Chondroitinase-mediated Degradation of Rare 3-O-Sulfated Glucuronic Acid in Functional Oversulfated Chondroitin Sulfate K and E*

Chondroitin sulfate K (CS-K) from king crab cartilage rich in rare 3-O-sulfated glucuronic acid (GlcUA(3S)) displayed neuritogenic activity and affinity toward various growth factors like CS-E from squid cartilage. CS-K-mediated neuritogenesis of mouse hippocampal neurons in culture was abolished by digestion with chondroitinase (CSase) ABC, indicating the possible involvement of GlcUA(3S). However, identification of GlcUA(3S) in CS chains by conventional high performance liquid chromatography has been hampered by its CSase ABC-mediated degradation. To investigate the degradation process, an authentic CS-E tetrasaccharide, Δ4,5HexUA-GalNAc(4S)-GlcUA(3S)-GalNAc(4S), was digested with CSase ABC, and the end product was identified as GalNAc(4S) by electrospray ionization mass spectrometry (ESI-MS). Putative GalNAc(6S) and GalNAc(4S,6S), derived presumably from GlcUA(3S)-GalNAc(6S) and GlcUA(3S)-GalNAc(4S,6S), respectively, were also detected by ESI-MS in the CSase ABC digest of a CS-E oligosaccharide fraction resistant to CSases AC-I and AC-II. Intermediates during the CSase ABC-mediated degradation of Δ4,5HexUA(3S)-GalNAc(4S) to GalNAc(4S) were identified through ESI-MS of a partial CSase ABC digest of a CS-K tetrasaccharide, GlcUA(3S)-GalNAc(4S)-GlcUA(3S)-GalNAc(4S), and the conceivable mechanism behind the degradation of the GlcUA(3S) moiety was elucidated. Although a fucose branch was also identified in CS-K, defucosylated CS-K exhibited greater neuritogenic activity than the native CS-K, excluding the possibility of the involvement of fucose in the activity. Rather, (3S)-containing disaccharides are likely involved. These findings will enable us to detect GlcUA(3S)-containing disaccharides in CS chains to better understand CS-mediated biological processes.

Chondroitin sulfate (CS) 4 is a ubiquitous component of the cell surface and extracellular matrix, belongs to the glycosaminoglycan (GAG) family, and participates in diverse biological processes such as cell growth, neuronal development, and viral invasion (1)(2)(3)(4). The ability of CS to regulate these processes is attributed to its complex structure, which arises from extensive modifications of a nonsulfated precursor with the structure (-4GlcUA␤1-3GalNAc␤1-) n by specific modifying enzymes such as sulfotransferases and epimerases (for review, see Ref. 5). The resulting sulfated chains show enormous structural diversity depending on the type of cell and tissue.
To evaluate the biological significance of GlcUA(3S)-containing structures, oligosaccharides containing GlcUA(3S) structures were previously isolated from CS-K derived from king crab cartilage after digestion with hyaluronidase (11), because preliminary studies by Seno et al. (12) showed that a CS-K preparation from king crab cartilage was rich in GlcUA(3S)-containing disaccharides. However, GlcUA(3S)containing disaccharides cannot be quantified using conventional high performance liquid chromatography (HPLC) because of their unexpected degradation following digestion with chondroitinase (CSase) ABC (11). Thus, GlcUA(3S)-containing disaccharides might have been overlooked in CS chains derived from various tissues including mammalian samples after CSase ABC treatment. Therefore, it is essential to re-evaluate the disaccharide composition of oversulfated CS chains, which is estimated currently by means of CSase ABC treatment. With the growing interest in the fine structure of CS chains (4), there is a need to understand the fate of GlcUA(3S)-containing disaccharides upon digestion with CSase ABC. This will enable us to decipher the domain structures of CS and to understand their interactions with growth/neurotrophic factors by which CS coordinates the diverse aspects of biological events and could reveal new therapeutic opportunities.
In this study, we showed the binding activity toward growth/ neurotrophic factors and neurite outgrowth-promoting (NOP) activity of CS-K. In addition, we developed a method of detecting the GlcUA(3S)-containing disaccharides of CS-E and CS-K rich in GlcUA(3S) by monitoring intermediates after CSase ABC digestion and elucidated the possible mechanism by which the disaccharides are degraded. Preliminary results have been reported in an abstract form (13).
