Characterization of oligosaccharides from the chondroitin/dermatan sulfates

¹H-NMR and ¹³C-NMR studies of reduced trisaccharides and hexasaccharides

Abstract

Chondroitin and dermatan sulfate (CS and DS) chains were isolated from bovine tracheal cartilage and pig intestinal mucosal preparations and fragmented by enzymatic methods. The oligosaccharides studied include a disaccharide and hexasaccharides from chondroitin ABC lyase digestion as well as trisaccharides already present in some commercial preparations. In addition, other trisaccharides were generated from tetrasaccharides by chemical removal of nonreducing terminal residues. Their structures were examined by high-field ¹H and ¹³C NMR spectroscopy, after reduction using sodium borohydride. The main hexasaccharide isolated from pig intestinal mucosal DS was found to be fully 4-O-sulfated and have the structure: ΔUA(β1–3)GalNAc4S(β1–4)l-IdoA(α1–3)GalNAc4S(β1–4)l-IdoA(α1–3)GalNAc4S-ol, whereas one from bovine tracheal cartilage CS comprised only 6-O-sulfated residues and had the structure: ΔUA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S-ol. No oligosaccharide showed any uronic acid 2-sulfation. One novel disaccharide was examined and found to have the structure: GalNAc6S(β1–4)GlcA-ol. The trisaccharides isolated from the CS/DS chains were found to have the structures: ΔUA(β1–3)GalNAc4S(β1–4)GlcA-ol and ΔUA(β1–3)GalNAc6S(β1–4)GlcA-ol. Such oligosaccharides were found in commercial CS/DS preparations and may derive from endogenous glucuronidase and other enzymatic activity. Chemically generated trisaccharides were confirmed as models of the CS/DS chain caps and included: GalNAc6S(β1–4)GlcA(β1–3)GalNAc4S-ol and GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S-ol. The full assignment of all signals in the NMR spectra are given, and these data permit the further characterization of CS/DS chains and their nonreducing capping structures.

Abbreviations

CS

chondroitin sulfate

DS

dermatan sulfate

GAG

glycosaminoglycan

GalNAc(-ol)

2-deoxy-2-N-acetylamino-d-galactose (-galactitol)

GlcA

d-glucuronic acid

GlcA-ol

d-glucuronic acid alditol

IdoA

α-L-iduronic acid

6S/4S

O-ester sulfate group on C6/C4

ΔUA

4,5-unsaturated hexuronic acid (4-deoxy-α-l-threo-hex-4-enepyranosyluronic acid)

SAX

strong anion-exchange

SEC

size exclusion chromatography

There is considerable interest in the detailed molecular structure [1,2] and function [3] of glycosaminoglycans (GAGs) including chondroitin and dermatan sulfate (CS/DS) [4–6]. These structurally diverse polymers are abundant components of extracellular matrices and cell surfaces in humans and other mammals. Data are emerging that show roles for CS/DS in a variety of fundamental biological processes including neurite outgrowth [7], disease development [8] and growth factor binding [9]. CS has also been found in invertebrates [10–12] including Drosophila melanogaster[13] and Caenorhabditis elegans[14], where it has been shown to have fundamental roles in development [15].

CS/DS chains comprise a linkage region, a chain cap and a repeat region [4,5]. The repeat region of CS is a repeating disaccharide of glucuronic acid (GlcA) and N-acetylgalactosamine (GalNAc) [-4)GlcA(β1–3)GalNAc(β1-]_n, which may be O-sulfated on the C4 and/or C6 of GalNAc and C2 of GlcA. GlcA residues of CS may be epimerized to iduronic acid (IdoA) forming the repeating disaccharide [-4)IdoA(α1–3)GalNAc(β1-]_n of DS. Thus, CS and DS may be found as pure polymers or a mixed copolymer in which the DS residues and GalNAc sulfation isoforms may be located together in large blocks or distributed throughout the chain. These will have very different effects on molecular interactions and biological function.

Both the concentrations and locations of sulfate ester substituents vary with GAG source [4,16]. For example, in adult human articular cartilage, extensive 6-O-sulfation of GalNAc is observed (≈ 95%) [4]. In shark cartilage, lower levels of 6-O-sulfation are found (≈ 70%), with 4-O-sulfation making up the balance along with ≈ 25% 2-sulfation of the uronic acid residues [4], and in tracheal cartilage lower levels of 6-O-sulfation are found (≈ 20–40%), with the balance being mainly GalNAc 4-O-sulfation. In D. melanogaster, 4-O-sulfation, but not 6-O-sulfation, is observed whereas in C. elegans the chondroitin is unsulfated [13,14].

