Anionic Adducts of Oligosaccharides by Matrix-Assisted Laser

signals (with characteristic isotope patterns) allowing subpicomole detection in the best cases. The matrixes that can aid the formation of chloride a...
21 downloads 9 Views 147KB Size
Download PDF
Anal. Chem. 2003, 75, 1638-1644

Anionic Adducts of Oligosaccharides by Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry Yang Cai, Yanjie Jiang, and Richard B. Cole*

Department of Chemistry, University of New Orleans, 2000 Lakeshore Drive, New Orleans, Louisiana 70148

The formation and decomposition (postsource decay, PSD) of anionic adducts in negative ion matrix-assisted laser desorption/ionization (MALDI) time-of-flight mass spectrometry have been studied. Chloride, a small inorganic anion, has been found to form stable anionic adducts with a variety of neutral oligosaccharides that can survive the MALDI process to give readily identifiable signals (with characteristic isotope patterns) allowing subpicomole detection in the best cases. The matrixes that can aid the formation of chloride adducts of oligosaccharides have gas-phase acidities lower than or close to that of HCl (1373 kJ/mol). In PSD experiments, precursor chloride adducts of oligosaccharides yield fragment ions that retain the charge on the sugar molecule rather than solely forming Cl-, and these fragments can provide structurally informative product ions. In negative ion MALDI, highly acidic oligosaccharides do not form adducts with chloride anions, but mildly acidic saccharides (e.g., containing a carboxylic acid group) form both deprotonated molecules and chloride adducts, and each may provide structural information concerning the oligosaccharide upon decomposition. Matrix-assisted laser desorption/ionization time-of-flight (MALDITOF) mass spectrometry (MS) was introduced by Karas and Hillenkamp1,2 and by Tanaka,3 and it very quickly became a powerful tool for the analysis of biopolymers such as peptides,1 proteins,4 oligonucleotides,5 and carbohydrates.1 For the analyses of neutral oligosaccharides, most often MALDI experiments are performed in the positive ion mode, and alkali metal ions (Li+, Na+, K+) are especially employed as attaching cations to form detectable cationic adducts with the sugars.6-8 Still, a potential * To whom correspondence should be addressed. E-mail: [email protected]. (1) Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Processes 1987, 78, 53. (2) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299. (3) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T. Rapid Commun. Mass Spectrom. 1988, 2, 151-153. (4) Hillenkamp, F.; Karas, M. Methods Enzymol. 1990, 193, 280. (5) Tang, K.; Allman, S. L.; Chen, C. H. Rapid Commun. Mass Spectrom. 1992, 6, 365. (6) Hofmeister, G. E.; Zhou, Z.; Leary, J. A. J. Am. Chem. Soc. 1991, 113, 59645970. (7) Ngoka, L. C.; Gal, J.-F.; Lebrilla, C. B. Anal. Chem. 1994, 66, 692-698. (8) Cancilla, M. T.; Penn, S. G.; Carroll, J. A.; Lebrilla, C. B. J. Am. Chem. Soc. 1996, 118, 6736-6745.

