Anal. Chem. 1996, 68, 3215-3223
Analysis of Acidic Oligosaccharides and Glycopeptides by Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry Damon I. Papac,* Anissa Wong, and Andrew J. S. Jones
Department of Analytical Chemistry, Genentech, Inc., 460 Pt. San Bruno Boulevard, South San Francisco, California 94080
2,5-Dihydroxybenzoic acid (DHB) is the most commonly used matrix for matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI/TOF) of oligosaccharides. Because acidic, sialylated oligosaccharides are detected at only the low picomole level with DHB, alternative matrices were screened to identify a matrix with a lower limit of detection. Negative-ion spectra of pure mono-, di-, and trisialylated oligosaccharides were acquired with either 6-aza-2-thiothymine or 2′,4′,6′-trihydroxyacetophenone (THAP) in the linear mode. Detection limits of less than 50 fmol with signalto-noise ratios of better than 5:1 were achieved with both matrices. THAP was the preferred matrix because it provided a lower limit of detection and gave less prompt fragmentation. Incorporation of ammonium citrate into the matrix, along with vacuum drying of the sample, was required in order to obtain maximum sensitivity with THAP. No evidence of competition for ionization was found when a mixture of mono-, di-, and trisialylated oligosaccharides was analyzed with THAP. These findings suggest that MALDI/TOF analysis may provide a rapid means to identify changes in carbohydrate composition in glycoproteins. In addition, THAP offered improved sensitivity for detection of acidic glycopeptides over r-cyano-4-hydroxycinnamic acid. Evaluation of glycosylation patterns is fundamental to understanding not only the structure-function relationships of glycoproteins but also the mechanisms and control of their biosynthesis. The latter is especially important in the development and characterization of glycoprotein pharmaceuticals produced by recombinant DNA methods. While the sites for potential N-linked glycosylation may be predicted from the consensus sequence (the asparagine residue in -Asn-X-Ser/Thr, where X is any amino acid other than proline), it is not possible to predict whether the site will actually be glycosylated.1 For O-linked glycosylation sites, there is no known consensus sequence.2 Therefore, for both types of glycosylation, the sites and structures must be identified and characterized experimentally. The complex biosynthetic pathways for glycan structures on glycoproteins usually result in microheterogeneity and the pro(1) Bause, E. Biochem. J. 1983, 209, 331-336. (2) Elhammer, A. P.; Poorman, R. A.; Brown, E.; Maggiora, L. L.; Hoogerheide, J. G.; Kezdy, F. J. J. Biol. Chem. 1993, 268, 10029-10038. S0003-2700(96)00324-1 CCC: $12.00
© 1996 American Chemical Society
duction of a range of glycoforms (molecules with identical polypeptide sequence but with different glycans present at the sites of glycosylation). It is also known that different cell types from a given organism may produce different distributions of glycoforms.3 For recombinant DNA production of glycoproteins, an additional complicating factor which influences the glycoform distribution is the environment in which the host cells are grown during cell culture.4 The possibility that the binding affinity,5 function,6 immunogenicity,7 pharmacokinetic properties,8 or tertiary structure9 may differ between glycoforms makes it critical that the microheterogeneity be understood and controlled. One analytical method that has recently shown promise for the analysis of glycopeptides10 and carbohydrates11,12 is matrixassisted laser desorption/ionization time-of-flight mass spectrometry (MALDI/TOF). MALDI/TOF provides the molecular weight of the oligosaccharides