Structural Characterization of Oligosaccharides Using Maldi-TOF

For example, both α(1−4)- and α(1−6)-linked isobaric structures produce ions ... Fragmentation in MALDI-MS can result from (a) the post-source d...
0 downloads 0 Views 281KB Size
Download PDF
Anal. Chem. 2003, 75, 4895-4903

Structural Characterization of Oligosaccharides Using MALDI-TOF/TOF Tandem Mass Spectrometry Yehia Mechref and Milos V. Novotny*

Department of Chemistry, Indiana University, Bloomington, Indiana 47405 Cheni Krishnan

Applied Biosystems, 500 Old Connecticut Path, Framingham, Massachusetts 01701

Extensive cross-ring fragmentation ions, which are very informative of the linkages of the monosaccharide residues constituting these molecules, were readily observed in the MALDI-TOF/TOF/MS/MS spectra of oligosaccharides. These ions, in some cases, were more intense than the commonly observed Y and B ions. The A-type ions observed for the simple oligosaccharides allowed the distinction between r(1-4)- and r(1-6)-linked isobaric structures. The distinction was based not merely on the differences in the type of ions formed, but also on the ion intensities. For example, both r(1-4)- and r(1-6)-linked isobaric structures produce ions resulting from the loss of ∼120 m/z units, but with different intensities, as a result of the fact that they correspond to two different ions (i.e., 0,4A- and 2,4A-ions), requiring different energies to be formed. Abundant A- and X-type ions were also observed for high-mannose N-glycans, allowing the determination of linkages. In addition, the high resolution furnished by MALDI-TOF/TOF allowed determination of certain ions that were commonly overlooked by MALDITOF or MALDI-magnetic sector instruments as a result of their lower resolution. Moreover, as a result of the fact that MS/MS spectra for parent ions and all fragment ions are acquired under the same experimental conditions, accurate determination of the molar ratios of isomeric glycans in a mixture analyzed simultaneously by MALDITOF/TOF tandem MS becomes possible. During the past decade, mass spectrometry (MS) has become a key methodology in the structural elucidation of glycoconjugates. With the rapidly increasing importance of protein glycosylation in life sciences, various MS techniques currently assist structural studies in the area, although the sensitivity of glycan measurements falls somewhat behind that experienced in peptide sequencing work. The overall success of MS-based methodologies in glycoanalysis is dependent on (a) the use of appropriate ionization technique; (b) extent of fragmentation and sensitive detection of diagnostically important fragment ions; and (c) perhaps to a lesser degree, accuracy of mass measurements. A succession from fast-atom* Corresponding author address: 800 E. Kirkwood Ave., Bloomington, IN 47405. Tel: (812) 855-4532. Fax: (812) 855-8300. E-mail: [email protected]. 10.1021/ac0341968 CCC: $25.00 Published on Web 08/15/2003

