Determination of the Primary Structures of Heparin-and Heparan

Anal. Chem. 2001, 73, 2310-2316

Determination of the Primary Structures of Heparin- and Heparan Sulfate-Derived Oligosaccharides Using Band-Selective Homonuclear-Decoupled Two-Dimensional 1H NMR Experiments Wei-Lien Chuang, Marie Dvorak Christ, and Dallas L. Rabenstein*

Department of Chemistry, University of California, Riverside, California 92521

Band-selective homonuclear-decoupled (BASHD) twodimensional NMR experiments are applied to the assignment of 1H NMR spectra of oligosaccharides, using as an example a heparin-derived hexasaccharide. The anomeric (H1) region of the 1H NMR spectrum is band-selected in the F1 dimension. With the increased resolution that results from less truncation of interferograms in the t1 dimension, finer digital resolution in the F1 dimension, and collapse of multiplets to singlets in the F1 dimension, cross-peaks to the anomeric protons of the two iduronic acid residues, which overlap in normal two-dimensional total correlation spectroscopy (TOCSY) and rotating frame Overhauser enhancement spectroscopy (ROESY) spectra of the hexasaccharide, are resolved in BASHD-TOCSY and BASHD-ROESY spectra, leading to an unequivocal assignment of the 1H NMR spectrum of the hexasaccharide. Incorporation of the water attenuation by transverse relaxation method for the complete and selective elimination of the water resonance into two-dimensional BASHD experiments makes it possible to observe oligosaccharide resonances at the frequency of the water resonance, as demonstrated with the observation of cross-peaks to resonances at the frequency of the water resonance in BASHD-TOCSY spectra of a second heparin-derived hexasaccharide. Heparin and heparan sulfate are linear copolymers of alternating 1 f 4 linked uronic acid and glucosamine residues. Both are synthesized in mammalian tissues from glucuronic acid and N-acetylglucosamine. Regions of the copolymer are then variously modified by N-deacetylation followed by N-sulfation, epimerization of glucuronic acid to iduronic acid, 2-O-sulfation of uronic acid residues, and 6-O-sulfation of glucosamine residues.1 The result is a family of polysaccharides with considerable sequence microheterogeneity (Figure 1).2-4 (1) Lindahl, U.; Kusche, M.; Lidholt, K.; Oscarsson, L.-G. Ann. N. Y. Acad. Sci. 1989, 566, 36-50. (2) Casu, B. In Heparin: Chemical and Biological Properties, Clinical Applications; Lane, D. A., Lindahl, U., Eds.; CRC Press: Boca Raton, FL, 1989; pp 25-49.