present. Because most oligosaccharides derived from mammalian glycoproteins are comprised of relatively few different monosaccharides with unique incremental masses (e.g., fucose, 146 Da; hexose, 162 Da; N-acetylhexosamine, 203 Da; N-acetylneuraminic acid, 291 Da), a molecular weight with an accuracy of e0.05% can allow determination of the composition. Furthermore, MALDI/TOF can also provide the sequence of the oligosaccharide when fragment ions are analyzed with the approach called postsource decay analysis.13 Unfortunately, MALDI/ TOF cannot directly distinguish anomerity, branching configurations, or epimeric forms of oligosaccharides. Even with these limitations, combining molecular weight information with knowledge of the oligosaccharide source allows reasonable confidence in the assignment of oligosaccharide structures. (3) Parekh, R. B. Biologicals 1994, 22, 113-119. (4) Jenkins, N.; Curling, E. M. Enzyme Microb. Technol. 1994, 16, 354-64. (5) Cerpa, P. A.; Bishop, L. A.; Hort, Y. J.; Chin, C. K.; DeKroon, R.; Mahler, S. M.; Smith, G. M.; Stuart, M. C.; Schofield, P. R. Endocrinology 1993, 132, 351-356. (6) Sairam, M. R. FASEB J. 1989, 3, 1915-1926. (7) Noguchi, A.; Mukuria, C. J.; Suzuki, E.; Naiki, M. J. Biochem. Tokyo 1995, 117, 59-62. (8) Hotchkiss, A.; Refino, C. J.; Leonard, C. K.; O’Connor, J. V.; Crowley, C.; McCabe, J.; Tate, K.; Nakamura, G.; Powers, D.; Levinson, A.; Mohler, M.; Spellman, M. W. Thromb. Haemostas. 1988, 60, 255-261. (9) Kaushal, S.; Ridge, K. D.; Khorana, H. G. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 4024-4028. (10) Huberty, M. C.; Vath, J. E.; Yu, W.; Martin, S. A. Anal. Chem. 1993, 65, 2791-2800. (11) Mock, K. K.; Davey, M.; Cottrell, J. S. Biochem. Biophys. Res. Commun. 1991, 177, 644-651. (12) Stahl, B.; Steup, M.; Karas, M.; Hillenkamp, F. Anal. Chem. 1991, 63, 14631466. (13) Kaufmann, R. J. Biotechnology 1995, 41, 155-175.
Analytical Chemistry, Vol. 68, No. 18, September 15, 1996 3215
The most commonly used matrix for MALDI/TOF analysis of oligosaccharides has been 2,5-dihydroxybenzoic acid (DHB).14 The detection limit typically achieved with this matrix for analysis of neutral oligosaccharides is 50-100 fmol. Unfortunately, this matrix has not been generally useful for the analysis of the acidic, sialylated oligosaccharides. Sialylated oligosaccharides analyzed with DHB in either the positive- or negative-ion mode lose sialic acid. In addition, sialylated oligosaccharides analyzed in the positive-ion mode yield a mixture of cation adducts. These events account for the poorer detection limit (1-10 pmol) of acidic oligosaccharides compared to neutral oligosaccharides (50-100 fmol). Moreover, as the number of sialic acid residues incorporated into the oligosaccharide increases, these problems are further aggravated. Because many N- and O-linked oligosaccharides contain sialic acid, this limitation in sensitivity needs to be addressed. One frequently tried approach to enhance sensitivity is to derivatize the oligosaccharides either by placing a fixed charge on the free reducing end15,16 or by peracetylation of the hydroxyl groups.17 This approach is reasonable for the structural elucidation of sialylated oligosaccharides. However, quantitation of a mixture of derivatized oligosaccharides can be compromised, unless it can be shown that the derivatization efficiency is comparable for each oligosaccharide species present and that sialic acid is not lost during derivatization. The easiest way to lower the detection limit without having to perform derivatization is to identify an alternative matrix. In this paper, we will describe the preparation and use of two matrices for the analysis of acidic oligosaccharides and glycopeptides. Both 6-aza-2-thiothymine