Structural Analysis of Glycoconjugates by On-Target Enzymatic

Dec 4, 1998 - Under these conditions, a series of sequential on-target exoglycosidase treatments was carried out in one single analyte spot in the pre...
0 downloads 10 Views 111KB Size
Download PDF
Anal. Chem. 1999, 71, 476-482

Structural Analysis of Glycoconjugates by On-Target Enzymatic Digestion and MALDI-TOF-MS Hildegard Geyer, Sigrid Schmitt, Manfred Wuhrer, and Rudolf Geyer*

Institute of Biochemistry, University of Giessen, Friedrichstrasse 24, D-35392 Giessen, Germany

Exoglycosidase digestion combined with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) has been demonstrated to be an effective method for the structural characterization of glycoconjugates and oligosaccharides in picomolar amounts. A sample preparation method is described, in which 6-aza-2-thiothymine (ATT) in water is used as matrix and enzymes are dialyzed before use against a low concentration of volatile buffer such as ammonium acetate. Under these conditions, a series of sequential on-target exoglycosidase treatments was carried out in one single analyte spot in the presence of ATT matrix. Subsequent mass spectrometric analysis of the resulting products yielded information on both the completeness of the reaction and structural features of the glycoconjugates such as monosaccharide sequence, branching pattern, and anomeric configurations of the corresponding glycosidic linkages. The results show that all exoglycosidases used retain their activity in the presence of ATT matrix. Hence, structural analysis of carbohydrates or mixtures thereof can be performed very fast, without intermediate desalting steps or sample splitting. This approach is illustrated by the analysis of underivatized glycans, oligosaccharide derivatives, glycopeptides, and glycolipids. Depending on the analyte, amounts of sample required could be limited to a few picomoles. Glycoconjugates such as glycoproteins and glycolipids exhibit multiple functions in biological systems. To correlate functional features with defined structural parameters, detailed structural analyses of the carbohydrate chains are required. Chemical and biochemical studies, however, are often impeded by the limited amounts of sample available as well as the structural complexity and the microheterogeneity of these glycans. A suitable tool for the analysis of glycoconjugates and free glycans is matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS), which provides the molecular masses of the compounds present.1-4 Since carbohydrate moieties are often composed of a relatively small number of * Corresponding author: (phone) +49-641-99-47400; (fax) +49-641-99-47409; (e-mail) [email protected]. (1) Stahl, B.; Steup, M.; Karas, M.; Hillenkamp, F. Anal. Chem. 1991, 63, 14631466. (2) Rouse, J. C.; Vath, J. E. Anal. Biochem 1996, 238, 82-92. (3) Harvey, D. J.; Ku ¨ ster, B.; Naven, T. J. Glycoconjugate J. 1998, 15, 333338. (4) Harvey, D. J. J. Chromatogr., A 1996, 720, 429-446.