2310 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

Heparin and heparan sulfate interact with and modulate the activity of other biological molecules.5,6 With the finding that the anticoagulant activity of heparin resides in a unique pentasaccharide sequence with antithrombin III binding properties,7-11 there has been intense interest in relating the primary structure of heparin to its biological activity.5,12-16 Likewise, the structural variability of heparan sulfate, which is widely distributed in basement membranes of various tissues and the plasma membranes of most cells, is the basis for its wide range of biological activities.6,17 However, identification of oligosaccharide sequences in heparin and heparan sulfate with specific biological activities has been difficult due to their structural heterogeneity. One approach has been to isolate and sequence oligosaccharide fragments produced by enzymatic or chemical depolymerization of heparin or heparan sulfate.13-16,18-23 These oligosaccharides can then be used in structure-activity studies. (3) Gatti, G.; Casu, B.; Hamer, G. K.; Perlin, A. S. Macromolecules 1979, 12, 1001-1007. (4) Lindahl, U.; Kjelle´n, L. Thromb. Haemostasis 1991, 66, 44-48. (5) Linhardt, R. J.; Loganathan, D. In Biomimetic Polymers; Gebelein, C. G., Ed.; Plenum Press: New York, 1990; pp 135-173. (6) Lindahl, U. Pure Appl. Chem. 1997, 69, 1897-1902. (7) Lam, L. H.; Silbert, J. E.; Rosenberg, R. D. Biochem. Biophys. Res. Commun. 1976, 69, 570-577. (8) Ho ¨o ¨k, M.; Bjo ¨rk, I.; Hopwood, J.; Lindahl, U. FEBS Lett. 1976, 66, 90-93. (9) Rosenberg, R. D.; Lam, L. Proc. Natl. Acad. Sci. U.S.A. 1979, 76, 12181222. (10) Lindahl, U.; Backstrom, G.; Ho ¨o ¨k, M.; Thunberg, L.; Fransson, L.-A.; Linker, A. Proc. Natl. Acad. Sci. U.S.A. 1979, 76, 3198-3202. (11) Lindahl, U.; Thunberg, L.; Ba¨ckstro¨m, G.; Riesenfeld, J.; Nordling, K.; Bjo¨rk, I. J. Biol. Chem. 1984, 259, 12368-12376. (12) Spillman, D.; Lindahl, U. Curr. Opin. Struct. Biol. 1994, 4, 677-682. (13) Gettins, P.; Horne, A. Carbohydr. Res. 1992, 223, 81-98. (14) Pervin, A.; Gallo, C.; Jandik, K. A.; Han, X. J.; Linhardt, R. J. Glycobiology 1995, 5, 83-95. (15) Larnkjaer, A.; Hansen, S. H.; Lstergaard, P. B. Carbohydr. Res. 1995, 266, 37-52. (16) Yamada, S.; Murakami, T.; Tsuda, H.; Yoshida, K.; Suguhara, K. J. Biol. Chem. 1995, 270, 8696-8705. (17) Kjelle´n, L.; Lindahl, U. Annu. Rev. Biochem. 1991, 60, 443-475. (18) Linker, A.; Hovingh, P. Carbohydr. Res. 1984, 127, 75-94. (19) Huckerby, T. N.; Sanderson, P. N.; Nieduszynski, I. A. Carbohydr. Res. 1986, 154, 15-27. (20) Linhardt, R. J.; Rice, K. G.; Merchant, Z. M.; Kim, Y. S.; Lohse, D. L. J. Biol. Chem. 1986, 261, 14448-14454. (21) Horne, A.; Gettins, P. Carbohydr. Res. 1991, 225, 43-57. (22) Yamada, S.; Yoshida, K.; Sugiura, M.; Sugahara, K. J. Biochem. 1992, 112, 440-447. 10.1021/ac0100291 CCC: $20.00

© 2001 American Chemical Society Published on Web 04/03/2001

Figure 1. Structures of heparin and heparan sulfate. (A) Major disaccharide repeating unit in heparin (iduronic acid 2-sulfate (1 f 4) glucosamine2,6-sulfate). (B) Minor-abundance disaccharide repeating unit in heparin. The uronic acid can be either iduronic acid or glucuronic acid. Disaccharide A accounts for at least 85% of heparins from beef lung and 75% of those from porcine intestinal mucosa.2 (C) Heparan sulfate consists of blocks of heparin-like disaccharide repeating units and blocks of relatively unmodified glucuronic acid (1 f 4) N-acetylglucosamine disaccharide repeating units. There is some microheterogeneity in both blocks.

Chart 1

1H

NMR spectroscopy is one of several methods24-26 used for determining the chemical structures of heparin- and heparan sulfate-derived oligosaccharides.13-16,18-23,27 A standard strategy involves first identification of the monosaccharide residues present in the oligosaccharide using scalar connectivities in two-dimensional correlation spectroscopy (COSY) and total correlation spectroscopy (TOCSY) spectra followed by determination of the sequence of the monosaccharides in the oligosaccharide using dipolar connectivities in nuclear Overhauser enhancement spectroscopy (NOESY) and rotating frame Overhauser enhancement spectroscopy (ROESY) spectra. The presence of sulfate and/or N-acetyl groups at specific sites on each residue is then inferred from chemical shift data. However, even with the increased resolution achieved with two-dimensional experiments, resonance overlap due to the rather narrow spectral dispersion typical of 1H NMR spectra of carbohydrates can still make determination of the complete structural formula difficult or ambiguous. Also, oligosaccharide resonances at the water resonance frequency are suppressed or completely eliminated in 1H NMR spectra measured (23) Tsuda, H.; Yamada, S.; Yamane, Y.; Yoshida, K.; Hopwood, J. J.; Sugahara, K. J. Biol. Chem. 1996, 271, 10495-10502 (24) Liu, J.; Desai, U. R.; Han, X.-J.; Toida, T.; Linhardt, R. J. Glycobiology 1995, 5, 765-774. (25) Venkataraman, G.; Shriver, Z.; Raman, R.; Sasisekharan, R. Science 1999, 286, 537-542. (26) Turnbull, J. E.; Hopwood, J. J.; Gallagher, J. T. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 2698-2703. (27) Sugahara, K.; Tohno-oka, R.; Yamada, S.; Kho, K.-H.; Morris, H. R.; Dell, A. Glycobiology 1994, 4, 535-544.

