Anal. Chem. 2003, 75, 2445-2455
Tandem Mass Spectrometry of Sulfated Heparin-Like Glycosaminoglycan Oligosaccharides Joseph Zaia* and Catherine E. Costello
Department of Biochemistry, Boston University School of Medicine, 715 Albany Street, R-806, Boston, Massachusetts 02118
The structural characterization of heparin-like glycosaminoglycans (HLGAGs) is a major challenge in glycobiology. These linear, sulfated oligosaccharides are expressed on animal cell surfaces, in extracellular matrixes, basement membranes, and mast cell granules and bind with varying degrees of specificity to families of proteases, growth factors, chemokines, and blood coagulation proteins. Cell surface HLGAGs bind growth factors and growth factor receptors and serve as coreceptors in these interactions. Understanding of the mechanism and regulation of growth factor-receptor binding requires efficient determination of cell surface HLGAG structures and the variations in their expression in response to the cellular environment. The solution to this problem entails rapid, sensitive structural analysis of these molecules. To date, HLGAG sequencing requires multistep processes that combine chemical and enzymatic degradation with gel-based or mass spectrometry-based detection systems. Although tandem mass spectrometry has revolutionized proteomics, the fragility of sulfate groups has limited its usefulness in the analysis of HLGAGs. This work demonstrates that tandem mass spectrometry can be effectively used to determine HLGAG structures while minimizing losses of SO3. First, collision-induced dissociation (CID) is shown to produce abundant backbone cleavage ions for HLGAG oligosaccharides, provided that most sulfate groups are deprotonated. Fragmentation of different precursor ion charge states produces complementary data on the structure of the HLGAG. Second, calcium ion complexation of HLGAGs stabilizes the sulfate groups, increases the relative abundances of backbone cleavage ions, and decreases the abundances of ions produced from SO3 losses. Although tandem mass spectrometry has revolutionized the sequencing of proteins and enabled the growth of proteomics, its use in the study of heparin-like glycosaminoglycans (HLGAG) structure has been quite limited. Like proteins, HLGAGs are linear biopolymers, the sequence elements of which arise through substitutions along the backbone units. Unlike proteins, these and other carbohydrates are biosynthesized without nucleic acid templates, greatly increasing the potential variety of expressed structures. The CID pathways for carbohydrates have been * Corresponding author. Phone: 617-638-6267. Fax: 617-638-6761. E-mail:
[email protected]. 10.1021/ac0263418 CCC: $25.00 Published on Web 04/17/2003
© 2003 American Chemical Society
described,1 and many classes of carbohydrates and glycoconjugates have been extensively studied using tandem mass spectrometry2. HLGAGs remain an exception, primarily because of the challenges posed by the high density of sulfate groups present on the repeating disaccharide backbone, the pattern of which is an important determinant of the activity of these biopolymers. Although the ionization of HLGAG oligosaccharides has been achieved using fast atom bombardment (FAB),3,4 electrospray (ESI),5,6 and matrix-assisted laser desorption/ionization (MALDI),7,8 losses of SO3 result in the destruction of information carried in the backbone sulfation pattern and have limited the usefulness of tandem mass spectrometry in this area. Effective use of tandem mass spectrometry for HLGAGs entails a greater understanding of the parameters affecting sulfate lability and elucidation of means by which mass spectrometric cleavage of glycosidic bonds can be favored to enable rapid, instrument-based sequencing of these molecules. Heparin-like glycosaminoglycans are assembled from repeating disaccharide units consisting of [HexA(1,3)GlcN(1,3)]n. In heparin, an HLGAG synthesized in mast cells, most of the HexA residues are C5 epimerized to IdoA and are sulfated at the 2 position. Most of the GlcN residues are sulfated at the N and 6 positions. The predominant heparin disaccharide can be written as [IdoA2S(1,3)GlcNS6S]. Other disaccharides exist in a typical heparin preparation but are expressed at a low level with respect to the predominant disaccharide structure.9 Thus, the pentameric sequence GlcNS6S-GlcA-GlcNS3S6S-IdoA2S-GlcNS6S that binds antithrombin III is expressed in heparin.10 Heparan sulfate (HS) is bound to proteoglycans expressed on the surfaces of all vertebrate cells and in basement membranes. HS consists of blocks of disaccharides containing primarily GlcAGlcNAc, having a low degree of sulfation, and blocks containing primarily IdoAGlcNS with a high degree of sulfation. The expression of (1) Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5, 397-409. (2) Reinhold: V. N.; Reinhold: B. B.; Costello, C. E. Anal. Chem. 1995, 67, 1772-1784. (3) Reinhold: V. N.; Carr, S. A.; Green, B. N.; Petitou, M.; Choay, J.; Sinay, P. Carbohydr. Res. 1987, 161, 305-313. (4) Dell, A.; Rogers, M.; Thomas-Oates, J.; Huckerby, T.; Sanderson, P.; Nieduszynski, I. Carbohydr. Res. 1988, 179, 7-19. (5) Takagaki, K.; Nakamura, T.; Izumi, J.; Saitoh, H.; Endo, M.; Kojima, K.; Kato, I.; Majima, M. Biochemistry 1994, 33, 6503-6507. (6) Chai, W.; Kogelberg, H.; Lawson, A. M. Anal. Biochem. 1996, 237, 88102. (7) Juhasz, P.; Biemann, K. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 4333-4337. (8) Juhasz, P.; Biemann, K. Carbohydr. Res. 1995, 270, 131-147. (9) Conrad, H. E. Heparin Binding Proteins; Academic Press: New York, 1998. (10) Lindahl, U.; Backstrom, G.; Hook, M.; Thunberg, L.; Fransson, L. A.; Linker, A. Proc. Natl. Acad. Sc. U.S.A. 1979, 76, 3198-3202.
