3-Aminoquinoline Acting as Matrix and Derivatizing Agent for MALDI

Apr 13, 2010 - Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) is a widely used method in oligosaccharide ...
0 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 3719–3726

3-Aminoquinoline Acting as Matrix and Derivatizing Agent for MALDI MS Analysis of Oligosaccharides Marion Rohmer,† Bjoern Meyer,† Marko Mank,‡ Bernd Stahl,‡ Ute Bahr,† and Michael Karas*,† Institute of Pharmaceutical Chemistry, Goethe-University Frankfurt, Max-von-Laue-Strasse 9, 60438 Frankfurt am Main, Germany and Danone Research Centre for Specialised Nutrition, Bahnstrasse 14-30, 61381 Friedrichsdorf, Germany Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) is a widely used method in oligosaccharide analysis. Underivatized oligosaccharides are not well-suited for that purpose due to their low ionization efficiency; however, derivatization requires tedious sample purification steps which may lead to sample losses, thereby decreasing its benefit. On-target derivatization performed by the matrix 3-aminoquinoline does not require such purification and yields Schiff bases which can be measured in positive and negative ion mode from one single spot. In negative ion mode, spectra from anionic adducts of the derivatives can be acquired from 1 fmol of oligosaccharide. Furthermore, postsource decay (PSD) fragmentation in positive and negative ion mode is enhanced, providing information on oligosaccharide sequence, linkage, and branching. Optimization of reaction conditions and matrix solution led to a complete and reproducible derivatization for all tested standard oligosaccharides. Finally, the method was applied to trifucosyllacto-N-hexaose and trifucosyl-para-lacto-N-hexaose, two isomers occurring in human breast milk samples, which were easily identified and distinguished. Glycosylation is by far the most common post-translational modification of proteins.1 Pattern and amount of the attached glycans are able to influence a protein’s bioactivity, immunogenicity and other properties.2-7 Due to previous discoveries on the versatile biological roles of oligosaccharides,2,4,8,9 their analysis has attracted continuously increasing interest. Especially, when * To whom correspondence should be addressed. Tel: +49 (0) 69 798-29916. Fax: +49 (0) 69 798-29918. E-mail: [email protected]. † Goethe-University Frankfurt. ‡ Danone Research Centre for Specialised Nutrition. (1) Lis, H.; Sharon, N. Eur. J. Biochem./FEBS 1993, 218 (1), 1–27. (2) Varki, A. Glycobiology 1993, 3 (2), 97–130. (3) Opdenakker, G.; Rudd, P. M.; Ponting, C. P.; Dwek, R. A. FASEB J. 1993, 7 (14), 1330–1337. (4) Dwek, R. A. Biochem. Soc. Trans. 1995, 23 (1), 1–25. (5) Dwek, R. A.; Lellouch, A. C.; Wormald, M. R. J. Anat. 1995, 187 (2), 279– 292. (6) Prescher, J. A.; Dube, D. H.; Bertozzi, C. R. Nature 2004, 430 (7002), 873–877. (7) Taylor, A. D.; Hancock, W. S.; Hincapie, M. R.; Taniguchi, N.; Hanash, S. M. Genome Med. 2009, 1 (6), 57. (8) Paulson, J. C.; Blixt, O.; Collins, B. E. Nat. Chem. Biol. 2006, 2 (5), 238– 248. (9) Varki, A. Cell 2006, 126 (5), 841–845. 10.1021/ac1001096  2010 American Chemical Society Published on Web 04/13/2010

