Heparin-binding growth factor, pleiotrophin, mediates neuritogenic activity of embryonic pig brain-derived chondroitin sulfate/dermatan sulfate hybrid chains.

Chondroitin sulfate (CS) and dermatan sulfate (DS) chains play roles in the central nervous system. Most notably, CS/DS hybrid chains (E-CS/DS) purified from embryonic pig brains bind growth factors and promote neurite outgrowth toward embryonic mouse hippocampal neurons in culture. However, the neuritogenic mechanism is not well understood. Here we showed that pleiotrophin (PTN), a heparin-binding growth factor, produced mainly by glia cells, was the predominant binding partner for E-CS/DS in the membrane-associated protein fraction of neonatal rat brain. The CS/DS chains were separated on a PTN column into unbound, low affinity, and high affinity fractions. The latter two fractions promoted outgrowth of dendrite- and axon-like neurites, respectively, whereas the unbound fraction showed no such activity. The activity of the low affinity fraction was abolished by an anti-PTN antibody or when glia cells were removed from the culture. In contrast, the high affinity fraction displayed activity under both these conditions. Hence, PTN mainly from glia cells mediated the activity of the low affinity but not the high affinity fraction. The anti-CS antibody 473HD neutralized the neuritogenic activities of both fractions. Interaction analysis indicated that the 473HD epitope and PTN-binding domains in the E-CS/DS chains largely overlap. The three affinity subfractions differed in disaccharide composition and the distribution of l-iduronic acid-containing disaccharides along the chains. Oversulfated disaccharides and nonconsecutive iduronic acid-containing units were the requirements for the E-CS/DS chains to bind PTN and to exhibit the neuritogenic activities. Thus, CS subpopulations with distinct structures in the mammalian brain play different roles in neuritogenesis through distinct molecular mechanisms, at least in part by regulating the functions of growth factors.

tinct structures in the mammalian brain play different roles in neuritogenesis through distinct molecular mechanisms, at least in part by regulating the functions of growth factors.
Chondroitin sulfate (CS) 1 and dermatan sulfate (DS) proteoglycans (PGs), as well as heparan sulfate PGs (HS-PGs), have been implicated in the processes of neural development in the brain such as neuronal adhesion, migration, and neurite formation (for reviews see Ref. [1][2][3]. CS and DS are synthesized as glycosaminoglycan (GAG) side chains of PGs and consist of repeating disaccharide units of -GlcUA-GalNAc-and -IdoUA-GalNAc-, respectively, and also exist as CS/DS hybrid chains composed of both units in varying proportions. Recent studies have demonstrated that the disaccharide composition and the GlcUA/IdoUA ratio of brain CS/DS chains change during development (4 -7) and that certain CS/DS hybrid epitopes are found only in specific regions of the mammalian brain (6,8), suggesting that CS/DS hybrid chains or their subpopulations play distinct roles in the development of the brain.
The effects of CS/DS chains on neurite formation are controversial. During development, strong immunostaining for CS is often localized at the boundaries of brain subregions such as the roof plate and midline dorsal tectum acting as barriers to migrating neurons or extending axons (9 -12). In vitro, CS chains inhibit the migration of neurons and outgrowth of neurites on defined growth-promoting substrata (13)(14)(15), and the enzymatic removal of CS chains in vivo permits both axonal regeneration after nigrostriatal tract axotomy and spinal cord injury (16 -18). However, tissues expressing CS do not always exclude the entry of axons, and in some cases immunostained CS coincides with developing axon pathways (19,20). Several studies suggest that some CS-PGs and CS/DS chains promote rather than inhibit neurite outgrowth under certain conditions (21). DSD-1-PG (22), the mouse homologue of rat phosphacan, * This work was supported in part by the Science Research Promotion Fund from the Japan Private School Promotion Foundation and Grants-in-aid for Exploratory Research 15659021 and Scientific Research-B 16390026 from MEXT, HAITEKU (2004-2008. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. promotes neurite outgrowth toward embryonic rat hippocampal neurons, and this effect is neutralized by the monoclonal antibody (mAb) 473HD, which recognizes the DSD-1 epitope embedded in the CS side chains (23).