1638 Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

disadvantage that has been reported for positive ion analysis is that certain alkali metal ion adducts lack stability; hence, they are quite susceptible to decomposition, including facile loss of terminal fucose units9 and sialic acid moieties.10 Moreover, highly acidic forms of carbohydrates, which tend to exist as anions, can be profoundly less sensitive to positive mode analyses. Although far fewer papers have appeared that report negative ion mass spectrometric analyses of oligosaccharides, the negative mode approach is quickly becoming recognized as an important tool for the analysis of carbohydrates. For example, positional isomers of deprotonated disaccharides, [M - H]-, have been shown to offer distinctive decomposition spectra in cases where no isomeric differentiation was possible in the positive ion mode.11,12 However, neutral oligosaccharides that lack acidic sites may not readily form deprotonated molecules; thus, the ability to observe [M - H]- signals is substantially reduced. An alternative means of charging a neutral molecule to render it detectable in negative ion MS is via the attachment of a small inorganic anion, A-, to create an anionic adduct of the form [M + A]-. In negative ion electrospray mass spectrometry,13,14 anion attachment has been observed, even though it is not as commonly employed as positive mode cation attachment. In MALDI-MS, however, anion attachment is rarely reported, possibly because the MALDI process imparts more than enough energy to dissociate a very weakly bound anionic adduct.15 Nevertheless, Breuker et al.16 were able to generate halide ion attachment under special conditions for MALDI using tetrabutylammonium salt-silicon binary matrixes. Wong et al.12,17 observed hydrogen sulfate and alkylsulfonate adducts of oligosaccharides using more conventional matrixes such as 2,5-dihydroxyacetophenone (DHAP) and 2,4,6-trihydroxyacetophenone (THAP). Upon collision-induced dissociation (CID), however, hydrogen sulfate or alkylsulfonate adducts of oligosac(9) Penn, S. G.; Cancilla, M. T.; Lebrilla, C. B. Anal. Chem. 1996, 68, 23312339. (10) Powell, A. K.; Harvey, D. J. Rapid Commun. Mass Spectrom 1996, 10, 10271032. (11) Mulroney, B.; Traeger, J. C.; Stone, B. A. J. Mass Spectrom. 1995, 30, 12771283. (12) Wong, A. W.; Cancilla, M. T.; Voss, L. R.; Lebrilla, C. B. Anal. Chem 1999, 71, 205-211. (13) Dole, M.; Mack, L. L.; Hines, R. L.; Mobley, R. C.; Ferguson, L. D.; Alice, M. B. J. Chem. Phys. 1968, 49, 2240-2249. (14) Yamashita, M.; Fenn, J. J. Phys. Chem. 1984, 88, 4671-4675. (15) Karas, M.; Glu ¨ ckmann, M.; Scha¨fer, J. J. Mass. Spectrom. 2000, 35, 1-12. (16) Breuker, K.; Knochemnuss, R.; Zenobi, R. Int. J. Mass Spectrom. Ion Processes 1998, 176, 149-159. (17) Wong, A. W.; Wang, H.; Lebrilla, C. B. Anal. Chem. 2000, 72, 1419-1425. 10.1021/ac0205513 CCC: $25.00

© 2003 American Chemical Society Published on Web 03/01/2003

Table 1. Matrixes for Formation of Chloride Adducts of Sucrose in MALDI-TOF MS (Linear Mode)

*The gas-phase basicities of deprotonated matrixes; see ref 36. **See text. ***Adduct intensities were evaluated relative to the base peak in the spectrum (100%, usually [matrix - H]-); ++ indicates signal intensity is 5-10%, +++ indicates 10-20%, +++++ indicates >40%.

charides were reported to yield hydrogen sulfate or alkylsulfonate ion only; i.e., the anion detaches from the molecule and holds on to the charge. The ability to gain structural information from consecutive decompositions of the analyte molecule is thereby lost. On the other hand, the in situ sulfate-derivatized ion [M + HSO4 - H2O]- was found to be able to provide structural information regarding the analyte upon CID.12 Recent progress in the study of anion attachment in negative ion ES-MS has shown that some anionic adducts can be very stable in the gas phase.18,19 The similarity in gas-phase basicities of the deprotonated analyte molecule [M - H]- and the anion moiety was found to play an important role in stabilizing anionic adducts of the form [M - H]-‚‚‚H+‚‚‚[A]-.18 For small inorganic anion adducts of saccharides, multiple hydrogen bonding was found to be the predominate form of interaction between the analyte molecule and the attaching anion, and such multiple hydrogen bonding can further stabilize this type of anionic adduct.19 Although these gas-phase factors that contribute to the stability of anionic adducts were deduced in ES-MS experiments, the lessons that were learned may also apply to MALDI experiments, which also involve ion desorption from a condensed phase into the gas phase. In this report, we have focused on establishing conditions that favor the formation and the postsource decay (PSD)20,21 of Cladducts of oligosaccharides in MALDI. Previous ES-MS studies18,19,22-24 have shown that chloride anion not only has a high affinity for saccharides, but it is also a good proton scavenger that (18) Cai, Y.; Cole, R. B. Anal. Chem. 2002, 74, 985-991. (19) Cai, Y.; Concha, M. C.; Murray, J. S.; Cole, R. B. J. Am. Soc. Mass Spectrom. 2002, 13, 1360-1369. (20) Spengler, B.; Kirsch, D.; Kaufmann, R. Rapid Commun. Mass Spectrom. 1991, 5, 198. (21) Spengler, B.; Kirsch, D.; Kaufmann, R. Rapid Commun. Mass Spectrom. 1992, 5, 105. (22) Cole, R. B.; Zhu, J. Rapid Commun. Mass Spectrom. 1999, 13, 607-611. (23) Zhu, J.; Cole, R. B. J. Am. Soc. Mass Spectrom 2000, 11, 932-941. (24) Zhu, J.; Cole, R. B. J. Am. Soc. Mass. Spectrom. 2001, 12, 1193-1204.