Within a tissue, the sulfation profile and levels of epimerization of GlcA to IdoA change with age. For example, the level of GalNAc 6-O-sulfation reported above for human articular cartilage applies only to the adult; at birth this level is close to zero but rises significantly during the first 20 years of life [5,17].

Not only does CS/DS structure change with tissue source and age, but within a single chain there is variability. The chain cap of CS is a GalNAc or GlcA residue; a 4,6-disulfated GalNAc residue, rare in the repeat region of human articular cartilage CS, represents over 50% of the chain caps for a normal adult, but only ≈ 30% at the termini of CS chains from osteoarthritic cartilage [18,19]. Whereas the CS chain caps may be highly sulfated, the linkage regions, via which these pendant polymers are attached to a protein core, have been shown to exhibit low levels of sulfation relative to residues within the repeat region [5,6,20], with preferential localization of unsulfated and 4-O-sulfated GalNAc residues at linkage regions [5,6,20].

The overall structure of CS chains is thus highly complex, showing significant variation in composition across materials from diverse tissue sources, from tissues of differing ages and also within a single chain. Enhanced availability of data to facilitate the characterization of these structures is therefore of value.

GAGs are not primary gene products and therefore their analysis cannot rely on genomic approaches; structural analysis requires their isolation followed by a complex characterization process. In our previous work we have used the paradigm of isolation and depolymerization of GAG chains to generate oligosaccharides, the structures of which are determined using NMR spectroscopy [6,16]. These oligosaccharides are then integrated into a chromatographic fingerprinting method which can be used for the analysis of biological samples [4].

Chondroitin lyase enzymes are eliminases which cleave the -3)GalNAc(β1–4)GlcA(β1-/IdoAα(1- bond in CS/DS in the case of chondroitin ABC lyase (EC 4.2.2.4), whereas chondroitin AC lyases act on CS alone. Chondroitin AC and ABC lyases generate disaccharides and tetrasaccharides [4] and have been widely used for the analysis of CS/DS composition. These studies have yielded crucial data allowing an understanding of species, tissue, age and pathology related differences and the estimation of changes in CS/DS abundance and composition. However, the reduction of the polymer to its individual disaccharide units removes any possible sequence data that would allow the reconstruction of biologically important functional motifs. In addition, the action of chondroitin lyase enzymes generates a 4,5-unsaturated hexuronic acid (ΔUA) from the uronic acid of the cleaved bond. Thus, the distinction between IdoA and GlcA, epimerization at C5, is lost and it is impossible to distinguish between disaccharides derived from DS and those derived from CS.

We have previously reported ¹H-NMR data for disaccharides and tetrasaccharides from CS/DS [16], and Sugahara et al. [21] have examined, by ¹H NMR and MS, a series of chondroitin ABC lyase-resistant fragments derived from CS or DS. Several of these were trisaccharides, including ΔUA(β1–3)GalNAc4,6diS(β1–4)GlcA, terminated by an unreduced GlcA ring, which could have been derived from polymer chain-reducing termini by a peeling reaction, or, through the action of a tissue endo-β-d-glucuronidase. A series of reduced and unreduced oligosaccharides obtained from DS were previously characterized by ¹H and ¹³C NMR [22], including the trisaccharide GalNAc4S(β1–4)l-IdoA(α1–3)GalNAc4S-ol.

More recently, the preparation and structural characterization of unreduced DS oligosaccharides of up to dodecasaccharide in size has been discussed [23]; a combination of 1D and 2D NMR together with electrospray MS was employed.

The employment of nondestructive analytical methods in the characterization of GAGs is becoming more important now that full structural information for large domains is required as part of the examination of function for these species. Data from other GAG fragments has already proved valuable in the study of intact parent polymers; the architecture of keratan sulfate chains has been explored in this manner [24,25]. There have already been attempts to examine intact CS chains isolated from various sources. Considerable difficulties were met when specific assignments of structural components were sought. It is thus important that polymer characterizations should be facilitated through the availability of comprehensive parameters describing the structures of a wide range of oligosaccharide structures. The complete assignments of ¹H and ¹³C NMR spectra from a series of disaccharides and tetrasaccharides derived from CS/DS chains have already been given [16]. In this report we present ¹H-NMR and some ¹³C-NMR data for trisaccharides and hexasaccharides from CS/DS. These have the structures shown below:

GalNAc6S(β1–4)GlcA(β1–3)GalNAc4S-ol: CS#604 GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S-ol: CS#606 ΔUA(β1–3)GalNAc4S(β1–4)GlcA-ol: CS040# ΔUA(β1–3)GalNAc6S(β1–4)GlcA-ol: CS060# GalNAc6S(β1–4)GlcA-ol: CS#60# ΔUA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S-ol: CS060606 ΔUA(β1–3)GalNAc4S(β1–4)l-IdoA(α1–3)GalNAc4S(β1–4)l-IdoA(α1–3)GalNAc4S-ol: DS040404

Results and Discussion

Isolation of oligosaccharides

After depolymerization by chondroitin ABC endolyase, oligosaccharides are isolated as disaccharides, trisaccharides, tetrasaccharides and, because of incomplete depolymerization, as hexasaccharides (Fig. 1) [4,16]. The heterogeneous pools of oligosaccharides generated in this way were purified by size exclusion chromatography (SEC) to yield individual oligosaccharides [4,16]. In addition, trisaccharides lacking nonreducing terminal unsaturated uronic acids have been successfully prepared from related tetrasaccharides; full ¹H-NMR and some ¹³C-NMR data for these tetrasaccharides have been reported elsewhere [16].

Figure 1. Isolation of trisaccharides. Insert: SEC of reduced oligosaccharides generated after depolymerization of bovine tracheal CS chains by chondroitin ABC endolyase on a Toyapearl HW40s column (50 cm × 1 cm) eluted in 0.5 m ammonium acetate at 0.4 mL·min⁻¹, the eluate was monitored by measuring A₂₃₂. Disaccharide and tetrasaccharide pools are indicated by 2 and 4, respectively. Main chromatogram: after removal of the nonreducing terminal unsaturated uronic acids (Experimental procedures) from the tetrasaccharide mixture (see insert), the crude trisaccharides were purified by SAX chromatography on a Spherisorb S5 column (25 cm × 1 cm) at 2 mL·min⁻¹. Bound material was eluted by a linear gradient of 2 mm LiClO₄ (buffer A) to 250 mm LiClO₄ (buffer B), pH 5.0, according to the following gradient profile: after a 10-min isocratic phase of 100% buffer A, a gradient of 0–100% buffer B was introduced over 240 min, followed by 10 min of 100% buffer B. The column eluate was monitored online at 206 nm. Individual fractions were pooled as indicated.

Trisaccharides

Two trisaccharides were prepared from the corresponding CS repeat unit tetrasaccharides ΔUA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc4S-ol (CST0604) and ΔUA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S-ol (CST0606) [16] by chemical removal [26] of the unsaturated nonreducing terminal residues (Fig. 1). The ¹H-NMR data for these species, designated CS#604 and CS#606, are summarized in Table 1. Partial 600-MHz gradient-COSY-45 and TOCSY 2D-NMR spectra for the former are shown in Figs 2 and 3, respectively, and a partial gradient-COSY-45 spectrum for the latter is given in Fig. 4. For both species, the presence of two anomeric proton sites, together with two N-acetyl methyl signals and the lack of responses corresponding to ΔUA protons all confirm that these are reduced trisaccharides derived from tetrasaccharides generated by chondroitin ABC endolyase. CS#604 and CS#606 are confirmed as containing GalNAc4S-ol and GalNAc6S-ol termini, respectively, through examination of the residue A (galactosaminitol) chemical shifts, which clearly indicate ester sulfate location and have values closely similar to those found for the corresponding tetrasaccharides CST0604 and CST0606 [16]. Similarly, the internal uronic acid residues (ring B) exhibit shift patterns almost identical with those for GlcUA in the tetramers. As would be expected, the nonreducing terminal GalNAc6S protons of both species (ring C) have chemical shift positions that are perturbed from those found for the corresponding site in the ΔUA-terminated tetrasaccharides. That for H3 is displaced by −0.21 p.p.m., together with movements of −0.185 and −0.12 p.p.m., respectively, for H4 and H2. All other changes are less than 0.04 p.p.m. in magnitude.

Table 1.  Proton chemical shifts for the trisaccharides CS#604 and CS#606. Shift values are from COSY-45 data except for those marked with an asterisk ( *).