Here we identified a specific ligand, which regulates the neuritogenic activity of E-CS/DS chains using affinity chromatography on an E-CS/DS-coupled matrix, as pleiotrophin (PTN). The structural characteristics of the functional E-CS/DS chains for the binding of PTN and the recognition by mAb 473HD were also clarified. CS/DS subpopulations of the embryonic pig brain with distinct structures were shown to play different roles in neuritogenesis through distinct molecular mechanisms at least in part by regulating the functions of a heparin-binding growth factor PTN.

EXPERIMENTAL PROCEDURES
Materials-Embryonic pigs (body weight 570 -620 g) were purchased from a local pig-raising company, and neonatal Wistar rats and pregnant ddY mice were obtained from SLC Inc (Shizuoka, Japan). CS-A from whale cartilage, CS-B from porcine skin, CS-C and CS-D from shark cartilage, CS-E from squid cartilage, mouse mAb CS-56 and MO-225, amino-cellulofine gel, chondroitinase ABC, chondroitinase AC-I, chondroitinase B and Streptomyces hyaluronidase were obtained from Seikagaku Corp. (Tokyo, Japan). Actinase E was purchased from Kaken Pharmaceutical Co. (Tokyo, Japan). Sep-Pak Plus Accell TM anion-exchange cartridges were supplied by Waters (Milford, MA), and a Superdex 75HR column (10 ϫ 300 mm), a Superdex Peptide HR column (10 ϫ 300 mm), prepacked disposable PD-10 columns, a silver staining kit, a HiTrap N-hydroxysuccinimide-activated affinity column (1 ml), and protein G-Sepharose 4FF were obtained from Amersham Biosciences. Streptavidin-coated 96-well plates were purchased from Nalge Nunc International (Roskilde, Denmark). Recombinant human (rh) PTN for interaction assays was purchased from RELIA Tech GmbH (Braunschweig, Germany). rh-PTN for preparing an affinity column was produced by the yeast and purified to homogeneity as in the case of recombinant midkine (31), and was a gift from Dr. Sakuma (Cell Signal Inc., Yokohama, Japan). Polyclonal goat anti-rh-PTN IgG and polyclonal goat anti-rh-midkine IgG were obtained from Genzyme/Techne (Cambridge, MA). Horseradish peroxidase-conjugated donkey anti-goat IgG and goat anti-mouse Ig(G ϩ M) were purchased from Chemicon (Temecula, CA). Rat mAb 473HD (IgM) was raised against mouse brain DSD-1-PG/phosphacan as described (23), and goat anti-rat IgM was from StressGen Biotechnology Corp. (San Diego, CA). HS and heparin from bovine intestinal mucosa, a protease inhibitor mixture, poly-DLornithine (P-ORN), and anti-neurofilaments were purchased from Sigma, and anti-microtubule-associated protein 2 (MAP2) was from Leico Technologies Inc. (St. Louis, MO). A Vectastain ABC kit was obtained from Vector Laboratories, and a BCA kit was from Pierce. All other chemicals and reagents were of the highest quality available.