with suppression of the water resonance or the residual HDO resonance by presaturation. In this paper, we describe the application of band-selective homonuclear-decoupled (BASHD) two-dimensional 1H NMR experiments28,29 to the assignment of 1H NMR spectra of oligosaccharides, using as an example heparin-derived hexasaccharide I (Chart 1). The 1H NMR spectrum of I has been assigned previously in four separate studies using standard 2D NMR methods; however, there are disagreements in the assignments due to the near coincidence of resonances for two of the anomeric protons.14,15,21,30 We demonstrate that, with the significantly increased resolution of the BASHD experiments, it is possible to completely and unequivocally assign the 1H NMR spectrum of I. We also report pulse sequences for two-dimensional BASHD 1H NMR experiments which incorporate the water attenuation by transverse relaxation (WATR) method for suppression of the water resonance,31-33 and we demonstrate their application with the observation of resonances at the water resonance frequency in the two-dimensional 1H NMR spectra of a second heparin-derived hexasaccharide (II) (Chart 2). (28) Krishnamurthy, V. V. Magn. Reson. Chem. 1997, 35, 9-12. (29) Kaerner, A.; Rabenstein, D. L. Magn. Reson. Chem. 1998, 36, 601-607. (30) Chai, W.; Hounsell, E. F.; Bauer, C. J.; Lawson, A. M. Carbohydr. Res. 1995, 269, 139-156. (31) Rabenstein, D. L.; Fan, S. Anal. Chem. 1986, 58, 3178-3184. (32) Rabenstein, D. L.; Larive, C. K. J. Magn. Reson. 1990, 87, 352-356. (33) Larive, C. K.; Rabenstein, D. L. Magn. Reson. Chem. 1991, 29, 409-417.

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2311

Chart 2

In addition, chemical shift-substituent correlations for use in determining the structures of heparin- and heparan sulfate-derived oligosaccharides are reported. EXPERIMENTAL SECTION Preparation of Heparin-Derived Hexasaccharides. The heparin-derived hexasaccharides were isolated from depolymerized porcine intestinal mucosal heparin. The depolymerization was catalyzed by heparin lyase I (EC 4.2.2.7). Both the heparin and the heparin lyase I were obtained from Sigma Chemical Co. One gram of heparin was dissolved in 50 mL of a pH 7.5 depolymerization solution containing 25 mM sodium acetate and 10 mM calcium acetate.13 One vial of heparin lyase I (250 units) was added to the heparin solution, and the progress of the reaction was monitored by measuring the absorbance at 232 nm. When there was no further reaction, the solution was lyophilized and the oligosaccharide mixture was size-fractionated by low-pressure (gravity flow) gel permeation chromatography on a BioRad BioGel P6 column (3 × 200 cm) using 0.5 M NH4HCO3 as eluent at a flow rate of 8-12 mL/h. The hexasaccharides used in this study were isolated from the size-uniform hexasaccharide fraction by strong anion exchange HPLC on a semipreparative scale Dionex CarboPac PA1 column. A Dionex 500 ion chromatography system equipped with a GP40 gradient pump and an AD20 UV/visible detector was used. The hexasaccharides were eluted with a linear gradient of 70 mM pH 3 phosphate buffer (solvent A) and 70 mM pH 3 phosphate buffer plus 2 M NaCl (solvent B) at a flow rate of 3.3 mL/min. 1H NMR Spectroscopy. 1H NMR spectra were recorded at 500 MHz on a Varian Unity Inova 500 spectrometer equipped with waveform generators, a Performa X, Y, Z gradient module, and a 1H{13C, 15N} triple-resonance, X, Y, Z triple-axis pulsed field gradient probe. Two-dimensional TOCSY and ROESY spectra were measured with the F1 spectral window set to 3600 Hz in both dimensions. Typically, 2D data sets were acquired with 256-512 increments in the t1 dimension and 4K points in the t2 dimension at a temperature of 25 °C. BASHD-TOCSY and BASHD-ROESY 1H NMR spectra were measured with literature pulse sequences.28,29 The band-selective pulses used in the BASHD-TOCSY and BASHD-ROESY experiments were Gaussian cascade Q3 pulses34 phase modulated to shift the center of excitation 300 Hz to the anomeric region of 1H NMR spectra of the heparin-derived hexasaccharides. An F1 spectral window of 600 Hz was used for band selection of the anomeric resonances. A total of 256 or 512 × 4K complex points were acquired in the t1 and t2 dimensions. A mixing time of 110 ms was used in both the BASHD-TOCSY and BASHD-ROESY experiments. Pulsed field gradients were (34) Emsley, L.; Bodenhausen, G. J. Magn. Reson. 1992, 97, 135-148.