Residue	Proton	CS#604 (p.p.m.)	CS#606 (p.p.m.)
GalNAc (C)	H1	4.558*	4.589*
	H2	3.917*	3.930*
	H3	3.743	3.745
	H4	3.988*	4.002*
	H5	3.967*	3.976*
	H6	∼4.24	∼4.24
	H6′	∼4.24	∼4.24
	CH3	2.056*	2.056*
GlcA (B)	H1	4.628*	4.556*
	H2	3.480*	3.451*
	H3	3.638	3.633*
	H4	∼3.77	∼3.75
	H5	∼3.77	∼3.75
GalNAc-ol (A)	H1	∼3.70	3.717
	H1′	∼3.72	3.784*
	H2	4.295	4.390*
	H3	4.257	4.077
	H4	4.492*	3.570*
	H5	4.146*	4.308*
	H6	∼3.705	∼4.07
	H6′	∼3.705	∼4.09
	CH3	2.018*	2.051*

Figure 2. Partial 600-MHz gradient-COSY-45 spectrum for CS#604 at 43 °C. The spectral width was 1750.7 Hz, and eight acquisitions for each of 1024 increments were sampled into 1024 complex points. The array was zero-filled to 2048 × 2048 complex points and transformed in each dimension after application of a (sinebell)² window function. CS#604 has the structure: GalNAc6S(β1–4)GlcA(β1–3)GalNAc4S-ol. C B A

Figure 3. Partial 600-MHz TOCSY spectrum for CS#604 at 43 °C. The spectral width was 1750.7 Hz, and eight acquisitions for each of 512 pairs of increments were sampled into 1024 complex points using a mixing time of 70 ms. The array was zero-filled to 2048 × 2048 complex points and transformed in each dimension after application of a 1.0 Hz exponential window function. CS#604 has the structure: GalNAc6S(β1–4)GlcA(β1–3)GalNAc4S-ol. C B A

Figure 4. Partial 600-MHz gradient-COSY-45 spectrum for CS#606 at 43 °C. The spectral width was 1750.7 Hz, and eight acquisitions for each of 1024 increments were sampled into 1024 complex points. The array was zero-filled to 2048 × 2048 complex points and transformed in each dimension after application of a (sinebell)² window function. CS#606 has the structure: GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S-ol. C B A

Two unusual trisaccharides were isolated after chondroitin lyase digestion of commercial CS preparations which are related to CS tetrasaccharides, containing a ΔUA residue as would normally be expected, but lacking the N-acetylgalactosaminitol reducing terminal moiety. These are designated CS040# and CS060#, respectively, and exhibit the chemical shift values summarized in Table 2. A partial 600-MHz gradient-COSY-45 NMR spectrum for CS060# is shown in Fig. 5. Comparisons of chemical shift values for the ΔUA residues in repeat-unit tetrasaccharides show only minor perturbations [16]. For CS040#, all (ring C) signals fall within 0.02 p.p.m. of the corresponding (ring D) locations for CST0404; in the case of CS060#, all are within 0.01 p.p.m. of the shift values seen in either CST0604 or CST0606 with the sole exception of H3 (+ 0.016 p.p.m. relative to CST0606).

Table 2.  Proton chemical shifts for the oligosaccharides CS040#, CS060# and CS#60#. Shift values are from COSY-45 data except for those marked with an asterisk ( *).

Residue	Proton	CS040# (p.p.m.)	CS060# (p.p.m.)	CS#60# (p.p.m.)
ΔUA (C)	H1	5.297*	5.199	–
	H2	3.860	3.805*	–
	H3	3.963*	4.120*	–
	H4	5.971*	5.884*	–
GalNAc (B)	H1	4.835*	4.763*	4.696*
	H2	4.092*	4.034*	3.912
	H3	4.201*	3.978	3.762
	H4	4.649*	4.173*	3.996*
	H5	3.876	3.968	3.938
	H6	3.783	4.205	4.214
	H6′	3.811	4.238	4.236
	CH3	2.113*	2.080*	2.073*
GlcA-ol (A)	H1	3.644*	3.679*	3.672*
	H1′	3.724*	3.762*	3.758
	H2	3.903	3.922*	3.917
	H3	3.845	3.855*	3.848*
	H4	4.154*	4.092*	4.073*
	H5	4.247*	4.214	4.201*