Neurite Outgrowth Promotion Assays-Cultures of mouse hippocampal neurons were established from embryonic day 16 animals as previously described (16,17). Briefly, 2 g/well of the CS-K and CS-E preparations were individually coated on to coverslips precoated with poly-DL-ornithine (P-ORN) (Sigma) at 4°C overnight. To investigate the structural characteristics of CS-K responsible for the neuritogenic activity, an aliquot (10 g as GAG) was digested with 10 mIU of CSase ABC, a mixture of CSases AC-I and AC-II, or ␣-L-fucosidase, and a 2-g aliquot of each digest was coated onto the coverslips precoated with P-ORN. Although the binding efficiency of oligosaccharides in the enzyme digests to P-ORN has not been evaluated, a recent study by Sotogaku et al. (18) showed the efficient binding of diand tetrasacchrides to P-ORN. Control experiments were carried out using inactivated enzymes. The hippocampal neuronal cells freshly isolated from E16 mouse embryos were suspended in Eagle's minimum essential medium containing supplements described previously (16,17). Subsequently, the cells were seeded on coverslips at a density of 20,000 cells/cm 2 and allowed to grow in a humidified atmosphere for 24 h at 37°C, 5% CO 2 . Thereafter, the cells were fixed using 4% (w/v) paraformaldehyde for 30 min at room temperature, and the neurites were visualized by immunochemical staining using anti-microtubule-associated protein-2 (Lieco Technologies Inc., St. Louis, MO) and anti-neurofilament (Sigma) as described previously (19). The antibodies were then detected using a Vectastain ABC kit (Vector Laboratories Inc., Burlingame, CA) with 3,3Ј-diaminobenzidine as a chromogen. The stained cells on each coverslip were scanned and digitalized with a ϫ20 objective lens on an optical microscope (BH-2; Olympus, Tokyo, Japan) equipped with a digital camera (HC-300Z/OL; Olympus). One hundred cells with at least one neurite longer than the cell body were chosen at random to determine the length of the longest neurite using morphological analysis software (Mac SCOPE; Mitani Corp., Tokyo, Japan). At least three independent experiments per parameter or condition were carried out.
Interaction Analysis-The interaction of various growth/ neurotrophic factors with CS-K was examined using a BIAcore J system (BIAcore AB, Uppsala, Sweden). The CS-K-immobilized sensor chip was prepared as reported earlier (20). For the kinetic analysis, various concentrations of growth/neurotrophic factors were injected onto the surface of the sensor chip in running buffer, pH 7.4 (HBS-EP; BIAcore AB), containing 10 mM HEPES, 0.15 M NaCl, 3 mM EDTA, and 0.005% (w/v) Tween 20. The flow rate was kept at a moderate speed (30 l/min) as the manufacturer recommended. Each growth/neurotrophic factor was allowed to interact with the CS-K-immobilized sensor chip for 2 min of association and dissociation. Before each injection, the base-line stability was achieved by injecting 1 M NaCl for 2 min. The kinetic parameters were evaluated with BIA evaluation software 3.1 (BIAcore AB) using a 1:1 binding model with mass transfer.
Enzyme Digestion and Fluorophore Labeling-CSase ABC digestion was carried out using 1 g of GAG or 1nmol of tetrasaccharide and 5 mIU of the enzyme in a total volume of 30 l of the appropriate buffer at 37°C for 60 min as described (21) unless otherwise specified. To generate the oligosaccharides, which are resistant to CSases AC-I and AC-II, CS-E was subjected to digestion with a mixture of CSases AC-I and AC-II in a 50 mM Tris-HCl buffer, pH 7.3, at 37°C for 60 min as described previously (15). ␣-L-Fucosidase treatment of CS-K (50 g as GAG) was performed using 50 mIU of the enzyme in a total volume of 500 l of 0.05 M citrate buffer, pH 4.5, for 120 min at 37°C (22). After incubation, each reaction mixture was boiled at 100°C for 1 min, cooled to room temperature, vacuum-dried, and derivatized with 2AB as described previously (21). These products were used for the structural analysis.