can effectively abstract a proton from saccharides upon CID, departing as HCl and yielding [M - H]-. Subsequent consecutive fragmentations of [M - H]- can provide structural information concerning oligosaccharide molecules. EXPERIMENTAL SECTION All neutral oligosaccharides were purchased from Sigma (St. Louis, MO) and prepared as 1 mM aqueous solutions. DGlucuronic acid was purchased from Aldrich (Milwaukee, WI), and 3-sialyllactose was purchased from Glyko (Novato, CA); these two compounds were also prepared as 1 mM aqueous solutions. Ammonium chloride was obtained from Aldrich and prepared at 1 mM in a 9:1 solution of methanol/water. The matrixes listed in Table 1 were purchased from Aldrich and were prepared at 50 mg/mL (or saturated) in ethanol. All chemicals were used without further purification. All MALDI experiments were performed on a Voyager-Elite MALDI rTOF mass spectrometer (Applied Biosystems, Framingham, MA) equipped with a pulsed N2 laser (λ ) 337 nm). The accelerating voltage was set at 20 kV. The sample solutions were deposited into the wells of a gold-coated 100-well sample plate. Each acquired mass spectrum represents an average of 100-150 laser shots. RESULTS AND DISCUSSION Small saccharides such as R-D-glucose and sucrose have been found to form very stable anionic adducts with chloride and hydrogen sulfate in negative ion ES-MS.18,19 Chloride and hydrogen sulfate adducts of sucrose even survived under conditions that favored the cleavage of the glycosidic bond in sucrose.19 Because hydrogen sulfate adducts of oligosaccharides have been observed in MALDI experiments,12 we decided to investigate the feasibility of generating observable chloride adducts of oligosacAnalytical Chemistry, Vol. 75, No. 7, April 1, 2003

1639

charides via the MALDI approach. We began our study by attempting to generate and optimize conditions enabling observation of chloride adducts of sucrose. Sample Preparation. Sample preparation can be the key to success or failure in MALDI experiments. Various ionic chloride salts can serve as sources of chloride anions to form adducts with oligosaccharides during the ionization process. However, if the countercations in the ionic chloride salts are metal ions, which are well known for their strong binding tendency toward oligosaccharides,25-30 these cations will compete to attach with the oligosaccharide and thus lower the propensity to observe negative ion adducts. For this reason, metal ion chloride salts were avoided, and instead, ammonium chloride was employed as the source of attaching chloride anions. Two approaches were employed in the preparation of MALDI targets in view of optimizing the yields of chloride adducts of sucrose. The first sample preparation approach is the “drieddroplet” method:2,31 1 µL of 1 mM sucrose/water solution was mixed with 2 µL of 1 mM NH4Cl in water:methanol ) 9:1, and 1 µL of this mixture solution was further added in to 20 µL of matrix solution to form the final sample solution. A 0.5-1-µL aliquot of this final solution was deposited into the well of a sample plate and allowed to dry. The second sample preparation approach is the “thin-layer” method:32,33 1 mM sucrose/water solution and 1 mM NH4Cl in water:methanol ) 1:9 were mixed in 1:2 ratio (volume). A 1-µL aliquot of matrix solution was first deposited into the well on the sample plate and dried to form a matrix layer. A 1-µL aliquot of the sucrose and chloride mixture was then deposited on top of the matrix layer and allowed to dry. The thinlayer method was found to yield higher intensity signals for chloride adducts of sucrose than the dried-droplet method (the latter yielding almost no adduct signal). This finding is in accordance with a recent study in positive ion MALDI-TOF MS, which revealed that carbohydrates that were not incorporated into matrix crystal could be ionized in higher efficiency.34 Subsequently, the thin-layer method was applied in the generation of the chloride adducts of all analytes reported in this work. Matrix Selection. Using sucrose as a model compound, ammonium chloride as the source of attaching anion, and the thinlayer sample preparation method, we compared the signals obtained for the sucrose chloride adduct (m/z 377, 379) enabled by the following matrixes: 3-aminoquinoline, 4-aminophenol, 4-nitroaniline, 7-methoxy-9H-pyrido[3,4-b]indole (harmine),35 and 2,5-dihydroxybenzoic acid (2,5-DHB). MALDI mass spectra were acquired in the linear mode using delayed extraction with a very (25) Hofmeister, G. E.; Zhou, Z.; Leary, J. A. J. Am. Chem. Soc. 1991, 113, 59645970. (26) Ngoka, L. C.; Gal, J.-F.; Lebrilla, C. B. Anal. Chem. 1994, 66, 692-698. (27) Striegel, A. M.; Piotrowiak, P.; Boue´, S. M.; B., C. R. J. Am. Soc. Mass Spectrom. 1999, 10, 254-260. (28) Zhou, Z.; Ogden, S.; Leary, J. A. J. Org. Chem. 1990, 55, 5444-5446. (29) Fura, A.; Leary, J. A. Anal. Chem. 1993, 65, 2805-2808. (30) Kohler, M.; Leary, J. A. Int. J. Mass Spectrom. Ion Processes 1997, 162, 17-34. (31) Xiang, F.; Beavis, R. C. Org. Mass Spectrom. 1993, 28, 1424. (32) Xiang, F.; Beavis, R. C. Rapid Commun. Mass Spectrom. 1994, 8, 199. (33) Vorm, O.; Roepstorff, P.; Mann, M. Anal. Chem. 1994, 66, 3281. (34) Gluckmann, M.; Pfenninger, A.; Kruger, R.; Thierolf, M.; Karas, M.; Horneffer, V.; Hillenkamp, F.; Strupat, K. Int. J. Mass Spectrom. 2001, 210/ 211, 121-132. (35) Nonami, H.; Fukui, S.; Erra-Balsells, R. J. Mass Spectrom. 1997, 32, 287296.