2312

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

applied in the Z-direction. Data were processed with linear prediction to 2K or 4K points in the t1 dimension and, for the 2K data, zero filling to 4K points. A 90° shifted sine-bell apodization was applied to the data in both dimensions. The difference between the transmitter offset and the center of the excitation window was applied to the t1 interferograms prior to Fourier transformation for the BASHD-TOCSY and BASHD-ROESY data. The residual HDO resonance was suppressed in both the BASHD and standard two-dimensional experiments with a selective presaturation pulse. The BASHD-TOCSY-WATR spectrum was measured using the BASHD-TOCSY pulse sequence modified by addition of a transverse relaxation period just prior to the spin-lock mixing period. A Carr-Purcell-Meiboom-Gill (CPMG) train of 180° pulses ([τ -180°- τ]n; τ ) 0.3 ms)35,36 was applied during the transverse relaxation period to avoid loss of signal due to magnetic field inhomogeneity effects and to suppress dephasing of the transverse magnetization due to homonuclear spin-spin coupling effects. Samples were contained in 5-mm Shigemi tubes (Shigemi Co., Ltd. Tokyo, Japan), which have matched magnetic susceptibility plugs above and below the sample solution to reduce the required sample volume to ∼300 µL. Sodium 3-(trimethylsilyl)propionate2,2,3,3-d4 (TMSP) was added to samples as a chemical shift reference. A trace amount of EDTA-d12 was also added to reduce line broadening caused by traces of paramagnetic cations. RESULTS AND DISCUSSION Determination of the Chemical Structures of Heparin- and Heparan Sulfate-Derived Oligosaccharides. The usual procedure for determining the chemical structures of heparin- and heparan sulfate-derived oligosaccharides by 1H NMR takes advantage of the relatively high dispersion of resonances for the anomeric (H1) protons to assign the spectrum. In the first step, a 2D TOCSY spectrum is measured to assign each H1 resonance to one of the possible types of monosaccharides present in the oligosaccharide. For oligosaccharides produced by heparinasecatalyzed depolymerization of heparin, these are 4,5-unsaturated uronic acid (∆UA) at the nonreducing end of the oligosaccharide, iduronic acid (IdoA), glucuronic acid (GlcA), and glucosamine (GlcN). The pattern of TOCSY cross-peaks between the resonances for the H1 proton and the other carbon-bonded protons of each monosaccharide serves as a fingerprint, which can be used to identify the monosaccharide. The next step is to determine the position of each monosaccharide in the oligosaccharide. Sequential assignments are made using cross-peaks between resonances for each H1 proton and (35) Carr, H. Y.; Purcell, E. M. Phys. Rev. 1954, 94, 630-638. (36) Meiboom, S.; Gill, D. Rev. Sci. Instrum. 1958, 29, 688-691.

