore, a simple method using MS for the analysis of fragment patterns is desirable to estimate the carbohydrate structure. Thus, we applied the post source decay (PSD) mode to the structural analysis of 2AB-oligosaccharides in order to obtain monosaccharide sequence information. Measurements of 2AB-Oligosaccharide Derivatives Using the PSD Mode. We measured the PSD mode spectra of 2,4-branched triantennary complex-type oligosaccharide I (Figure 3A). Ions originating from both the reducing (Figure 4A) and nonreducing end (Figure 4C) are clearly differentiated in the spectra. The fragment scheme used was according to the nomenclature of Domon and Castello.28 Ions retaining the reducing terminus were termed Y and Z, whereas those retaining the nonreducing terminus were termed B and C. A mass-to-charge ratio of 2148.0 was detected as the sodium-adduct ion of the triantennary sugar chain (Figure 3A). Most of the other peaks were also reasonably interpretable as fragments (Figure 4B). As the fragment ions retaining the reducing end (Figure 4A), the ions at m/z 1812.1 and 1607.4 were assigned as being derived from the sodium-adduct ion at m/z 2148.0 by sequential loss of HexNAc-2AB and HexNAc residues (chitobiosyl core structure). Then, the following fragmentations occurred from the nonreducing terminal. The ions at m/z 1443.8 and 1242.8 are produced by sequential loss of hexose and HexNAc (Gal-GlcNAc), and an additional series of two ions at m/z 1078.7 and 874.9 could be interpreted as being due to the loss of another Gal-GlcNAc group. Therefore, the ion at m/z 874.9 is assigned as Gal-GlcNAc-Man(Man)-Man (Figure 4B). We could not determine which lactosamine chain of the triantennary outer portion remains in this pentasaccharide. Further fragmentation of the pentasaccharide, with sequential loss of three hexoses, could be interpreted by the appearance of a series of ions at m/z 713.6, 567.8 or 551.2, and 389.9, produced from the ion at m/z 874.9. As concerns the deletion of three hexoses, two ways should be considered: (a) three mannose residues and (b) one galactose and two mannose residues. In the case of (a), the two possible ways of loss of three mannoses, Y4 f Y3 f C3 or B3 f B4 as shown in Figure 4A,B, occurred in accordance with cleavage-types. It suggested the (26) Paulson, J. C.; Prieels, J. P.; Glasgow, L. R.; Hill, R. L. J. Biol. Chem. 1978, 253, 5617-5624. (27) Yamashita, K.; Ohkura, T.; Yoshima, H.; Kobata, A. Biochem. Biophys. Res. Commun. 1981, 100, 226-232. (28) Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5, 397-409.
Figure 3. PSD spectra of isomeric 2,4-branched triantennary oligosaccharides obtained from fetuin. (A) Oligosaccharide I; (B) oligosaccharide II. [M + Na]+ was selected as the parent ion.
existence of Gal-GlcNAc as a final product. In the case of (b), on the other hand, there were four possible ways in accordance with the order of loss of galactose and mannose residues and cleavagetypes, Y4 f Y3 f C3 or B3 f Z5 or Y5, Y4 f Y3 f Y5 f B3, Y4 f Y5 f Y3 f B3, respectively, as shown in Figure 4A,B. This suggested the existence of GlcNAc-Man as a final product. As the fragment ions retaining the nonreducing end (Figure 4C), the ions at m/z 1989.1 and 1788.4 are derived from the
pseudomolecular ion by sequential loss of the Gal-GlcNAc group. Additional loss of Gal-GlcNAc could be interpreted with the detection of ions at m/z 1625.2 and 1420.1. The ion at m/z 1259.3 is due to the loss of one hexose residue. The ions at m/z 1078.7 and 874.9 are produced by sequential loss of hexose and HexNAc residues, respectively. In regard to these two hexoses, however, it could not be determined which hexose corresponds to the galactose or mannose residue. Furthermore, we could not Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1211
determine the order of loss of three lactosamine structures from the trimannosyl core structure. The ion at m/z 874.9 was interpreted as Man-Man-GlcNAc-GlcNAc-2AB, via either sequential fragmentation process (Figure 4D). The loss of another mannose residue was interpreted as causing the appearance of the ion at m/z 713.6. Further fragmentation of this 2AB-derivatized trisaccharide, with sequential loss of hexose and HexNAc, produced the ion at m/z 364.2. The loss of mannose could be 1212
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
interpreted in two ways on the basis of the fragment ions. First, the appearance of the ion at m/z 567.8 was produced by Y-type cleavage (Y2 in Figure 4C,D), and second, the appearance of the ion at m/z 551.2 was produced by Z-type cleavage (Z2 in Figure 4C,D). Glycosyl-Linkage-Dependent Fragment Patterns. To investigate the applicability of PSD analysis to discriminate isomers, oligosaccharides having the same molecular weight but carrying
Figure 4. Generation of the fragment ions in the MALDI-PSD/MS of a 2,4-branched triantennary oligosaccharide. (A) Fragmentation ions from the reducing end. (B) Formation of the fragment ions of oligosaccharide I shown in (A). (C) Fragmentation ions from the nonreducing end. (D) Formation of the fragment ions of oligosaccharide I shown in (C). See text for a description of the fragment ion series and nomenclature. For illustrative purposes, ions labeled with R, β, or γ in (B) and (D) are formed by the cleavage of one of the branches, while the spectra do not allow these branches to be differentiated. See text for details.