HPLC-The analysis of monosaccharides/disaccharides/ oligosaccharides was carried out by HPLC on an amine-bound silica PA-03 column (YMC Co., Kyoto, Japan) using a linear gradient of NaH 2 PO 4 from 16 to 798 mM at a flow rate of 1 ml/min at room temperature as described (21). Because ⌬HexUA-GalNAc (⌬O-unit) and GalNAc(4S) were co-eluted with the linear gradient of NaH 2 PO 4 , the conditions for HPLC were modified to separate the peaks. Namely, isocratic conditions with 16 mM of NaH 2 PO 4 were employed for the first 20 min followed by a linear gradient from 16 to 798 mM NaH 2 PO 4 over 70 min at a flow rate of 1.0 ml/min at room temperature. To fractionate the CSase ABC digest, an aliquot was subjected to gel filtration chromatography on a Superdex peptide TM HR column (10 ϫ 300 mm; Amersham Biosciences) using 0.2 M NH 4 HCO 3 as an eluent at a flow rate of 0.4 ml/min. Elution was monitored using a RF-10A XL fluorometric detector (Shimadzu Co., Kyoto, Japan) with excitation and emission wavelengths of 330 and 420 nm, respectively. Each fraction was collected and desalted by repeated evaporation.
ESI-MS-The mass spectral analysis was performed with mass spectrometers operated in negative ion mode (API-3000, PE Biosystems, CA, at Kobe Pharmaceutical University; or JMS-700TZ, JEOL Ltd., Tokyo, Japan, at the Center for Instrumental Analysis, Hokkaido University) and equipped with an electrospray ion source. The CS-derived disaccharides were used to optimize the acquisition parameters. The fractionated and desalted samples were dissolved in 50% acetonitrile containing 0.1% acetic acid in water and introduced into an ion source at a flow rate of 5 l/min. The nebulizer pressure was set to 8 p.s.i. at room temperature. The samples were examined over a mass range of m/z 50 -650. 500-MHz 1 H NMR Spectroscopy-An aliquot of an unidentified fraction (40 nmol as disaccharides) derived from CS-K after CSase ABC digestion was repeatedly exchanged in D 2 O with intermediate lyophilization. 1 H NMR spectra were measured on a Varian VNMRS-500 ( 1 H: 499.7 MHz) with a Nano gHX probe (sample volume 40 l) at a spinning rate of 2,000 Hz. One-dimensional spectrum with presaturation of the HOD signal, correlation spectroscopy, and two-dimensional homonuclear Hartmann-Hahn spectroscopy were measured at 26 and 55°C. Chemical shifts are given relative to sodium 4,4-dimethyl-4-silapentane-1-sulfonate but were actually measured indirectly relative to acetone (␦ 2.225) in D 2 O (23-25).

RESULTS
NOP Activity of the CS-K Preparations-CS-E derived from squid cartilage and containing a GlcUA(3S)-disaccharide is involved in several intriguing biological events such as the binding of growth factors and differentiation of neurons (8 -10, 26). To understand the possible involvement of GlcUA(3S)-containing disaccharides in NOP activity, a CS-K preparation derived from king crab cartilage rich in K-unit was assessed for NOP activity. Hippocampal neuronal cells from E16 mouse embryos were cultured on CS-K or CS-E (a positive control) immobilized onto coverslips precoated with P-ORN or on P-ORN alone (a negative control). Both CS-K and CS-E promoted neurite outgrowth as compared with a P-ORN-control ( Fig. 1, A-C). Notably, the NOP activity of CS-K was slightly yet significantly stronger than that of CS-E, being principally axonic in nature, and similar in this respect to what was observed with CS-E (Fig. 1, B and E). The greater NOP activity of CS-K compared with CS-E may reflect the importance of the GlcUA(3S)-containing structures in addition to E-unit (GlcUA␤1-3GalNAc(4S,6S)). CS-E from squid cartilage consists of ϳ10% GlcUA(3S)-containing disaccharides in addition to the major (66%) oversulfated disaccharide (E-unit) (10). In E, the mean length of the longest neurite was measured for 100 randomly selected neurons cultured on various substrates (see "Experimental Procedures"). The values obtained from two independent experiments are expressed as the means Ϯ S.E. Mann-Whitney's U test was used to evaluate the significance of differences between means (**, p Ͻ 0.01). DECEMBER (20,27). Thus, signaling of various growth/neurotrophic factors appears to be involved in the NOP activity of CS-K. Hence, we analyzed the interaction using the BIAcore system with various growth/neurotrophic factors expressed in the brain during embryonic development. The purified CS-K preparation was biotinylated and immobilized on the streptavidinprecoated sensor chip. To determine the association and dissociation rate constants (k a and k d ) as well as the dissociation equilibrium constants (K d ), various growth/neurotrophic factors (Table 1) at different concentrations were injected individually onto the surface of a sensor chip coated with CS-K. The overlaid sensorgrams shown in Fig. 2, were analyzed collectively using the 1:1 Langmuir binding model with mass transfer to calculate kinetic parameters, which are summarized in Table  1. The direct binding of CS-K to MK, PTN, FGF-18, HGF, BDNF, and GDNF was demonstrated. FGF-18, HGF, PTN, and MK displayed K d values in the low nanomolar (nM) range, signifying their strong affinity for CS-K (Table 1). In contrast, BDNF and GDNF showed weaker affinity as reflected in the K d values. CS-K bound MK, PTN, and FGF-18 more efficiently than did CS-E (9).