© 2003 American Chemical Society

bombardment (FAB) MS1-3 and infrared-laser desorption4,5 techniques to the currently widespread matrix-assisted laser desorptionionization (MALDI) MS, and to a somewhat lesser degree, electrospray ionization (ESI), underlines the sensitivity needs of modern glycoanalysis. Although mass spectrometers using single mass analyzers can now readily assist the acquisition of compositional data in terms of isobaric monosaccharides, the structural analysis of glycans additionally requires determination of branching, linkage position, and monomer anomericity. With a relative inefficiency of most post-source decay (PSD) ion fragmentation processes, these structural attributes necessitate a use of tandem MS techniques, although some structural studies can also be aided by the use of exoglycosidase mixtures in concert with MALDI-MS.6 Not surprisingly, the fragmentation mechanisms for glycans have been studied extensively for a number of years, utilizing different combinations of ionization techniques and mass analyzers: laser-desorption Fourier transform mass spectrometry,7,8 FAB-MS,1-3 infrared-laser desorption mass spectrometry,4,5 MALDImagnetic sector-,9-11 ESI-MS,13,14 MALDI-time-of-flight-MS,15-17 ESI-ion-trap-MS,18,19 MALDI/Fourier transform ion cyclotron (1) Domon, B.; Mu ¨ ller, D. R.; Richter, W. J. Biomed. Environ. 1990, 19, 390392. (2) Domon, B.; Mu ¨ ller, D. R.; Richter, W. J. Int. J. Mass Spectrom. Ion Processes 1990, 100, 301-311. (3) Garozzo, D.; Giuffrida, M.; Impallomeni, G.; Ballistreri, A.; Montaudo, G. Anal. Chem. 1990, 62, 279-286. (4) Spengler, B.; Dolce, J. W.; Cotter R. J. Anal. Chem. 1990, 62, 1731-1737. (5) Carroll, J. A.; Ngoka L.; McCullough S.; Gard E.; Jones A. D.; Lebrilla, C. B. Anal. Chem. 1991, 63, 2526-2529. (6) Mechref, Y.; Novotny, M. V. Anal. Chem. 1998, 70, 455-463. (7) Coates, M. L.; Wilkins, C. L. Biomed. Mass Spectrom. 1985, 12, 424-428. (8) Coates, M. L.; Wilkins, C. L. Anal. Chem. 1987, 59, 197-200. (9) Carroll, J. A.; Ngoka L.; Beggs, C. G.; Lebrilla, C. B. Anal. Chem. 1993, 65, 1582-1587. (10) 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, 753-766. (11) Harvey, D. J.; Naven, T. J. P.; Kuster, B.; Bateman, R. H.; Green, M. R.; Critchley, G. Rapid. Commun. Mass Spectrom. 1995, 9, 1556-1561. (12) Harvey, D. J.; Bateman, R. H.; Green, M. R. J. Mass Spectrom. 1997, 32, 167-187. (13) Viseux, N.; deHoffmann, E.; Domon, B. Anal. Chem. 1997, 69, 3193-3198. (14) Weiskopf, A. S.; Vouros, P.; Harvey, D. J. Rapid Commun. Mass Spectrom. 1997, 11, 1493-1504. (15) Naven, T. J. P.; Harvey, D. J.; Brown, J.; Critchley, G. Rapid Commun. Mass Spectrom. 1997, 11, 1681-1686. (16) Harvey, D. J.; Hunter, A. P.; Bateman, R. H.; Brown, J.; Critchley, G. Int. J. Mass Spectrom. 1999, 188, 131-146.

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003 4895

resonance-MS,20 and, very recently, ESI21,22 and MALDI-quadrupole-time-of-flight-MS instruments.23,24 However, it is still extremely difficult to deduce some structural details on a glycan from MS data alone. The considerations of combining an effective ionization technique and energetic collision-induced dissociation, together with our current efforts25 to combine microcolumn separations of glycans with MALDI-MS, have led us to investigate the merits of a recently available MALDI-time-of-flight/time-of-flight instrumentation in elucidating detailed structures of glycans without the use of enzymes. Fragmentation of glycans observed in MALDI-MS is similar to that observed earlier in FAB-MS and ESI-MS, being dependent on several factors, such as a type of ion formation, its charge state, the energy deposited into an ion, and the time available for fragmentation. These situations have been discussed extensively in several recent reviews.26-28 In general, glycans undergo two types of cleavages: glycosidic cleavages that result from breaking the bond linking two sugar residues and cross-ring cleavages that involve rupturing two bonds on the same sugar residue. The former provide information pertaining mainly to the sequence and branching, while the latter may reveal additionally some details on a linkage. Fragmentation in MALDI-MS can result from (a) the postsource decay (PSD), which designates the fragments formed after ion extraction from the ion source; (b) in-source decay (ISD), which designates the fragments formed within the ion source; and (c) collision-induced dissociation (CID), which designates the fragments formed in a collision cell filled typically with a gas. PSD spectra of the sodiated ions from neutral carbohydrates tend to be dominated by the glycosidic and internal cleavages with very weak cross-ring ions.29 Major ions are usually the result of B and Y cleavages (according to the nomenclature introduced by Domon and Costello30), providing the information related to sequence and branching. However, a lack of abundant cross-ring cleavages limits the linkage information to be deduced. This lack of cross-ring fragmentation in PSD is attributed to the high-energy requirements. High-energy CID provides more readily such energies, and accordingly, cross-ring fragmentation is often observed. The very abundant 1,5X-ions are of particular significance, since their masses can be used to determine the branching patterns of glycans, as is readily seen in the high-mannose sugars.12 Never(17) Mechref, Y.; Baker, A.; Novotny, M. V. Carbohydr. Res. 1999, 313, 145155. (18) Weiskopf, A. S.; Vouros, P.; Harvey, D. J. Anal. Chem. 1998, 70, 44414447. (19) Sheeley, D. M.; Reinhold, V. N. Anal. Chem. 1998, 70, 3053-3059. (20) Solouki, T.; Reinhold, B. B.; Costello, C. E.; O’Malley, M.; Guan, S. H.; Marshall, A. G. Anal. Chem. 1998, 70, 857-864. (21) Harvey, D. J. J. Am. Soc. Mass. Spectrom. 2000, 11, 900-915. (22) Harvey, D. J. J. Mass Spectrom. 2000, 35, 1178-1190. (23) Harvey, D. J.; Bateman, R. H.; Bordoli, R. S.; Tyldesley, R. Rapid Commun. Mass Spectrom. 2000, 14, 2135-2142. (24) Hunnam, V.; Harvey, D. J.; Priestman, D. A.; Bateman, R. H.; Bordoli, R. S.; Tyldesley, R. J. Am. Soc. Mass. Spectrom. 2001, 12, 1220-1225. (25) Tegeler, T. J.; Mechref, Y. S.; Boraas, K.; Reilly, J. P.; Novotny, M. V. Anal. Chem., to be submitted. (26) Reinhold, V. N.; Reinhold, B. B.; Costello, C. E. Anal. Chem. 1995, 67, 1772-1784. (27) Harvey, D. J. Mass Spectrom. Rev. 1999, 18, 349-451. (28) Mechref, Y.; Novotny, M. V. Chem. Rev. 2002, 102, 321-370. (29) Spengler, B.; Kirsch, D.; Kaufmann, R.; Lemoine, J. J. Mass Spectrom. 1995, 30, 782-787. (30) Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5, 397-409.