Figure 5. Partial 600-MHz gradient-COSY-45 spectrum for CS060# at 43 °C. The spectral width was 1750.7 Hz, and 24 acquisitions for each of 1024 increments were sampled into 1024 complex points. The array was zero-filled to 2048 × 2048 complex points and transformed in each dimension after application of a (sinebell)² window function. CS060# has the structure: ΔUA(β1–3)GalNAc6S(β1–4)GlcA-ol. C B A

When the internal GalNAc (ring B) data are examined, they clearly distinguish between sulfation at C4 for CS040# (H4, 4.649 p.p.m.; H6,6′ at 3.783 and 3.811 p.p.m.) and at C6 for CS060# (H4, 4.173 p.p.m.; H6,6′ at 4.205 and 4.238 p.p.m.), but other signal positions are strongly perturbed compared with their locations in the related tetrasaccharides. There is one further connected set of protons present for both of these species. This represents a residue consisting of a nonequivalent methylene group connected to a series of four further single proton sites. The chemical shift values found for both CS040# and CS060# are quite similar, the minor perturbations reflecting the different influences of 4-sulfated and 6-sulfated neighbouring rings. This residue is derived from a GlcA ring, which has been reductively opened by alkaline borohydride treatment, forming the corresponding glucuronitol. The methylene proton pair represent the hydrogens at the reduced C1 site, and the H2–H5 series does not terminate (as would be the situation for a GalNAc-ol residue) with a further nonequivalent methylene group at C6.

A novel disaccharide, related to these trisaccharides has also been characterized, and the chemical shift data for this species, designated CS#60#, are also summarized in Table 2. In this molecule, the ΔUA ring is absent and the presence of GalNAc 6-sulfation is confirmed through the presence of H6,6′ protons at 4.214 and 4.236 p.p.m. The GalNAc H4 response is found at 3.996 p.p.m.; this, together with H3 which is displaced by over −0.2 p.p.m. relative to the corresponding signal in CS060#, confirms that the nonreducing ΔUA residue, which would have been attached at C3, is no longer present.

For these novel oligomers, ¹³C-NMR data are also available and are summarized in Table 3. The influence of the sulfation position on the observed signal positions within rings B and C is very similar to that observed previously [16] for the corresponding tetrasaccharides, as is seen in Table 4 where difference data are presented together with the calculated values.

Table 3.  Carbon chemical shifts for the oligosaccharides CS040#, CS060# and CS#60#.

Residue	Carbon	CS040# (p.p.m.)	CS060# (p.p.m.)	CS#60# (p.p.m.)
ΔUA (C)	C1	102.83	104.25	–
	C2	71.54	72.60	–
	C3	67.51	68.89	–
	C4	109.41	110.11	–
	C5	147.01	147.62	–
	C6	–	172.21	–
GalNAc (B)	C1	102.97	103.70	103.81
	C2	55.02	54.12	55.27
	C3	78.24	82.50	73.86
	C4	79.04	70.51	70.49
	C5	77.41	75.60	75.49
	C6	63.91	70.36	69.91
	CH3	25.27	25.30	25.29
	C = O	177.79	178.02	–
GlcA-ol (A)	C1	65.61	65.46	65.38
	C2	74.58	74.78	74.73
	C3	73.04	73.16	73.01
	C4	82.89	83.16	82.89
	C5	76.11	76.59	76.46

Table 4.  Carbon chemical shift Δδ values between CS040# and CS060#.

Residue	Carbon	Calculated (p.p.m.)	Observed (p.p.m.)
ΔUA (C)	C1	+1.1	+1.4
	C2	+0.9	+1.1
	C3	+1.2	+1.4
	C4	+0.6	+0.7
	C5	+0.4	+0.6
GalNAc (B)	C1	+0.6	+0.7
	C2	−1.0	−0.9
	C3	+4.0	+4.3
	C4	−8.9	−8.5
	C5	−1.9	−1.8
	C6	+6.2	+6.4

The chondroitin A lyase activity, which acts on bonds containing 4-sulfated GalNAc residues, is an order of magnitude greater in chondroitin ABC endolyase than the chondroitin C lyase which acts on bonds containing 6-sulfated GalNAc residues. Thus, during cleavage, bonds involving a 4-sulfated GalNAc residue will be cleaved more often, resulting in a predominance of 4-sulfated GalNAc residues at the reducing terminus of the resulting oligosaccharides. This significantly reduces the number of oligosaccharides actually observed. However, it also means that the oligosaccharide CS#406, found at the chain cap of articular cartilage CS [19], is very challenging to produce as the required tetrasaccharide CS0406 is not abundant after chondroitin ABC endolyase digestion [4]. Other trisaccharides generated in this study are chain cap model structures; these data will aid identification of chain cap signals in large, or indeed intact, CS/DS chains.