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003 2445
N-acetylated and N-sulfated domains and variations in sulfation and epimerization within each domain make HS the most complex type of glycosaminoglycan structure. The fragmentation of sulfated carbohydrates using CID MS depends to a large degree on two factors: (1) the charge state on the ion and (2) the ions pairing the sulfate negative charge. For chondroitin sulfate (CS) oligosaccharides, consisting of repeating units of glucuronic acid and sulfated N-acetylgalactosamine residues, the observation of charge states wherein each sulfate group is deprotonated is typical in an electrospray mass spectrum.11,12 Sulfates are the most acidic groups in these carbohydrates and, to a first approximation, are assumed to carry the negative charges. CID fragmentation of CS oligosaccharides cleaves glycosidic bonds without losses of neutral SO3 or H2SO4 so long as all sulfate groups are deprotonated.12 For CS ions in which sulfate groups do not bear a net charge because of pairing with protons, losses of neutral SO3 from the precursor ion are more likely than is formation of product ions via backbone cleavage.13 For CS tetramers and hexamers with a charge state of 1-, we have observed intramolecular transfer of SO3 and believe this results from the relatively compact conformations of singly charged ions relative to those of multiply charged ions.13 The fact that ESI results in more highly charged ions than those observed for FAB ionization or MALDI has important implications for the analysis of highly sulfated carbohydrates. Using negative ESI and solutions containing ammonium salts,13-15 heparin oligosaccharides generated by heparin lyase digestion can be ionized as multiply charged deprotonated ions without SO3 losses, provided that mild desolvation conditions are used. ESI is appropriate to detect sulfated glycosaminoglycan oligosaccharides eluting on-line from size-exclusion11 and reversed-phase16 chromatography columns. When negative ESI is used in the absence of ammonium ions, abundant sodium adducted ions are observed.14,17 Fragmentation of singly charged monosulfated18 or doubly charged disulfated heparin disacchrides19 results in abundant ions from glycosidic bond and cross ring cleavages. Fragmentation of a doubly charged, trisulfated, heparin disaccharide results in an abundant loss of one SO3 molecule from the precursor in addition to glycosidic and cross ring cleavages.18 In summary, SO3 losses are minimized and glycosidic bond and cross ring cleavages are maximized when all sulfate groups are negatively charged or paired with a metal cation. Differentiation of positional sulfation isomers by CID MS has been demonstrated for 4- and 6-sulfated CS disaccharides.11,15,20,21 (11) Zaia, J.; Costello, C. E. Anal. Chem. 2001, 73, 233-239. (12) Zaia, J.; McClellan, J. E.; Costello, C. E. Anal. Chem. 2001, 73, 6030-6039. (13) McClellan, J. M.; Costello, C. E.; O’Connor, P. B.; Zaia, J. Anal. Chem. 2002, 74, 3760-3771. (14) Chai, W.; Luo, J.; Lim, C. K.; Lawson, A. M. Anal. Chem. 1998, 70, 20602066. (15) Desaire, H.; Leary, J. J. Am. Soc. Mass Spectrom. 2000, 11, 916-920. (16) Kuberan, B.; Lech, M.; Zhang, L.; Wu, Z. L.; Beeler, D. L.; Rosenberg, R. J. Am. Chem. Soc. 2002, 124, 8707-8718. (17) Kim, Y. S.; Ahn, M. Y.; Wu, S. J.; Kim, D. H.; Toida, T.; Teesch, L. M.; Park, Y.; Yu, G.; Lin, J.; Linhardt, R. J. Glycobiology 1998, 8, 869-877. (18) Ruiz-Calero, V.; Moyano, E.; Puignou, L.; Galceran, M. T. J. Chromatogr., A 2001, 914, 277-291. (19) Pope, R. M.; Raska, C. S.; Thorp, S. C.; Liu, J. Glycobiology 2001, 11, 505513. (20) Lamb, D. J.; Wang, H. M.; Mallis, L. M.; Linhardt, R. J. J. Am. Soc. Mass Spectrom. 1992, 3, 797-803. (21) Desaire, H.; Sirich, T. L.; Leary, J. A. Anal. Chem. 2001, 73, 3513-3520.
2446
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
These results showed that the abundances of ions produced from glycosidic bond cleavage are indicative of the position of sulfation on the GalNAc residue. We have demonstrated that CS oligosaccharides dissociate to produce abundant glycosidic bond product ions and that the abundances of these ions are indicative of the position of sulfation of GalNAc residues along the backbone.12 The pattern of glycosidic bond cleavages produced is strongly influenced by the charge state of the precursor ion13 and, presumably, its gas phase conformation. An approach combining ESI CID MS of CS oligosaccharides with capillary electrophoresis has also been described.22 Because heparin contains 2-3 sulfate groups per disaccharide residue, charge-charge repulsion makes it likely that the highest charge state observed will contain a mixture of protonated and deprotonated sulfate groups. As a result, it is expected that the sulfate groups with the highest degree of charge localization will have the least tendency to be lost as SO3. It is also expected that the pattern of product ions generated by CID of heparan sulfate oligosaccharides will vary, depending on the charge state of the selected precursor ion. The present work shows how deprotonated HLGAG ions fragment to produce product ion patterns that are highly dependent on the charge state. For low charge states, several losses of SO3 occur before abundant glycosidic bond fragment ions are observed. For higher charge states, however, glycosidic bond fragmentation competes favorably with losses of SO3. To obtain complete coverage of the HLGAG sequence from CID of deprotonated ions requires the combination of data from different charge states. It is also shown that the likelihood of glycosidic bond fragmentation is enhanced by adducting the HLGAGs with sodium or calcium ions. The resulting mass spectra contain, relative to deprotonated ions, more complete glycosidic bond fragmentation and some cross-ring cleavages, providing valuable information on the HLGAG sulfation pattern. These results show that highly sulfated carbohydrates, such as HLGAGs, may be effectively analyzed using negative ESI tandem MS of either deprotonated or metal cationized ions. EXPERIMENTAL SECTION Materials. Synthetic heparin oligosaccharides H1, H2, and H3 (See Figures 3, 5, and 7 for structures) were donated by SanofiSynthelabo (Toulouse, France). Quadrupole Ion Trap Mass Spectrometry. Nanoelectrospray mass spectra were acquired using a Bruker Daltonics (Billerica, MA) Esquire 3000 instrument equipped with a Picoview source. Nanospray needles (1-2-µm i.d.) were pulled from thinwall borosilicate glass capillaries (1.2-mm o.d., 0.90-mm i.d.; World Precision Instruments, Sarasota, FL) using a Sutter Instrument P 80/PC micropipet puller (San Rafael, CA). Ion source conditions were set so as to desolvate the ions without causing losses of SO3 (skimmer 1-10 V, capillary exit offset -20 V). The nitrogen dry gas was flowed at 1 L/min, and the drying temperature was 150 °C. Scans were acquired at 1650 u/s, and the target was set to 6000 with a maximum accumulation time of 50 ms. The distribution of ion charge states could be manipulated by controlling the position of the needle with respect to the mass spectrom(22) Zamfir, A.; Seidler, D. G.; Kresse, H.; Peter-Katalinic, J. Rapid Commun. Mass Spectrom. 2002, 16, 2015-2024.