4896

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

theless, these cross-ring fragmentations are still very limited, even in the case of conventional CID,17 and accordingly, exoglycosidases are still being utilized to acquire linkage information.6 In this paper, MALDI-TOF/TOF tandem MS has been employed for investigating first the linear oligosaccharides and branched oligosaccharides, including high-mannose-type N-glycans derived from ribonuclease B. The early part of our studies has been aimed at evaluating the extent of fragmentation these oligosaccharides experience and the potential of utilizing this type of mass spectrometer for the detailed characterization of different oligosaccharides. EXPERIMENTAL SECTION Materials. Pancreatic bovine ribonuclease B, dextran (8000 av MW), dextrin from maize starch, and maltohexaose were purchased from Sigma Chemical Company (St. Louis, MO). Recombinant PNGase F was received from Glyko, Inc. (Novato, CA). The MALDI matrix, 2,5-dihydroxybenzoic acid (DHB), and all other common chemicals were purchased from Aldrich (Milwaukee, WI). Enzymatic Release of N-Glycans from Ribonuclease B. The enzymatic release of N-glycans from ribonuclease B was carried out according to our previously published procedure.6 Briefly, 1 µg glycoprotein was suspended in 1 µL of an incubation buffer consisting of 10 mM sodium phosphate, pH 7.0, and 1% mercaptoethanol. The sample was then thermally denatured by incubation at 95 °C for 5 min. Next, the sample was allowed to cool to room temperature prior to the addition of 5 mU of PNGase F and incubation at 37 °C in a water bath for 3 h. Finally, the sample was spotted directly on the MALDI plate and mixed with the DHB matrix. MALDI Spot Preparation. The DHB matrix was prepared by suspending 10 mg in 1 mL of water to produce 10 mg/mL matrix concentration. Sample and matrix were mixed on the MALDI plate at a ratio of 1:1 and allowed to dry at room temperature. All ions observed in the spectra were sodiated. Instrumentation. An Applied Biosystems 4700 Proteomics Analyzer (Applied Biosystems, Framingham, MA) was utilized in this study. This TOF/TOF instrument is equipped with an Nd:YAG laser with 355-nm wavelength of 10 000), while the observed profile reflects the right distribution of the different glycan structures known to exist on ribonuclease B, including high-mannose glycans extend(31) Manners, D. J.Carbohydr. Polym. 1989, 11, 87-112.

Figure 3. MS/MS recording of dextrin DP6 from maize starch.

Figure 4. MALDI-TOF/TOF spectrum of N-glycans derived from ribonuclease B. Symbols: (9) N-acetylglucosamine, (O) mannose.

