Novel 70-kDa Chondroitin Sulfate/Dermatan Sulfate Hybrid Chains with a Unique Heterogenous Sulfation Pattern from Shark Skin, Which Exhibit Neuritogenic Activity and Binding Activities for Growth Factors and Neurotrophic Factors*

Chondroitin sulfate (CS) and dermatan sulfate (DS) hybrid chains of proteoglycans are critical in growth factor binding, neuritogenesis, and brain development. Here we isolated CS/DS hybrid chains from shark skin aiming to develop therapeutic agents. Digestion with various chondroitinases showed that both GlcUA- and IdoUA-containing disaccharides are scattered along the polysaccharide chains with an unusually large average molecular mass of 70 kDa. The CS/DS chains were separated into major (80%) and minor (20%) fractions by anion-exchange chromatography. Both fractions had relatively low degrees of sulfation (sulfate/disaccharide molar ratio (cid:1) 1.17 versus 0.87), showing a unique feature compared with the marine CS and DS isolated to date, most of which are oversulfated. They were highly heterogeneous and characterized by multiple disaccharides including GlcUA-GalNAc, GlcUA-GalNAc(6S), GlcUA-GalNAc(4S), IdoUA-GalNAc(4S), GlcUA-GalNAc-(4S,6S), IdoUA-GalNAc(4S,6S), GlcUA(2S)-GalNAc(6S), and/or washing the cuvette surface with phosphate-buffered saline with Tween 20 (running buffer), it was blocked with 20 (cid:3) g of bovine serum albumin. The interaction with growth factors, namely FGF-1, FGF-2, FGF-10, FGF-16, HB-EGF, PTN, and MK, was then studied. A single binding assay consisted of two phases: (i) an association phase obtained as a result of the binding of the growth factor to the immobilized DS preparation following addition of the growth factor in 200 (cid:3) l of the running buffer, and (ii) a dissociation phase characterized by dissocia- tion of the bound ligate initiated by addition of 200 (cid:3) l of the running buffer. The cuvette was regenerated by washing with 200 (cid:3) l of 1 M NaCl. The stirrer speed was maintained at 80% and the temperature at 25 °C throughout the experiment. The distribution of immobilized SS-DS and of the bound growth factor on the surface of the biosensor cuvette was inspected by the examination of resonance scans, which showed that these molecules were distributed on the sensor surface at all times and therefore were not microaggregated. To determine the kinetics of binding, growth factors were applied to the cuvette at varying concentrations in the running buffer, and disso- ciation and regeneration carried out as described earlier. Binding parameters were calculated from the association and dissociation phases of the binding reactions using the FASTfit software (Affinity Sensors). Assay for Anti-factor IIa Activity— Anti-factor IIa activity or anti-Hep cofactor II (HC-II) activity was measured by incubating 1–10 (cid:3) g/ml of the SS-DS preparations in 100 (cid:3) l of 50 m M Tris-HCl buffer, pH 8.3 containing 227 m M NaCl with 60 (cid:3) l of normal human plasma and 40 (cid:3) l of human thrombin (1.2 NIH units/ml) at 25 °C for 1 min in a disposable cuvette (44). The amidolytic activity of thrombin was determined as a change in absorbance at 405 nm recorded for 100 s after the addition of 100 (cid:3) l of 1.9 (cid:3) mol/ml chromozym TH. The rate of change of absorbance was proportional to the remaining thrombin activity. Hep and DS from porcine skin were used as positive controls. Determination of Uronic Acid and Neutral Sugars— Uronic acid was estimated by the carbazole reaction of Bitter and Muir (45). Neutral sugars were estimated by the orcinol method (46).

Dermatan sulfate (DS) 1 proteoglycans are widely distributed in most tissues including skin and have been implicated in various biological processes including cell adhesion (1), proliferation (2), interactions with various growth factors (3), and wound repair (4). DS glycosaminoglycan (GAG) is particularly recognized for its antithrombotic activity, exhibited through a mechanism different from that of heparin (Hep) (5). DS often occurs in co-polymeric form with chondroitin sulfate (CS) consisting of varying proportions of -4GlcUA␤1-3GalNAc␤1-and -4IdoUA␣1-3GalNAc␤1-variably sulfated on both hexuronic acid and GalNAc residues, thus forming a hybrid structure (6). CS and DS have the potential to display enormous structural diversity, comparable to that of HS, thereby embedding multiple overlapping sequences constructed with distinct disaccharide units modified by different patterns of sulfation (7). This heterogeneity is the structural basis for the diverse biological functions of DS (8).
Our laboratory has been elucidating the structural and functional aspects of CS isoforms, showing the importance of CS GAGs from a simple molecule such as chondroitin in the cell division of a nematode (9) to differentially oversulfated CS (10 -12) and DS chains (13) in neural development, thus highlighting the importance of the rare oversulfated disaccharide units, D (GlcUA(2S)-GalNAc(6S)), iD (IdoUA(2S)-GalNAc(6S)), E (GlcUA-GalNAc(4S,6S)), and iE (IdoUA-GalNAc(4S,6S)), of CS and DS in various biological functions, where 2S, 4S, and 6S represent 2-O-, 4-O-, and 6-O-sulfate, respectively, and the i in iE stands for IdoUA. The neuritogenic properties exhibited by CS and DS chains are controversial with both inhibitory (14,15) and promotive effects (16,17) observed for various neurons. The discrepancies are most likely caused by their spatiotemporal distributions (18,19) and developmental changes in structure and function (21,65). The critical importance of IdoUA in neuritogenesis and growth factor binding and the developmental change of its expression in the brain have recently been demonstrated for CS/DS hybrid chains of pig embryos (22).