476 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

different monosaccharide constituents with unique incremental masses (e.g., fucose, 146 Da; hexose, 162 Da; N-acetylhexosamine, 203 Da; N-acetylneuraminic acid, 291 Da), a molecular mass with an accuracy of e0.05% allows the determination of the composition of a glycan in terms of deoxyhexose, hexose, N-acetylhexosamine, and N-acetylneuraminic acid. MALDI-TOF-MS itself, however, can establish neither anomerity of the monosaccharide components nor branching configurations or epimeric forms of oligosaccharides. This information is most commonly provided by digestion of glycans with highly specific exoglycosidases either in sequence or in array,5-7 which selectively release monosaccharides from the nonreducing termini based on their stereochemistry, anomeric configuration, and linkage to the remainder of the carbohydrate side chain. Employing MALDI-TOF-MS to monitor the enzymatic digestion, the high specificity of the enzymes can be, therefore, combined with the high resolution and sensitivity of mass spectrometry. Using a panel of well-defined enzymes, the molecular mass information after each digestion step thus reveals the sequence of the monosaccharide constituents and, at least in part, information on the primary structure of the carbohydrate chains.8-13 In the case of oligosaccharides, usually low-picomole amounts of analyte are required to obtain satisfactory mass spectra whereas glycopeptides and glycolipids can be detected at even lower levels. In most cases, portions of the total sample are set aside for analysis after each digestion step which cannot be used for further treatments.11,12,14,15 Alternatively, samples are split into aliquots for simultaneous incubations with different arrays of enzymes.16 In both cases, the overall sensitivity is reduced. (5) Geyer, H.; Geyer, R. Acta Anat. 1998, 161, 18-35. (6) Edge, C. J.; Rademacher, T. W.; Wormald, M. R.; Parekh, R. B.; Butters, T. D.; Wing, D. R.; Dwek, R. A. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 63386342. (7) Mizuochi, T.; Yonemasu, K.; Yamashita, K.; Kobata, A. J. Biol. Chem. 1978, 253, 7404-7409. (8) Sutton, C. W.; O’Neill, J. A.; Cottrell, J. S. Anal. Biochem 1994, 218, 3446. (9) Zhao, Y.; Kent, S. B.; Chait, B. T. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 1629-1633. (10) Mechrev, Y.; Novotny, M. V. Anal. Chem. 1998, 70, 455-463. (11) Ku ¨ ster, B.; Naven, T. J. P.; Harvey, D. J. J. Mass Spectrom. 1996, 31, 11311140. (12) Mortz, E.; Sareneva, T.; Julkunen, I.; Roepstorff, P. J. Mass Spectrom. 1996, 31, 1109-1118. (13) Rudd, P. M.; Dwek, R. A. Curr. Opin. Biotechnol. 1997, 8, 488-497. (14) Harvey, D. J.; Rudd, P. M.; Bateman, R. H.; Bordoli, R. S.; Howes, K.; Hoyes, J. B.; Vickers, R. G. Org. Mass Spectrom. 1994, 29, 753-765. (15) Yang, Y.; Orlando, R. Anal. Chem 1996, 68, 570-572. (16) Glocker, M. O.; Bauer, S. H. J.; Kast, J.; Volz, J.; Przybylski, M. J. Mass Spectrom. 1996, 31, 1221-1227. 10.1021/ac980712w CCC: $18.00

© 1999 American Chemical Society Published on Web 12/04/1998

One important factor that determines the quality of a MALDI mass spectrum is sample preparation. Although this technique is comparatively tolerant to the presence of low salt concentrations, ionization of oligosaccharide samples seems to be easily affected.11 Most buffers currently used for enzymatic digestions as well as certain contaminants such as detergents are detrimental to MALDI-TOF-MS2. Thus, various sample pretreatment procedures such as droplet dialysis on a floating membrane or on-target desalting with ion-exchange beads have been developed to circumvent these problems.2,11 In recent studies, however, evidence has been provided that the high buffer concentrations commonly used are not necessary for effective enzymatic digestion.10,11 Yang and Orlando15 even demonstrated that 25 mM ammonium acetate solutions adjusted to the proper pH can replace the normal digestion buffers. All of the enzymes tested retained both their activity and their specificity under these conditions. Since these ammonium acetate solutions are volatile, the majority of the buffer is evaporated when the sample is dried. Hence, direct MALDI-TOF-MS analyses of the digestion mixtures should be feasible without desalting. In UV MALDI-TOF-MS, solid samples are prepared by the addition of a large excess of matrix with suitable absorption properties. Analyte molecules incorporated into matrix crystals after evaporation can then be ionized upon UV laser irradiation.4 Since most of the commonly used matrixes for MALDI-TOF-MS of glycans such as 2,5-dihydroxybenzoic acid (DHB) or 2-(phydroxyphenylazo)benzoic acid (HABA)4,17-19 are highly acidic, complete denaturation of protein structures has to be generally expected under these conditions. Therefore, these substances are not compatible with the buffer requirements for exoglycosidase digestions and have to be removed before enzyme incubation. As an alternative nonacidic matrix, 6-aza-2-thiothymine (ATT) has been recently introduced for the analysis of protein complexes, acidic oligosaccharides, and glycopeptides.16-18,20 It could be demonstrated that proteins maintain their intact tertiary structures and that, for instance, trypsin retains its enzymatic activity in the presence of ATT.16 The aim of our study was the development of a sample preparation and exoglycosidase incubation protocol that enables numerous digestion and molecular mass determination cycles using only one single sample spot without intermediate desalting or purification steps. In the method described here, ATT, dissolved in water, is used as matrix and exoglycosidases are predialyzed against low concentrations of ammonium acetate buffer. Since the actual sample consumption during MALDI-TOF-MS is estimated to be very low (i.e., in the attomole range), the low-picomolar amounts of glycans or glycoconjugates usually required to obtain satisfactory spectra are sufficient for the performance of numerous digestion and mass determination cycles. EXPERIMENTAL SECTION Materials. 6-Aza-2-thiothymine was purchased from Sigma (Deisenhofen, Germany). ATT was dissolved in distilled water (17) Juhasz, P.; Costello, C. E. J. Am. Soc. Mass Spectrom. 1992, 3, 785-796. (18) Juhasz, P.; Costello, C. E.; Biemann, K. J. Am. Soc. Mass Spectrom. 1993, 1993, 399-409. (19) Strupat, K.; Karas, M.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Processes 1991, 111, 89-102. (20) Papac, D. I.; Wong, A.; Jones, A. J. Anal. Chem 1996, 68, 3215-3223.