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

4899

Figure 5. MS/MS recording of Man5.

ing from Man5 to Man9. Only Man5, 6, and 7 were examined in this work to evaluate the effectiveness of MALDI-TOF/TOF for characterization of branched glycans. The abundance of cross-ring fragmentation in the case of highmannose N-glycans was similar to that of the other structures discussed so far. Moreover, the extent of cross-ring fragmentation was very much similar to that recorded on a magnetic sector instrument fitted with a collision cell and an orthogonal TOF analyzer.12 The MS/MS spectrum of Man5 is shown in Figure 5. Unlike a spectrum acquired on a MALDI-TOF instrument,17 this spectrum exhibits extensive cross-ring fragmentation ions with a relatively high intensity. In addition to the highly abundant and intense cross-ring fragmentation ions, MALDI-TOF/TOF/MS/ MS spectra exhibit high resolution (>5000), permitting, for example, the designation of Y4x and 2,4A5 ions in Figure 5. These two ions differ from each other by ∼2 m/z units and are not readily resolved in the MALDI-TOF/MS/MS spectrum.17 Although the MS/MS spectrum acquired by MALDI-TOF/ TOF resembles that acquired by a magnetic sector instrument fitted with orthogonal TOF analyzer, the MALDI-TOF/TOF/MS/ MS spectrum is more abundant with the ions resulting from internal fragmentation and more intense cross-ring fragmentation ions. One of the most intense ions observed in the MS/MS spectrum of Man5 is that corresponding to ion D, which results from the loss of 3-linked antenna and two reducing-end Nacetylglucosamine residues (i.e., B3/Y3β), a mechanism proposed for its formation by Harvey.27 The detection of this ion allows determination of the monosaccharide composition of the 6-linked antenna, because it contains a complete structural arrangement. 4900

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

Moreover, the composition of a 3-linked antenna could be deduced from the formation of this ion by difference. Other ions that are important for the determination of structural linkages were observed, including 1,5X3R, 3,5A3 and 0,4A3. Among these, the 3,5A3 and 0,4A3 ions produced by cleavage of the core-branching mannose residue are, in general, the most useful cross-ring cleavage ions in the spectra of N-glycans. Since these ions contain only the antenna attached to the 6-position of the core mannose residue, they provide valuable information on the composition of each antenna. In addition, a complete 1,5X-ion series extending throughout all the Man5 monosaccharide residues is observed in the spectrum (Figure 5), along with many other A-type ions that permit the determination of the linkages of the N-glycan monosaccharide residues. The commonly observed Y-, B- ,and some C-ion series were also observed in the MS/MS spectrum of Man5. A detailed designation of the cross-ring fragmentations observed for Man 5 is illustrated on the structure shown in Figure 5. The MS/MS spectrum of Man6 is depicted in Figure 6; it closely resembles that of Man5. An abundance of ions corresponding to cross-ring fragmentation was observed with relatively high intensities. Several A-ion and X-ion series were observed, aiding in the determination of the monosaccharide linkages on the N-glycan structure. As in the case of Man5, the ions resulting from the cross-ring fragmentation were as intense as those resulting from the cleavage of the glycosidic bond (if not higher) in some cases. Ion D was also observed at m/z 671.22, along with 1,5X , 3,5A , and 0,4A ions. The cross-ring fragmentations observed 3R 3 3 for Man 6 are designated on the structure depicted in Figure 6.

Figure 6. MS/MS recording of Man6.

The inability of MALDI-TOF/MS to produce spectra that reflect the relative intensity of the different ions (produced as a result of PSD or CID experiments) is one of the major disadvantages of conventional tandem instruments. This is due to the fact that the MS/MS spectra acquired on a MALDI-TOF instrument are the result of data compilation from several experiments not necessarily acquired under strictly identical conditions. An alternative has been to employ a curved-filed reflectron, which was exploited recently for PSD fragmentation analyses of the linkage isomers of oligosaccharides.32,33 A reflectron of the curved voltage gradient type34,35 allows acquisition of a total fragment spectrum in a single experiment. This type of reflectron employs a modified single-stage reflector; the gradient differential increases of its axial voltage produce an alignment of energy focal points for product ions. Therefore, the parent ion and all the fragment ions are detected under the same measurement conditions. A major advantage of this reflectron is an accurate comparison of the intensity of the differentially observed fragments, since they are all detected under the same conditions. The MALDI-TOF/TOF instrument allows the production of MS/MS spectra in which the parent ion and all fragment ions (32) Yamagaki, T.; Nakanishi, H. J. Mass Spectrom. 2000, 35, 1300-1307. (33) Yamagaki, T.; Nakanishi, H. Proteomics 2001, 1, 329-339. (34) Cornish, T. J.; Cotter, R. J. Rapid Commun. Mass Spectrom. 1993, 7, 10371040. (35) Cornish, T. J.; Cotter, R. J. Rapid Commun. Mass Spectrom. 1994, 8, 781785.