characterizing recombinant proteins produced by the pharmaceutical industry, glycan profiling and monitoring is highly important. Over the last decades, numerous methods have been developed to address this issue.10,11 Among those, mass spectrometry has emerged as one of the most important due to its sensitivity and rapidity.12-18 MALDI MS is often used to provide a rapid overview of glycan profiles or complex mixtures, while electrospray ionization (ESI) MS is applied for structural analysis via MS/MS or MSn. As ionization efficiency of native oligosaccharides is structure dependent and often low,19 derivatization is a common tool.20 The most popular derivatization method is reductive amination, including Schiff base formation and its subsequent reduction to a secondary amine. Unfortunately, the resulting improvement in sensitivity is associated with prevalently tedious purification which inevitably leads to sample losses.21 Since reduction of the Schiff base is usually most inconvenient, methods were developed which avoided this step, instead forming stable oximes,22 hydrazones,21,23,24 or Schiff bases.25,26 However, there is always an equilibrium between these (10) Dwek, R. A.; Edge, C. J.; Harvey, D. J.; Wormald, M. R.; Parekh, R. B. Annu. Rev. Biochem. 1993, 62, 65–100. (11) Wada, Y.; Azadi, P.; Costello, C. E.; Dell, A.; Dwek, R. A.; Geyer, H.; Geyer, R.; Kakehi, K.; Karlsson, N. G.; Kato, K.; Kawasaki, N.; Khoo, K.; Kim, S.; Kondo, A.; Lattova, E.; Mechref, Y.; Miyoshi, E.; Nakamura, K.; Narimatsu, H.; Novotny, M. V.; Packer, N. H.; Perreault, H.; Peter-Katalinic, J.; Pohlentz, G.; Reinhold, V. N.; Rudd, P. M.; Suzuki, A.; Taniguchi, N. Glycobiology 2007, 17 (4), 411–422. (12) Stahl, B.; Steup, M.; Karas, M.; Hillenkamp, F. Anal. Chem. 1991, 63 (14), 1463–1466. (13) Bahr, U.; Pfenninger, A.; Karas, M.; Stahl, B. Anal. Chem. 1997, 69 (22), 4530–4535. (14) Harvey, D. J. Mass Spectrom. Rev. 1999, 18 (6), 349–450. (15) Harvey, D. J. J. Chromatogr., A 1996, 720 (1-2), 429–446. (16) Zaia, J. Mass Spectrom. Rev. 2004, 23 (3), 161–227. (17) Harvey, D. J.; Royle, L.; Radcliffe, C. M.; Rudd, P. M.; Dwek, R. A. Anal. Biochem. 2008, 376 (1), 44–60. (18) Nilsson, B. Mol. Biotechnol. 1994, 2 (3), 243–280. (19) Naven, T. J.; Harvey, D. J. Rapid Commun. Mass Spectrom. 1996, 10 (11), 1361–1366. (20) Lamari, F. N.; Kuhn, R.; Karamanos, N. K. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2003, 793 (1), 15–36. (21) Naven, T. J. P.; Harvey, D. J. Rapid Commun. Mass Spectrom. 1996, 10, 829–834. (22) Zhao, Y.; Kent, S. B. H.; Chait, B. T. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 1629–1633. (23) Lattova, E.; Perreault, H. J. Chromatogr., B 2003, 793, 167–179. (24) Kapkova´, P. Rapid Commun. Mass Spectrom. 2009, 23, 2775–2784. (25) Snovida, S. I.; Chen, V. C.; Perreault, H. Anal. Chem. 2006, 78, 8561– 8568. (26) Snovida, S. I.; Rak-Banville, J. M.; Perreault, H. J. Am. Soc. Mass Spectrom. 2008, 19, 1138–1146.