a linkage difference were subjected to PSD analysis. It should be noted that although quantitation of the MALDI-TOF MS spectrum
was a controversial subject, the relative intensity of each peak is remarkably reproducible using the PSD mode. Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1213
Figure 5. Generation of the fragment ions in the MALDI-PSD/MS of sialyllactoses. (A) Fragmentation ions from 3′-SL-2AB, (B) those from 6′-SL-2AB. See text for a description of the fragment ion series and nomenclature. Caption of the fragment ions of 3′-SL-2AB and 6′-SL-2AB are the same as in Figure 4. 1214
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
Figure 6. Generation of the fragment ions in the MALDI-PSD/MS of tetrasaccharides. (A) Fragmentation ions from LNT-2AB, (B) those from LNnT-2AB. See text for a description of the fragment ion series and nomenclature. Caption of the fragment ions of LNT-2AB and LNnT-2AB are the same as in Figure 4.
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1215
We compared the PSD patterns of two triantennary complextype oligosaccharides obtained from fetuin. It is well-known that fetuin has two isomeric triantennary oligosaccharides carrying Galβ1-3GlcNAcβ1-4Man and the Galβ1-4GlcNAcβ1-4Man, as shown in Table 1.29,30 Both oligosaccharide I and oligosaccharide II gave similar fragmentation patterns as shown in Figure 3A,B. However, the relative intensities of the ions at m/z 1242.8 and 1443.8 in Figure 3A were larger than the corresponding ions at m/z 1242.5 and 1443.5 in Figure 3B. When their values were compared with the intensity of ion at m/z 1607.4(6) (derived from the loss of chitobiosyl core structure), the former two gave 1.0 and 0.9 (Figure 3A) and the latter two gave 0.6 and 0.4 (Figure 3B), respectively. As described already, the ions m/z 1242.8 and 1443.8 are due to the loss of the Gal and Gal-GlcNAc residues, respectively, in the outer chain moieties. These results suggest that the structural difference between the Galβ1-4GlcNAcβ1-4Man and Galβ1-3GlcNAcβ1-4Man groups (Table 1) causes the change in the relative intensities of these two peaks. To pursue this possibility, we analyzed other molecules that were smaller than triantennary complex-type oligosaccharides. When the PSD spectra of 3′-SL-2AB and 6′-SL-2AB were compared, the relative intensities of the fragment ions were clearly different as shown in Figure 5A,B. In the case of PSD measurement, we selected a protonated molecular ion instead of the sodium adduct, because it is known that not only the assignment of each fragment ion obtained from the former is easier than the latter, but also fragmentation of the former occurs more frequently than that of the latter. The mass-to-charge ratio of 754.0 matched the protonated molecular ion, [M + H]+. With fragmentation of these trisaccharides by sequential loss of Neu5Ac and Hex (Gal), m/z 463.3 and 301.1 are obtained. However, the relative intensity of the ion at m/z 301.2 is higher than that of the ion at m/z 463.3 in the case of 3′-SL-2AB (the ratio, 5.2), but lower than that of the ion at m/z 463.3 in the case of 6′-SL-2AB (the ratio, 0.4), as shown in Figure 5. It is possible that the R2-3 linkage was cleaved more frequently than the R2-6 linkage under these experimental conditions. It should be noted that the appearance of the ion at m/z 474.6, produced from the parent ion by the loss of Glc-2AB by C-type cleavage, suggested that only the β1-4 linkage of 3′SL-2AB was cleaved by this type of fragmentation. These results suggest that the sialyl linkage of the outer moiety affects the fragmentation of oligosaccharides in PSD analysis and that the detection of the ion at m/z 474.6 discriminates the two isomers. The PSD spectra of LNT(lacto-N-tetraose)-2AB and LNnT(lacto-N-neotetraose)-2AB were compared as shown in Figure 6A,B. Fragments retaining the nonreducing end were detected at m/z 667.3, 463.2, and 301.4, which are generated from the m/z 828.0 parent ion of LNT-2AB (protonated ion, [M + H]+) by sequential elimination of Hex, HexNAc, and Hex residues, respectively, allowing the identification of the sequence of the GalGlcNAc-Gal group. The same series of ions at m/z 667.1, 463.1, and 301.3 were detected in the case of LNnT-2AB. However, the (29) Townsend, R. R.; Hardy, M. R.; Wong, T. C.; Lee, Y. C. Biochemistry 1986, 25, 5716-5725. (30) Takasaki, S.; Kobata, A. Biochemistry 1986, 25, 5709-5715. (31) Ohbayashi, H.; Endo, T.; Mihaesco, E.; Gonzales, M. G.; Kochibe, N.; Kobata, A. Arch. Biochem. Biophys. 1989, 269, 463-475. (32) Takasaki, S.; Mizuochi, T.; Kobata, A. Methods Enzymol. 1982, 83, 263268. (33) Kobata, A. Methods Enzymol. 1972, 28, 262-271.