1640 Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

Figure 1. MALDI-TOF mass spectra of chloride adducts of neutral oligosaccharides obtained using the thin-layer sample preparation method with harmine as matrix. (a) Linear mode spectrum of sucrose + ammonium chloride. (b) Reflectron mode spectrum of stachyose + ammonium chloride showing isotopic distributions. (c) Reflectron mode spectrum of γ-cyclodextrin + ammonium chloride with isotopic distributions.

short delay time (1 ns). Nonami et al.35 had found that β-carbonline alkaloids were good matrixes for MALDI experiments. Recently Wong et al.12 employed harmane, one of the β-carbonline alkaloids, in their efforts to generate hydrogen sulfate adducts of oligosaccharides. Table 1 ranks the results of the five tested matrixes in yielding the chloride adduct of sucrose. Harmine, a β-carbonline alkaloid, gave the best results (Figure 1a). The peaks corresponding to chloride adducts of sucrose at m/z 377 and 379 are very strong, and the [sucrose - H]- (m/z 341) signal is barely detectable. The chloride anions at m/z 35 and 37, [matrix - H]at m/z 211, and [matrix + Cl]- at m/z 247 and 249 are the other major peaks. The adduct peaks appear in higher signal-to-noise ratios with fewer extraneous peaks compared to the mass spectra generated by the other matrixes listed in Table 1. Compared to hydrogen sulfate, which only forms adducts with oligosaccharides larger than trisaccharides,12 chloride appears to have some advantages when serving as the attaching anion to smaller molecules in negative ion MALDI. The gas-phase basicity of harmine is not available in the literature, but a “bracketing method”18 was employed to determine its gas-phase basicity. The 4-chlorophenoxide adduct of harmine at m/z 339 and the chloride adduct of harmine at m/z 247 were generated by electrospray-MS. Upon CID, [harmine + 4-chlorophenoxide]- yields [harmine - H]- at m/z 211 as the sole product (Figure 2a). This indicates that [harmine - H]- has a lower gasphase basicity than 4-chlorophenoxide (1407 ( 8.436,37 or 1409 ( 8.4 kJ/mol36,38); [harmine + Cl]-, on the other hand, yields Clat m/z 35 as the sole product upon CID (Figure 2b), indicating

anion- + H+ f conjugate acid of anion

Figure 2. Electrospray MS/MS spectra of (a) 4-chlorophenoxide adduct of harmine at m/z 339. The product ion at m/z 211 corresponds to [matrix - H]- in MALDI experiments; (b) chloride adduct of harmine at m/z 247. The product ion at m/z 35 is Cl-. The gas-phase basicity of [harmine - H]- is between that of 4-chlorophenoxide (1407 kJ/ mol) and chloride (1373 kJ/mol).