Figure 2. Strategy for the sequential assignment of 1H NMR spectra of heparin- and heparan sulfate-derived oligosaccharides. See text for discussion.

Figure 3. A 500-MHz 1H NMR spectrum of a 2.5 mM solution of heparin-derived hexasaccharide I in D2O at pD 5.11. The residual HDO resonance was suppressed by presaturation.

the H4 and/or H3 protons of the next residue in the sequence in 2D ROESY and/or NOESY spectra. The usual starting point is the H1 resonance of the ∆UA residue, which can be identified from its TOCSY cross-peak to the unique H4 resonance of the ∆UA residue (step 1 in Figure 2). The glucosamine residue linked to ∆UA is then identified by the ROESY cross-peak between its H4 resonance and the ∆UA H1 resonance (step 2 in Figure 2), the H1 resonance of the glucosamine by its TOCSY cross-peak to the H4 resonance (Step 3), and so on until the position of each monosaccharide in the oligosaccharide sequence has been determined. The presence of O- and N-sulfate groups on each residue is then inferred from the chemical shifts of the carbon-bonded protons, which can be determined from subspectra obtained by taking traces through the TOCSY spectrum at the H1 chemical shifts. The sequential assignment procedure is based on the observation of assignable cross-peaks between the resonances for each H1 proton and the H4 proton of the same residue in 2D TOCSY spectra and the H4 proton of the next residue in the sequence in 2D ROESY or NOESY spectra. However, unequivocal assignment of all the H1 T H4 cross-peaks is not always possible due to the relatively narrow spectral dispersion of oligosaccharide resonances and the low resolution of standard 2D TOCSY, ROESY, and NOESY spectra. This is the case with hexasaccharide I.

Figure 4. Portion of the regular 2D TOCSY spectrum of I that contains cross-peaks to the overlapped H1 resonances of the two IdoA(2S) residues. Also shown is the subspectrum obtained by taking a slice through the TOCSY spectrum at the center of the H1 crosspeaks on the F1 dimension. The resonances at 4.835 and 4.781 ppm for the H5 protons of the two IdoA(2S) residues are reduced in intensity due to presaturation of the HDO resonance at 4.8 ppm.

Figure 5. Portion of the BASHD-TOCSY spectrum of I that contains cross-peaks to the overlapped H1 resonances of the two IdoA(2S) residues. Also shown are subspectra obtained by taking slices through the BASHD-TOCSY spectrum at the chemical shifts of the two singlet H1 resonances on the F1 dimension. A total of 256 × 4K complex points were acquired in the t1 and t2 dimensions. Data were processed by linear prediction to 2K and zero filling to 4K points in the t1 dimension, and a 90° shifted sine-bell apodization was applied in both dimensions.

The 1H NMR spectrum of I is shown in Figure 3. The spectrum can be divided into three regions: the resonance at 5.992 ppm for the H4 proton of the ∆UA residue, the region from 5.6 to 5.2 ppm for the six H1 protons, and the region from 4.85 to 3.2 ppm for the 28 other carbon-bonded protons. The chemical shifts of cross-peaks to the H1 protons in the TOCSY spectrum of I indicate the resonance at 5.505 ppm to be for the H1 proton of the ∆UA residue, the resonances at 5.443, 5.421, and 5.395 ppm to be for the H1 protons of three GlcN residues, and the two overlapping resonances at ∼5.22 ppm to be for the H1 protons of two IdoA residues. However, because the TOCSY cross-peaks between resonances for the H1 protons and the other carbon-bonded protons of the two IdoA residues all fall at 5.22 ppm on the H1 chemical shift dimension (Figure 4), the two IdoA H1 T H4 TOCSY cross-peaks cannot be assigned to specific H1 resonances. The same region of the BASHD-TOCSY spectrum measured with band selection of the H1 region in the F1 dimension is shown in Figure 5. The significantly increased resolution in the BASHDTOCSY spectrum is a result of less truncation of the interferogram in the t1 dimension and finer digital resolution in the F1 dimension, since a smaller spectral width is observed with band selection, and most importantly to collapse of 1H-1H spin-coupled multiplets Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2313