It has also been our endeavor to look for sources of GAGs with unique structures and prominent activity, which have potential as a therapeutic agent. This pursuit revealed the ability of structurally characterized oversulfated DS, named CS-H, from hagfish notochord (12,23), which promotes neuritogenesis and growth factor binding, as well as the essentiality of both GlcUA-and IdoUA-containing moieties for these activ-ities (25). It is also noteworthy that DS is being used for other applications such as the preparation of artificial tissues (26).
In furthering the understanding of DS in relation to its structure and function, especially in light of the recent finding that relatively low sulfated CS/DS hybrids (the sulfate/disaccharide molar ratio (S/unit) ϭ 0.83ϳ0.84) with a considerable proportion (23ϳ25%) of non-sulfated units and small proportions (1ϳ2%) of oversulfated disaccharide units could promote growth factor binding and neurite outgrowth (22), we looked for other possible sources of CS/DS chains that are less sulfated than CS-H. Toward that end, DS chains with unique structural features and multiple biological activities were isolated from shark skin, which is an industrial waste with an immense potential to be exploited for pharmaceutical purposes. While a classical preliminary work on GAGs from the skin of blue shark and sandbar shark showed oversulfated DS with an S/unit of 1.42-1.62 and 1.26 -1.37, respectively, in addition to hyaluronan (HA) and CS-C by conventional analyses of amino sugars, hexuronic acid, infrared spectrum, rotation etc. (27), no detailed structure or biological activities have been reported. from Streptomyces hyalurolyticus, unsaturated disaccharides and bovine serum albumin were obtained from Seikagaku Corp. (Tokyo, Japan). Chondroitinase B preparations were sometimes contaminated with chondro-4-sulfatase and hence each lot was examined for contamination. In addition, chondroitinase B was also obtained from Sigma. Actinase E was from Kaken Pharmaceutical Co. (Tokyo, Japan). 2-Aminobenzamide (2AB) was purchased from Nacalai Tesque (Kyoto, Japan). Sodium cyanoborohydride (NaBH 3 CN) and 1,9-dimethylmethylene blue (DMMB) were from Aldrich Chemical Co. (Milwaukee, WI). EZ-Link TM biotin-LC-hydrazide was obtained from Pierce. Prepacked disposable PD-10 columns containing Sephadex G-25 (medium) were purchased from Amersham Biosciences (Tokyo, Japan). Sep-Pak plus Accell TM QMA anion-exchange cartridges were from Waters Corp. (Milford, MA). Recombinant human (rh)-midkine (MK) and rh-fibroblast growth factor-1 (FGF-1 or acidic FGF), expressed in Escherichia coli, were from PeproTech EC LTD (London, England). rh-FGF-2 (basic FGF-2) expressed in E. coli was from Genzyme TECHNE (Minneapolis). rh-FGF-10 expressed in E. coli was provided by Takashi Katsumata (Sumitomo Pharmaceutical Research Center, Osaka). rh-Pleiotrophin (PTN) expressed in E. coli was from RELIA Tech GmbH (Braunschweig, Germany). rh-Hep-binding epidermal growth factor-like growth factor (HB-EGF) expressed in Sf 21 insect cells, rh-brain-derived neurotrophic factor (BDNF) and rh-glial cell line-derived neurotrophic factor (GDNF) were from R&D systems (Minneapolis). Recombinant rat (rr)-FGF-16 and recombinant mouse (rm)-FGF-18 were prepared as described earlier (28,29). Human plasma was obtained from Cosmo Bio Company Ltd. (Tokyo). Human thrombin and chromozym TH were obtained from Roche Applied Science (Tokyo).
Isolation of GAGs-Skin of the so-called blue shark Prionace glauca (85 g dry weight) was dehydrated and delipidated by extraction with acetone three times, and air-dried thoroughly. It was suspended in water to obtain a slurry and kept in a boiling water bath for 30 min to inactivate proteolytic enzymes. To the slurry, borate-NaOH buffer and CaCl 2 were added to give a final concentration of 0.1 M and 10 mM, respectively, and subjected to digestion with a protease (actinase) (2% by weight of the sample) at 60°C for 24 h, after which fresh actinase (1% by weight of the sample) was added at the end of 24 h and 48 h and digestion continued. After a total of 95 h of digestion, proteins were precipitated by adding 50% trichloroacetic acid to a final concentration of 5%, the mixture was centrifuged, and the resultant precipitate was resuspended in 5% trichloroacetic acid, and centrifuged again. The supernatants were pooled. Excess trichloroacetic acid in the supernatant was removed by extraction with diethyl ether, and the GAGs were precipitated by adding 4 volumes of 80% ethanol containing 5% sodium acetate and left at 4°C overnight, after which the precipitate was collected by centrifugation and dried.
Purification of DS-The above precipitate containing GAGs was solubilized in 0.02 M Na 2 SO 4 and precipitated by addition of 10% (w/v) of cetylpyridinium chloride (CPC) in 0.02 M Na 2 SO 4 and left overnight at room temperature (30). The flocculent precipitate obtained was redissolved in a solution containing 100: 15 (v/v) 2 M NaCl/ethanol and precipitated by adding 3 volumes of 99.5% ethanol. The above step was repeated three times and finally precipitated from water and dried. A second precipitation was done as above, keeping the critical electrolyte concentration at 0.5 M NaCl to remove HA and enrich the GAGs (31).