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3719

unreduced derivatives and the corresponding underivatized oligosaccharide. As this equilibrium can be shifted to either side by external conditions, complete derivatization is difficult to achieve, which often leads to an increased number of peaks in MS spectra. Moreover, fragmentation of these derivatives provides spectra highly differing from those of underivatized oligosaccharides.27-29 Since the well-established fragmentation rules of underivatized species cannot be applied without restrictions, structural characterization is more complex. For MALDI MS analysis of oligosaccharides, the most widely used matrix is still 2,5-dihydroxybenzoic acid (DHB), one of the first matrices ever applied in this field.12 However, restriction to positive ion mode and inhomogeneous crystallization limit its application. Regardless of these disadvantages, no other matrix has been successfully established as an adequate alternative.14 Ideally, a matrix for oligosaccharide analysis should be able to combine high sensitivity, suitability for positive as well as negative ion mode, and good fragmentation patterns. Although 3-aminoquinoline (3-AQ), first applied by Metzger et al. for the analysis of plant inulins, is described as superior to DHB in terms of sensitivity and resolution,30,31 this matrix never gained high popularity because it forms Schiff bases (imines) with the reducing end of oligosaccharides. The occurrence of these derivatives was considered as an undesired side reaction, as it complicated the resulting mass spectra.14 Our approach is to take advantage of this reaction to combine both, a potent MALDI matrix and the benefits of derivatization for structural analysis of oligosaccharides. We report here the optimization of reaction conditions leading to the quantitative formation of Schiff bases with 3-AQ directly on the MALDI target. We will show that this method allows for the structural analysis of different classes of neutral oligosaccharides in positive and negative ion mode. EXPERIMENTAL SECTION Materials and Reagents. Ten different commercially available oligosaccharides (maltoheptaose, maltopentaose, maltotetraose, cellopentaose, 3R,6R-mannopentaose, panose, N-acetyl-D-lactosamine, N-acetyl-allolactosamine, gentiobiose, palatinose) were tested in the course of method development. All standard oligosaccharides, 3-aminoquinoline, harmine, and all acids were purchased from Sigma (St. Louis, MO). Additionally, two human milk oligosaccharides, trifucosyllacto-N-hexaose and trifucosylpara-lacto-N-hexaose, were obtained from Danone Research Centre for Specialised Nutrition (Friedrichsdorf, Germany). All oligosaccharides were dissolved in ultrapure water produced by a Milli-Q system (Millipore, Billerica, MA) to a final concentration of 1 mM and further diluted depending on the kind of application (see text). Acetonitrile (ACN, gradient grade), acetone, and ethanol were obtained from Carl Roth (Karlsruhe, Germany). DHB was supplied by Bruker Daltonics (Bremen, Germany). Optimization of Reaction Conditions. Experimental conditions which may influence reaction conditions and crystallization (27) Ku ¨ster, B.; Naven, T. J.; Harvey, D. J. Rapid Commun. Mass Spectrom. 1996, 10, 1645–1651. (28) Lattova´, E.; Snovida, S.; Perreault, H.; Krokhin, O. J. Am. Soc. Mass Spectrom. 2005, 16, 683–696. (29) Harvey, D. J. J. Am. Soc. Mass Spectrom. 2000, 11, 900–915. (30) Metzger, J. O.; Woisch, R.; Tuszynski, W.; Angermann, R. Fresenius J. Anal. Chem. 1994, 349 (6), 473–474. (31) Stahl, B.; Thurl, S.; Zeng, Z. R.; Karas, M.; Hillenkamp, F.; Steup, M.; Sawatzki, G. Anal. Biochem. 1994, 223 (2), 218–226.

3720

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

were varied; among those were concentration of matrix ((5/10/ 20/30) mg/mL), percentage and type of organic solvent (ACN, acetone, ethanol (0/10/33/50/100)%), type of acid (nitric acid, hydrochloric acid, trifluoroacetic acid, phosphoric acid, sulfuric acid, acetic acid), hence resulting pH values (3/4/5/6/7), drying conditions (ambient temperature or elevated temperature), and different types of preparation (dried-droplet versus thin-layer). For best possible comparison of all reaction conditions, only one parameter was varied while all other parameters were kept constant at their optimum. The respective matrix solutions were spotted five times each on a polished steel MALDI target together with maltoheptaose solution (10 pmol per spot). All preparations were examined concerning completeness of derivatization and crystallization behavior. Derivatization rate was calculated by dividing peak intensity of 3-AQ-derivatized maltoheptaose by peak intensities of both underivatized and 3-AQ-derivatized maltoheptaose. Average values and standard deviations were determined for each preparation type. Mass Spectrometric Analysis of Oligosaccharides. MALDI MS spectra were recorded on an Ultraflex I MALDI TOF/TOF mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with a N2 laser (337 nm) in positive and negative reflector ion mode. Laser intensity was set marginally above the threshold of ionization to avoid fragmentation. All MALDI MS/MS PSD spectra were acquired with an ABI 4800 MALDI TOF/TOF Analyzer (Applied Biosystems, Darmstadt, Germany) with a Nd:YAG laser (355 nm). Laser energy was set to 5000 and 5500 au for derivatized and underivatized oligosaccharides, respectively. Oligosaccharide and matrix solution (0.5-1 µL) were mixed together on a polished steel target. The m/z range was chosen according to the individual structure of the examined oligosaccharide. With both instruments, 500-1000 shots were summarized for each spectrum. Default calibration was conducted using maltooligosaccharides (maltoheptaose, maltopentaose, and maltotetraose) for MS measurements and their wellcharacterized fragments in positive and negative ion mode for MS/MS experiments. Resulting spectra were interpreted manually assisted by the GlycoWorkbench software (EuroCarbDB).32 RESULTS AND DISCUSSION On-Target Derivatization. The variation of reaction conditions as described in the Experimental Section yielded preparations highly differing in derivatization rate (see Figure 1) and crystallization. As to be expected, the most influential parameter turned out to be the pH value of the matrix solution (see Figure 1A). Imine formation is catalyzed by a weakly acidic environment; above pH 7, derivatization yield is close to zero (data not shown). On the other hand, pH values lower than 3 result in acidic hydrolysis. A pH of 5 proved to be optimal, which was achieved by adjusting the solution with 0.07% of a strong, inorganic acid. In the course of the experiments, the choice of acid proved to be an important parameter for the particular type of measurement and will be discussed later. Furthermore, a concentration of 20 mg/mL of 3-AQ was optimal in terms of crystallization and (32) Ceroni, A.; Maass, K.; Geyer, H.; Geyer, R.; Dell, A.; Haslam, S. M. J. Proteome Res. 2008, 7 (4), 1650–1659.