without any addition of organic solvents to give a matrix concentration of 5 mg/mL. β-Galactosidase from Diplococcus pneumoniae, β-N-acetylglucosaminidase from D. pneumoniae, R-mannosidase from jack beans, O-glycosidase (endo-R-N-acetylgalactosaminidase) from D. pneumoniae, and R-galactosidase from green coffee beans were obtained from Boehringer Mannheim (Mannheim, Germany), β-N-acetylhexosaminidase from jack beans and β-mannosidase from snail acetone powder were purchased from Sigma, and sialidase from Arthrobacter ureafaciens, and R1-3,4-L-fucosidase were from Calbiochem (Bad Soden, Germany). All enzymes were dialyzed against 25 mM ammonium acetate buffer adjusted to the suggested pH for each enzyme (i.e., pH 6.0 for R-galactosidase, β-galactosidase, β-N-acetylglucosaminidase, sialidase, and O-glycosidase, pH 5.0 for β-N-acetylhexosaminidase, R-mannosidase, and R-fucosidase, and pH 4.0 for β-mannosidase). After dialysis, the activity of the enzymes was determined as described elsewhere21 using the respective p-nitrophenylglycosides and 25 mM ammonium acetate buffer. Enzymes were aliquoted and kept at -20 °C. Diantennary oligosaccharide standard (NA2, Galβ4GlcNAcβ2ManR3[Galβ4GlcNAcβ2ManR6]Manβ4GlcNAcβ4GlcNAc) was purchased from Oxford GlycoSciences (Abingdon Oxfordshire, UK). Pyridylamino (PA)- and 2-aminobenzamide (2-AB) derivatives of NA2 were a kind gift of A. Sauer (Institute of Biochemistry, University of Giessen, Germany). The amount of desalted pure derivatives was determined by carbohydrate constituent analysis of the peracetylated compounds.22 Oligosaccharides from R1-acid glycoprotein were released by peptide-N4-(N-acetyl-β-glucosaminyl)asparagine amidase F, reduced to the respective alditols, desialylated, and fractionated by HPLC using a LiChrosorb-NH2 column (5 µm, 4.6 × 250 mm) and a gradient of acetonitrile in 15 mM potassium phosphate buffer, pH 5.3, as eluent as described earlier.23 In this study, two subfractions of R1-acid glycoprotein glycans have been analyzed. Isomaltosyl oligosaccharides were prepared from dextran 4 (Serva, Heidelberg, Germany) by partial acid hydrolysis and preparative high-pH anion-exchange chromatography.24 The glycopeptide (calculated mass of the peptide moiety, 3023.4 Da) was isolated from hepatitis B virus M protein25 by tryptic digestion and reversed-phase HPLC. Tetraosylgloboside (GalNAcβ3GalR4Galβ4Glcβ1Cer) was purchased from Sigma. The glycolipid fraction from Ascaris suum, comprising the ceramidepentahexoside GalR3GalNAcβ4GlcNAcβ3Manβ4Glcβ1Cer and the respective ceramidetetra- and ceramidetrihexosides was prepared from corresponding zwitterionic glycolipids (kind gift of C. Friedl, Institute of Biochemistry, University of Giessen) by treatment with hydrofluoric acid as described previously.26 Glycolipids were dissolved in chloroform/methanol/water (10:10:1, v/v/v) to give a concentration of 10 pmol/µL. (21) Li, Y. T.; Li, S. C. Methods Enzymol. 1972, 28, 702-713. (22) Geyer, R.; Geyer, H.; Ku ¨ hnhardt, S.; Mink, W.; Stirm, S. Anal. Biochem. 1982, 121, 263-274. (23) Geyer, H.; Jacobi, J.; Linder, D.; Stirm, S.; Bialojan, S.; Strube, K.-H.; Geyer, R. Eur. J. Biochem. 1996, 237, 113-127. (24) Pfeiffer, G.; Geyer, H.; Geyer, R.; Kalsner, I.; Wendorf, P. Biomed. Chromatogr. 1990, 4, 193-199. (25) Kann, M.; Gerlich, W. H. In Topley & Wilsons Microbiology and Microbial Infections; Collier, L., Balows, A., Sussman, M., Eds.; Arnold: London, 1998; pp 745-774. (26) Lochnit, G.; Dennis, R. D.; Ulmer, A. J.; Geyer, R. J. Biol. Chem. 1998, 273, 466-474.