In addition, trisaccharides may be isolated from chains that have been partially depolymerized by prior endogenous glucuronidase activity. It is noteworthy that the trisaccharides isolated from tissue sources, i.e. CS040# and CS060#, have only been derived from commercial samples of CS and not from those produced in a research laboratory environment, suggesting that these may not represent major components in vivo. This suggests that commercial samples of CS contain significant levels of chains that are not intact and that do not represent an appropriate material for the study of intact CS chains.

Hexasaccharides

Although data from disaccharides, trisaccharides and tetrasaccharides are valuable for the detailed structural characterizations of unknown segments derived from the CS families of polymers, there are many important features that reside within larger oligosaccharide units. Data are therefore presented in Table 5 giving comprehensive ¹H-NMR signal assignments for the primary reduced hexasaccharide repeat units derived from chondroitin 6-sulfate and DS polymers, namely CS060606 and DS040404. In addition, a partial 600-MHz gradient-COSY-45 NMR spectrum for CS060606 is shown in Fig. 6, and that for DS040404 is shown in Fig. 7.

Table 5.  Proton chemical shifts for the hexasaccharides CS060606, and DS040404. Shift values are from COSY-45 data except for those marked with an asterisk ( *).

Residue	Proton	CS060606 (p.p.m.)	DS040404 (p.p.m.)
ΔUA (F)	H1	5.193*	5.268*
	H2	3.796*	3.843*
	H3	4.108*	3.952*
	H4	5.885*	5.961*
GalNAc (E)	H1	4.593	4.701
	H2	4.027	4.066
	H3	3.949	4.165*
	H4	4.176	4.633
	H5	3.996	∼3.87
	H6	4.218	∼3.80
	H6′	4.242	∼3.80
	NAc	2.063*	2.123*
UA (D)	H1	4.511*	4.886*
	H2	3.387*	3.540*
	H3	3.587	3.887
	H4	3.745	4.115*
	H5	3.709	4.728*
GalNAc (C)	H1	4.618	4.670
	H2	4.031	4.054
	H3	3.860	4.024
	H4	4.187	4.675
	H5	3.984	∼3.87
	H6	4.218	∼3.80
	H6′	4.242	∼3.80
	NAc	2.019*	2.082*
UA (B)	H1	4.558*	4.975*
	H2	3.451*	3.661
	H3	3.646*	4.052
	H4	∼3.754	4.132*
	H5	∼3.754	4.634
GalNAc-ol (A)	H1	3.718	3.690
	H1′	3.789	3.722
	H2	4.388*	4.260
	H3	4.076	4.280
	H4	3.573	4.451*
	H5	4.300*	4.013*
	H6	4.076	∼3.677
	H6′	4.085	∼3.677
	NAc	2.054*	2.037*

Figure 6. Partial 600-MHz gradient-COSY-45 spectrum for CS060606 at 43 °C. The spectral width was 1750.7 Hz, and 24 acquisitions for each of 1024 increments were sampled into 1024 complex points. The array was zero-filled to 2048 × 2048 complex points and Fourier transformed in each dimension after application of a 5% offset (sinebell)² window function. CS060606 has the structure: ΔUA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S(β1–4)GlcA(β1–3)GalNAc6S-ol. F E D C B A

Figure 7. Partial 600-MHz gradient-COSY-45 spectrum for DS040404 at 43 °C. The spectral width was 1750.7 Hz, and 56 acquisitions for each of 512 increments were sampled into 1024 complex points. The array was zero-filled to 2048 complex points in t₂ and Fourier transformed after application of a 10% offset sinebell window function. Data were then extended to 1024 points in t₂ by forward linear prediction with order 12 before application of a 10% offset sinebell window function, zero filling to 2048 points and Fourier transformation. DS040404 has the structure: ΔUA(β1–3)GalNAc4S(β1–4)L-IdoA(α1–3)GalNAc4S(β1–4)L-IdoA(α1–3)GalNAc4S-ol. F E D C B A