1216
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
relative intensities of these three ions were different: the order was 301.4 > 463.2 > 667.3 for LNT-2AB (the ratio 2.1:1.3:1.0) and 463.1 > 301.3 > 667.1 (the ratio 13.1:9.3:1.0) for LNnT-2AB. The ion at m/z 301.4 is due to the loss of Galβ1-3GlcNAcβ1-3Gal in LNT-2AB, and the ion at m/z 463.1 is due to the loss of Galβ14GlcNAc in LNnT. These results indicate that type 1 (Galβ13GlcNAc linkage) and type 2 (Galβ1-4GlcNAc linkage) structures affect the cleavage of the GlcNAcβ1-3Gal linkage, suggesting that the analysis of the relative intensities of these three ions could discriminate the two isomers. It is noteworthy that the fragment ion derived by C-type cleavage from the reducing terminal, which was detected only in the case of 3′-SL-2AB (Figure 5A), was not observed among these oligosaccharides. Two interesting features are pointed out. The first is that an unidentified ion at m/z 625.5 was detected only in the case of LNT-2AB, and the second is that the relative intensities of three ions, at m/z 463.1, 475.9, and 497.4, were different between LNT-2AB and LNnT-2AB: the order was 475.9 > 463.2 > 497.4 (the ratio 1.0:0.8:0.5) for LNT-2AB and 463.1 >497.4 > 475.9 (the ratio 5.9:2.1:1.0) for LNnT-2AB. Among them, the ion at m/z 463.1 is assigned as described above. Although the ion at m/z 475.9 could not be assigned at this stage, the ion at m/z 497.4 (indicated as peak a) was probably due to X-type cleavage as shown in Figure 6. Based on these results, we conclude that the structure of oligosaccharides influences the relative intensities of PSD peaks in terms of differences in glycosidic linkages. These differences could be excellent reporter data for estimating the monosaccharide sequence and the glycosidic linkages in oligosaccharides. CONCLUSIONS This article describes a new approach for the identification of carbohydrates isolated in limited amounts from biological sources. In conclusion, the PSD spectra of glycans showed characteristic fragmentations, which are quite useful for structural characterization. The major advantages of this technique are its simplicity and sensitivity. The accumulation of PSD spectra of glycans will be informative for the estimation of glycosidic linkages in glycan structures in the future. Progress in this area and accumulation of data significantly enhance the usefulness of mass spectrometry by reducing the difficulty and time-consuming steps of total structure determination. Abbreviations Used: AAL, Aleuria aurantia lectin; 2AB, 2-aminobenzamide; Con A, concanavalin A; DHBA, 2,5-dihydroxybenzoic acid; E-PHA, erythroagglutinin; MALDI-TOF MS, matrixassisted laser desorption ionization time-of-flight mass spectrometry; PSD, post source decay. All sugars mentioned in this paper have the D configuration except for fucose, which has the L configuration. ACKNOWLEDGMENT This work was supported in part by the Grant-in-Aids for Scientific Research (09558083, 10771313, 10178102) from the Ministry of Education, Science, Sports and Culture of Japan and by Special Coordination Funds of the Science and Technology Agency of the Japanese Government. Received for review August 3, 1999. Accepted January 5, 2000. AC9908750