Figure 1. Negative ion electrospray mass spectra of synthetic heparin oligosaccharides (a) H1, (b), H2, (c) H3. All mass spectra were acquired using a QIT instrument using solutions of 30% methanol and 0.1% ammonium hydroxide.
eter orifice. The highest charge states were achieved by setting the needle 2-3 mm off-axis and 2-3 mm away from the source orifice. HLGAG oligosaccharides were dissolved at 5 pmol/µL in 30% methanol and 0.1% ammonium hydroxide, and 3-µL portions were sprayed by grounding the solution and ramping the orifice potential to 600-800 V positive relative to the solution. For CID, 3-u windows were selected and excited for 40 ms. The collision amplitude was adjusted so that fragment ions were produced without obliterating the precursor ion. Quadrupole Orthogonal Acceleration Time-of-Flight Mass Spectrometry. Mass spectra were acquired in the negative ion mode using an Applied Biosystems/MDS-Sciex API QSTAR Pulsar quadrupole orthogonal acceleration time-of-flight mass spectrometer. Samples were dissolved in 30% methanol to achieve a 1 pmol/ µL solution. Aliquots (3 µL) were infused into the mass spectrometer source using nanospray23 tips pulled in-house (see above). Steady ion signal was typically observed using a needle potential of -1000 V, and all spectra were calibrated externally. The needle was positioned 1 cm away from the mass spectrometer orifice and slightly off-axis. Divalent calcium ions were added to the nanospray solutions as acetate salts to a final concentration of 100 µM using a 10 mM stock solution. Conditions for CID were set so that the precursor ion remained abundant in the tandem mass spectra of HLGAG oligosaccharide ions. RESULTS AND DISCUSSION Highly sulfated heparin oligosaccharides can be ionized and detected using ESI MS without degradation and without the use of basic peptide or other pairing agents. Figure 1 shows ESI mass spectra of three synthetic heparin oligosaccharides using a 30% (23) Wilm, M. S.; Mann, M. Int. J. Mass Spectrom. Ion Processes 1994, 136, 167180.
Figure 2. Yields of product ions resulting from CID of different charge states of H1. Stages of MS were acquired on [M - nH m(SO3)]n- ions where n ) 2, m ) 0 (m/z 424); n ) 2, m ) 1 (m/z 384); n ) 2, m ) 2 (m/z 344); n ) 3, m ) 0 (m/z 282); n ) 3, m ) 1 (m/z 255); and n ) 4, m ) 0 (m/z 211). The yield of glycosidic bond cleavage ions as a percent of all product ions for each MS stage is shown by the shaded bars. The percent of all glycosidic bond cleavage product ions that were detected fully sulfated for each stage of MS is shown by the open bars.
methanol, 0.1% ammonium hydroxide solution (see Figures 3, 5, and 7 for structures). The top spectrum shows that the heparin trimer H1, a tetrasulfated trisaccharide, produces an abundant 4- ion. This ion corresponds to a species in which all four sulfate groups are deprotonated and, therefore, charged (denoted [M 4H]4-). Ionization of pentasulfated tetramer H2 results in the formation of three charge states, 5-, 4-, and 3-. The fact that the 4- ion is the most abundant in the spectrum probably indicates that charge-charge repulsion limits the abundance of the 5- ion. The ESI mass spectrum of the octasulfated pentamer H3 indicates an order of abundance of charge states of 6- > 5> 4- . 3- under the conditions used. No 7- or 8- ions were detected, indicating again that charge-charge repulsion limits the number of sulfate groups that are observed in deprotonated form. Thus, we show that highly sulfated oligosaccharides can be ionized in deprotonated form with high charge states and without losses of sulfate from precursor ions. CID of Heparin Oligosaccharides in a Quadrupole Ion Trap MS. CID of sulfated oligosaccharides produces fragment ions from two major pathways: (1) loss of SO3 and (2) glycosidic bond cleavage. The ions observed in Figure 1 were subjected to CID using a QIT mass spectrometer using a 3-u selection window, a 3-u fragmentation window, and resonant excitation. Under such conditions, newly formed product ions are no longer excited, and therefore, primarily single bond cleavages are observed. Note that the observed charge state distribution can be shifted by adjusting the position of the nanospray tip relative to the mass spectrometer orifice. It is, thus, possible to isolate and fragment the H1 [M 2H]2- ion (m/z 423.5) in sufficient abundance to investigate its fragmentation properties. CID of the [M - 2H]2- ion results in an abundant ion at m/z 383.5 corresponding to [M - 2H - SO3]2and other fragment ions in very low abundance (Figure 2). Selection and fragmentation of the [M - 2H - SO3]2- ion results in an abundant [M - 2H - 2SO3]2- ion (m/z 343.6), in addition to ions produced through glycosidic cleavage. Figure 2 shows the yield of glycosidic bond product ions from CID of the ions Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
2447
produced by successive losses of SO3 from the precursor. The yield of glycosidic bond product ions increases and reaches 100% in the MS4 stage. Isolation and fragmentation of the H1 [M 3H]3- ion (m/z 281.9, Figure 1) results in an abundant ion corresponding to [M - 3H - SO3]3- (m/z 255.2), in addition to abundant glycosidic bond cleavage product ions. The yield of such ions is low in the MS2 stage, but is nearly 100% in the MS3 stage. Isolation and fragmentation of the [M - 4H]4- ion from H1 (m/z 211.0) results in 100% yield of glycosidic bond product ions in the MS2 stage. CID of the H1 [M - 2H]2- ion results in facile loss of SO3 with very little glycosidic bond cleavage. For the 2- ion, a comparatively small ion population undergoes glycosidic bond cleavage in the MS3 stage, and only a small percentage of these are fully sulfated. After the MS4 stage, only 19% of glycosidic bond product ions retain all sulfate groups (Figure 2). Although CID of the [M - 3H]3- ion (m/z 281.9) results in a comparatively low yield of glycosidic bond product ions, approximately one-half of these are fully sulfated, indicating the existence of a population of precursor ions for which glycosidic bond cleavages are favored over SO3 loss. The yield of sulfated product ions increases during the MS3 stage. The yield of sulfated ions from glycosidic bond cleavage is 78% for the MS2 stage from the [M - 4H]4- ion. These results are consistent with the conclusion