Figure 1. Derivatization rates obtained by variation of reaction conditions. Influential parameters were pH value of matrix solution (A), matrix concentration (B), percentage (C) and type (D) of organic solvent, preparation type (E), and temperature (F).

derivatization (see Figure 1B). More than 20 mg/mL of 3-AQ showed no further improvement of derivatization, but crystallization was impaired. In addition to that, the solvent composition plays an important role (see Figure 1C,D). A percentage of 33% organic solvent was selected, as solutions with lower organic content turned out to be unfavorable for crystallization. However, increasing rates of organic solvent resulted in decreasing derivatization rates, probably because accelerated drying of the spots reduced the reaction time. This may be contrary to expectations, since Schiff base formation involves dehydration and should be best when the water content is as low as possible. However, water is withdrawn by evaporation, shifting the equilibrium toward the reaction product. The excess of 3-AQ adds to this effect. Out of different organic solvents, ACN was favored as it promoted the best crystallization behavior. Taking all this into account, matrix solution was found to be optimal in terms of derivatization and crystallization when 20 mg/mL of 3-AQ in an ACN-water mixture (1:2 v/v) with 0.07% acid were used. Scheme 1 illustrates that the Schiff base formation is a reaction involving dehydration and that the product is at equilibrium with the respective glycosylamine.33 For derivatization directly on the MALDI target, the dried-droplet method was preferred to the thin-layer method since it offers a better contact between oligosaccharide and derivatizing agent and, therefore, improves the derivatization rate (see Figure 1E). Immediately after spotting, a highly viscous, almost gel-like fluid is formed, which subsequently crystallizes rapidly at elevated temperatures or in approximately 1 h at room temperature. Although drying at ambient temperature is more time-consuming, it ensures complete derivatization since reaction time is extended (see Figure 1F). Subsequent to the outlined optimazition process, 10 different standard oligosaccharides were selected which cover different structural elements and a wide mass range to prove the general applicability of this method. With the optimized sample preparation procedure, all tested oligosaccharides were derivatized (33) Ojala, W. H.; Ostman, J. M.; Ojala, C. R. Carbohydr. Res. 2000, 326 (2), 104–112.

Scheme 1. Schiff Base Formation from 3-Aminoquinoline and the Reducing End of Oligosaccharides

quantitatively and reproducibly (see Supporting Information). The symbols and abbreviations used in all following figures and the oligosaccharide masses are listed in Chart 1. (Symbols are according to the Nomenclature Committee of the Consortium for Functional Glycomics (CFG); see http://www.functionalglycomics.org/static/consortium/Nomenclature.shtml.) Oligosaccharides derivatized with 3-AQ are denoted as 3-AQ-OS in the following text. MS and MS2 Spectra of 3-AQ-Derivatized Standard Oligosaccharides. Maltopentaose is chosen as a standard oligosaccharide to exemplify the benefits of derivatization with 3-AQ. Figure 2A,B illustrates the difference between spectra of underivatized maltopentaose with standard matrix DHB and 3-AQmaltopentaose, both acquired in positive ion mode. While the molecules are ionized by Na+ and K+ attachment using DHB, the Schiff base appears as protonated species due to the basic nature of the quinoline system introduced by 3-AQ. The predicted increase in molecular weight of 126 Da is observed, and no signals of the underivatized oligosaccharides remain, Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3721

Chart 1.

a

Abbreviations, Symbols, and Masses of Monosaccharidesa

Symbols are depicted according to the consortium for functional glycomics, CFG.