3-O-Sulfated Glucuronic Acid in CS-E and CS-K
Identification of the End Product of the Digestion of a GlcUA(3S)-containing Disaccharide with CSase ABC-The above results clearly showed the functional properties of CS-K. However, the quantification of GlcUA(3S)-containing disaccharides is not feasible using conventional anion exchange HPLC because of degradation upon digestion with CSase ABC (11). Therefore, we have isolated oligosaccharides containing GlcUA(3S) from CS-K after digestion with testicular hyaluronidase as models and identified them using 1 H NMR spectroscopy (11). Here we investigated the probable intermediates/ end products because of the degradation of GlcUA(3S)containing disaccharides upon CSase ABC digestion. The structurally defined ⌬A-K tetrasaccharide ⌬HexUA-GalNAc(4S)-GlcUA(3S)-GalNAc(4S) isolated from CS-E (15) was used. It was subjected to CSase ABC digestion and labeled with the fluorophore 2AB. HPLC of the CSase ABC digest gave rise to ⌬A (⌬HexUA␣1-3GalNAc(4S)) in addition to an unidentified peak eluting shortly after ⌬O (⌬HexUA␣1-3GalNAc) in a molar ratio of 46:54 (Fig. 3A). To identify this peak, an

TABLE 1 Kinetic parameters for the interaction of growth or neurotrophic factors with immobilized CS-K
The k a , k d , and K d values were determined using a 1:1 Langmuir binding model with mass transfer as described under "Experimental Procedures." The value for each growth factor is expressed as the mean Ϯ S.E. of five different concentrations.

3-O-Sulfated Glucuronic Acid in CS-E and CS-K
aliquot of a 2AB-labeled CSase ABC digest of ⌬A-K was fractionated on a Superdex peptide TM column, yielding fractions A and B as shown in Fig. 3B. Each fraction was collected, desalted, and subjected to ESI-MS in negative ion mode. The ESI-MS spectrum of fraction A showed a major signal at m/z 578, consistent with the molecular mass of a monosulfated disaccharide ⌬A-unit (data not shown). Interestingly, fraction B afforded a molecular ion at m/z 420 corresponding to the molecular mass of a monosulfated N-acetylhexosamine labeled with 2AB (Fig.  3C). Thus, it is reasonable to assign this product as GalNAc(4S), resulting from the degradation of K-unit (GlcUA(3S)␤1-3GalNAc(4S)) by the action of CSase ABC. Identification of Intermediates during the CSase ABC-mediated Degradation of K-unit-To elucidate the mechanism by which K-unit is degraded to GalNAc(4S) after CSase ABC treatment and to identify the intermediates during the digestion, a structurally defined K-K tetrasaccharide, GlcUA (3S)␤1-3GalNAc(4S)␤1-4GlcUA(3S)␤1-3GalNAc(4S), which was isolated from king crab cartilage (11), was partially digested with CSase ABC and subjected to ESI-MS. As expected, the ESI-MS profile in the Q1 mode of the CSase ABC digest showed no molecular ion signal corresponding to ⌬K (⌬HexUA(3S)␣1-3GalNAc(4S)) (molecular weight, 539) derived from the reducing end (data not shown). To evaluate possible intermediates generated from the degradation of ⌬K, a precursor ion mode was used. Results with ⌬A-K showed GalNAc(4S) as an end product after the degradation of ⌬HexUA(3S)␣1-3GalNAc(4S) followed by CSase ABC digestion. Because GalNAc(4S) contains a sulfate