Both of the hexasaccharides showed the expected presence of three N-acetyl methyl singlet resonances. Other regions of the spectra were complex, but the H1 and H4 sites in the ΔUA nonreducing terminal rings (F) were readily assignable; 2D correlations to H2 and H3 indicated that for these unsaturated residues the observed signal positions were closely similar to the corresponding ring D values in the tetrasaccharides CST0606 and DST0404 [16]. In a similar manner, the positions of signals corresponding to the GalNAc-ol (residue A) reducing termini are all found to correspond closely to those of the parallel tetrasaccharides and those for ring B, residing between a uronic acid and a GalNAc-ol unit in both the tetrasaccharides and the hexasaccharides, are likewise little perturbed. The other ring likely to exhibit only small differences relative to the corresponding tetrasaccharide is E (C in the smaller oligomer). This is indeed found to be the case; only for H1 of ring E in CS060606 is there a relatively strong change, of −0.07 p.p.m., relative to H1 of ring C in CST0606. For all of the other ring E signals in both hexasaccharides the movements are marginal.

The remaining two pairs of residues may now be assigned; the uronic acids at position D both comprise a set of five coupled spins. The GalNAc rings located at C are present as the typical seven-spin systems, and both are strongly second order, even at 600 MHz, which is typical for these carbohydrate oligomers. Rings C and D are located in regions of the hexasaccharides that are beginning to approximate the environments of polymeric chain segments, therefore the shift values observed are also becoming closer to those observed in macromolecular species. These are important data to take forward to the study of intact CS/DS chains.

It is becoming clear that a full understanding of the structure-function relationships of the chondroitin/dermatan sulfates will require detailed data on sulfation sequence. It is not sufficient to gain data on the average ratio of 4-sulfation to 6-sulfation in a population of chains. It is likely that many functions will be associated with domains or capping sequences of specific structure, and further studies of specific oligosaccharides from CS/DS chains are required to enhance our understanding of the biological activities of CS and DS.

Experimental Procedures

Materials

A Mono-Q 10/10 column was from Pharmacia (Uppsala, Sweden), the Spherisorb S5 SAX column was from Phase Separations Ltd (Deeside, Clwyd, UK), the Toyapearl HW-40 s resin was from Anachem (Luton, UK) and the Bio-Gel P2 resin was from Bio-Rad (Watford, Herts., UK). Papain was from Sigma Chemical Co. (Poole, Dorset, UK), chondroitin ABC endolyase (protease free; Proteus vulgaris; EC 4.2.2.4) was from Seikagaku Corp. (Tokyo, Japan) via ICN Biomedicals Ltd (High Wycombe, Bucks., UK), and lithium perchlorate (ACS grade) and piperazine were from Aldrich Chemical Co. (Gillingham, Dorset, UK). All other chemicals were of analytical grade.

Isolation of CS from cartilage and DS from lung

CS was isolated from fresh articular and tracheal cartilages as previously described [4,5], and DS was isolated from bovine lung as previously described [22]. Briefly, the diced cartilage, or lung, was digested by papain (1 U per 100 mg tissue) in 0.1 m sodium acetate, pH 6.8, with 2.4 mm EDTA and 10 mm cysteine HCl, added just before digestion, for 24 h at 65 °C. The GAGs were precipitated from the soluble fraction by the addition of 4 vol. ethanol, and the solution was cooled to 6 °C and allowed to stand overnight. The precipitate was resuspended in a minimum volume of 50 mm sodium acetate, and the CS, or DS, precipitated by the dropwise addition of 2 vol. ethanol while the solution was stirred. The solution was again cooled to 6 °C and allowed to stand overnight before recovery of the CS, or DS, rich precipitate, which was dialyzed overnight against distilled water and lyophilized.

The CS, or DS, chains were released from the attached amino acids by β-elimination with 0.05 m NaOH containing 1 m sodium borohydride at 45 °C for 48 h [27]. The reaction was terminated by the careful addition of 1 m acetic acid, and the solution dialyzed extensively against distilled water and lyophilized.

Purification of CS/DS on Mono-Q

CS/DS was separated from any remaining non-GAG material using a Mono-Q (10/10) column (1 cm × 10 cm) in a gradient of 2–500 mm LiClO₄/10 mm piperazine, pH 5, at a flow rate of 1 mL·min⁻¹. The elution of bound chains was monitored online by measuring A₂₃₂.

Depolymerization of GAGs

Aliquots, 1–10 mg, of CS or DS chains were depolymerized with 1 U per 100 mg chondroitin ABC endolyase at 20 mg·mL⁻¹ in 0.1 m Tris/HCl, pH 8, at 37 °C for 15 h. The enzyme was inactivated by heating at 100 °C for 1 min, and the oligosaccharides generated were reduced by the addition of 25 mm NaBH₄.