Preparation of E-CS/DS Chains-The E-CS/DS chains were purified from a brain homogenate from 28 embryonic pigs and were prepared with detergent-free phosphate-buffered saline (PBS) as described previously with some modifications (7). Briefly, the PBS-soluble PG fraction was digested with actinase E to degrade proteins, and the resulting GAG chains were recovered by ethanol precipitation and then desalted by gel filtration on a PD-10 column. The GAG mixture was treated with nitrous acid at pH 1.5 at room temperature for 45 min to remove HS followed by dialysis. The dialysate was subjected to an anion-exchange chromatography on an Accell QMA Plus cartridge using a stepwise elution with 0.3 M phosphate buffers, pH 6.0, containing 0.15, 0.5, and 1.5 M NaCl. The 1.5 M NaCl-eluted fraction, which contains Ͼ90% of all the CS/DS chains recovered, was further used for purification. To remove hyaluronate, this fraction was digested with hyaluronidase, and the hyaluronidase-resistant polysaccharides were recovered by gel filtration on a Superdex 75 column as described (7). Trace amounts of free peptides were removed by C-18 hydrophobic chromatography. E-CS/DS could be completely digested with chondroitinase ABC (data not shown), confirming the purity.
Preparation of a Membrane-bound Protein Fraction-Neonatal (P1) Wistar rats were anesthetized, and the whole brain was dissected. Tissues (15.1 g) were mixed with 25 mM HEPES buffer (2.5 ml/g tissue), pH 7.2, containing 2.5 mM CaCl 2 , 1 mM MgCl 2 , and a protease inhibitor mixture and homogenized in a glass-Teflon Potter homogenizer. The homogenate was centrifuged at 2,000 ϫ g for 10 min at 4°C, and the precipitate was homogenized and centrifuged again under the same conditions. The combined supernatant was subjected to ultracentrifugation at 105,000 ϫ g for 1 h at 4°C. The resulting residue was mixed with a 2% CHAPS containing 10 mM Tris-HCl buffer, pH 7.4, supplemented with 0.15 M NaCl, 5 mM EDTA and protease inhibitors and agitated at 4°C overnight. After extraction, the concentrations of CHAPS and CaCl 2 were adjusted to 0.5% and 5 mM, respectively, by addition of the Tris-HCl buffer and a 1 M CaCl 2 solution. The mixture was centrifuged at 40,000 ϫ g for 20 min at 4°C, and the supernatant was subjected to affinity chromatography using the CS/DS-coupled column.
Affinity Chromatography of Membrane-bound Proteins-The coupling of CS/DS chains to amino-cellulofine gel was carried out as described by Funahashi et al. (32). The amino-cellulofine gel (1 ml) was suspended in an aqueous E-CS/DS solution (6 mg/800 l), and the pH value was then adjusted to ϳ4.5 with 1 M HCl. The reaction was initiated by addition of 200 l of an 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride solution (6.9 mg/200 l, pH ϳ4.5), and the pH was kept between 4.5 and 5.5 by intermittent addition of 2 M HCl over a 1-h period. The incubation was allowed to continue for 24 h at 4°C with gentle rotation and then stopped by addition of 3 ml of 0.1 M NaHCO 3 , pH 8.5. The mixture was centrifuged to recover the supernatant. An aliquot was used to quantify the uronic acid content by the carbazole reaction (33), which showed that ϳ1.9 mg of E-CS/DS chains was coupled onto the gel. The gel was washed, and the remaining free amino groups on the beads were acetylated with anhydrous acetic acid as reported (32). Likewise, chondroitin, CS-A and CS-B were coupled to the gel to prepare control columns.
An equivalent aliquot of a CHAPS extract was mixed with the E-CS/ DS-coupling gel or each control gel. After rotation of the gel overnight at 4°C, the treated gel was poured into a small open column and washed with 20 ml of 10 mM Tris-HCl buffer (buffer A), pH 7.4, containing 5 mM CaCl 2 , 0.5% CHAPS, and 0.2 M NaCl. The bound proteins were eluted stepwise with 10 ml each of 10 mM EDTA-or 1 M NaCl-containing buffer A and 4 M urea. Likewise, an affinity chromatography was carried out using each control column. All steps were done at 4°C.