group, the precursor ions for the sulfate group at m/z 97 were measured in the negative ion mode (Fig. 4A). The ESI-MS analysis afforded a signal predominantly at m/z 300.3, corresponding to the molecular mass of GalNAc(4S). To detect signals of the intermediates generated after CSase ABC digestion, the precursor ions for the GalNAc(4S) signal at m/z 300 were measured. Four major signals at m/z 357.9, 382.0, 416.0, and 439.8 were detected as possible intermediates (Fig. 4B). For their identification, the product ion MS/MS mode was used, and particular ions were selected for collision-activated dissociation. The product ion mode at m/z 382 displayed a major signal at m/z 300.3 corresponding to GalNAc(4S) and a signal at m/z 358.4 (Fig. 4C). The signal at m/z 382.0 was assigned to 1-hydroxy-1-[GalNAc(4S)-3-oxy]-but-3-yn-2-one (S4) (Scheme 1). The signal at m/z 439.8 can be assigned to 5,6-dihydroxy-6-[GalNAc(4S)-3-oxy]-hex-1-yne-3,4-dione (S2), whose product ion gave a major signal at m/z 300.1 and detectable peaks at m/z 381.9 and 96.8 (Fig. 4D). Based on these results, we propose a possible mechanism for the degradation of K-unit to GalNAc(4S) via ⌬HexUA(3S)␣1-3GalNAc(4S) as depicted in Scheme 1.   DECEMBER 21, 2007 • VOLUME 282 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 36899
CS-E sample, an oligosaccharide fraction, which is supposed to contain GlcUA(3S), was first prepared by digesting CS-E exhaustively with a mixture of CSases AC-I and AC-II, because N-acetylgalactosaminidic linkages to GlcUA(3S) are resistant to these enzymes. This oligosaccharide fraction was subjected to digestion with CSase ABC, labeled with 2AB, and fractionated on a Superdex peptide TM column as shown in Fig. 5B. Fractions E1-E5 were collected separately, and each fraction was subjected to ESI-MS, which showed an unidentified molecular ion signal for fraction E1 in the negative mode, whereas fractions E2 and E3 showed signals at m/z 658 and 578 corresponding to the molecular masses of disulfated and monosulfated unsaturated CS disaccharides labeled with 2AB, respectively (data not shown). In contrast, fraction E4 eluted at the position of a 2AB-labeled nonsulfated disaccharide showed a molecular ion signal at m/z 522 (Fig. 5C) instead of the one expected at m/z 498. This molecular mass corresponds to the monosodiated disulfated GalNAc labeled with 2AB presumably derived from M-unit. Although the disulfated GalNAc and ⌬O disaccharide labeled with 2AB have similar masses, 501 and 499, respectively, the signal at m/z 522 can be explained by GalNAc(4S,6S), which is stabilized by a sodium ion via two sulfate groups. Based on this observation, it is reasonable to assign this product as GalNAc(4S,6S) resulting from the degradation of ⌬A-M or ⌬E-M after CSase ABC digestion (Fig. 5C). Fraction E5 showed a major signal at m/z 420 (data not shown), which corresponds to the molecular mass of a monosulfated N-acetylhexosamine labeled with 2AB, indicating that the product is either GalNAc(4S) or GalNAc(6S) or both, resulting from the degradation of ⌬A-K and ⌬E-K or ⌬A-L.

3-O-Sulfated Glucuronic Acid in CS-E and CS-K
GalNAc(4S) from ⌬A-K and ⌬E-K after digestion with CSase ABC.