that the abundances of ions resulting from glycosidic bond cleavage increase with the number of charged sulfate groups on the parent trisaccharide. The fact that the yield of fully sulfated glycosidic bond cleavage product ions is much higher for the 3- ion than for the 2- precursor ion implies that the presence of more than one protonated sulfate group in the trimer structure markedly decreases the production of backbone cleavages during CID. Figure 3a-d show the yield of individual glycosidic bond product ions as a percent of all product ion abundance for structure H1, [M - 2H]2- ion. The yield for Bn and Yn ion is close to zero in the MS2 stage (not shown) and is quite low for the MS3 stage (Figure 2, 3a,b) for which the major population of ions fragment to lose SO3 and B22- ions are detectable in fully sulfated form. This indicates that a small population of ions having two charged sulfate groups on the nonreducing terminal GlcN residue exists in the MS3 stage. Knowing this, it can be inferred that this population has lost SO3 on the reducing terminal GlcN residue during the MS2 stage. The much higher percent total ion abundance observed for this ion in the MS4 stage indicates that the same population of ions loses a second molecule of SO3 from the reducing terminal GlcN residue. Another population exists with charge on both GlcN residues that fragments to produce [B2 - SO3]1- and [Y2 - SO3]1- ions in the MS4 stage. Interestingly, the absence of an ion corresponding to Y22- in the MS3 or MS4 stages indicates the absence of an ion population with two charges on the reducing terminal GlcN residue. Although the major population of triply charged H1 ions fragments to lose SO3 in the MS2 stage, a significant abundance of glycosidic bond product ions is observed (Figure 3e,f). The ions whose production is accompanied by losses of SO3, [B1 SO3]1- and [Y1 - SO3]1-, indicate that a small fraction of the [M - 3H - SO3]3- ion population produced in the MS2 stage (Figure 3e,f) have enough internal energy to undergo subsequent glyco2448 Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
Figure 3. Glycosidic bond product ions observed from tandem mass spectra of H1. Stages of MS were acquired on [M - nH - m(SO3)]nions where n ) 2, m ) 1 (a, b); n ) 2, m ) 2 (c, d); n ) 3, m ) 0 (e, f); n ) 3, m ) 1 (g, h); and n ) 4, m ) 0 (i, j). The percent total ion abundances for fully sulfated glycosidic bond cleavage product ions are shown by the shaded bars. Those for ions undergoing losses of SO3 are shown by the open bars. Two charge states are indicated for observed product ions, desulfated ion/fully sulfated ion in the denominator, that is, B21-/2-. Percent total ion abundances for Bn and Yn ions are shown for charge states and MS stages as indicated in a-j. The observed glycosidic bond cleavage product ions are shown structurally in k.
sidic bond cleavage. The complementary cleavages (B22- and Y22-) are observed with no losses of SO3. The MS3 stage, acquired by exciting the [M - 3H - SO3]3- ion, results in the formation of abundant Y12- and complementary [B2 - SO3]1- product ions (Figure 3g,h). A significant abundance of the B22- ion indicates
Figure 4. The yields of product ions resulting from CID of different charge states of H2. Stages of MS were acquired on [M - nH m(SO3)]n- ions where n ) 3, m ) 0 (m/z 367); n ) 3, m ) 1 (m/z 340); n ) 3, m ) 2 (m/z 313); n ) 4, m ) 0 (m/z 275), 1 (m/z 255); and n ) 5, m ) 0 (m/z 220). The yield of glycosidic bond cleavage ions as a percent of all product ions for each MS stage is shown by the shaded bars. The percent of all glycosidic bond cleavage product ions that were detected fully sulfated for each stage of MS is shown by the open bars.
the existence of a smaller population of ions that has lost SO3 from the reducing terminal GlcN residue in the MS2 stage. These results demonstrate that the major population of triply charged H1 precursor ions have one charge on the nonreducing terminal GlcN residue and two on the reducing terminal GlcN residue. The uncharged sulfate group is lost during the MS2 stage, and the [M - 3H - SO3]3- ion fragments in the MS3 stage to produce abundant Y12- and [B2 - SO3]1- ions. A minor population of precursor ions exists with two charges on the nonreducing terminal GlcN residue and loses SO3 from the reducing terminal GlcN residue in the MS2 stage and produces the B22- and [Y1 - SO3]1- ions. Isolation and CID of the H1 [M - 4H]4- ion from the spectrum shown in Figure 1b results in abundant fully sulfated product ions corresponding to Y12-, B22-, and C12-, with B12- in lower abundance (Figure 3i,j). These results show that CID of the H1 molecular ion, in which all sulfate groups are charged, results in abundant, fully sulfated product ions arising via backbone cleavages. Figure 4 shows the yields of glycosidic bond fragment ions resulting from stages of MS generated from the 3-, 4-, and 5charge states, respectively, of structure H2. Despite the fact that the 3- charge state contains 2 protonated (and, therefore, chargeneutral) sulfate groups, the yield of glycosidic bond cleavage product ions in the MS2 stage is relatively high (32%, Figure 5a,b). Of these product ions, most are fully sulfated. Both the yield of product ions formed by glycosidic bond cleavages and the percent of fully sulfated product ions diminishes in the MS3 stage (Figure 5c,d). Having lost two molecules of SO3, the ions fragment in the MS4 stage to produce a 100% yield of cleavage product ions that retain a low (14%) extent of sulfation (Figure 5e,f). The MS2 stage results in abundant Y32- and B11- ions (Figure 5a,b), from cleavage adjacent to the nonreducing terminal GlcN residue. It is significant that such abundant backbone cleavages are observed in the MS2 stage without losses of SO3, despite the fact that the H2 [M 3H]3- ion contains two protonated sulfate groups. Although the