Figure 6. Portions of two-dimensional ROESY spectra of I that contain cross-peaks between the resonances for the H1 protons of the two IdoA(2S) residues and the H3 and H4 protons of the glucosamine residues linked to their reducing ends. (A) Measured with the standard ROESY pulse sequence. (B) Measured with the BASHD-ROESY pulse sequence. Cross-peaks between the H1 resonances of the two IdoA(2S) residues and the H3 and H4 protons of the glucosamine residues linked to their reducing ends are labeled. The IdoA(2S) and GlcNS(6S) residues are labeled I and A, respectively, and the sequential position of the monosaccharide residues is labeled a-f, starting with the ∆UA(2S) residue. For the BASHD-ROESY spectrum, 256 × 4K complex points were acquired in the t1 and t2 dimensions, and a 90° shifted sine-bell apodization was applied in both dimensions.

to singlets in the F1 dimension.28,29 With the increased resolution in the F1 dimension, two sets of cross-peaks are clearly observed for the two IdoA residues. The chemical shifts of the H1 protons of the two IdoA residues are determined to be 5.217 and 5.219 ppm from the BASHD-TOCSY spectrum; that is, the two H1 resonances are separated by only 1 Hz in the 500-MHz spectrum. Nevertheless, with the superior resolution in the BASHD-TOCSY spectrum, the cross-peaks can be resolved and subspectra can be obtained for the two iduronic acid residues (Figure 5). Although each subspectrum contains residual resonances from the other IdoA residue, chemical shift data for the two IdoA residues can be obtained from the two subspectra. Having assigned the six H1 resonances to specific monosaccharide residues, the next step is to determine their sequence in the hexasaccharide. Starting with the resonance at 5.505 ppm for the H1 proton of ∆UA (residue a), the resonance at 3.843 ppm can then be assigned to H4 of the GlcN residue next in the sequence (residue b), the resonance at 5.395 ppm to H1 of residue b, the resonance at 4.086 ppm to H4 of the IdoA residue next in the sequence (residue c), and the resonance at 5.217 ppm to H1 of residue c by the procedure outlined in Figure 2. However, the assignment stops at residue c because cross-peaks between the two IdoA H1 resonances and the H4, and H3, protons of the two “downstream” GlcN residues are all at 5.22 ppm on the H1 dimension in the standard ROESY spectrum (Figure 6A). Thus, the IdoAc(H1) f GlcNd(H4) and IdoAc(H1) f GlcNd(H3) ROESY cross-peaks cannot be distinguished from the IdoAe(H1) f GlcNf(H4) and IdoAe(H1) f GlcNf(H3) ROESY cross-peaks, where the superscripts indicate residue position in the sequence. The same portion of the BASHD-ROESY spectrum, measured with band selection in the H1 region on the F1 dimension, is shown in Figure 6B. With the significantly increased resolution in the BASHD-ROESY spectrum, cross-peaks between the resonances for the H1 protons of the two IdoA residues and the H3 and H4 protons of the two 1 f 4 linked GlcN residues can be assigned 2314 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

Table 1. 1H Chemical Shift Data for Hexasaccharide Ia residue proton

a

b

c

d

e

f

H1 H2 H3 H4 H5 H6 H6′

5.505 4.623 4.315 5.992

5.395 3.297 3.655 3.843 4.032 4.250 4.355

5.217 4.340 4.235 4.086 4.835

5.421 3.278 3.659 3.759 4.020 4.270 4.386

5.219 4.321 4.192 4.116 4.781

5.443 3.260 3.696 3.753 4.127 4.300 4.365

a

Chemical shifts are reported in ppm relative to internal TMSP.