that [harmine - H]- has a higher gas-phase basicity than that of chloride (1372.8 ( 0.4236,39 or 1374 ( 8.4 kJ/mol36,37). The gasphase basicity of [harmine - H]- is thus determined to be between 1373 and 1407 kJ/mol. In addition to having high molar absorptivities and good vacuum stabilities, the three matrixes that were most effective at generating high signals for the sucrose chloride adduct, 3-aminoquinoline, 4-aminophenol, and harmine, have one more property in common; that is, they all have gasphase basicities for [matrix - H]- higher than that of chloride (see Table 1). Even though the exact details of the ionization mechanism in MALDI are still under debate, it is clear that the matrix plays an essential role in facilitating ionization. Previous studies have proposed that proton transfer is a major phenomenon occurring in both positive40,41 and negative42 ion MALDI. In considering the possibilities to form a stable anionic adduct in MALDI, the ease of proton transfer between neutral matrix molecules and anionic species [A]- (small inorganic anion) must be taken account.

matrix + A- f [matrix - H]- + HA

(1)

Upon laser desorption, both neutral matrix molecules and anionic species will be transferred into the MALDI plume where a complex series of ion-molecule reactions takes place. The gasphase basicity of the anion ([matrix - H]- or A-), defined as the negative of the free energy change (-∆Ganion-) of the reaction (36) Bartmess, J. E. Negative Ion Energies Data. In NIST Chemistry WebBook; NIST Standard Reference Database 69; Mallard, W. G., Linstrom, P. J., Eds.; National Institute of Standards and Technology: Gaithersburg, MD, 20899, February 2000; http://webbook.nist.gov. (37) Fujio, M.; McIver, R. T., Jr.; Taft, R. W. J. Am. Chem. Soc. 1981, 103, 4017. (38) Viggiano, A. A.; Henchman, M. J.; Dale, F.; Deakyne, C. A.; Paulson, J. F. J. Am. Chem. Soc. 1992, 114, 4299. (39) Martin, J. D. D.; Hepburn, J. W. J. Chem. Phys. 1998, 109, 8139-8142. (40) Ehring, H.; Karas, M.; Hillenkamp, F. Org. Mass Spectrom. 1992, 27, 472. (41) Gimon-Kinsel, M.; Preston-Schaffter, L. M.; Kinsel, G. R.; Russell, D. H. J. Am. Chem. Soc. 1997, 119, 2534-2540. (42) Knochemuss, R.; Karbach, V.; Wiesli, U.; Breuker, K.; Zenobi, R. Rapid Commun. Mass Spectrom. 1998, 12, 529-534.

(2)