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

477

On-Target Enzymatic Cleavages. All digestions were performed directly on the stainless steel target according to the following procedure: Oligosaccharide and glycopeptide samples (0.2-1 µL) were deposited on the plate and dried in a gentle stream of cold air. ATT solution (0.4-1 µL) was added and again air-dried. After determination of the molecular mass by MALDITOF-MS, the sample matrix mixture was redissolved in 0.2-1 µL of the first dialyzed enzyme solution. The target was placed in a screw-capped jar containing ammonium acetate buffer at the bottom. The covered jar was incubated at 37 °C for 2-24 h, depending on the enzyme used. Subsequently, spots were dried in a cold stream of air and the mass profile of the digestion products was recorded. After addition of the next enzyme solution, the reaction was similarly allowed to proceed in the moistured chamber. The mass profile was again determined without addition of new matrix solution. In total, the sample underwent as many cycles of exoglycosidase treatment and MALDI analysis as necessary to establish the sequence of its monosaccharide constituents. When glycolipids were analyzed, first 0.5 µL of ATT solution was spotted onto the target and 1 µL of glycolipid was added under a stream of warm air. After MALDI-TOF-MS, the spot was reconstituted with 2 µL of dialyzed enzyme solution and the target was incubated as described above. The spot was dried and recrystallized under a stream of warm air with 1 µL of chloroform/ methanol/water (10:10:1, v/v/v). In this way, an amorphous film was obtained with a homogeneous distribution of glycolipid. The film was overlaid, under a stream of warm air, with 1 µL of ATT matrix and the mass profile was recorded. Thereafter, the glycolipid sample could undergo the next digestion cycle in the presence of the ATT matrix. Mass Spectrometry. The MALDI-MS experiments were performed on a Vision 2000 time-of-flight mass spectrometer (Finnigan/MAT, Bremen, Germany) equipped with a UV-nitrogen laser (337 nm). The instrument was operated in the positive-ion reflectron mode throughout. Ions generated were accelerated at a potential of 5 kV. The laser power density was about 106 W cm-2. All spectra shown represent accumulated spectra obtained by 5-100 laser shots and given molecular masses represent the average masses of the Na+ adducts ([M + Na]+). The instrument was calibrated with an external mixture of isomaltosyl oligosaccharides containing 5-15 glucose units or with angiotensin I and bovine insulin (both from Sigma). Mass accuracy was e0.025%. RESULTS AND DISCUSSION General Aspects. The aim of this study was to use a MALDITOF mass spectrometer as a sensitive detection system after sequential enzymatic degradation of minute quantities of glycans and glycoconjugates without intermediate purification or desalting procedures. All enzymatic cleavages were, therefore, carried out exclusively on the target of the instrument in the presence of ATT matrix. The first step implies the deposition of an aqueous solution of oligosaccharides or glycopeptides or an organic solution of glycolipids on the target together with ATT, dissolved in water (pH 4.2). Since (1) signal intensities did not change when aqueous buffer solutions such as ammonium citrate or acetate were used instead of water, (2) high buffer concentrations commonly used for enzymatic reactions have been reported to be not necessary for effective cleavage,10,11,15 and (3) most exoglycosidases retain 478