(Figure 6B), which makes it possible to complete the unequivocal assignment of the 1H NMR spectrum of I. The chemical shifts for all the carbon-bonded protons of I, as determined from the 1D spectrum in Figure 3 or from subspectra obtained by taking traces through TOCSY or BASHD-TOCSY spectra, are reported in Table 1. The substituents on each monosaccharide residue can be determined from the chemical shifts of the carbon-bonded protons. Average reference chemical shift values, calculated from literature chemical shift data for heparin-derived oligosaccharides,14-16,21,23,30,37-40 are presented in Table 2. Starting at the nonreducing end, the chemical shift of the resonance for H2 of ∆UA indicates a 2-O-sulfate group (∆UA(2S)), the chemical shift of 3.297 ppm for H2 of residue b indicates an N-sulfate group, (37) Yamada, S.; Yamane, Y.; Tsuda, H.; Yoshida, K.; Sugahara, K. J. Biol. Chem. 1998, 273, 1863-1871. (38) Yamada, S.; Sakamoto, K.; Tsuda, H.; Yoshida, K.; Sugiura, M.; Sugahara, K. Biochemistry 1999, 38, 838-847. (39) Hileman, R. E.; Smith, A. E.; Toida, T.; Linhardt, R. J. Glycobiology 1997, 7, 231-239. (40) Toida, T.; Hileman, R. E.; Smith, A. E.; Vlahova, P. I.; Linhardt, R. J. J. Biol. Chem. 1996, 271, 32040-32047.

Table 2. Reference 1H Chemical Shift Data for Monosaccharides in Heparin- and Heparan Sulfate-Derived Oligosaccharidesa

a The reference chemical shift values are averages calculated from chemical shift data reported in refs 14-16, 21, 23, 30, and 37-40 unless otherwise indicated. b For GlcNS(3S,6S): H1, 5.46 ( 0.02; H2, 3.44 ( 0.03; H3, 4.50 ( 0.03; H4, 4.02 ( 0.05; H5, 4.27 ( 0.07; H6, 4.39 ( 0.05; H6′, 4.42 ( 0.01 ppm. c For the GlcNAc or GlcNAc(6S) residue at the reducing terminus: H1(R), 5.24 ( 0.10; H1(β), 4.72 ( 0.02; H2(R), 3.70 ( 0.03; H2(β), 3.88 ( 0.03; H3(R), 3.89 ( 0.07; H3(β), 3.78 ( 0.04; H4(R), 3.81 ( 0.06; H4(β), 3.81 ( 0.02; H5(R, R ) H), 3.95 ( 0.02; H5(β, R ) H), 3.58 ( 0.01; H5(R, R ) SO3 ), 4.16 ( 0.04; H5(β, R ) SO3 ), 3.80 ( 0.05; H6, H6′(R, R ) H), 3.85 ( 0.02, 3.80 ( 0.05; H6, H6′(β, R ) H), 3.89 ( 0.04, d 3.80 ( 0.03; H6, H6′(R, R ) SO3 ), 4.28 ( 0.01, 4.32 ( 0.11; H6, H6′(β, R ) SO3 ), 4.26 ( 0.03, 4.30 ( 0.08 ppm. For GlcNS(6S) of the sequence GlcNS(6S)-GlcA: H1, 5.61 ( 0.03 ppm. e Calculated from chemical shift data reported in refs 27 and 37. f For GlcA at the reducing terminus: H1(R), 5.21; H1(β), 4.61; H2(R), 3.57; H2(β), 3.28; H3(R), 3.83; H3(β), 3.63; H4(R), not available; H4(β), 3.74.