Hyaluronidase Digestion-The precipitate obtained with CPC as above was dissolved in 0.02 M acetate buffer containing 0.15 M NaCl, pH 6, and digested with 50 turbidity reducing units of Streptomyces hyaluronidase at 60°C for 5 h in a water bath with intermittent shaking (32). An aliquot of 50 turbidity reducing units of hyaluronidase was further added at the end of 5 h, and digestion continued for another 12 h. After ascertaining the digestion of HA by cellulose acetate membrane electrophoresis, the digest was treated with trichloroacetic acid to remove proteins and adjusted to 64% ethanol to precipitate the remaining GAGs.
Nitrous Acid Treatment-The precipitate obtained after hyaluronidase digestion was treated with freshly prepared nitrous acid, which was generated by mixing equal volumes of 0.5 mmol of sulfuric acid and 0.5 mmol of barium nitrite, and left at room temperature for 40 min (33). An aliquot of freshly prepared nitrous acid was added again and the incubation continued for a further 40 min. The treated sample was neutralized with 0.5 M Na 2 CO 3 and desalted on a column (1 ϫ 56 cm) of Sephadex G-50 using 0.2 M ammonium bicarbonate as the eluent at a flow rate of 0.6 ml/min. The elution was monitored at 210 nm and the fraction eluting at the void volume was pooled and freeze-dried repeatedly by reconstituting in water.
Purification of DS Using a C18 Cartridge and an Anion-exchange Cartridge-The desalted, HS-free GAG was passed through a Sep-Pak C18 cartridge and eluted with water followed by 100% methanol. This peptide-free DS has been referred to as SS-DS (Native), an abbreviation for shark skin DS. SS-DS (Native) was passed through the anion exchanger cartridge (Waters Sep-Pak plus Accell QMA) and eluted stepwise with 300 mM phosphate buffer containing 0.15, 0.5, 1.0, 1.5, or 2.0 M NaCl and desalted on a PD-10 column using 50 mM pyridine acetate buffer, pH 5.0 as an eluent.
For digestion with 4-or 6-sulfatases, 1.6 g of the SS-DS preparation was at first digested with 10 milli-international units of chondroitinase ABC in a total volume of 10 l. A one-third portion of the chondroitinase ABC digest was subjected to digestion with 10 milli-international units of chondro-4-or chondro-6-sulfatase, and the digests were labeled with 2AB and analyzed by HPLC employing conditions mentioned above.
Gel Filtration Analysis of the Chondroitinase Digests of SS-DS on a Superdex Peptide Column-The SS-DS (Native) preparation (1.0 g each) was digested with chondroitinase AC-I or B. An equal amount was also sequentially digested with either chondroitinase AC-I followed by B or chondroitinase B followed by AC-I. All the digests were individually labeled with 2AB and excess reagents removed by extraction with a water/chloroform mixture as above. The digests were made up with 0.2 M ammonium bicarbonate containing 7% 1-propyl alcohol and analyzed on a Superdex peptide column using the same buffer as eluent at a flow rate of 0.4 ml/min using fluorescence detection.
Molecular Mass Determination-The molecular mass of DS preparations was determined by gel filtration using a column of Superdex 200 (10 ϫ 300 mm) calibrated with molecular mass markers including dextran preparations (average mass: 18.1, 37.5, and 65.5 kDa) and HS from bovine intestinal mucosa (average mass: 6 kDa) (39). V o and V t were determined using dextran (average mass: 170 -200 kDa) and NaCl, respectively. SS-DS preparations (40 g) were loaded onto the column and eluted with 0.2 M ammonium acetate at a flow rate of 0.3 ml/min, and the fractions collected at a 3-min interval, evaporated to dryness, and reconstituted in 100 l of water. An aliquot was taken for estimating GAG using 1,9-dimethylmethylene blue (DMMB) according to the method of Chandrasekhar et al. (40), except that the absorbance was read at 525 nm.
Interaction Analysis of SS-DS Preparations with Various Growth Factors-This was carried out using IAsys (Affinity Sensors, Cambridge, UK) as previously reported (41). Initially, the DS preparations were biotinylated at the C terminus using biotin-LC-hydrazide and 1-ethyl-3-(3-dimethyl-aminopropyl)carbodiimide hydrochloride as described previously (42). Excess biotinylating reagents were removed by gel filtration on a PD-10 column, and the degree of biotinylation was estimated by the modified protocol of Green (43), which showed that approximately 2% of the uronic acid residues were derivatized with the biotin tag. Biotinylated DS preparations (10 g) were immobilized onto the respective cuvettes after coating the surface with streptavidin. The immobilization level measured around 46.3 and 52.4 arc seconds for SS-DS (1.0 M) and SS-DS (1.5 M), which corresponded to 0.31 ng (4.4 fmol) and 0.35 ng (5.0 fmol) of GAG immobilized, respectively, since 600 arc seconds correspond to 1 ng of protein/mm 2 in a 4-mm 2 cuvette. After washing the cuvette surface with phosphate-buffered saline with Tween 20 (running buffer), it was blocked with 20 g of bovine serum albumin. The interaction with growth factors, namely FGF-1, FGF-2, FGF-10, FGF-16, HB-EGF, PTN, and MK, was then studied. A single binding assay consisted of two phases: (i) an association phase obtained as a result of the binding of the growth factor to the immobilized DS preparation following addition of the growth factor in 200 l of the running buffer, and (ii) a dissociation phase characterized by dissociation of the bound ligate initiated by addition of 200 l of the running buffer. The cuvette was regenerated by washing with 200 l of 1 M NaCl. The stirrer speed was maintained at 80% and the temperature at 25°C throughout the experiment. The distribution of immobilized SS-DS and of the bound growth factor on the surface of the biosensor cuvette was inspected by the examination of resonance scans, which showed that these molecules were distributed on the sensor surface at all times and therefore were not microaggregated.