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

their activity even in ammonium acetate solution,15 all enzymes used in this study were dialyzed prior to use against 25 mM ammonium acetate buffer of the appropriate pH. As a consequence, the concentration of buffer ions, possibly affecting mass determination of glycans or glycoconjugates, could be kept to a minimum. Enzyme/substrate ratios, pH of respective digestion buffers, and incubation times were chosen in accordance with recent incubation protocols.11,15 In agreement with literature data,10,11 most enzymatic digestions were accomplished much faster than initially reported, thus allowing a considerable reduction of the incubation times in many cases. For example, incubation with R- or β-galactosidase, β-N-acetylglucosaminidase from D. pneumoniae, sialidase, and R-fucosidase resulted in a complete conversion into the respective products within 2-6 h of incubation at 37 °C. On the other hand, digestion with R-mannosidase and β-N-acetylhexosaminidase from jack beans, β-mannosidase, and O-glycosidase required a minimal incubation time of about 8 h or overnight. If the cleavage turned out to be incomplete, samples were reconstituted with 0.25-0.5 µL of enzyme solution and re-incubated. Hence, depending on the number and nature of the exoglycosidases applied, a whole digestion sequence of five cycles could be performed in 1-2 days without any sample manipulation other than repeated mass determination and enzyme addition. Since all sample transfer or dilution steps were avoided, low-picomolar quantities of glycans and glycoconjugates could thus be analyzed and structurally characterized. In the following sections, examples for the successful application of this technique to free oligosaccharides, oligosaccharide derivatives, glycopeptides, and glycolipids are presented. Enzymatic Sequencing of Free Oligosaccharides and Oligosaccharide Derivatives. Initial studies were carried out with different concentrations of an underivatized diantennary standard oligosaccharide (NA2). In addition, two popular derivatives of NA2 in which reducing termini were reductively aminated with 2-AB or PA were employed. Although derivatization of sugars is not necessary for the MALDI process, fluorophore-tagged glycan derivatives are often prepared for sensitive HPLC detection of glycans. Starting from 20 pmol of NA2 and 6 pmol of NA2-2AB or NA2-PA, respectively, three sequential cycles of enzyme digestions were performed with intermediate molecular mass determinations for establishing the monosaccharide sequence of the glycans. The results are shown in Figure 1. Treatment of the glycans with β-galactosidase from D. pneumoniae caused shifts in the molecular masses of about 324 Da, demonstrating the release of two hexose residues. Subsequent treatment with β-Nacetylglucosaminidase from D. pneumoniae resulted in the loss of two GlcNAc residues and corresponding mass shifts of 406 Da yielding the molecular masses of the common core pentasaccharide of N-linked oligosaccharides or the respective derivatives. After digestion with R-mannosidase, the molecular masses were further reduced by two hexose units, leading to the trisaccharides HexHexNAc2 and HexHexNAc2-2-AB(or -PA). In the case of the PA derivatives, products were characterized by doublet signals in the mass spectra. In addition to the sodium adducts of the glycans, signals of [M + H]+ pseudomolecular ions were registered, which represented the major ions in the spectra, similar to

Figure 1. MALDI-TOF mass spectra of the products of sequential exoglycosidase digests of diantennary standard oligosaccharide: (A) underivatized glycan (20 pmol); (B) oligosaccharide, reductively aminated with 2-aminobenzamide (6 pmol); (C) pyridylamino derivative (6 pmol). Glycans were digested with 0.08 munit of β-galactosidase (2 h), 0.05 munit of β-N-acetylglucosaminidase from D. pneumoniae (3 h), and 5 munits of R-mannosidase from jack beans (overnight). In (C), [M + H]+ and [M + Na]+ pseudomolecular ions are detected. Peaks marked by asterisks reflect unknown impurities; no oligosaccharide composition could be deduced from these masses. b, Gal; 9, GlcNAc; O, Man.

the findings reported by Takao et al.27 for 4-aminobenzoic derivatives. In all cases, enzymatic cleavages were found to be complete, since no traces of incomplete or partial digestion products were observed. The signals of the respective end products were strong enough to be unambiguously identified despite decreasing signalto-noise ratios. Attempts to further reduce the quantity of analyte to only half the amount, i.e., 10 pmol of NA2 and 3 pmol of NA22-AB or NA2-PA, resulted in clear and unambiguous signals in the spectra after treatment with β-galactosidase and β-N-acetylglucosaminidase (data not shown). The cleavage of R-linked mannosyl residues after treatment with R-mannosidase, however, could be only traced by the disappearance of the pentasaccharide core signal at m/z 934.3 (1054.5 or 990.8), whereas signals indicative for the respective end products were not detected. In (27) Takao, T.; Tambara, Y.; Nakamura, A.; Yoshino, K. J.; Fukuda, H.; Fukuda, M.; Shimonishi, Y. Rapid Commun. Mass Spectrom. 1996, 10, 637-640.