Figure 2. MS spectra of underivatized and derivatized maltopentaose. The MALDI-TOF MS spectrum of underivatized maltopentaose with DHB in positive ion mode (A) shows adducts of different metal cations, whereas 3-AQ-maltopentaose is ionized as protonated species in positive ion mode (B). With HNO3 as an additive, the nitrate adduct of 3-AQ-maltopentaose is formed in negative ion mode (C).

demonstrating complete conversion into 3-AQ-OS. Due to the enhanced proton affinity of the Schiff base and the fact that the signal is no longer split into Na+- and K+-adducts, the detection limit for standard oligosaccharides such as maltoheptaose is lowered from about 1 pmol with DHB to about 100 fmol with 3-AQ. Furthermore, 3-AQ-OS form anionic adducts in negative ion mode as published for underivatized oligosaccharides.34-45 As already reported for ESI MS in negative ion mode,39 nitrate ions were found to form the most stable adducts also for MALDI compared to chloride, sulfate, and phosphate and are, therefore, the best choice for low sample amounts. The nitrate (34) Chai, W.; Piskarev, V.; Lawson, A. M. Anal. Chem. 2001, 73 (3), 651–657. (35) Pfenninger, A.; Karas, M.; Finke, B.; Stahl, B. J. Am. Soc. Mass Spectrom. 2002, 13 (11), 1331–1340. (36) Pfenninger, A.; Karas, M.; Finke, B.; Stahl, B. J. Am. Soc. Mass Spectrom. 2002, 13 (11), 1341–1348. (37) Yamagaki, T.; Suzuki, H.; Tachibana, K. J. Am. Soc. Mass Spectrom. 2006, 17 (1), 67–74. (38) Yamagaki, T.; Suzuki, H.; Tachibana, K. J. Mass Spectrom. 2006, 41 (4), 454–462. (39) Harvey, D. J. J. Am. Soc. Mass Spectrom. 2005, 16 (5), 622–630. (40) Harvey, D. J. J. Am. Soc. Mass Spectrom. 2005, 16 (5), 631–646. (41) Harvey, D. J. J. Am. Soc. Mass Spectrom. 2005, 16 (5), 647–659. (42) Wong, A. W.; Cancilla, M. T.; Voss, L. R.; Lebrilla, C. B. Anal. Chem. 1999, 71 (1), 205–211. (43) Cai, Y.; Jiang, Y.; Cole, R. B. Anal. Chem. 2003, 75 (7), 1638–1644. (44) Harvey, D. J. Proteomics 2005, 5 (7), 1774–1786. (45) Guan, B.; Cole, R. B. J. Am. Soc. Mass Spectrom. 2008, 19 (8), 1119–1131.

3722

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

adduct of 3-AQ-maltopentaose is displayed in Figure 2C. The limit of detection for the [M + NO3]- ions of 3-AQ-OS was determined with a dilution series of maltoheptaose ranging from 1 nmol to 1 fmol (see Supporting Information). Negative ion mode mass spectra could be acquired from 1 fmol of sample on the target with a signal-to-noise (S/N) ratio of greater than 3. Subsequently, fragmentation of 3-AQ-OS in MALDI-TOF/TOF analysis was investigated. Fragmentation of the typical Na+adducts generated with DHB was used as a comparison in positive ion mode. Figure 3A shows the MS2 spectrum with [M + Na]+ of underivatized maltopentaose as precursor ion. Fragments are designated according to the nomenclature proposed by Domon and Costello46 (see Supporting Information), and water losses are indicated with (*) in all following MS2 spectra. As cleavage products from both the reducing and the nonreducing end are able to retain the charge during the fragmentation process, it is difficult to assign a peak to a particular fragment. On the contrary, PSD fragmentation of 3-AQ-OS in positive ion mode benefits from charge localization at the reducing end (see Figure 3B). Due to the proton affinity of the quinoline system, only fragments including the derivatized reducing end retain the charge when the molecule is cleaved. This (46) Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5 (4), 397–409.

Figure 3. MALDI-TOF/TOF fragmentation patterns and MS2 spectra of underivatized and 3-AQ-derivatized maltopentaose. On the left, positive ion mode MALDI MS2 spectra are shown for the underivatized oligosaccharide with DHB ([M + Na]+ as precursor) (A) and the derivatized oligosaccharide with 3-AQ ([M + H]+ as precursor) (B). On the right, negative ion mode MALDI MS2 spectra with harmine + NaCl (C) and with 3-AQ + HCl (D) are compared; [M + Cl]- ions are selected as precursors in both cases.