To determine the kinetics of binding, growth factors were applied to the cuvette at varying concentrations in the running buffer, and dissociation and regeneration carried out as described earlier. Binding parameters were calculated from the association and dissociation phases of the binding reactions using the FASTfit software (Affinity Sensors). A plot of the on-rate constant, k on (obtained from the association analysis) versus the ligand concentration was obtained by a monophasic fit. The slope of the line gives the association rate constant, k a , and the intercept value on the y-axis gives the dissociation rate constant, k d . The equilibrium dissociation constant, K d , was obtained from the ratio of the dissociation and association rate constants (k d /k a ).
Interaction of SS-DS with BDNF and GDNF-Interaction analysis of the binding of BDNF and GDNF to SS-DS (1.5 M) was carried out in a BIAcore J system (BIAcore AB, Uppsala, Sweden) by immobilizing the biotinylated SS-DS (1.5 M) onto a streptavidin-coated sensor chip. Varying concentrations of the respective neurotrophic factors were injected onto the sensor chip at a high flow rate (60 l/min) and given a period of 3 min for the association phase and 2 min for the dissociation phase. The sensorgrams obtained by injecting various concentrations of neurotrophic factors were overlaid and collectively fitted by global fit using the 1:1 Langmuir binding with mass transfer model of the BIAevaluation 3.1 software for obtaining the kinetic parameters, k a , k d , and K d .
Neurite Outgrowth Promotion Assays of the Purified SS-DS-This was done as reported earlier (13,16). Briefly, the coverslips were coated with 600 l of 1.5 g/ml poly-DL-ornithine (P-ORN) for 2 h at 37°C and then incubated with 2 g/well of the SS-DS preparations or equivalent amounts of chondroitinase ABC, AC-I, or B digested-SS-DS preparations and left at 37°C overnight. The hippocampal neuronal cells freshly isolated from E16 mouse embryos were seeded in Eagle's minimum essential medium at a density of 10,000 cells/mm 2 and allowed to grow for 24 h at 37°C, 5% CO 2 . Thereafter the cells were fixed and subjected to immunochemical staining using monoclonal antibodies directed against neurofilament and microtubule-associated protein-2. The immunostained 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). The length of the longest neurite and the number of primary neurites of cells chosen at random was calculated using a morphological analysis software (Mac SCOPE; Mitani Corp., Tokyo, Japan).
Assay for Anti-factor IIa Activity-Anti-factor IIa activity or anti-Hep cofactor II (HC-II) activity was measured by incubating 1-10 g/ml of the SS-DS preparations in 100 l of 50 mM Tris-HCl buffer, pH 8.3 containing 227 mM NaCl with 60 l of normal human plasma and 40 l of human thrombin (1.2 NIH units/ml) at 25°C for 1 min in a disposable cuvette (44). The amidolytic activity of thrombin was determined as a change in absorbance at 405 nm recorded for 100 s after the addition of 100 l of 1.9 mol/ml chromozym TH. The rate of change of absorbance was proportional to the remaining thrombin activity. Hep and DS from porcine skin were used as positive controls.
Determination of Uronic Acid and Neutral Sugars-Uronic acid was estimated by the carbazole reaction of Bitter and Muir (45). Neutral sugars were estimated by the orcinol method (46).

Isolation and Purification of DS from Shark Skin-GAGs
were isolated from shark skin by exhaustive protease digestion followed by ethanol precipitation. The precipitate comprising a crude mixture of GAGs consisted of HA, DS, and a small amount of HS as examined by electrophoresis on a cellulose acetate membrane and disaccharide analysis after digestion with various GAG lyases (data not shown). It was purified further by CPC precipitation followed by digestion with Streptomyces hyaluronidase to remove the bulk of the HA, and nitrous acid treatment to degrade the HS. Since the preparation still showed small amounts of tetra-and hexasaccharides characteristic of HA on HPLC analysis of the hyaluronidase digest, HA was eventually removed by anion-exchange chromatography (see "Experimental Procedures"). Fractions obtained during each step of purification were monitored by electrophoresis on a cellulose acetate membrane. Trace amounts of GAGs such as HA and low sulfated CS/DS were detected in the 0.15 and 0.5 M NaCl-eluted fractions by digestion with Streptomyces hyaluronidase or chondroitinase ABC followed by anion-exchange HPLC. The minor and major fractions amounting to 20 and 80%, which were eluted with 1.0 and 1.5 M NaClcontaining buffers, were referred to as SS-DS (1.0 M) and SS-DS (1.5 M), respectively. The yields of various fractions obtained during the course of purification were determined by the carbazole reaction for uronic acid, and are presented in Table I. Both the SS-DS (1.0 M) and SS-DS (1.5 M) preparations contained small amounts of amino acids (0.6 and 1.2% by weight,  respectively) including serine, aspartic acid, glutamic acid, glycine, alanine, and arginine at a ratio of 1.0:0.4:0.7:1.7:0.3:0.3. All the three SS-DS preparations were free of HS/Hep, which was confirmed by HPLC analysis of the heparanse/heparitinase digest (data not shown).