will determine the availability of the respective anions. If [matrix - H]- has a lower gas-phase basicity than that of A-, then reaction 1 is favored, because A- can easily abstract a proton from the neutral matrix molecule, leaving [matrix - H]- as the major observed ion. To suppress the formation of [matrix -H]- and to increase the yield of the anionic adducts of interest in MALDI, we conclude that the deprotonated form of the matrix of choice should have a higher gas-phase basicity than that of the anion used to form the adduct with the neutral analyte. In other words, the neutral matrix should be a weaker acid in the gas phase than the protonated attaching anion. Hydrogen sulfate12 and deprotonated alkylsulfonate17 adducts of oligosaccharides have been generated in MALDI with the aid of commonly used matrixes such as DHAP and THAP. The gasphase basicity of hydrogen sulfate is 1265 ( 10 kJ/mol,36,38 which is lower than that of THAP (1324 kJ/mol43). This means that hydrogen sulfate and hydrogen sulfate adducts of oligosaccharides can exist in the plumes of matrixes having lower gas-phase acidities than sulfuric acid. Chloride, on the other hand, has a higher gas-phase basicity (1372.8 ( 0.4236,39 or 1374 ( 8.4 kJ/ mol36,37) than that of hydrogen sulfate (i.e., HCl is a weaker acid in the gas phase than sulfuric acid), and this places a tighter restriction on the ability to observe chloride adducts in MALDI. In a special case where chloride adducts were generated using a binary liquid matrix composed of tetrabutylammonium chloride and silicon particulates,16 silicon particulates served to absorb laser energy. Because silicon particles contain no protons, BrønstedLowry acid-base considerations were less relevant. Chloride Adducts of Neutral Oligosaccharides. Using the thin-layer method in sample preparation, ammonium chloride as the source of attaching anion, and harmine as matrix and employing the same experimental conditions mentioned above, we examined the formation and stability of chloride adducts of various neutral oligosaccharides in MALDI-TOF MS. From small monosaccharides such as D-xylose and R-D-glucose to larger oligosaccharides such as γ-cyclodextrin (eight sugar units), all of the neutral saccharides listed in Table 2 form anionic adducts with chloride. Figure 1b shows the MALDI reflectron spectrum of stachyose while Figure 1c gives the MALDI reflectron spectrum of γ-cyclodextrin. Previous studies12,17 have indicated that laser intensity is not a highly influential factor in the formation of alkylsulfonate adducts of oligosaccharides, as these anionic adducts could survive laser fluences 100% higher than threshold. In these latter studies, anionic adducts were detected in a Fourier transform ion cyclotron instrument that requires significantly longer analysis times relative to TOF detectors. Our own experiments indicate that the intensities of peaks corresponding to anionic adducts do not change significantly when the extraction delay is prolonged. Under optimized conditions, 1.0-pmol sucrose preparations were detected in the reflectron mode at S/N levels of 14-20, indicating a limit of detection of ∼0.2 pmol. Postsource Decay of Chloride Adducts of Neutral Oligosaccharides. Small neutral oligosaccharides often have gasphase basicities of [M - H]- close to that of chloride.18,24 Upon (43) Breuker, K.; Knochenmuss, R.; Zenobi, R. Int. J. Mass Spectrom. 1999, 184, 25-38.

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

1641

Table 2. Chloride Adducts of Neutral Oligosaccharides Observed by MALDI (Matrix, Harmine; Source of Chloride, NH4Cl)a oligosaccharides

no. of monosaccharide units

av MW

chloride adduct m/z

signal intensityb

D-xylose R-D-glucose sucrose D-turanose raffinose stachyose R-cyclodextrin β-cyclodextrin γ-cyclodextrin

1 1 2 2 3 4 6 7 8

150.1 180.1 342.3 342.3 504.5 666.6 972.9 1135.0 1297.2

185.1/187.1 214.9/216.9 377.2/379.2 377.2/379.2 539.4 702.1 1008.4 1170.5 1332.7

+++++ +++++ +++++ +++++ +++++ +++++ +++ ++ +++

a The experiments were performed in linear mode negative ion MALDI at low resolution such that, above m/z 500, isotopes were not resolved and average masses were obtained. Samples were prepared using the thin-layer method. Extraction delay was set at 1 ns. All adduct signals are unambiguously identified with S/N > 100. bCompared to the intensity of the base peak in the spectrum (100%, usually [matrix - H]-), ++ indicates 5-10%, +++ indicates 10-20%, and +++++ indicates >40%.

Figure 3. MALDI-TOF mass spectra of D-turanose + ammonium chloride using the thin-layer sample preparation method and harmine as matrix. (a) Conventional linear mode spectrum, ions at m/z 377 and 379 correspond to chloride adducts; (b) PSD spectrum of [D-turanose + Cl]-, product ions m/z 161 and 179 result from glycosidic bond cleavage, and m/z 251 is the result of cross-ring cleavage of D-turanose. The chlorine atom has been lost with the neutral fragment in all three cases.

decomposition of the chloride adduct of a neutral oligosaccharide, chloride can abstract a proton from the analyte to form HCl, leaving product ions representing the analyte that are observable in negative ion MS. In cases where the deprotonated saccharides have somewhat higher gas-phase basicities than that of chloride anion, the yields of [M - H]- and its fragments can be increased by employing higher collision energies.22,23 D-Turanose (3-O-R-Dglucopyranosyl-D-fructose) is a neutral dissaccharide. [D-turanose + Cl]- at m/z 377 and 379 can be generated in MALDI-TOF MS (Figure 3a) under conditions similar to those that favor formation of the sucrose chloride adduct as detailed above. Figure 3b is the PSD spectrum of [D-turanose + Cl]-. Product ions ar m/z 161 and 179 result from the glycosidic bond cleavage of D-turanose, 1642 Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