Isolation of oligosaccharides

Reduced oligosaccharides were subjected to SEC on a Toyapearl HW40s column (50 cm × 1 cm) eluted in 0.5 m ammonium acetate at 0.4 mL·min⁻¹, the eluate being monitored by measuring A₂₃₂. Disaccharides, trisaccharides, tetrasaccharides and hexasaccharides were separately pooled as previously described [4], subjected to repeated lyophilization, and then stored at −20 °C.

The individual oligosaccharides were purified, from the CS trisaccharide, tetrasaccharide and hexasaccharide pools recovered after HW40 SEC, by strong anion-exchange (SAX) chromatography as previously described [6,16,20]. In each case a 10-mg aliquot of oligosaccharide was resuspended in 500 µL 2 mm LiClO₄, pH 5.0, and chromatographed on a Spherisorb S5 column (25 cm × 1 cm) at 2 mL·min⁻¹. Bound material was eluted by a linear gradient of 2 mm LiClO₄ (buffer A) to 250 mm LiClO₄ (buffer B), pH 5.0, according to the following gradient profile: after a 10-min isocratic phase of 100% buffer A, a gradient of 0–100% buffer B was introduced over 240 min, followed by 10 min of 100% buffer B. The column eluate was monitored online at 232 nm or, in the case of trisaccharides, at 206 nm. Individual fractions were pooled, desalted by SEC on a column of Bio-Rad P2 resin (1 × 12 cm) running at 0.4 mL·min⁻¹ and 50 °C, and then lyophilized.

Trisaccharide preparation by chemical removal of unsaturated chain termini from tetrasaccharides

For removal of the nonreducing terminal unsaturated uronic acids [26], an aliquot of reduced tetrasaccharide mixture was incubated at room temperature with 500 µL 35 mm mercuric acetate, pH 5, prepared as previously described [26]. After 1 h, excess reagent was removed by mixing with 2 mL Dowex AG-50X (H⁺ form) which had previously been washed with 5 mL 5% HCl followed by 50 mL distilled water. The oligosaccharides were separated from the resin by centrifugation through a 0.45-µm nylon filter, and the resin was subsequently washed with 2 mL distilled water followed by 500 µL 1 m NH₄HCO₃ and the sample lyophilized. The crude trisaccharides were purified by SAX chromatography on a Spherisorb S5 column as described above, and the purified trisaccharides were desalted by SEC on Bio-Gel P2 as described above and then lyophilized.

NMR spectroscopy

Samples were dissolved in 0.5 mL 99.8%²H₂O, buffered to pH 7 with phosphate (10 mm) and referenced with sodium 3-trimethyl[²H₄]propionate as internal standard. After microfiltration through 0.45-µm nylon filters, samples were lyophilized using a rotary concentrator and exchanged several times with 0.5 mL 99.8%²H₂O and then once with 99.96%²H₂O before final dissolution in 0.7 mL 99.96%²H₂O.

Preliminary ¹H-NMR spectra and all ¹³C-NMR spectra were obtained at 400 MHz (100 MHz for ¹³C) on a JEOL GSX400 spectrometer fitted with a 5 mm probe. For 1D ¹³C-NMR spectra, 50 000–250 000 acquisitions were performed, using 60° pulses at 1 s intervals. High-field 1D and 2D correlation (gradient-COSY-45 and TOCSY) ¹H-NMR spectra were determined at 600 MHz on a Varian Unity INOVA spectrometer fitted with a 5 mm triple nucleus probe capable of field-gradient experiments. All spectra were determined at 43 °C, and ¹H and ¹³C chemical shifts are quoted relative to internal sodium 3-trimethylsilyl[²H₄]propionate at 0.0 p.p.m. Experimental details for 2D spectra are given in the legends to the Figures. The C/H-correlation ¹³C-NMR spectrum was obtained using similar conditions to those described previously by Huckerby et al.[28–30].

Spectra were reprocessed for presentation using the software packages Gifa V4.2 [31], obtained from Dr M.-A. Delsuc (University of Montpellier, France), and nmrPipe [32].

Acknowledgements

We thank the Arthritis Research Campaign (arc) (grant N0528) for support, and the Engineering and Physical Sciences Research Council are acknowledged for provision of 600-MHz NMR facilities at the University of Edinburgh.