Disaccharide Composition Analysis of SS-DS-The disaccharide composition of the purified SS-DS preparations was determined by digestion with chondroitinases followed by anionexchange HPLC. The DS preparations were individually digested with chondroitinase ABC, AC-I, or B, and each digest was labeled with a fluorophore 2AB for a high sensitivity analysis and high resolution (see "Experimental Procedures").
Digestion of the SS-DS preparations with chondroitinase B resulted in ⌬ 4,5 HexUA␣1-3GalNAc(4S) and ⌬ 4,5 HexUA(2S)-␣1-3GalNAc(4S) along with oligosaccharides. The enzyme digest of SS-DS (Native) is shown in Fig. 1B as an example. The disaccharides accounted for only 2% of the total amount of disaccharides obtained by chondroitinase ABC digestion of the equivalent amounts for all three SS-DS preparations (Table II). The release of only small proportions of ⌬ 4,5 HexUA␣1-3-GalNAc(4S) (2-7%) and ⌬ 4,5 HexUA(2S)␣1-3GalNAc(4S) (11-23%) probably suggests that a majority of their parent units are present as GlcUA-GalNAc(4S) (A unit) and GlcUA(2S)-GalNAc(4S) (B unit) rather than IdoUA-GalNAc(4S) (iA unit) or IdoUA(2S)-GalNAc(4S) (iB unit), which are present only in smaller proportions (For the naming of the units, see Refs. 7 and 8). It is also assumed that a majority of the existing iA units and iB units do not form a cluster, and hence are resistant to digestion with chondroitinase B. Chondroitinase B did not release ⌬ 4,5 HexUA␣1-3GalNAc(6S), suggesting either that all 6-O-sulfated units exist as GlcUA-GalNAc(6S) (C units) as some were detected after digestion with chondroitinase AC-I (see below) or that they exist as IdoUA-GalNAc(6S) (iC units), which may be resistant to the enzyme action or embedded in the resistant sequences. In addition to disaccharides, major oligosaccharides were observed at 29 and 55 min along with a number of minor oligosaccharides (Fig. 1B), probably indicating the hybrid nature of the SS-DS chains. The non-sulfated disaccharide observed in the chondroitinase B digest was apparently generated by the action of 4-sulfatase contaminating the enzyme preparation on ⌬ 4,5 HexUA␣1-3GalNAc(4S) units, which was confirmed using the chondroitinase B preparation devoid of 4-sulfatase (Table II).
Digestion with chondroitinase AC-I of SS-DS (Native) gave little or no disulfated disaccharides, but yielded non-sulfated and monosulfated disaccharides including ⌬ 4,5 HexUA␣1-3-GalNAc, ⌬ 4,5 HexUA␣1-3GalNAc(6S), and ⌬ 4,5 HexUA␣1-3-GalNAc(4S) (Fig. 1C), which again accounted for only 10% of all the disaccharides in the SS-DS preparations (Table II). The results suggest that at least some of the 4-O-sulfate and 6-Osulfated units exist as A and C units. As in the case of the chondroitinase B digest, differences in the proportion of disaccharides were dependent on the SS-DS preparations. Along with the above mentioned disaccharides, a number of oligosaccharides were also observed, supporting the notion of the hybrid nature of SS-DS with a mixed sequential arrangement of GlcUA-and IdoUA-containing disaccharide units. Although it was assumed that more GlcUA is present than IdoUA in view of the greater susceptibility to chondroitinase AC-I (10%) than to chondroitinase B (2%), the precise molar ratio of IdoUA to GlcUA remains to be clarified.
Gel Filtration Analysis of Chondroitinase AC-I and B Digests-To investigate the distribution of IdoUA-and GlcUAcontaining units, SS-DS (Native) (1.0 g) was digested with chondroitinase AC-I or B, and each digest was analyzed by gel filtration chromatography on a column of Superdex Peptide. Oligosaccharides corresponding to hexa-, octa-, and decasaccharides were observed in both digests, suggesting again a mixed distribution of the GlcUA-and IdoUA-containing units along the polysaccharide chains, thus forming a CS/DS hybrid structure (Fig. 2, A and B). Sequential digestion of the chondroitinase AC-I or B digest with chondroitinase B or AC-I, respectively, largely resulted in the formation of disaccharides except for some resistant fragments corresponding to tetra-and hexasaccharides, suggesting the presence of oligosaccharides with contiguous IdoUA-or GlcUA-containing disaccharide units (Fig. 2C). A smaller number of deca-and larger oligosaccharides were produced with chondroitinase AC-I ( Fig. 2A) than with chondroitinase B (Fig. 2B), supporting a somewhat higher proportion of GlcUA than IdoUA.
In strong contrast, ⌬ 4,5 HexUA(2S)␣1-3GalNAc(6S) was not obtained in either the chondroitinase AC-I or B digest (Fig. 1, B and C, Table II), suggesting the indigestibility of D (GlcUA(2S)-GalNAc(6S))-or iD (IdoUA(2S)-GalNAc(6S))-unit-containing sequences by either chondroitinase AC-I or B as previously reported (50). It was not detected upon digestion with chondroitinase AC-II either, although D and/or iD units are present in appreciable proportions (3%) ( Table II) and chondroitinase AC-II efficiently releases D units from CS oligosaccharides  c ⌬Di-0S obtained in the chondroitinase B digest was suggested to be obtained due to the action of a chondro-4-sulfatase contaminant present in the enzyme preparation on ⌬Di-4S units. Hence, the amount has been added to ⌬Di-4S (See the legend to Fig. 1 and "Results"). d ND, not detected. e The peaks were not detected clearly in Fig. 1, B and C, but were identified in Fig. 2, A and B. f nd, not detected in Fig. 1, B and C, but are probably present in very small proportions as in SS-DS (Native). g S/unit, a molar ratio of sulfate to disaccharide. (51), which tempted us to speculate that most, if not all, of the ⌬ 4,5 HexUA(2S)␣1-3GalNAc(6S) produced by chondroitinase ABC might de derived from iD, which are amenable to chon-droitinase ABC but not to chondroitinase B. 2 However, it remains to be clarified whether iD units are indeed present in SS-DS and D units are not.