H1 [M - 2H]2- ion also contains two protonated sulfate groups, the fragmentation pattern in the MS2 step produces much lower backbone cleavage yield. The H2 [M - 3H]3- ion contains 60% deprotonated sulfate and the H1 [M - 2H]2- 50%, indicating that as the percent of charged residues increases, the abundances of glycosidic bond product ions increase relative to those resulting from losses of SO3. These results are consistent with chargecharge repulsions causing strain on glycosidic bonds, resulting in increased abundances of ions produced from cleavages of these bonds as the percent of charged sulfate groups increases. The results are consistent with the ions’ adopting a gas phase conformation that separates charge and predisposes the glycosidic bond adjacent to the nonreducing terminus to rupture during CID. The observation of B21- and Y32- ions (Figure 5a,b) indicates that, for this population, one charge resides on the nonreducing terminus and two, on the internal GlcN residue. Subsequent selection and fragmentation of the [M - 3H - SO3]3- ion results in much lower yield of backbone cleavage ions, consistent with the conclusion that two populations of [M - 3H]3- ion exist. One population fragments in the MS2 stage primarily to lose SO3; the other results in rupture of the bond adjacent to the nonreducing GlcN residue. The former population does not exhibit abundant bond ruptures that correspond to backbone cleavages until the MS4 stage (Figure 5e,f), at which point only charged sulfate groups remain. The latter population evidently has a conformation with enough strain in one glycosidic bond to make its rupture more energetically favorable than loss of SO3. These two populations may correspond to the different charge patterns on the interior GlcN residue. The B22- ion is abundant in the MS4 stage, a fact that indicates that a population of ions with two charges on the nonreducing terminal GlcN residues favors the loss of two molecules of SO3 before abundant backbone cleavages are observed. Two populations of H2 [M - 4H]4- ions fragment to produce an abundant [M - 4H - SO3]4- ion (Figure 4) and backbone cleavages (Figure 5g,h), respectively. The Y22- ion is observed without substantial losses of SO3, and the Y33- ion is more abundant than the [Y3 - SO3]3- ion, indicating that backbone cleavage is energetically favored. Fragmentation of the [M - 4H - SO3]4- ion in the MS3 stage results in B12- and B22ions, showing that the nonreducing terminal GlcN residue is doubly charged (Figure 5i,j). The [Y3 - SO3]3- and [Y2 - SO3]2ions indicate that SO3 loss occurs to the internal GlcN residue during the MS2 stage for this population of ions. Formation of the [B3 - 2SO3]3- occurs with loss of SO3 from the internal GlcN residue. The H2 [M - 5H]5- ion has one charge per sulfate group. Its fragmentation results in no losses of SO3 from the precursor ion but produces abundant ions via backbone cleavages (Figure 5k,l). Detection of an abundant Y11- ion indicates that a charge resides on the reducing terminal IdoA residue, consistent with the conclusion that, because of repulsive forces, all five sulfate groups cannot be simultaneously charged. A fully charged B3 ion is absent, and abundant ions corresponding to [B3 - H2SO4]3-, [B3 - H2SO4]2-, and [B3 - SO3 - H2SO4]3- are observed. This is consistent with loss of H2SO4 under conditions in which a high degree of like-charge repulsion exists, such as with CID of the H2 [M - 5H]5- ion. Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
2449
Figure 5. Glycosidic bond product ions observed from tandem mass spectra of H2. Stages of MS were acquired on [M - nH - m(SO3)]nions where (a, b) n ) 3, m ) 0; (c, d) n ) 3, m ) 1; (e, f) n ) 3, m ) 2; (g, h) n ) 4, m ) 0; (i, j) n ) 4, m )1; (k, l) n ) 5, m ) 0. The percent total ion abundances for fully sulfated glycosidic bond cleavage product ions are shown by the shaded bars. Those for ions undergoing losses of SO3 are shown by the open bars. The gray bars depict B3 - 2(SO3) (lower) and B3 - SO3 - H2SO4 (upper), respectively. Percent total ion abundances for Bn and Yn ions are shown for charge states and MS stages, as indicated in a-j. Two charge states are indicated for observed product ions, desulfated ion/fully sulfated ion in the denominator, that is, B21-/2-. The observed glycosidic bond cleavage product ions are shown structurally in k.
Figure 6a shows an MSn experiment in which ions produced from successive losses of SO3 from the H3 [M - 3H]3- ion are selected and fragmented. Backbone cleavage ions are abundant only in the MS7 stage (Figure 7c,d). Despite the fact that five molecules of SO3 have been removed, fully sulfated, structurally useful B22- and Y12- ions are observed. Although the H3 [M - 4H]4- ion requires six stages of MS before reaching 100% yield of backbone cleavage, such ions are produced in significant yield in the MS2-MS5 stages (Figure 6b, Figure 7e-l). The abundant Y4 and B1 ions indicate that cleavage between residues 4 and 5 is more favorable in the MS2 stage than loss of SO3 for a population of ions constituting a substantial fraction of all precursor ions (Figure 7e,f). The [M - 4H - SO3]4ion fragments in the MS3 stage to produce B1 and Y4 ions with a 2450 Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
greater degree of SO3 loss to the reducing end of the structure (Figure 7g,h). The [M - 4H - 2(SO3)]4- population fragments in the MS4 stage to produce B2 and Y3 ions with losses of SO3 (Figure 7i,j). Interestingly, the yield of fully sulfated B22ions increases in the MS5 and MS6 stages (Figure 7k-n). Clearly, the addition of a charge increases the lability of the glycosidic bonds of the H3 [M - 4H]4- ion relative to that for the [M - 3H]3- ion. Fragmentation of the H3 [M - 5H]5- and [M - 6H]6- ions, respectively, results in very high yields of backbone cleavage product ions (Figure 6c). While the [M - 5H]5- ion fragments to produce an abundant Y44- ion (Figure 7o,p), the [M - 6H]6- ion produces abundant B4 and C4 ions (Figure 7q,r). Evidently, the gas phase conformations for the two ions differ significantly,
Figure 6. Yields of product ions resulting from CID of different charge states of H3. Stages of MS were acquired on [M - nHm(SO3)]n- ions, where (a) n ) 3, m ) 0 (m/z 501), 1 (m/z 474), 2 (m/z 448), 3 (m/z 421), 4 (m/z 394), and 5 (m/z 368); (b) n ) 4, m ) 0 (m/z 376), 1 (m/z356), 2 (m/z 336), 3 (m/z 315), and 4 (m/z 394); (c) n ) 5, m ) 0 (m/z 300); and n ) 6, m ) 0 (m/z 250). The yield of glycosidic bond cleavage ions as a percent of all product ions for each MS stage is shown by the shaded bars. The percent of all glycosidic bond cleavage product ions that were detected fully sulfated for each stage of MS is shown by the open bars.