Identification of GalNAc Derivatives by Anion Exchange
Chromatography-To quantify the amount of GlcUA(3S)-containing disaccharide in CS-E, the elution positions of mono-and disulfated GalNAc residues were next determined under conventional HPLC conditions. An aliquot of fraction E4 or E5 (Fig. 5B) was analyzed by anion exchange HPLC. The HPLC profile of fraction E4 showed a single major peak X near the elution positions of ⌬C and ⌬A as shown in Fig. 6A. Co-chromatographic analysis with a mixture of standard 2AB-labeled ⌬CS disaccharides and GalNAc(4S) showed fraction E4 eluting distinctively between ⌬C and ⌬A as shown in Fig. 6B. The molar percentage of this peak corresponded to 29% of all disaccharides in the oligosaccharide fraction resistant to a mixture of CSases AC-I and AC-II, although the value is arbitrary because no authentic GalNAc(4S,6S) is available. The detection of GalNAc(4S,6S) derived presumably from ⌬A-M and ⌬E-M units is reported here for the first time. In contrast, the anion exchange HPLC profile of fraction E5 showed two peaks (data not shown). On co-chromatography with a mixture of standard unsaturated disaccharides and GalNAc(4S), one peak eluted prior to ⌬O and the other peak co-eluted with GalNAc(4S) (data not shown). They most likely correspond to GalNAc(6S) and GalNAc(4S), respectively, because an N-acetylgalactosaminidic linkage to GlcUA sulfated at either C2 or C3 is resistant to CSases AC-I and AC-II. This assumption was sup-

FIGURE 5. Analysis of the CSase ABC digest of the CS-E-derived oligosaccharide fraction resistant to a mixture of CSases AC-I and AC-II.
A, the oligosaccharide fraction resistant to a mixture of CSases AC-I and AC-II, which was prepared from CS-E as described under "Experimental Procedures," was subjected to CSase ABC digestion, labeled with the flurophore 2AB, purified to remove excess 2AB-derivatizing reagents, and analyzed by anion exchange chromatography as described in the legend to Fig. 3. B, an aliquot of the 2AB-deravative of the CSase ABC digest was fractionated on a column of Superdex peptide TM using 0.2 M NH 4 HCO 3 as an eluent at a flow rate of 0.4 ml/min as described in the legend to Fig. 3. C, ESI-MS analysis of fraction E4 (from panel B) in negative ion mode. For details, refer to "Experimental Procedures."  DECEMBER 21, 2007 • VOLUME 282 • NUMBER 51 JOURNAL OF BIOLOGICAL CHEMISTRY 36901 ported by the detection of a lone molecular ion at m/z 420 for fraction E5 by ESI-MS (data not shown). Thus, the combination of HPLC and ESI-MS allowed us to detect a putative GalNAc(6S) derived from ⌬A-L and GalNAc(4S,6S) derived from ⌬A-M and ⌬E-M in addition to GalNAc(4S) derived from ⌬A-K and ⌬E-K after CSase ABC digestion, in a molar ratio of 1.00: 0.98: 2.42 in CS-E. Because the resolution between the GalNAc(4S) and ⌬O disaccharide was low under conventional HPLC conditions, where a linear gradient of NaH 2 PO 4 was used over 0 -60 min, the conditions were modified to resolve the peaks as described under "Experimental Procedures." Thus, a better separation was achieved using an isocratic elution with low salt during the first 20 min (Fig. 6B).

3-O-Sulfated Glucuronic Acid in CS-E and CS-K
Analysis of the Disaccharide Composition of CS-K-To ascertain the results obtained with CS-E, the disaccharide composition of the CS-K preparation was also examined. The preparation was exposed to CSase ABC and then labeled with the fluorophore 2AB, which showed three peaks, a, b, and c (Fig.  7A). Peak c accounted for 48% of the sum of peaks a-c and corresponds to peak K2 on gel filtration (Fig. 7B). ESI-MS of this fraction showed a major signal at m/z 578, corresponding to the molecular mass of a 2AB-labeled monosulfated disaccharide ⌬A (data not shown). Peak a, which was eluted at the position of the authentic GalNAc(4S) and was derived from K-unit, accounted for 33% of the sum of peaks a-c, corresponding to peak K3 on gel filtration (Fig. 7B). The ESI-MS profile of this peak showed a lone molecular ion at m/z 420 (data not shown), consistent with the previous finding that CS-K predominantly consisted of K-units (11). Because mono-or di-sulfated Gal-NAc residues have been reported as nonreducing terminal modifications in CS chains (28 -30), the possibility exists that GalNAc(4S) was released from the nonreducing terminal because of the action of CSase ABC. Hence, the CS-K preparation was digested with a mixture of CSases AC-I and AC-II, which gave no significant peak at the position where GalNAc(4S) was eluted upon anion exchange HPLC (data not shown), indicating that the CS-K preparation contains no significant amount of GalNAc(4S) at the nonreducing ends of the polysaccharide chains. The unidentified peak b (Fig. 7A), which accounted for 18.6% of the sum of peaks a-c, was eluted between the positions of ⌬C and ⌬A. To characterize the compound in this peak, an aliquot of the 2AB derivative of the CSase ABC digest was fractionated on a Superdex peptide TM column into peaks K1-K3 as shown in Fig. 7B. Peak K1, corresponding to peak b in Fig. 7A, was collected and further subjected to ESI-MS, which showed a molecular ion at m/z 724.2 (data not shown), corresponding to a 2AB derivative of monosulfated unsaturated CS disaccharide substituted with a methyl pentose (data not shown). This prompted us to characterize the structure of this compound by 1 H NMR spectroscopy.