Molecular Size Determination of SS-DS-
The average molecular masses of the purified SS-DS preparations were determined by gel filtration using a column of Superdex 200, which had been calibrated using markers of known molecular mass as detailed under "Experimental Procedures." All three preparations were included in the column as monitored using a metachromatic dye DMMB (Fig. 3) (52), and pig brain (40 kDa) (22). Since CS-C chains from shark cartilage are also rather large (43ϳ70 kDa) (53), the large molecular mass of SS-DS may be characteristic of CS and DS chains of certain tissues of shark.

Interaction of SS-DS with Various Hep-binding Growth Factors-Recently
, it was demonstrated that IdoUA-containing units in CS/DS hybrid chains isolated from the pig brain were critical for binding growth factors (22), and oversulfated CS-E from squid cartilage interacted with high affinity with various Hep-binding growth factors in a comparable fashion to that of Hep (41). The CS-H preparation purified from hagfish notochord, which had a high proportion of IdoUA-containing iE units/GlcUA-containing E units, also interacted with multiple growth factors with high affinity (25). In light of these findings, we investigated using an IAsys system whether the SS-DS preparation, which contains substantial proportions of IdoUA-2 X. Bao and K. Sugahara, unpublished results.  Table III in comparison with the previously reported binding capacity of CS-H and Hep. The results revealed that SS-DS chains can accommodate a larger number of these Hep-binding growth factors than CS-H chains, suggesting the importance of the sequence information. Even when compared with Hep, SS-DS bound a larger number of most growth factors per polysaccharide chain. The specific binding capacity, which is expressed by the number of growth factor molecules bound to the same length of each GAG chain (equivalent to for example 30 disaccharide units), was still severalfold higher for SS-DS than for CS-H (the right column in Table III). Despite the lower values for the binding of these growth factors to SS-DS as compared with Hep, the highly specific binding capacity of SS-DS is indicative of the high sequence specificity considering the lower sulfation degree of SS-DS (S/unit ϭ 1.17) than Hep (S/unit ϭ 2.55).
To determine the binding affinity, growth factors at varying concentrations were allowed to interact with the immobilized SS-DS (1.5 M) preparation, which represented the major fraction (80%) and was completely free of other GAGs such as HA or HS. The sensorgrams obtained with various growth factors are shown in Fig. 5, except for FGF-1 with a low binding response (see above). All the growth factors tested exhibited binding patterns typical of a growth factor. A plot of on-rate association (k on ) versus ligate concentration was obtained, from which k a , k d , and K d were calculated, and showed the high affinity binding of all the growth factors with SS-DS (Table IV). The highest affinity was exhibited toward FGF-18 (K d ϭ 4.4 nM) and HB-EGF (K d ϭ 4.5 nM) followed by PTN (K d ϭ 7.6 nM) and FGF-10 (K d ϭ 8.6 nM), reflecting higher association and lower dissociation rates, which resulted in lower equilibrium dissociation constants. MK showed the lowest association rate among the growth factors tested but the dissociation rate was comparable to that exhibited by FGF-10, PTN and HB-EGF, resulting in a somewhat lower affinity (K d ϭ 58.5 nM). Interaction of PTN with DS (porcine skin), which contains exclusively IdoUA, has been reported, wherein a K d of 51 nM was obtained in an IAsys system (54), whereas we have obtained a K d of 7.6 nM for SS-DS (Table IV). The differences could be due to the structural differences in the DS preparations used and suggest the importance of the hybrid nature of SS-DS for the binding of PTN, being consistent with the recent observation about the CS/DS chains from embryonic pig brain (22).
Interestingly, the affinity of SS-DS (1.5 M) for various growth factors was higher than that exhibited by Hep (Table IV) except for FGF-2 and HB-EGF, which showed a stronger or comparable affinity for Hep than for SS-DS, respectively. The high capacity, affinity and specificity of the binding of SS-DS toward these growth factors strongly suggest the importance of the hybrid nature of CS/DS.
Binding Analysis of SS-DS with BDNF and GDNF-Neurotrophic factors play important roles as key regulators of cell fate and cell shape in the vertebrate nervous system (55). A transforming growth factor ␤ superfamily member GDNF binds to Hep with high affinity as was shown using ELISA (56). Previously, we showed the binding of GDNF and a neurotrophin family member BDNF to CS-H from hagfish notochord and HS from bovine intestinal mucosa. In both instances, the binding affinity was higher with CS-H than with HS, suggesting that DS might play a role in binding to these neurotrophic factors (25). With that background, we determined if binding could occur with SS-DS having a lesser degree of sulfation. Toward that end, the binding and the kinetic parameters were determined using a BIAcore system after immobilizing the biotinylated SS-DS (1.5 M) and CS-H (a control) onto a streptavidin-coated sensor chip. The binding to SS-DS, initially tested with an excess amount of BDNF and GDNF, was higher for GDNF than BDNF (Fig. 6A). Although the binding responses of SS-DS were only 1.5ϳ2.2-fold higher than those of CS-H, the calculated binding capacity of SS-DS toward BDNF and GDNF was much greater (7.9-and 19-fold, respectively) than that of CS-H presumably due largely to the CS/DS hybrid nature and partly to the larger chain size (Table III).