resulting in stress being placed on different glycosidic bonds and different resulting cleavage patterns. Tandem Mass Spectrometric Fragmentation of Heparin Oligosaccharides and Their Metal Complexes in a Q-oTOF MS. Using a Q-o-TOF or any tandem-in-space mass spectrometer, precursor ions are selected using the first mass analyzer and fragmented, by virtue of the ion kinetic energy, through collisions with gas molecules. This type of fragmentation is nonspecific in that product ions can further fragment by subsequent collisions with gas molecules, resulting in multiple bond cleavages. Multiple backbone cleavages of peptide ions commonly occur during lowenergy CID on tandem-in-space mass spectrometers.24-26 CS oligosaccharide ions subjected to CID on a Q-oTOF instrument produce multiple backbone cleavage ions, the abundances of which are minimized through the use of collision energies with low absolute values.12 Fragmentation of the H2 [M - 3H]3- and [M - 4H]4precursor ions, respectively, by Q-oTOF CID results in abundant ions from backbone cleavages (Figure 8a,b). Although more glycosidic bonds are cleaved than are observed for the same ions using the QIT instrument (Figure 5), the extent of SO3 loss is also greater. The data for fragmentation of [M (Na) - 3H]3-, [M (2Na) - 3H]3- and [M (4Na) - 3H]3- demonstrate that the losses of SO3 from backbone cleavage product ions diminishes as the (24) Bean, M. F.; Carr, S. A.; Thorne, G. C.; Reilly, M. H.; Gaskell, S. J. Anal. Chem. 1991, 63, 1473-1481. (25) Burlet, O.; Yang, C.-Y.; Gaskell, S. J. J. Am. Soc. Mass Spectrom. 1992, 3, 337-344. (26) Papayannopoulos, I. A. Mass Spectrom. Rev. 1995, 49-73.
number of sodium adducts increases (Figure 8c-e) (Ions formed by metal ion exchanges for hydrogen ions, [Ion + nMx - (nx + y)H]y-, will be written as [Ion (nM) - yH]y-. This nomenclature assumes that each metal, nMx, displaces nx hydrogen ions. Proton displacement will also be assumed for CID product ions). The [M (4Na) - 3H]3- precursor ion corresponds to deprotonation of all H2 sulfate and carboxylate groups and results in low abundance of backbone cleavage ions produced with loss of SO3, with the exception of B1. It has been observed that the lability of sulfate groups in the gas phase follows the order SO3H > SO3- > SO3Na.27 In the absence of ammonium ions, HLGAG oligosaccharides are typically observed as polysodiated ions as a result of the affinity of sulfate groups for alkali cations. Because sodium is usually present at a background level, the degree of sodiation of ions corresponding to HLGAG oligosaccharides can easily be manipulated by varying the concentration of ammonium ions in the electrospray solution. Deprotonated ions are typically observed using ammonium acetate concentrations of 10-100 µM or with 0.1% ammonium hydroxide (see Figure 1). The determination of linkage position in pentasaccharides has been demonstrated using cobalt complexes.28 Cobalt complexes have also been used to differentiate diastereomeric HexNAc monosaccharides using tandem MS.29 Differences in product ion abundances have also been used in the quantification of multicomponent hexosamine30 and N-acetylhexosamine31 mixtures. Product ions specific to stereochemical differences at C2 and C4 are observed in tandem mass spectra generated from diastereomeric N-glycoside zinc (II) complexes.32 HLGAGs are known to bind divalent metals, including calcium,33,34 copper,35,36 zinc,37 iron,38 and platinum.39 Tandem mass spectra of divalent metal adducts of synthetic heparin oligosaccharides were, therefore, examined in order to facilitate glycosidic bond cleavage during MS/MS. The ESI mass spectrum of H2 diluted in 100 µM calcium acetate, 30% methanol was characterized by an abundant ion at m/z 392.98, corresponding to [M (2Ca) - 3H]3- (calcd m/z 392.961), indicating that all five sulfate groups and both carboxylates are deprotonated. This ion was selected and subjected to CID using -15-V collision energy and was shown to produce abundant ions from backbone cleavages accompanied by only lowabundance SO3 losses (Figure 8f). The observed exceptions, B1 and B3, are consistent with losses of SO3 occurring in greater abundance for Bn ions to the reducing side of N-sulfated GlcN (27) Yagami, T.; Kitagawa, K.; Aida, C.; Fujiwara, H.; Futaki, S. J. Pept. Res. 2000, 56, 239-249. (28) Konig, S.; Leary, J. A. J. Am. Soc. Mass Spectrom. 1998, 9, 1125-1134. (29) Desaire, H.; Leary, J. A. Anal. Chem. 1999, 71, 1997-2002. (30) Desaire, H.; Leary, J. A. Anal. Chem. 1999, 71, 4142-4147. (31) Desaire, H.; Leary, J. A. J. Am. Soc. Mass Spectrom. 2000, 11, 1086-1094. (32) Gaucher, S. P.; Leary, J. A. Anal. Chem. 1998, 70, 3009-3014. (33) Grant, D.; Long, W. F.; Moffat, C. F.; Williamson, F. B. Biochem. J. 1992, 282, 601-604. (34) Grant, D.; Long, W. F.; Williamson, F. B. Biochem. Soc. Trans. 1996, 24, 203S. (35) Mukherjee, D. C.; Park, J. W.; Chakrabarti, B. Arch. Biochem. Biophys. 1978, 191, 393-399. (36) Grant, D.; Long, W. F.; Moffat, C. F.; Williamson, F. B. Biochem. J. 1992, 283, 243-246. (37) Grant, D.; Long, W. F.; Williamson, F. B. Biochem. J. 1992, 287, 849-853. (38) Grant, D.; Long, W. F.; Williamson, F. B. Biochem. Soc. Trans. 1992, 20, 361S. (39) Grant, D.; Long, W. F.; Williamson, F. B. Biochem. Soc. Trans. 1996, 24, 204S.