Identification of a Fucosylated Trisaccharide in CS-K-To characterize peak K1 (Fig. 7B), an aliquot of the CSase ABC digest of CS-K was fractionated on a Superdex peptide TM column (data not shown), and the fraction corresponding to the peak was subjected to 1 H NMR spectroscopy, which allowed us to identify this component as a monosulfated CS disaccharide fucosylated at the C3 position of GlcUA, namely Fuc(␣1-3)⌬HexUA(␣1-3)GalNAc(4-O-sulfate), as shown in supplemental Table S1 and supplemental Fig. S1. Although the trisaccharide structure Fuc(␣1-3)⌬HexUA(␣1-3)GalNAc(4-O-sulfate) was previously detected in the CSase ABC digest of the pentasaccharide, GlcUA(3S)-GalNAc(4S)-(Fuc-)GlcUA-GalNAc(4S) isolated from CS-K (24), the NMR data for the fucosylated trisaccharide was obtained here for the first time. Based on the peak area, ϳ18% of the disaccharide units appear to be modified by fucosylation in the CS-K preparation.
Investigation of the Involvement of Fucosylation in CS-K-mediated Neuritogenesis-Because CS-K contained a significant proportion of Fuc, we investigated the possible involvement of fucsoylation in the NOP activity of CS-K. The CS-K was subjected to ␣-L-fucosidase treatment. The disaccharide analysis of the ␣-L-fucosidase-treated CS-K after digestion with CSase ABC and 2AB labeling showed a 43% decrease in Fuc content (data not shown). The partially defucosylated CS-K showed greater NOP activity toward hippocampal neurons than native CS-K (Fig. 1, D and E). These results suggest that GlcUA(3S)containing K-units rather than Fuc residues in CS-K contribute to the NOP activity. A, the CS-K polysaccharide preparation purified from gill cartilage of king crab was digested with CSase ABC, labeled with 2AB, and analyzed by anion exchange HPLC on an amine-bound silica column using a linear gradient of Na 2 HPO 4 from 16 mM to 798 mM over 60 min, as indicated by dashed lines. The peaks obtained before 10 min were due to 2AB-derivatizing reagents. B, the CSase ABC digest of the CS-K labeled with 2AB was analyzed on a Superdex peptide TM column as described in the legend to Fig. 3B. The asterisk indicates a ghost peak observed in the mock injections.

DISCUSSION
In this study, we demonstrated NOP activity and strong affinity toward growth/neurotrophic factors expressed in the brain during development for a CS-K preparation rich in GlcUA(3S)-containing disaccharides. Both CS-K and CS-E preparations are characterized by GlcUA(3S)-containing disaccharides (10,11). Although CS-E contains more heterogeneous GlcUA(3S)-containing disaccharides than CS-K (11), the two exhibited similar neuronal cell morphological features: a relatively small cell soma and a few prominent long neurites. Most of the CS variants derived from marine animals, which are neuritogenic in vitro, contain oversulfated disaccharide units such as D or E. However, CS-K, which is devoid of D or E units, still exhibited potent neuritogenic activity, signifying the role of GlcUA(3S)-containing disaccharides. Furthermore, the complete abolishment of NOP activity upon digestion with CSase ABC or with a mixture of CSases AC-I and AC-II indicates that an intact GlcUA(3S)-containing structure of sufficient length is essential for the neuritogenic activity of CS-K chains.