Kinetic analysis was done after overlaying the sensorgrams obtained for BDNF and GDNF (Fig. 6, B and C, respectively) by injecting at varying concentrations and fitting the sensorgrams globally as mentioned in "Experimental Procedures" to obtain the kinetic parameters, which are presented in Table V, reveal-ing the high affinity binding of both factors to SS-DS. Their binding to SS-DS was characterized by higher association and dissociation rates compared with the binding to CS-H, which may indicate that SS-DS is a better binding partner than CS-H as described below in the "Discussion." BDNF exhibited higher association and dissociation rates than GDNF.

of SS-DS toward growth factors and neurotrophic factors in comparison with that of CS-H and Hep
The binding capacity of immobilized SS-DS (1.5 M) toward various protein factors was calculated based on the data obtained using the response of each protein factor at the saturation level at the highest protein concentration used in Fig. 5 (for the growth factors) and in Fig. 6   The preparations were immobilized on coverslips precoated with P-ORN. The hippocampal neuronal cells were separated from E16 mouse embryos and seeded. After 24 h incubation at 37°C, the cells were fixed and stained to visualize the neurites as described in "Experimental Procedures." CS-E, derived from squid cartilage, was used as a positive control. All three SS-DS preparations promoted neurite outgrowth (Fig. 7A). Neurite outgrowth-promoting activity was stronger for SS-DS (Native) and SS-DS (1.5 M) than SS-DS (1.0 M). The neuronal cells cultured on SS-DS (Native)-coated coverslips exhibited outgrowth of a long axon-like neurite along with dendrite-like neurites (Fig. 7D) in contrast to cells cultured on P-ORN-coated control coverslips (Fig. 7E). On average, there were more than 3 primary neurites per cell in the case of SS-DS (Fig. 7C). The neurite outgrowth-promoting activity, in terms of the formation of the longest neurite, was stronger than that exhibited by CS-E (Fig. 7A). This is intriguing because, though not oversulfated to the extent of CS-E, SS-DS is still better able to promote neurite outgrowth than CS-E, suggesting that it is not due to the oversulfation alone but to the sequential arrangement of the disaccharide constituents, and that the IdoUA content might play an equally important role as recently suggested for embryonic pig brain CS/DS (22).
To investigate the neuritogenic activity due to the SS-DS preparations and their structures, neurite outgrowth-promoting activity was evaluated after digestion with chondroitinase ABC, AC-I, or B. All three enzymatic digestions resulted in an abolishment of the neuritogenic activities (Fig. 7B), suggesting that CS/DS chains were responsible for the activity and that both CS-and DS-like structures were required for promotion of the neurite outgrowth.
AntiHC-II Activity-DS from porcine skin along with Hep exhibit potent anti-coagulation activity (57). The anti-coagulation activity is due to DS acting as an antithrombotic agent by binding to HC-II (58). There are reports of an increase in anti-coagulation activity with oversulfation (48,59). On the other hand, there are also reports that oversulfated DS chains are not necessarily good promoters of anti-coagulation (44,48). In light of these findings, anti-coagulation activity was evaluated as a measure of inhibition of thrombin via SS-DS binding to HC-II, which was estimated colorimetrically as the rate of change of absorbance using an artificial substrate, chromozym TH. All the three SS-DS preparations inhibited inhibition of thrombin activity, which was comparable to that exhibited by CS-B (porcine skin). SS-DS (Native) showed slightly higher antiHC-II activity at lower concentrations than the other two preparations (Fig. 8). Hep exhibited higher antiHC-II activities even at one-tenth the concentrations used for DS (porcine skin) and SS-DS. There was no inhibition of thrombin when GAGs were not added. DISCUSSION In this study, DS purified from shark skin was structurally and functionally characterized. The isolated DS preparations had a copolymeric structure comprising CS-and DS-like moieties. SS-DS (Native) had an S/unit ratio of 1.08, and similar to that (S/Di ratio of 1.08) exemplified by the CS/DS-proteoglycan named endocan secreted from human endothelial cells and circulating in the human bloodstream (52), suggesting that CS/DS hybrid chains with a degree of sulfation comparable to SS-DS are present in the mammalian systems despite some differences in composition from SS-DS as described below. The existence of similar structures in mammals has wide implica-    Neurite outgrowth-promoting activity of SS-DS preparations. SS-DS preparations (0.67 g as uronic acid) were immobilized on individual coverslips precoated with P-ORN, and primary hippocampal neuronal cells derived from E16 mice were seeded and cultured in Eagle's minimum essential medium for 24 h (for details, see "Experimental Procedures"). In A, the mean length of the longest neurite was measured for 100 randomly selected individual neurons cultured on the three SS-DS preparations, CS-E from squid cartilage (positive control) or P-ORN alone (negative control). In B, equivalent amounts of the SS-DS preparations, digested with either chondroitinase ABC, AC-I or B, were individually coated on coverslips as above and neurite outgrowth-promoting activity was assessed in terms of the mean length of the longest neurites. Effects of chondroitinase digestion on the neurite outgrowth-promoting activity of the SS-DS (Native) preparation are given as an example. In C, the number of primary neurites per cell was also measured for all the three SS-DS preparations. The values obtained from two separate experiments are expressed as the mean Ϯ S.E. Statistical analysis was performed using Mann-Whitney's U test for the selected pairs of substrates indicated by brackets. The symbols indicate the following: in A, **, p Ͻ 0.001; *, p Ͻ 0.005; in B, n.s., not significant; ***, p Ͻ 0.001; *, p Ͻ 0.05; in C, n.s., not significant; **, p Ͻ 0.001. The morphology of the neuronal cells, cultured on SS-DS (Native)-coated coverslips, which bears both axonic and dendritic neurites is shown in D, compared with the morphology of cells grown on P-ORN-coated cover slips shown in E. Scale bar, 50 m. urchin (49) as well as iD units from ascidians (48), have been isolated, signifying the importance of such disaccharide units in development.