while the product ion at m/z 251 is the result of cross-ring cleavage (the glycosidic bond in D-turanose is preserved while one of the monosaccharide rings is broken). Each of the three product ions requires the loss of the chlorine atom, most likely in the form of HCl. These fragment ions likely represent consecutive decompositions of [M - H]- following HCl loss. Figure 3b thus clearly illustrates one of the most important advantages of the chloride attachment approach in the mass spectrometric analysis of oligosaccharides, i.e., that [M + Cl]-, like [M - H]-, can provide structural information concerning the analyte upon decomposition. This is in sharp contrast to the hydrogen sulfate and alkylsulfonate adducts of oligosaccharides, which only yield the attaching anion as product ions upon decomposition;12,17 hence, no structural information regarding the analyte oligosaccharide can be obtained from them in PSD or tandem mass spectrometry experiments. Formation of Chloride Adducts with Oligosaccharides of Different Acidities. The influence of oligosaccharide acidity and the competition for available chloride anions in MALDI have also been investigated. Three oligosaccharides, D-glucuronic acid, sucrose, and 3-sialyllactose, were chosen as model compounds. An aqueous mixture of D-glucuronic acid:sucrose:3-sialyllactose: NH4Cl:HCl ) 1:1:1:2.5:2.5 was prepared, and then a 1-µL aliquot of the mixture was deposited on a dried layer of harmine that had been predeposited on a MALDI sample plate. The negative ion reflectron MALDI spectrum of the mixture is shown in Figure 4, which clearly indicates that, under the same condition, 3-sialyllactose (containing one sialic acid moiety) exclusively forms [M - H]- (m/z 632), neutral sucrose forms only [M + Cl]-(m/z 377, 379), but D-glucuronic acid (containing one carboxylic acid moiety) forms both [M - H]- (m/z 193) and [M + Cl]-(m/z 229, 231). This observation confirms the conclusion that the gas-phase basicity of the deprotonated oligosaccharide plays an important role in the formation of [M + Cl]- in MALDI. Chloride adduct ions appearing at m/z 298 and 300 contain one nitrogen; they are the result of chloride attachment to neutral mass 263, which is the remainder of the N-acetylneuraminic acid (sialic acid) moiety after loss of HCOOH. A previous study24 has shown that, upon CID, the chloride adduct of sucrose yields chloride and [M - H]- in approximately equal abundances under low-energy collision conditions. This implies that [sucrose - H]- has a gas-phase basicity very close to that of chloride. The gas-phase basicities of [3-sialyllactose H]- and [D-glucuronic - H]- are not available in the literature. We have evaluated the gas-phase basicities of [3-sialyllactose H]- and [D-glucuronic - H]- via MALDI-PSD and ES-MS/MS approaches. 3-Sialyllactose does not form [M + Cl]- in MALDI with harmine as matrix, but it does form [M + HSO4]- with THAP as matrix. However, the abundance of [3-sialyllactose + HSO4]formed in MALDI is not high enough for a PSD experiment, so we turned to ES-MS and ES-MS/MS for the evaluation of the [M - H]- gas-phase basicity of 3-sialyllactose. In electrospray under the same conditions that favor adduct formation in Figure 2, 3-sialyllactose does not form observable adducts with chloride, but it does form adducts with hydrogen sulfate. ES-MS/MS experiments reveal that, upon CID, the precursor ion [3-sialyllactose + HSO4]- gives both [M - H]- and [HSO4]- as product ions, and the two fragment ions have nearly the same abundances. The fact that the abundances of [M - H]- and [HSO4]- product

Figure 4. MALDI-TOF (reflectron mode) mass spectrum of D-glucuronic acid, sucrose, and 3-sialyllactose mixed with HCl and ammonium chloride, using the thin-layer method in sample preparation and harmine as the matrix. Ions at m/z 193 and 229 correspond to [M - H]- and [M + Cl]- of D-glucuronic acid; m/z 377 corresponds to [sucrose + Cl]-; m/z 632 corresponds to [3-sialyllactose - H]-. [Harmine - H]- appears at m/z 211, while [harmine + Cl]- appears at m/z 247 (saturated).