promotes the formation of Y-ions, giving a very simple fragmentation pattern from which the oligosaccharide’s sequence can easily be deduced. Mass spectrometric analysis of oligosaccharides in negative ion mode has enjoyed increasing popularity due to its higher sensitivity and stability of fucoses and other labile residues compared to positive ion mode.31,34–36 The greatest drawback of negative ion mode is the lability of [M - H]- ions,39 which can be eluded by forming stable adducts with small, inorganic anions. Throughout our experiments, harmine spiked with sodium chloride (NaCl) proved to be suited best for the generation of anionic adducts and was, therefore, chosen as a reference in negative ion mode. As already described elsewhere,34–37 chloride adducts were best for MALDITOF/TOF fragmentation in our case and were used for all further fragmentation experiments. This phenomenon can be explained by the fact that the gas-phase basicity of chloride is slightly lower than or close to that of the deprotonated oligosaccharide [M H]-.43 Therefore, chloride can abstract a proton from the oligosaccharide, thereby generating [M - H]- ions of oligosaccharides which provide structural information. Figure 3C shows the MS2 spectrum obtained from [M + Cl]- of underivatized maltopentaose with harmine as a matrix. Fragmentation patterns of anionic adducts of underivatized oligosaccharides via ESI and MALDI MS have been the issue of earlier publications.35–45 Here, C-cleavages are preferred to B/Y-cleavages, as they are stabilized by the formation of a new reducing end.35 After initial proton abstraction, electron pair rearrangement leads to consecutive C-fragmentation so that no internal fragments are

produced.39 This behavior is very useful for structural elucidation, as the sequence of the oligosaccharide can, therefore, be read in fragmentation spectra from the reducing end. Besides, cross-ring cleavages, which are very important for structural characterization of oligosaccharides as they provide information on linkage and branching, are much more prominent than in positive ion mode. 2,4 A- and 0,2A-cleavages (often accompanied by fragments originating from water losses) are promoted by the electronrich center evolving from proton loss to form the [M - H]ion.45 Both fragments are specific for 1,4-linked hexose residues. As displayed in Figure 3D, fragment ions of 3-AQ-OS (precursor: [M + Cl]- of 3-AQ-maltopentaose) have the same mass as those of underivatized oligosaccharides (see Figure 3C). With the exception of one fragment (Z1 + 3-AQ), the derivatization does not influence the fragmentation pattern as exclusively fragments containing the nonreducing end are produced in negative ion mode. By elimination of HCl, the [M - H]- ion of 3-AQ-maltopentaose is generated. (Deprotonated species of 3-AQ-OS are observed in several spectra of standard oligosaccharides, data not shown.) This means that fragmentation rules reported for anionic adducts of underivatized oligosaccharides still apply and can readily be transferred to 3-AQOS. The only observed difference is that cross-ring cleavages are now more abundant, which can be highly important for structural elucidation. MS2 spectra containing all relevant fragments could be acquired from 5 pmol of sample with 3-AQ, which makes a 10-fold increase in sensitivity compared to harmine. To sum up the results of fragmentation analysis, Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3723

3724

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

Figure 4. TFLNH and TFpLNH: MS2 spectra of 3-AQ-HMOS in positive and negative ion mode. On the left, MALDI-TOF/TOF MS2 spectra in positive ion mode are displayed for 3-AQ-TFLNH (A) and 3-AQ-TFpLNH (B) ([M + H]+ as precursor). On the right, MALDI-TOF/TOF MS2 spectra in negative ion mode are shown for 3-AQ-TFLNH (C) and 3-AQ-TFpLNH (D) ([M + Cl]- as precursor). Fragments only appearing in the spectrum of one structure or with a distinct difference in intensity in the spectra of both structures are marked with black and white stars, respectively.