Amino Acid Sequence Analysis-Fractions obtained by affinity chromatography were treated with 80% acetone to precipitate proteins, which were separated by 12.5% SDS-PAGE and then transferred to a polyvinylidene difluoride membrane according to the method of Towbin et al. (34). Resolved proteins were stained with Coomassie Brilliant Blue R-250 for 5 min and then destained with 50% methanol for 15 min. Protein bands were excised, and the NH 2 -terminal amino acid sequences were determined in an Applied Biosystems Sequencer 492.
Affinity Fractionation of E-CS/DS-Bovine serum albumin-free rh-PTN (0.5 mg) was coupled to a HiTrap N-hydroxysuccinimide-activated column (1 ml) according to the manufacturer's protocol. To protect the GAG-binding sites of PTN from possible inactivation by the coupling, CS-D (0.5 mg) was incubated with PTN at room temperature for 30 min prior to the coupling. The efficiency of coupling was about 70%, as estimated by the quantification of uncoupled proteins using a BCA kit.
The affinity column was equilibrated with 5 ml of 10 mM Tris-HCl buffer (buffer B), pH 7.4, containing 0.15 M NaCl after a wash with buffer B containing 2.0 M NaCl. E-CS/DS (250 g) was dissolved in 500 l of buffer B containing 0.15 M NaCl and loaded onto the column. To maximize the absorbance, loading was repeated six times by recycling each unbound fraction. Subsequently, the column was washed stepwise with 3 ml of buffer B containing 0.15, 0.4, or 1.0 M NaCl. To obtain enough bound material, multiple affinity fractionations were performed at 4°C. All fractions were desalted on a PD-10 column.
Cell Culture-Hippocampal cells were cultured using embryonic day 16.5 (E16.5) mouse brains as described previously (23,24). Briefly, the hippocampi were obtained by microdissection and dissociated with trypsin treatment. Dissociated cells were suspended in Eagle's modified essential medium (EMEM) containing N2 supplement and then seeded on coverslips as described below. For the culture of purified hippocampal neurons, the dissociated cells were precultured in EMEM containing 10% fetal bovine serum for 3 h at 37°C to remove more adherent non-neuronal cells (35,36). After 3 h, the non-adherent cells (neurons) were harvested, rinsed with PBS, seeded on coverslips, and cultured in serum-free EMEM as described above.
Neurite Outgrowth Assays-Neurite outgrowth of hippocampal neurons from E16.5 mouse brains was assayed as reported (23,24). Briefly, plastic coverslips (10 ϫ 10 mm) were treated with 1.5 g/ml of P-ORN in 0.1 M borate buffer, pH 8.2, for 2 h at room temperature, then further coated with CS/DS chains at a dose of 1.0 -3.0 g/cm 2 per coverslip in PBS at 4°C overnight, and washed with PBS three times before cells were plated on these coverslips at 10,000 cell/cm 2 . Blocking with mAb 473HD was carried out by adding the antibody (diluted 100-fold in PBS) onto the coverslips 2 h before the seeding of the cells. When either polyclonal anti-PTN antibody (10 and 100 g/ml) or rh-PTN (0.1 and 1.0 g/ml) was used, it was added to the medium 2 h after the seeding. Rat IgM or goat IgG was run as a control depending on the primary antibody used.
After a 24-h culture, cells were fixed with 4% (w/v) paraformaldehyde and then immunostained with anti-MAP2 and anti-neurofilament as described (25). The antibodies were detected using a Vectastain ABC kit with 3Ј,3-diaminobenzidine as a chromogen. Neurite outgrowth was evaluated by determining total neurite length and the number of primary neurites per cell, where primary neurites represent the neurites longer than the cell body. One hundred clearly isolated cells with at least one neurite longer than the cell body were chosen at random per coverslip. The results were expressed as the mean Ϯ S.E., and the significance of differences between means was evaluated with the Student's t test.