ions are similar provides evidence that [3-sialyllactose - H]- has a gas-phase basicity very close to that of hydrogen sulfate (1265 kJ/mol).36,38 As for D-glucuronic acid, it forms adducts with chloride, bromide, and hydrogen sulfate in MALDI. PSD spectra in Figure 5 show that [D-glucuronic acid + Cl]- yields [D-glucuronic acid - H]- as the major product ion (Figure 5a), [D-glucuronic acid + HSO4]- yields hydrogen sulfate as the major product ion (Figure 5c), while [D-glucuronic acid + Br]- yields both bromide and [D-glucuronic acid - H]- as product ions (Figure 5b). ES-MS/ MS experiments were also performed on the three adducts of D-glucuronic acid, and corroborative results were obtained. It can thereby be concluded that D-glucuronic acid has a gas-phase basicity not far from that of bromide (1332 kJ/mol).36,44 In addition to allowing approximate evaluation of this gas-phase basicity, Figure 5 reinforces the point that chloride adducts may lose HCl much more readily than hydrogen sulfate adducts lose H2SO4. Further energy uptake by the respective product ions (e.g., via CID on a TOF-TOF or a FTICR instrument) could thus yield structurally informative decomposition products for the chloride adducts but not for the hydrogen sulfate adducts. For the three compounds 3-sialyllactose, D-glucuronic acid and sucrose, the [M - H]- gas-phase basicities follow the trend (44) Blondel, C.; Cacciani, P.; Delsart, C.; Trainham, R. Phys. Rev. A 1989, 40, 3698.

Figure 5. PSD spectra of D-glucuronic acid adducts with (a) chloride, (b) bromide, and (c) hydrogen sulfate anions. Upon decomposition, [D-glucuronic acid + Cl]- at m/z 229 and 231 yields [D-glucuronic acid - H]- at m/z 193; [D-glucuronic acid + Br]- at m/z 273 and 275 yields both [D-glucuronic acid - H]- at m/z 193 and bromide anion at m/z 79 and 81; while [D-glucuronic acid + HSO4]at m/z 291 yields only hydrogen sulfate anion at m/z 97.

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

1643

GB[surcrose-H]- > GB[D-glucuronic acid-H]- > GB[3-sialyllactose-H]-. From the pattern of chloride adduct formation for these three compounds shown in the conventional MALDI mass spectrum (Figure 4), it can be rationalized that, under the same condition, relatively acidic compounds (low [M - H]- gas-phase basicity), such as 3-sialyllactose, will not form stable adducts with chloride. Instead, they will form [M - H]-. On the other hand, neutral compounds such as sucrose will form adducts with chloride in overwhelming preference to [M - H]-. Last, mildly acidic compounds, such as D-glucuronic acid, can yield both chloride adducts and [M - H]in the conventional MALDI mass spectrum. CONCLUSIONS The conditions that favor the formation of anionic adducts in MALDI have been systematically investigated and rationalized. The sample preparation procedure plays an important role in enabling the formation of anionic adducts in MALDI. For promoting chloride attachment to neutral oligosaccharides, the thin-layer method works better than the dried-droplet method in MALDITOF MS. The choice of matrix is also crucial in determining the success of forming and observing anionic adducts. The matrixes that can aid the formation of anionic adducts should have a gasphase basicity of [matrix - H]- that is higher than that of the attaching anion. Chloride anions have a gas-phase basicity that is low enough such that in a variety of matrixes they can form stable

1644

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

adducts with various neutral oligosaccharides. However, the gasphase basicity of Cl- is still high enough such that, upon PSD, chloride adducts of neutral oligosaccharides can yield fragment ions other than Cl-; from the decomposition pattern of these product ions, structural information of the oligosaccharide can be derived. Employing negative ion MALDI in the presence of chloride anions, highly acidic oligosaccharides may form predominantly [M - H]-, but weakly acidic and neutral oligosaccharides can form [M + Cl]- in varying degrees of preference to [M - H]-. For MALDI-MS analyses of mixtures containing both acidic and neutral carbohydrates, while positive ion analyses may fail to detect the acidic components at high sensitivities, the negative ion approach with chloride anion attachment provides a simple means to simultaneously analyze neutral and acidic carbohydrates without switching the instrument polarity. ACKNOWLEDGMENT Financial support for this research was provided by the National Science Foundation through Grant CHE-9981948 and by the Louisiana Board of Regents through Grant HEF(2001-06)-08.

Received for review September 3, 2002. Accepted January 16, 2003. AC0205513