and 2′,4′,6′-trihydroxyacetophenone (THAP) enabled the detection of acidic oligosaccharides at the femtomole level. The potential use of THAP for quantitation of acidic oligosaccharide mixtures will also be addressed. EXPERIMENTAL SECTION Materials. The matrices R-cyano-4-hydroxycinnamic acid, 2,5dihydroxybenzoic acid, 5-methoxysalicylic acid, and 2′,4′,6′-trihydroxyacetophenone monohydrate were purchased from Aldrich (Milwaukee, WI). Dithiothreitol, 6-aza-2-thiothymine, ammonium citrate dibasic, and iodoacetic acid were obtained from Sigma Chemical Co. (St. Louis, MO). The oligosaccharides (Table 1) and peptide-N-glycosidase F (PNGase F) were purchased from Oxford GlycoSystems (Rosedale, NY). HPLC grade acetonitrile was obtained from Burdick and Jackson (Muskegon, WI), and punctilious ethanol was acquired from Quantum Chemical Co. (Newark, NJ). The glycopeptide fraction was obtained from a tryptic digest of a carboxymethylated recombinant glycoprotein separated by reversed-phase high-performance liquid chromatography. Oligosaccharide Isolation from tPA. Oligosaccharides were released from 2 mg of tissue plasminogen activator (tPA) with PNGaseF as previously described.18 The released oligosaccharides were recovered in the supernatant following protein precipitation with 75% ice-cold ethanol and centrifugation. The (14) Strupat, K.; Karas, M.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Processes 1991, 111, 89-102. (15) Whittal, R. M.; Palcic, M. M.; Li, L. Anal. Chem. 1995, 67, 3504-3514. (16) Levery, S. B.; Chen, L.; Nuwaysir, L.; Zaidi, I.; O’Neill, R. A. Glycobiology 1995, 5, 722. (17) Harvey, D. J. Rapid Commun. Mass Spectrom. 1993, 7, 614-619. (18) Basa, L. J.; Spellman, M. W. J. Chromatogr. 1990, 499, 205-220.
3216 Analytical Chemistry, Vol. 68, No. 18, September 15, 1996
Table 1. Structuresa of Oligosaccharides Used as Standards b
a Abbreviations for structures: Man, mannose; GlcNAc, N-acetylglucosamine; Fuc, fucose; Gal, galactose; Neu5Ac, N-acetylneuraminic acid. b The average molecular weight is provided.
recovered oligosaccharides, some existing as glycosylamines, were hydrolyzed to the free oligosaccharides by treatment with 13 mM acetic acid at room temperature for 2 h.19 Following hydrolysis, the samples were dried in a Savant Speed Vac (Farmingdale, NY) and diluted to a final concentration equivalent to 2 mg/mL (based on the original protein concentration) with deionized water. To remove cations, the released oligosaccharides were loaded onto 0.5 mL of AG 50W-X8 resin (hydrogen form, Bio-Rad, Hercules, CA) packed in 1.0 mL compact reaction columns (Amersham Life Sciences, Arlington Heights, IL). The columns were washed three times with 300 µL of deionized water. The sample effluent was collected and pooled with the effluent obtained from the water washes. The effluent containing the released oligosaccharides was dried in the Speed Vac and reconstituted to 2 mg/mL (initial protein concentration) with 0.2 M ammonium hydroxide to eliminate lactones formed between the sialic acid and the preceding galactose residue.20 High-pH Anion-Exchange Chromatography. HPAEC was performed on a DX-500 system (Dionex, Sunnyvale, CA) equipped with a CarboPac PA-100 anion-exchange column (4 mm × 250 mm, Dionex) and an ED-90 electrochemical detector (Dionex). Separation of the oligosaccharides was achieved with a gradient program at a flow rate of 1 mL/min. Solvent A consisted of 0.1 N NaOH, and solvent B consisted of 0.1 N NaOH containing 0.5 M sodium acetate. The column was equilibrated with 2.5% solvent (19) Hardy, M. R.; Townsend, R. R. Methods Enzymol. 1994, 230, 208-225. (20) Ando, S.; Yu, R. K.; Scarsdale, J. N.; Kusunoki, S.; Prestegard, J. H. J. Biol. Chem. 1989, 264, 3478-3483.