while the chemical shifts of 4.250 and 4.355 ppm for the resonances for the H6 protons of residue b indicate a 6-O-sulfate group (GlcNS(6S)), the chemical shift data for residue c indicates it to be 2-O-sulfated iduronic acid (IdoA(2S)), and so on to identify oligosaccharide I as 1 f 4 linked ∆UA(2S)-GlcNS(6S)-IdoA(2S)-GlcNS(6S)-IdoA(2S)-GlcNS(6S). Chemical shift assignments have been reported previously for hexasaccharide I.14,15,21,30 The chemical shift data in Table 1 are in reasonably good agreement with that reported previously. The major differences are due to the fact that it was not possible to resolve the resonances for the H1 protons of the two IdoA(2S) residues at ∼5.22 ppm in three of the previous studies,14,15,21 and in the fourth study, the two IdoA(2S) H1 resonances are reported to have a chemical shift difference of 0.06 ppm.30 This in turn has resulted in errors in the assignments of resonances for the other IdoA(2S) protons. As demonstrated here, it was possible with the superior resolution of the 2D BASHD experiments not only to resolve the two H1 resonances but also to obtain subspectra for the two IdoA(2S) residues and to determine unequivocally their positions in the oligosaccharide sequence. Measurement of Two-Dimensional 1H BASHD NMR Spectra with Elimination of the Water Resonance by the WATR Method. The water resonance falls in the middle of the oligosaccharide spectral region (Figure 3) and often overlaps some oligosaccharide resonances. The water resonance can be suppressed by the presaturation method. However, any oligosaccharide resonances under the water resonance are also suppressed by the presaturation pulse. We describe here use of the WATR method to eliminate the water resonance from two-dimensional BASHD-NMR spectra of oligosaccharides. The WATR method was developed to observe resonances at the water resonance frequency, the basis of the

method being selective elimination of the water resonance by transverse relaxation.31 This is accomplished in the BASHDTOCSY experiment by addition of a CPMG-type transverse relaxation period [(t/2 - 180° - t/2)n] just prior to the mixing period in the BASHD-TOCSY pulse sequence. The water resonance can be eliminated from BASHD-ROESY spectra by the WATR method by using the standard BASHD-ROESY pulse sequence with a sufficiently long mixing period during which the water resonance is eliminated by transverse relaxation. BASHD-TOCSY 1H NMR spectra of a 1 mM solution of hexasaccharide II in 90% H2O/10% D2O are shown in Figure 7. The solution also contained the water proton exchange reagent NH4Cl to selectively decrease T2* of the water resonance.31 Spectrum A was measured with suppression of the water resonance by a selective presaturation pulse; spectrum B was measured with elimination of the water resonance by the WATR method. Cross-peaks to the resonance at 4.83 ppm for the H5 proton of the IdoA(2S) residue are suppressed in spectrum A by the presaturation pulse, whereas they are clearly observed in spectrum B, even though the intensity of the water resonance is ∼100 000 times that of the oligosaccharide resonances in the normal single pulse 1H NMR spectrum. CONCLUSIONS The results presented here demonstrate that, with the significantly increased resolution of 2D TOCSY and ROESY spectra measured with the BASHD-TOCSY and BASHD-ROESY pulse sequences, it is possible to assign 1H NMR spectra of heparinderived oligosaccharides even when some of their H1 resonances have nearly identical chemical shifts. These experiments should be particularly useful for large oligosaccharides and oligosaccharides that contain multiple residues of the same monosaccharide. Results presented here also demonstrate that oligosaccharide Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2315

Figure 7. Portions of 500-MHz BASHD-TOCSY spectra of a 1.0 mM solution of hexasaccharide II in 90% H2O/10% D2O at pH 6.4. The solution also contained 0.45 M NH4Cl as a water proton exchange reagent. Spectrum A was measured by the normal BASHD-TOCSY pulse sequence with suppression of the water resonance with a 1-s presaturation pulse. Spectrum B was measured with suppression of the water resonance by the WATR method using a pulse sequence with a CPMG-type transverse relaxation period immediately prior to the mixing period. The length of the transverse relaxation period was 300 ms.

resonances overlapped by the water resonance can be observed by selective elimination of the water resonance by the WATR method, making it possible to observe cross-peaks that otherwise would be eliminated by the presaturation pulse. The WATR method requires addition of a water proton exchange reagent, which might affect the solution structure or conformation of an oligosaccharide. However, if the purpose is to assign the 1H NMR spectrum of an oligosaccharide or to determine its structural formula by 1H NMR, addition of the water proton exchange reagent should have no affect on the results, in which case the WATR method provides complete and selective elimination of the

2316 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

water resonance from one- and two-dimensional 1H NMR spectra of oligosaccharides. ACKNOWLEDGMENT This research was supported in part by the National Institutes of Health Grant HL56588. Funding for the Varian UnityInova 500 spectrometer was provided in part by NSF-ARI Grant 9601831. Received for review January 8, 2001. Accepted March 5, 2001. AC0100291