conclusion, three sequential cycles of on-target digestion could be easily monitored by MALDI-TOF-MS when 20 pmol of underivatized glycan or 6 pmol of oligosaccharide derivatives were employed, whereas products of two digestion cycles could be detected from as little as 10 or 3 pmol, respectively. Isomer Identification of Triantennary Glycans. Triantennary species of N-linked glycans may exist as different isomers. The third N-acetyllactosamine antenna can be linked either via a (β1-6) bond to the (R1-6)-linked mannose residue (2,6-branched isomer) or via a (β1-4) bond to the (R1-3)-linked mannose of the molecule (2,4-branched isomer). In addition, subterminal GlcNAc residues may be substituted by (β1-3)- or (β1-4)-linked Gal.28 Since all these isomers have the same monosaccharide composition, i.e., the same molecular mass, they cannot be distinguished by MALDI-TOF-MS. The on-target sequencing approach leads to a fast and unambiguous structural characteriza(28) Kobata, A. Eur. J. Biochem. 1992, 209, 483-501.

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

479

Figure 3. Mass spectra of a mixture of oligosaccharide alditols (80 pmol) before and after sequential enzymic treatment: native glycans (A), incubation with β-galactosidase from D. pneumoniae (B), β-Nacetylglucosaminidase from D. pneumoniae (C), β-N-acetylhexosaminidase from jack beans (D), 0.08 munit of R-fucosidase (5 h) (E), and R-mannosidase from jack beans (F). For digestion conditions and symbols, see legends to Figures 1 and 2. Peaks marked in parentheses reflect unknown impurities or unspecific degradation products; 1, Fuc.

Figure 2. MALDI-TOF mass spectra of triantennary oligosaccharide alditols (20 pmol) before (A) and after sequential treatment with β-galactosidase from D. pneumoniae (B), β-N-acetylglucosaminidase from D. pneumoniae (C), 0.7 munit of β-N-acetylhexosaminidase from jack beans (overnight) (D), and R-mannosidase from jack beans (E). For digestion conditions, asterisks and symbols, see legend to Figure 1.

tion of such isobaric glycans. As an example, the sequencing of a triantennary oligosaccharide alditol preparation, isolated from R1-acid glycoprotein, is shown in Figure 2. Three terminal Gal residues were removed by β-galactosidase from D. pneumoniae, hydrolyzing specifically Gal(β1-4) linkages (Figure 2B). After incubation with β-N-acetylglucosaminidase from D. pneumoniae, 480 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

two products were formed, differing in the number of residual GlcNAc residues (Figure 2C), all of which were removed by jack bean β-N-acetylhexosaminidase (Figure 2D). The typical core pentasaccharide alditol (Hex3HexNAcHexNAcOH) formed could be further degraded by R-mannosidase from jack beans cleaving both (R1-6)- and (R1-3)-linked Man (Figure 2E). Since β-Nacetylglucosaminidase from D. pneumoniae cleaves only GlcNAcβ2Man units, the mannosyl residue of which is not simultaneously substituted by a (β1-6)-linked GlcNAc,29 the two branching isomers can be clearly differentiated. From the results obtained it may be therefore concluded that the oligosaccharide preparation under study contained the 2,4-branched glycomer as the main component in addition to small amounts of the 2,6-branched isomer. Furthermore, both components were shown to carry exclusively β4-linked galactosyl residues. On-Target Analysis of an Oligosaccharide Mixture. Since most carbohydrates give similar peak heights as the result of an equivalent ionization method (mostly [M + Na]+ ions),3,4 mixtures of oligosaccharides may be similarly analyzed by this on-target sequencing technique. As an example, we have studied a fraction of desialylated oligosaccharide alditols derived from R1-acid glycoprotein. Three major signals were detected which fit the oligosaccharide compositions given in Figure 3A, comprising (29) Yamashita, K.; Ohkura, T.; Yoshima, H.; Kobata, A. Biochem. Biophys. Res. Commun. 1981, 100, 226-232.