are detected simultaneously under the same measurement conditions, thus permitting an accurate comparison of the intensity of the fragments observed. This aspect is very valuable when the MS/MS data are acquired for a mixture of isomers. This capability of the MALDI-TOF/TOF instrument was examined using Man7 cleaved from ribonuclease B. It is known that this N-glycan exists as three different isomers in which the seventh mannose residue is attached to either the upper (Man7I) or lower (Man7II) branches of the 6-linked antenna or to the 3-linked antenna (Man7III). According to an NMR study by Fu et al.,36 these isomers exist in a molar ratio of 25:37:38, constituting together 4 mol % of the total N-glycans derived from ribonuclease B. There are six ions observed in the MS/MS spectrum of Man7 (Figure 7), three of which are characteristic of Man7I and Man7II, whereas the other three are characteristic of the third isomer Man7III. Accordingly, the intensity ratio of these different ions should agree with the molar ratio at which these isomers are known to exist.34 The ions at m/z 570.20 and 731.22 correspond to 0,4A3 ions of Man7III and Man7I and II, respectively. The intensity ratio of these two ions is 0.625, which closely matches that determined by NMR (i.e., 0.613). In addition, ions at m/z 583.24 and 74 correspond to 3,5A3 ions of Man7III and Man7I and II, respectively. The intensity ratio of these two ions is also 0.625, which closely matches that determined by NMR (i.e., 0.613). Accordingly, the two sets of ions could be reliably used to calculate (36) Fu, D.; Chen, L.; O’Neill, A. Carbohydr. Res. 1994, 261, 173-186.

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

4901

Figure 7. MS/MS recording of Man7. Inset represents an enlarged section of the spectrum.

the molar ratios of isomers analyzed by MALDI-TOF/TOF tandem mass spectrometry. The third set of ions, which could be potentially used to estimate the molar ratio of an isomeric mixture analyzed by tandem MS, are the two ions at m/z 671.23 and 833.26, corresponding to D ions of Man7III and Man7I and II, respectively. The intensity ratio of these two ions is 0.536, which is less than the expected value. This is due to the fact that intensity of the signal observed at m/z 833.26 is not only due to ion D, but is also due to other ions produced as a result of the fragmentation of the parent ion. For example, the ion at 833.27 is an internal fragment observed in the MS/MS spectrum of Man6, and it is also expected to be present in the spectrum of Man7. Accordingly, the intensity of the signal observed at m/z 833 is higher than it should be (if it were only corresponding to ion D), and as a result, the molar ratio is underestimated. Therefore, these ions could not be used to quantitatively determine the molar ratios of a mixture of isomers. CONCLUSIONS MS/MS spectra of simple oligosaccharides and N-glycans derived from glycoproteins exhibit a high abundance of crossring fragmentation ions that are very informative of the linkages of the monosaccharide residues constituting these molecules. These ions were not only abundant, but also intense to the extent that, in some instances, they were more intense than the commonly observed Y and B ions. The A-type ions observed for the simple oligosaccharides allowed the distinction between R(1-4)- and R(1-6)-linked isobaric structures. The distinction was 4902

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

based not merely on the differences in the type of ions formed, but also on their intensities; for example, both R(1-4)- and R(1-6)-linked isobaric structures produce ions resulting from the loss of ∼120 m/z units, but with different intensities as a result of the fact that they correspond to two different ions (i.e., 0,4A and 2,4A ions). N-glycans were also comprehensively characterized using MALDI-TOF/TOF tandem mass spectrometry. Abundant A- and X-type ions were observed, allowing the determination of linkages. In addition, the high resolution furnished by MALDI-TOF/TOF also allowed determination of certain ions that were commonly overlooked by MALDI-TOF or MALDI-magnetic sector instruments as a result of their low resolution. The measurements of certain ions in the MALDI-TOF/TOF/ MS/MS spectra combined with the fact that MS/MS spectra for parent ions and all fragment ions are acquired under the same experimental conditions permits accurate determination of the molar ratios of isomeric glycans in a mixture. Although this study pertains to fairly simple oligosaccharide structures, a forthcoming communication37 addresses the structural information extracted from the spectra of fucosylated and sialylated structures. ACKNOWLEDGMENT This work was supported by Grant no. GM24349 from the National Institute of General Medical Sciences, U.S. Department (37) Mechref, Y.; Novotny, M. V. To be submitted.

of Health and Human Services (M.V.N.). In addition, the instrument used in this work was funded as a result of the Indiana Genomics Initiative (INGEN), which is funded in part by the Lilly Endowment, Inc. (M.V.N.).

Received for review February 26, 2003. Accepted June 18, 2003. AC0341968

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

4903