positive ion mode offers information on fragments which contain the reducing end and yield the oligosaccharide’s sequence, whereas negative ion mode produces fragments which include the nonreducing end and are important for linkage and branching analysis. If positive and negative ion modes are combined for structural analysis of oligosaccharides, the ability of both modes to complement each other can be exploited, and all required information can be gathered. Derivatization with 3-AQ is particularly advantageous in this context as both fragmentations can be executed from one single spot. As additional examples, the MS2 spectra of one small linear (gentiobiose) and one branched (3R,6R-mannopentaose) 3-AQ-OS in positive and negative ion mode and their evaluation by the GlycoWorkbench software are displayed in the Supporting Information section. Application to Human Milk Oligosaccharides. Two human milk oligosaccharides (HMOS), trifucosyllacto-N-hexaose (TFLNH) and trifucosyl-para-lacto-N-hexaose (TFpLNH), were selected to verify the ability of our method to differentiate between isomers. Both oligosaccharides consist of one lactose at the reducing end and two N-acetyl-lactosamine units altogether carrying three fucose (Fuc) residues. TFLNH has a branched core as both N-acetyl-lactosamine units are linked directly to the lactose. Figure 4 displays the oligosaccharides’ structures and the PSD fragmentation patterns of their 3-AQ-derivatives in both positive and negative ion mode. Branchings are indicated with R/β (first branching) or ′/′′ (second branching) according to the nomenclature by Domon and Costello.46 Some fragments appear in both spectra, although they may be labeled differently according to the different structures of the isomers. Other fragments allow a differentiation of both species and are marked with black and white stars in the respective spectra. Figure 4A,B clearly shows that B- and Y-ions are again most prominent in positive ion mode. For underivatized oligosaccharides, positive ion mode MS2 spectra are usually dominated by fragments arising from fucose losses, making it difficult to locate these residues in the context of structural characterization. It is striking that these fragments are much less prevalent for 3-AQ-HMOS; most prominent fragments still contain all originally attached fucoses. In the MALDI-TOF/TOF spectrum of 3-AQ-TFLNH (see Figure 4A), Y2R and Y2β stem from cleavage of the two branches, and the respective complementary ions are also present (B3R and B2β). Together with other Y-ions and YY-ions (the latter emerging from cleavage of two glycosidic bonds in branched structures), all necessary information for sequence analysis of TFLNH is provided. The sequence of TFpLNH can be derived from a complete series of Y-ions (Y1, Y2, Y3R, Y4R, Y5R’, Y6R’; see Figure 4B) as those are consecutively formed starting from the nonreducing end. Therefore, derivatization with 3-AQ highly increases the amount of structural information gained by positive ion mode MS2 spectra of HMOS. Moreover, differences in fragmentation of the two isomers are evident, out of which some examples shall now be explained. Figure 4A demonstrates that TFLNH produces fragments which do not appear in the MS2 spectrum of TFpLNH. Y2β (m/z ) 1126) arises from the elimination of the 6-antenna, confirming the branched structure of TFLNH. TFpLNH cannot produce such a fragment because the former 6-antenna is

incorporated between the former 3-antenna and the reducingend lactose. The complementary B-ion (B2β, m/z ) 512) is also present in the spectrum, revealing the mass of the 6-antenna. However, a fragment with the same mass is also observed for TFpLNH, as it can be formed by several BY-cleavages (B3RY6R’/ B3RY5R”/B5Y4R). Figure 4B illustrates that there are more characteristic fragments for TFpLNH. B5 (m/z ) 1169) can only be produced by the linear structure, since with TFLNH, the cleavage of the analogous bond would result in the already discussed B2β-fragment. Y3RY3β (m/z ) 672) emerges from Y3R (m/z ) 818) due to fucose loss. These two fragments are neither observed for TFLNH because their formation would require additional cleavages. Finally, Y2 (m/z ) 469) also gives a prominent peak in the spectrum of TFpLNH, while the corresponding fragment of TFLNH (Y2RY2β) is much less intense as two cleavages instead of one would be needed. In negative ion mode, fragmentation patterns are completely different from those in positive ion mode as now C-cleavages are preferred. The negative ion mode MS2 spectra of 3-AQ-TFLNH and 3-AQ-TFpLNH are displayed in Figure 4C,D, respectively. Obviously, the rules for negative ion mode fragmentation of underivatized, i.e., reducing, milk oligosaccharides described in previous articles also apply to 3-AQ-HMOS.34–38 Sequence information, although not as complete as in positive ion mode, can be gained by consecutive C-fragmentation starting from the reducing end. Additionally, linkages and branchings are revealed by specific fragments appearing in negative ion mode MS2 spectra. In both spectra, one single 2,4A-ion (m/z ) 1389) is observed due to the reducing-end glucose substituted at position 4. Besides, prominent peaks arise from CZ-double bond cleavages, indicating 1,3-linked N-acetyl-glucosamine (GlcNAc) residues.34 In the spectrum of 3-AQ-TFLNH (Figure 4C), B4Z2R (m/z ) 654) emerges from C4Z2R (m/z ) 672) by dehydration. Both of these ions are diagnostic for branched core structures as they contain solely the 6-antenna and the branching galactose (Gal) residue. The mass difference to the B4-ion (m/z ) 1329) reveals the mass of the 3-antenna. These ions consisting of 6-antenna and branching monosaccharide have been frequently observed in negative ion mode MS2 spectra of N-glycans and have been referred to as “D-ions”.44 Branch losses are caused by β-elimination, yielding a big gap in MS2 spectra from which the branching point is easily derived. Furthermore, C2βZ3β” is characteristic for a 3-substituted GlcNAc residue with an unsubstituted Gal residue in position 4 (m/z ) 203 + 162 - 1 ) 364).34 1,3-linkages at GlcNAc residues often undergo β-elimination in negative ion mode34 so that retention of the Gal residue during CZ-fragmentation means that Gal has to be 1,4-linked to GlcNAc, while the cleaved Fuc was initially 1,3-linked. This indicates that a differentiation of lacto-series (Gal 1,3-linked to GlcNAc) and lacto-neo-series (Gal 1,4-linked to GlcNAc) is possible. C1β’ (m/z ) 179) represents the unsubstituted Gal residue. All mentioned ions are specific for the branched core structure of TFLNH and are solely observed for this isomer. Figure 4D shows the negative ion mode MS2 spectrum of 3-AQ-TFpLNH. C3R, C4R, and C5Z3β (m/z ) 674, 836, and 1021), the latter resulting from β-elimination of the 1,3-linked Fuc, can only be formed by the linear core structure of TFpLNH. Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3725