triantennary oligosaccharide alditols (as in Figure 2), fucosylated triantennary glycans, and tetraantennary species. Based on the molecular masses of the native glycans and the digestion products as well as the specificities of the enzymes used, the given structures were deduced as follows. Removal of terminal (β14)-linked galactosyl residues by β-galactosidase from D. pneumoniae caused a mass shift of the three signals corresponding to the loss of three, two, or four hexoses (Figure 3B). In agreement with previous data,30 fucosylated N-acetyllactosamine antennae were resistant to β-galactosidase treatment. After digestion with β-N-acetylglucosaminidase from D. pneumoniae, four different signals were observed (Figure 3C). The agalacto triantennary species produced two products with one or two residual GlcNAc residues (compare Figure 2C) whereas the fucosylated glycan lost two and the tetraantennary species only one GlcNAc residue. Complete removal of terminal GlcNAc residues was achieved by β-N-acetylhexosaminidase from jack beans; signals at m/z 1139.2, 1342.5, and 1545.7 collapsed into a new peak at m/z 935.9 corresponding to the molecular mass of pentasaccharide core oligosaccharide alditols, whereas the signal of the fucosylated species did not change (Figure 3D). Fucosyl residues were subsequently removed by treatment with R3/4-specific fucosidase (Figure 3E). Finally, R-bound mannoses were removed, resulting in the signal of the trisaccharide alditol (HexHexNAcHexNAcOH) at m/z 611.7 in addition to a new signal at m/z 1139.4 derived from the formerly fucosylated species (Figure 3F). Although incubation with enzyme was repeated once, this cleavage seemed to be, in part, incomplete possibly due to steric hindrance. In agreement with literature data,31 the results clearly demonstrated the presence of two types of triantennary glycans (2,4-branched and 2,6-branched isomers), fucosylated triantennary species, the fucosyl residue of which is bound to the β4-linked N-acetyllactosamine antenna, and tetraantennary oligosaccharide alditols. Sequencing of Carbohydrates Linked to Peptides. The direct on-target sequencing strategy has been also applied to a glycopeptide carrying O-linked glycans. The results shown in Figure 4 demonstrate the presence of two major signals indicative for a peptide carrying either a HexNAcHex or a HexNAcHexSA unit (Figure 4A). Oligosaccharides and glycopeptides substituted with sialic acid are known to lose sialic acid very easily under MALDI conditions.20 Even with ATT as matrix, usually less than 50% of sialylated molecules pass through the reflector intact. Therefore, the mass spectrum of this glycopeptide was also recorded in the linear mode, which revealed that about 50-60% of the glycopeptides were sialylated (data not shown). Since HPLC fractionation as well as methylation data clearly demonstrated that the original glycopeptide has been completely sialylated (data not shown), it may be concluded that sialic acid is lost in a prompt as well as a metastable fragmentation mode. After digestion with sialidase from A. ureafaciens, the second peak shifted to the first one, indicating the removal of one sialic acid residue (Figure 4B) whereas the respective enzyme from Vibrio cholerae was hardly effective (data not shown). Subsequent treatment with O-glycosidase from D. pneumoniae, specifically releasing O-glycosidically linked Galβ3GalNAc chains from Ser/Thr residues, resulted in a (30) Wendorf, P.; Linder, D.; Sziegoleit, A.; Geyer, R. Biochem. J. 1991, 278, 505-514. (31) Yoshima, H.; Matsumoto, A.; Mizuochi, T.; Kawasaki, T.; Kobata, A. J. Biol. Chem. 1981, 256, 8476-8484.

Figure 4. MALDI-TOF mass spectra of an O-glycosylated glycopeptide (2 pmol): (A) native glycopeptide, (B) after digestion with 0.4 munit of sialidase from A. ureafaciens (6 h), and (C) after incubation with 0.5 munit of O-glycosidase from D. pneumoniae (overnight). Only [M + H]+ pseudomolecular ions are marked. For symbols, see legend to Figure 1; 2, sialic acid; 0, GalNAc.

corresponding mass shift of 365 Da, yielding the molecular mass of the unsubstituted peptide (Figure 4C). In conclusion, the results revealed that the peptide carried one O-glycosidically linked Galβ3GalNAc disaccharide chain substituted by one sialic acid. The whole analysis was performed with 2 pmol of analyte. On-Target Digestion of Glycolipids. To probe the general applicability of the present sample preparation procedure, glycolipids were similarly digested with exoglycosidases. Usually, glycolipids are enzymatically degraded in the presence of 0.1% of sodium taurodeoxycholate,32 which has to be removed by tedious sample purification steps after each cycle before mass determination of the reaction products. Using the direct on-target sequencing method, the presence of detergent is no longer required for effective cleavage. In Figure 5, two examples of such direct on-target glycolipid digestions are given. As a standard glycolipid, commercially available tetraosylgloboside (10 pmol) was used (Figure 5A). Multiple signals, seen in the mass spectra of the native analyte and digestion products, are due to heterogeneities in the ceramide portion of the molecule. The carbohydrate moiety of the glycolipid was successively degraded by β-Nacetylhexosaminidase, R-galactosidase, and β-galactosidase to yield the end product Glcβ1ceramide. Note that after digestion with β-N-acetylhexosaminidase two products occurred corresponding to the loss of one HexNAc residue and of one HexNAc + one Hex, respectively. Obviously, the enzyme preparation used contained a contaminating activity of R-galactosidase. The gly(32) Uda, Y.; Li, S. C.; Li, Y. T. J. Biol. Chem. 1977, 252, 5194-5200.