Compared with CS, DS has been implicated in a variety of biological processes by virtue of the presence of IdoUA, whose pyranose ring has the tendency to form various conformations, resulting in an inherent plasticity for interaction with various partners (60,61). The importance of IdoUA is appreciated by its ability to promote the antiproliferative activity of cultured fibroblasts and the activity increased with the increased IdoUA content (62). The binding of most of the growth factors elucidated till recently requires IdoUA (63), with CS-E binding to various growth factors (41) and CS-D binding to PTN (64), being notable exceptions completely lacking IdoUA-containing disaccharides. SS-DS was able to bind various Hep-binding growth factors as well as neurotrophic factors as was previously seen for CS-H (25). Notably, however, its binding capacity was far greater than that of CS-H (Table III), which will make SS-DS a useful source for preparing reactive oligosaccharides with a hybrid nature. All the tested growth factors and neurotrophic factors are expressed in the brain. While various growth factors and neurotrophic factors bind with high affinity to CS-H, the dissociation of growth factors such as FGF-10, FGF-18, MK, PTN and neurotrophic factors including BDNF and GDNF is very slow, which prompted us to speculate that they might get released as growth factor-CS/DS oligosaccharide complexes from a parent proteoglycan molecule carrying CS/DS chains with the structural feature of CS-H by the actions of endoglycosidases such as hyaluronidase and endo-␤glucuronidase and transferred to the cell surface receptor (25). In contrast, SS-DS, which is comparatively less sulfated, showed faster dissociation than CS-H but the affinity was still high, comparable to Hep (Table IV), suggesting that sugar sequences such as those found in SS-DS may act as a coreceptor in the signaling cascade.
SS-DS exhibited neurite outgrowth-promoting activities, promoting the outgrowth of neurites of both an axonic and a dendritic nature characteristic of the DS from embryonic sea urchin (13). These results are intriguing because there is a great difference in terms of the disaccharide composition between the DS from embryonic sea urchin rich in iE units (74%) and SS-DS, wherein E units account for only 10% of the disaccharides. This could imply that such activity is dependent on distinct domain structures rather than the content of these oversulfated disaccharides, and that neuritogenic activities are not solely dependent on the charge density but are dependent on the sequential arrangement of the disaccharide units. Although the molecular mechanism of the action of CS/DS chains in neuritogenesis is not well understood, accumulating evidence suggests that the neuroregulatory effects of these chains may be attributable at least in part to their binding of the growth factors and regulating of their signaling (8,20,65; also see "Discussion" in Ref. 25). The abolition of neurite outgrowthpromoting activity by digestion with chondroitinases suggests that for such activity, the CS/DS hybrid structure has functional domains consisting of both IdoUA and GlcUA, which are equally responsible.
DS shows anti-coagulation and anti-thrombotic activities (66,67) and displays less hemorrhagic effects than unfractionated Hep (57,66). SS-DS showed anti-HC-II activity comparable to that exhibited by porcine skin DS. It has been reported that the HC-II-binding domain of DS contains contiguous sequences of at least three iB units (57). DS from ascidians, Halocynthia pyriformis and Styela plicata, which were oversulfated containing a high proportion (66ϳ70%) of iB showed good anti-HC-II activity compared with mammalian DS (48). On the other hand, DS from Ascidian nigra characterized by a high proportion (80%) of oversulfated iD units showed no discernible anti-HC-II activity, indicating that sulfation patterns play an important role (48). The structural basis for the observed antithrombotic activity of SS-DS remains to be investigated to clarify whether consecutive iB units or any other sequential combinations of oversulfated units such as iE units (61) are involved in the activity.
In recent years therapeutics from non-mammalian sources, which reduce the risk of contamination with pathogenic agents, have attracted attention. SS-DS with its unique structure, potent biological activities and high binding capacity toward various growth factors and neurotrophic factors may well serve as a good candidate for therapeutic application. Notably, co-administration of insulin-like growth factor-1 and GAGs greatly delays motor neuron disease and affects expression of insulin-like growth factor-1 in the wobbler mouse (68). On the other hand, neurotrophic factors exert various trophic effects on neurons. BDNF levels are decreased in neurodegenerative diseases like Alzheimer's disease (69), implying a role in the prevention of neurodegeneration. A neuropoietic cytokine, GDNF, has been successfully tested in the prevention of neurodegeneration in Parkinson's disease (70). It is of interest to test a possible preventive role and/or cooperative role of SS-DS for neurotrophic factors. Further studies on DS varying in sulfation profiles, structural motifs required for growth factor/neurotrophic factor binding and neuritogenesis would pave the way for developing therapeutics for neuronal disease in view of the emerging concept of neurogenesis in adult hippocampus (24).