Figure 1. Effects of sample drying on crystal formation THAP (0.5 µL) was mixed on target with 0.5 µL of water: under ambient conditions, (B) dried under vacuum, and under vacuum and then allowed to absorb moisture environment. The spots are ∼2 mm in diameter.
of THAP. (A) dried (C) dried from the
B for 3 min. Using a linear gradient, the oligosaccharides were eluted by increasing solvent B to 40% in 50 min. Matrix Preparation. The R-cyano-4-hydroxycinnamic acid matrix (RCN) was prepared by dissolving 5 mg of recrystallized (in ethanol) R-cyano-4-hydroxycinnamic acid in 1 mL of acetonitrile/0.1% aqueous trifluoroacetic acid (1:1 v/v). The 2,5-dihydroxybenzoic acid matrix (sDHB) was prepared by dissolving 5 mg of 2,5-dihydroxybenzoic acid and 0.25 mg of 5-methoxysalicylic acid in 1 mL of ethanol/10 mM aqueous sodium chloride (1:1 v/v). The 6-aza-2-thiothymine (ATT) matrix was prepared by dissolving 1 mg of 6-aza-2-thiothymine in 1 mL of ethanol/20 mM aqueous ammonium citrate (1:1 v/v). The 2′,4′,6′-trihydroxyacetophenone matrix (THAP) was prepared by dissolving 1 mg of 2′,4′,6′-trihydroxyacetophenone in 1 mL of acetonitrile/20 mM ammonium citrate (1:1 v/v). Sample Preparation. Typically, 0.5 µL of analyte was applied to a polished stainless steel target, and 0.5 µL of matrix was then added to the analyte. For the oligosaccharides released from tPA, 0.5 µL of analyte was placed on the target and dried with a gentle stream of air. Once the analyte was dried, 0.5 µL of water was added to the spot, followed by 0.75 µL of THAP. All samples were dried under vacuum (50 × 10-3 Torr), except for samples prepared with the RCN matrix, which were dried under a gentle stream of air. When the THAP matrix was dried under ambient conditions, long needles formed (Figure 1A). Following vacuum drying of the THAP, the sample appeared translucent, as if an oil had formed
(Figure 1B). The best results were obtained if the sample was allowed to absorb moisture from the atmosphere following vacuum drying. Upon absorption of moisture, the appearance of the sample changed, forming small crystals (Figure 1C). Instrument Operating Parameters. The MALDI/TOF mass spectrometer used to acquire the spectra was a Voyager Elite (PerSeptive Biosystems, Framingham, MA). All samples were irradiated with UV light (337 nm) from a N2 laser. Neutral oligosaccharides were analyzed at 25 kV with a single-stage reflectron (3 m flight path) in the positive-ion mode using sDHB as the matrix. Acidic oligosaccharides were analyzed at 30 kV without the reflectron (2 m flight path) in the negative-ion mode using either ATT or THAP as the matrix. The oligosaccharides released from tPA were analyzed in THAP with the instrument operated at 20 kV in the negative-ion mode using delayed extraction. The delay time was set to 100 ns, in addition to the inherent 20 ns delay present in the electronics.21 The glycopeptides were analyzed in the linear negative-ion mode with delayed extraction (50 ns) using either THAP or RCN as the matrix. A two-point external calibration was used for mass assignment of the ions. When neutral oligosaccharides were being analyzed, structures V [(M + Na)+mono ) 1257.42 m/z] and VIII [(M + Na)+mono ) 2393.95 m/z] were used to calibrate the instrument (Table 1). When the acidic oligosaccharides were analyzed, structures I [(M - H)-avg ) 1931.8 m/z] and IV [(M - H)-avg ) 2879.6 m/z] were used to calibrate the instrument (Table 1). Generally, a mass accuracy of