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

481

and enzymes can be selectively removed by repetitive washes with water. Because of their hydrophobicity, glycolipids remain adsorbed to the target under these conditions. Compared to enzymatic sequencing in solution, this new technique affords an, at least, 100-fold increase in sensitivity. Due to the absence of detergents, the procedure allows a fast and sensitive glycolipid analysis without any intermediate sample purification.

Figure 5. MALDI-TOF mass spectra of the products of sequential exoglycosidase treatments of tetraosylgloboside (A) and a fraction of glycolipids from A. suum (B). Digestions were performed with 1.5 munits of R-galactosidase from green coffee beans (6 h), 80 munits of β-N-acetylhexosaminidase from jack beans (overnight), 3.5 munits of β-mannosidase from snail (overnight), and 0.4 munit of β-galactosidase from D. pneumoniae (overnight). In some cases, 1 µL of a 4 mM LiCl solution was added to the analyte spot to produce [M + Li]+ pseudomolecular ions instead of the Na+ adducts, resulting in an increased sensitivity. For symbols, see legends to Figures 1 and 4; [, Glc; Cer, ceramide.

colipid fraction from A. suum (Figure 5B, 10 pmol) comprising three components of differing carbohydrate chain length was first incubated with R-galactosidase, which resulted in a mass shift of the largest signal corresponding to the loss of one hexose. After treatment with β-N-acetylhexosaminidase, one or two HexNAc residues were split off leading to a new signal at m/z 982.9. Finally, incubation with β-mannosidase caused the removal of one hexose, as evidenced by the respective change in molecular mass. The final product again represented ceramide-monohexoside, which could not be further degraded. In both examples, the monosaccharide sequences and anomeric linkages could be established using 10 pmol of glycolipid. From the peak heights it may be, however, concluded that such experiments could be performed with even lower amounts of sample. To further increase signal intensities, buffer components

482 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

CONCLUSIONS MALDI-TOF-MS is a rapid and sensitive method for monitoring exoglycosidase digestions of oligosaccharides and glycoconjugates. In this study, a sample preparation technique has been developed in which ATT was used as matrix without addition of organic solvents and enzymes were predialyzed against a lowmillimolar volatile ammonium acetate buffer. Under these conditions, a series of enzymatic cleavages can be sequentially carried out directly on the MALDI target in one analyte spot, allowing intermediate mass determinations of the digestion products. Since the enzymes retain their activity and specificity in the presence of the ATT matrix, no additional sample purification steps are required. The present results suggest that the method seems to be applicable to a wide range of carbohydrates such as free oligosaccharides, oligosaccharide derivatives, and glycans bound to peptides or lipids. Since all reactions are performed in the same sample spot and the consumption of analyte during MALDI-TOFMS is minimal, sensitivity is highly improved, enabling structural analyses in the very low picomolar range. Depending on the ionization properties of the analyte, a complete analysis can be performed with 2-20 pmol of substance. Abbreviations: 2-AB-, 2-aminobenzamide derivative; ATT, 6-aza-2-thiothymine; HexNAcOH, N-acetylhexosaminitol; MALDITOF-MS, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry; NA2, diantennary standard oligosaccharide; PA-, pyridylamino derivative; SA, sialic acid. ACKNOWLEDGMENT This work was supported by the Deutsche Forschungsgemeinschaft (Sonderforschungsbereich 535, Teilprojekt Z1, and Graduiertenkolleg “Molekulare Biologie und Pharmakologie”). Part of the results were already presented at the International GlycoBioTechnology Symposium, held in Braunschweig, Germany, May 3-8, 1998. Received for review July 2, 1998. Accepted October 27, 1998. AC980712W