Altogether, it is obvious that the structures can easily be distinguished by MALDI-TOF/TOF fragmentation of their 3-AQ derivatives. Positive and negative ion mode MS2 spectra contain fragments revealing sequence, linkage, and branching of both species. CONCLUSIONS In this study, we achieved rapid, complete, and reproducible on-target derivatization of all tested neutral oligosaccharides with 3-AQ by optimization of reaction conditions. With this preparation, MS and MS2 spectra of the resulting 3-AQ derivatives can be acquired in positive and negative ion mode from one single spot. 3-AQ is applicable to various MS instruments, as it works both for N2 lasers with 337 nm and Nd:YAG lasers with 355 nm wavelength. To our knowledge, detection limits as low as 1 fmol have not been reported before in oligosaccharide analysis. PSD fragmentation of protonated and chlorinated 3-AQ-OS yields informative spectra determining the oligosaccharide sequence, linkage, and branching. Moreover, the applicability to complex multiply fucosylated neutral human milk oligosaccharides has been examined successfully. 3-AQ derivatization of N-glycans released from glycoproteins and of negatively charged oligosaccharides containing sialic acids or sulfate groups is currently under investigation and will be the subject of future publications. Advantages of Schiff base formation with 3-AQ over other derivatization methods are evident. The on-target reaction with 3-AQ offers low sample handling requirements and no need for purification. 3-AQ is available in high quality as a MALDI matrix substance, which makes it a good choice for a derivatizing agent. The use of toxic reagents for reduction, e.g., sodium cyanoborohydride (NaBH3(CN)), is avoided. Other nonreductive aminations, e.g., oxime or hydrazone formation, have been developed in order to sidestep the inconveniences associated with reductive amination, but no method offers as many possibilities as our approach. In 2006, another on-target derivatization method with a DHB/aniline mixture was reported by Snovida et al.25 involving conversion of glycans to Schiff bases with aniline. However, complete derivatization was difficult, and measure-

3726

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

ments were restricted to positive ion mode due to the use of DHB. Fragmentation of [M + Na]+ ions of aniline derivatives showed no advantage over underivatized oligosaccharides. Derivatization with 3-AQ as described in our approach is applicable to the analysis of all forms of neutral reducing-end oligosaccharides. If nonreducing structures shall be examined, 3-AQ is still a good choice as a matrix, as sensitivity is also high without derivatization.30 Using NaCl, nonreducing oligosaccharides can be measured as Na+-adducts in positive ion mode and Cl--adducts in negative ion mode. By omission of acidic additives to the matrix solution, an almost complete suppression of the Schiff base formation can also be achieved for oligosaccharides containing a reducing end. The various possibilities offered by this new derivatization approach should be of great benefit for oligosaccharide analysis, as the advantages of many different preparations are therein combined. Finally, we are confident that the high information content received by fragmentation of the Schiff bases should facilitate structural characterization of biological samples. ACKNOWLEDGMENT We thank the Cluster of Excellence Macromolecular Complexes for their financial support and our colleagues for helpful and inspiring discussions. SUPPORTING INFORMATION AVAILABLE (1) Nomenclature for carbohydrate fragmentation according to Domon and Costello. (2) Spectra of dilution series of maltoheptaose measured in negative ion mode with 3-AQ + HNO3. (3) MS spectra of all standard derivatized oligosaccharides in positive and negative ion mode. (4) Annotation reports and statistics for the MS2 spectra of two standard derivatized oligosaccharides in positive and negative ion mode created with GlycoWorkbench Software v1.1. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review January 14, 2010. Accepted March 9, 2010. AC1001096