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
2451
Figure 7. Glycosidic bond product ions observed from tandem mass spectra of H3. Stages of MS were acquired on [M - nH - m(SO3)]nions where (a, b) n ) 3, m ) 0; (c, d) n ) 3, m ) 5; (e, f) n ) 4, m ) 0; (g, h) n ) 4, m ) 1; (i, j) n ) 4, m ) 2; (k, l) n ) 4, m ) 3; (m, n) n ) 4, m ) 4; (o, p) n ) 5, m ) 0; and (q, r) n ) 6, m ) 0. The percent total ion abundances for fully sulfated glycosidic bond cleavage product ions are shown by the shaded bars. Those for ions undergoing losses of SO3 are shown by the open bars. Two charge states are indicated for observed product ions, desulfated ion/fully sulfated ion in the denominator, that is, B21-/2-. The observed glycosidic bond cleavage product ions are shown structurally in (s).
residues and may serve as indicator ions. These data clearly demonstrate that cation adduction effectively favors the production of backbone cleavage ions without SO3 losses. In the CID spectrum of the H2 [M (2Ca) - 3H]3- (Figure 9a), the precursor ion remains abundant, indicating that the collision 2452 Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
energy suffices to fragment the molecule but not to produce abundant internal fragment ions.12 Ions corresponding to [M (2Ca) - 3H- SO3]3- (m/z 366.32) and [M (2Ca) - 2H-H2SO4]2- (m/z 540.98) are observed at comparatively low abundances, indicating that loss of SO3 from the [M (2Ca) - 3H]3- ion occurs only to a
Figure 8. Graphs of ion abundances for product ions of metal adducts of H2 showing contributions from both intact ions and those undergoing loss of SO3. (a) [M - 4H]4-, (b) [M - 3H]3, (c) [M (Na) 3H]3-, (d) [M (2Na) - 3H]3-, (e) [M (4Na) - 3H]3-, (f) [M (2Ca) 3H]3-. M in the graphs refers to the quasimolecular ion.
Figure 9. (a) Q-oTOF CID product ion mass spectrum of H2 adducted with 2 Ca ions, [M (2Ca) - 3H]3- at -15 V collision energy and (b) H2 structure with major product ions indicated.
small degree. The observation of the [Y2 (Ca)]2- ion is consistent with one Ca2+ ion bound to residues 1 and 2 and one to residues 3 and 4. The Y11- ion without bound calcium indicates that one charge resides on the GlcNS3S6S residue. These conclusions are supported by the observed [B3 (2Ca)]2- ion, indicating that two Ca2+ ions are bound by residues 2, 3, and 4. The presence of the B12- ion at m/z 159.49 indicates no bound calcium in the structure. This observation is consistent with calcium ion binding by the
Figure 10. (a) Q-oTOF product ion mass spectrum of H3 [M (3Ca) - 4H]4- at -20 V collision energy and (b) H3 structure with major product ions indicated.
carboxylate group of residue 3 and a sulfate group in residue 4. Alternately, the binding of the Ca2+ ion by sulfate groups in residue 4 may be disrupted by collision-induced glycosidic bond cleavage forming the B12- ion. The [Y2 (Ca)]2- ion (obsd m/z 322.47, calcd m/z 322.464) is consistent with three sulfate groups modifying residues 1 and 2 and two modifying residues 3 and 4. Observation of the Y11- ion (obsd m/z 207.06, calcd m/z 207.050) indicates that residue 1 is not sulfated. This conclusion is supported by the mass difference observed between the [B3 (2Ca)]2- (m/z 485.93, calcd m/z 485.916) and B12- ions (m/z 159.49, calc. m/z 159.482), corresponding to the presence of two sulfate groups modifying residues 3 and 4. The 0,2A11- ion (m/z 199.00, calcd m/z 198.991) indicates that one sulfate group is located on the 3 or 6 position of residue 4. Figure 10 shows the tandem mass spectrum of heparin structure H3 complexed with three Ca2+ ions, 4- charge state (m/z 404.20) at -20 V collision energy. The collision energy was limited so as to generate product ions without obliterating the precursor ions. Similar conditions have been found to minimize the extent of internal fragmentation of chondroitin sulfate oligosaccharides.12 An ion corresponding to [M (3Ca) 4H - SO3]4- is observed at comparatively low abundance relative to the precursor ion, indicating that sulfate groups are not readily lost as neutral SO3. The abundant ion at m/z 479.96 corresponds to [0.4X4 (2Ca)]3- (calcd m/z 479.948), and it is likely that cleavage of the nonreducing terminal ring disrupts one of the Ca2+ ions. An ion corresponding to [Y3 (2Ca)]2- is observed at m/z 541.92 (calcd m/z 541.908), the presence of which indicates that two Ca2+ ions are bound to the reducing terminal trisaccharide. The presence of an ion at m/z 322.48, corresponding to [Y2 (Ca)]2- (calcd m/z 322.462), indicates that one Ca2+ ion is bound to the reducing terminal disaccharide. These results are confirmed by the observation of an ion at m/z 485.94, corresponding to [B3(2Ca)]2- (calcd m/z 485.916), that Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
2453
indicates that two Ca2+ ions are bound to the nonreducing terminal trisaccharide. The observation of the ion at m/z 445.94, corresponding to [B3 (2Ca) - SO3]2- (calcd m/z 445.938), is consistent with the conclusion that glycosidic bond cleavage on the reducing side of an N-sulfated GlcN residue may destabilize a sulfate group. If so, then the N-sulfate group is a likely candidate for this destabilization. The observed glycosidic bond and cross-ring cleavages provide valuable information regarding the positions of sulfate groups on the molecule, as can be seen in Figure 10b. Observation of the [0.4X4 (2Ca)]3- ion at m/z 479.96 (calcd m/z 479.948) is consistent with the presence of one sulfate group at the 6 position of GlcNAc residue 5. The cross-ring cleavage ion 0,2A11- (m/z 199.01) is consistent with the composition C4H7O9S (calcd m/z 198.991), indicating that a single sulfate group modifies the nonreducing terminal fragment that includes positions 3 and 6. The reconstructed mass difference between 0.4X4 (2Ca) and Y3 (2Ca) ions (357.01 u) determines the composition of the molecule from the reducing side of the cross-ring cleavage to the reducing side of the GlcNAc residue at position 4. This composition (C10H15NO11S, calcd 357.036 u) shows that one sulfate group is located at either the N or 3 position of residue 5 or the 2 position of residue 4. The reconstructed mass difference between the [Y3 (2Ca)]1and [Y2 (Ca)]1- (438.90 u) corresponds to the composition C6H9NO13S3Ca (calcd 438.886 u), consistent with the location of three sulfate groups on