Morphometric analysis of primary neurons grown on the CS-K substrata showed a good correlation between the neuritogenic property and growth/neurotrophic factor binding ability of CS-K. HGF, FGF-18, PTN, MK, BDNF, and GDNF examined here are broadly expressed in the brain and implicated in its development (31)(32)(33)(34)(35). The greater affinity of CS-K, particularly for HGF, FGF-18, PTN, and MK, may indicate that the NOP activity of CS-K involves at least one of these growth factors. Earlier, we demonstrated the involvement of PTN and HGF, through their receptors anaplastic lymphoma kinase and cMet, in the NOP activity of CS/DS hybrid chains isolated from embryonic pig brain and shark liver (27). The strong affinity of CS-K for PTN, HGF, FGF-18, or MK, which is greater than that of CS-E or Hep (9), implies that GlcUA(3S)-containing structures may be present on cell surfaces or in extracellular matrices in the embryonic brain and play a critical role in the regulation of biological functions of various Hep-binding growth factors. However, BDNF and GDNF showed low affinity toward CS-K, indicating that GlcUA(3S)-containing structures may not be involved in the binding of these neurotrophic factors to CS/DS. Earlier, Nandini et al. (36,37) reported the strong affinity of BDNF and GDNF for shark skin DS and CS-H from hagfish notochord containing a significant proportion of the DS structure, which in turn implies that iduronic acid-containing structures may be involved in the CS/DS-mediated BDNF and GDNF signaling pathways. However, precisely how CS/DS chains are involved in BDNF/GDNF signaling remains to be clarified.
In this study, we also demonstrated the degradation of GlcUA(3S)-containing disaccharides by CSase ABC to produce either mono-or disulfated GalNAc residues and proposed a pathway for the degradation process. Previously, we observed that digestion with CSase ABC of tetrasaccharides isolated from CS-K with GlcUA(3S) at the internal position unexpectedly destroyed the disaccharide unit containing GlcUA(3S) (11). This might explain why this structure has never been reported in any CS samples or in mammalian tissues. Therefore, here we developed a method to detect GlcUA(3S)-con-taining structures in CS samples. To understand the mechanism of the degradation of biologically active GlcUA(3S)containing structures after CSase ABC digestion, we used CS-K-or CS-E-derived oligosaccharides as model compounds generated by either sheep testicular hyaluronidase or a mixture of CSases AC-I and AC-II (10,15,24). HPLC and ESI-MS revealed that digestion of the structurally defined tetrasaccharide ⌬A-K, isolated from CS-E, with CSase ABC resulted in the generation of a GalNAc(4S) residue from the GlcUA(3S)-containing disaccharide (K-unit) in addition to the ⌬〈-unit from the nonreducing end (Fig. 3). Therefore, it is essential to reevaluate the disaccharide composition of CS chains from various biological sources, which can be achieved by anion exchange HPLC after CSase ABC treatment as described here or by fluorescence-assisted carbohydrate electrophoresis (38). In fact, fluorescence-assisted carbohydrate electrophoresis should be a good alternative to HPLC, because the separation of mono or di-sulfated GalNAc residues can be clearly achieved (38).
GlcUA(3S) has been detected in glycolipids isolated from human peripheral nerves using the antibody HNK-1 (39,40). The terminal GlcUA(3S) residue is essential for immunoreactivity with HNK-1, and this carbohydrate epitope is also expressed on glycoproteins (41). The HNK-1 epitope associates with neural crest cell migration (42), neuron-to-glial cell adhesion (43), outgrowth of astrocytic processes and migration of cell body (44), and the outgrowth of neurites from motor neurons (45). It has been proposed to serve as a ligand for selectins, which are leukocyte-endothelial cell adhesion molecules (46). Interestingly, a similar epitopic structure has been reported for CS-proteoglycans of mammalian tissues (47). Although its association with CS chains has not been elucidated, these findings suggest that GlcUA(3S) residues may occur on CS chains of proteoglycans in the mammalian nervous system and are potentially involved in the functions of CS polysaccharides. This possibility remains to be investigated.