residue 3. Since there are no options, these groups must be located on the N, 3, and 6 positions. The reconstructed neutral mass of the Y2 (Ca) species (646.94 u) is consistent with the composition C13H21NO20S3Ca (calcd 646.944 u). This composition indicates that three sulfate groups are distributed between the 2 position of residue 2 and the N, 3, and 6 positions of residue 1. CONCLUSIONS During electrospray ionization MS, heparin oligosaccharides produce abundant negatively charged molecular ions without losses of SO3. When subjected to CID in a QIT mass spectrometer, these precursors generate mixtures of product ions that result from losses of SO3 and from backbone fragmentation. The yields of these product ions depend on the charge state of the precursor ion relative to the number of sulfate groups. Generally, backbone cleavage product ions are abundant when at least half of the sulfate groups are charged. Under such conditions, like-charge repulsions place stress on glycosidic bonds, favoring their rupture over losses of SO3. It is significant that the pattern of glycosidic bond cleavages changes dramatically with the precursor ion charged state. This implies that the gas phase conformations adopted by different charge states differ enough to place strain on different glycosidic bonds. The most useful ions for the purposes of interpretation are the complementary Bm and Yn ions resulting from cleavage of the same bond. The sum of the deconvoluted neutral masses of such pairs equals the molecular mass and provides an accounting for all sulfate groups. Substantial structural information may be obtained by combining the data obtained from fragmentation of different charge states and MS stages. Losses of SO3 from both precursor and glycosidic bond product ions are more pronounced for tandem-in-space, relative to those 2454
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
observed for tandem-in-time mass spectrometers. Pairing the sulfated oligosaccharide anions with metal cations serves to increase sulfate stability such that abundant backbone cleavage fragments are observed using a Q-oTOF mass spectrometer, allowing more complete sequence-related data to be obtained in a single MS stage. The spectra serve to define the compositions of the molecular ions determined on the basis of the observed fragment ions. Because sulfate losses from backbone cleavage ions are of low abundance, the masses of product ions can be used to determine the compositions and positions of sulfation of individual residues. The number of sulfate groups on a given residue may be determined by the m/z values of glycosidic bond product ions (primarily Bm and Yn). The differentiation of sulfation position of mono- and disulfated GlcN residues requires the production of cross-ring cleavage ions, and these ions have been observed for the nonreducing terminal residue of a heparin tetramer and a pentamer. Although the information from a single tandem mass spectrum is not sufficient to completely define the sulfation pattern, it is likely that complementary information will be produced from CID of different charge and metal adduction states. It is even conceivable that metal ion adduction can differentiate uronic acid epimers by virtue of the stereoselective nature of the complex. Investigation of this feature is the subject of ongoing studies. These findings provide a framework for practical analysis of unknown HLGAG structures. It is recommended that negative nano-ESI be used with a solution of a low concentration of ammonium hydroxide. The analysis of deprotonated ions will entail obtaining CID mass spectra on all observed charge states. In some cases, it may be possible to shift the observed charged states by adjusting the position of the nano ESI emitter relative to the source orifice. Best results for deprotonated ions will be obtained from a trapped ion instrument, in which multiple stages of MS can be obtained. It is recommended that ions produced from SO3 loss in the MS2 stage be subjected to additional stages of CID. Alternatively, the excitation window can be broadened to include the m/z of the [M - SO3 - nH]n- ion, necessitating the use of fewer MS stages. When no such ions are observed, it can be assumed that only glycosidic bond fragmentation occurs. Combining the results from all charge states and stages, all m/z values should be converted to neutral equivalents, and pairs that sum to the neutral mass of the ion observed in the MS stage should be identified. These product ion pairs contain no losses of SO3 and are, therefore, structurally useful. For researchers using tandem-in-space instruments, such as the Q-oTOF, use of metal cations is recommended. Although the usefulness of different metal cations to facilitate tandem MS of HLGAGs is the subject of ongoing work in the authors’ laboratory, calcium is recommended at the present time. Such cationization can also be accomplished from 100 µM calcium acetate solutions and result in stabilization of sulfate groups and formation of structurally useful product ions. Although formation of the observed product ions can generally be assumed to occur without losses of SO3, it appears that some of the ions produced from fragmentation to the reducing side of sulfated GlcN resides are observed with loss of SO3. The observation of a Bm and a Bm SO3 (or Yn and Yn - SO3) pair may therefore serve as an indicator for this group. Because sodium is ubiquitously present in HLGAG
solutions, researchers may also consider the use of this cation. Use of ammonium hydroxide suppresses sodium cationization, but such ions are generally more abundant when using ammonium acetate-containing spray solutions. ACKNOWLEDGMENT This work was supported by NIH/NCRR Grant no. P41RR10888. The Esquire 3000 mass spectrometer was donated by Bruker Daltonics, Inc. Synthetic heparin oligosaccharides were donated by Sanofi-Synthelabo, Toulouse, France.
SUPPORTING INFORMATION AVAILABLE A listing of all quadrupole ion trap mass spectra of synthetic heparin oligosaccharides used in this manuscript is available as Supporting Information. This material is available free of charge via the Internet at http://pubs.acs.org.
Received for review November 23, 2002. Accepted March 18, 2003. AC0263418
Analytical Chemistry, Vol. 75, No. 10, May 15, 2003
2455