Western Blotting-Hippocampal cells or neurons (4 ϫ 10 6 cells) in N2-supplemented EMEM were plated on 6-cm culture plates coated with P-ORN and then E-CS/DS or with P-ORN alone as described above. After 24 h of culture under 5% CO 2 at 37°C, the cells were solubilized with 50 mM Tris-HCl buffer (buffer C), pH 7.5, containing 1% CHAPS, 0.15 M NaCl, 1 mM EDTA and a protease inhibitor mixture. After centrifugation at 15,000 ϫ g for 15 min, the supernatant was combined with the conditioned medium and then absorbed for 1 h at 4°C with 30 l of protein G-Sepharose 4FF beads (a 50% slurry) to remove nonspecific endogenous proteins. After a brief centrifugation, 1 g of anti-PTN antibody was added to the supernatants and incubated for 30 min at 4°C. Thereafter, 25 l of protein G-Sepharose 4FF beads was added to the solution to capture the antibody-antigen complex, and the tubes were gently rotated at 4°C overnight. The gels were washed with buffer C three times and mixed with a 3-fold concentrated sample buffer for SDS-PAGE containing 1 mM dithiothreitol. After boiling for 5 min, the samples were subjected to SDS-PAGE and processed for immunoblotting according to the manufacturer's directions.
Enzyme-linked Immunosorbent Assay (ELISA)-E-CS/DS was biotinylated as described (37). All steps of ELISA were performed at room temperature. Biotinylated E-CS/DS in 50 l of PBS was added to the streptavidin-coated 96-well plate (2 g/well) and incubated for 1 h. After a wash with 200 l of PBS containing 0.05% Tween 20 (PBS-T) three times, each well was blocked with 100 l of 1% bovine serum albumin in PBS for 1 h. Subsequently, 50-fold diluted antibody, 473HD, CS-56, or MO-225, in PBS (50 l) was added to the wells and incubated for 1 h. Wells were washed with 200 l of 0.05% Tween 20, 0.15 M NaCl, and 10 mM Tris-HCl buffer, pH 8.0 (TBS-T), three times and incubated with 50 l of alkaline phosphase-linked anti-rat IgM (diluted 1,000-fold for 473HD) or anti-mouse IgM (diluted 5,000-fold for CS-56 and MO-225) for 1 h. After the washing of plates with TBS-T, color was developed by adding 50 l of p-nitrophenyl phosphate in 0.1 M sodium carbonate buffer, pH 9.8, and the absorbance at 415 nm was recorded 1 h after the addition of p-nitrophenyl phosphate. As a negative control, the primary antibody was omitted. For the inhibition ELISA, 473HD (diluted 50-fold) was incubated with different concentrations of CS/DS preparations in PBS at room temperature for 1 h, and the mixture was applied to a 96-well plate. The inhibition efficiency was calculated from the reduced absorbance compared with that obtained from an incubation without GAGs.
Surface Plasmon Resonance Analysis-Inhibition assays against the interaction of PTN or 473HD with immobilized E-CS/DS was performed using a BIAcore system (BIAcore AB, Uppsala, Sweden) according to the manufacturer's directions. Biotinylated E-CS/DS was immobilized onto a streptavidin-derivatized sensor chip as described previously (7). To investigate the relation between the 473HD epitope and the PTNbinding domains in E-CS/DS, the E-CS/DS-immobilized sensor chip was first saturated with repeated injections of 473HD antibody or PTN, followed immediately by an injection of PTN (100 ng) or 473HD antibody (diluted 50-fold) in 10 mM HEPES containing 0.15 M NaCl, 3 mM EDTA, and 0.005% (w/v) Tween 20 (running buffer) onto the surface of the sensor chip. Response curves were recorded, and maximal values were used for calculation.
To investigate the structural characteristics of the PTN-binding domains in E-CS/DS, PTN (100 ng) was incubated with a wide range of concentrations of various GAGs at room temperature for 15 min in a total volume of 130 l before being applied to the sensor chip. Response curves were recorded, and the inhibition efficiency was calculated from the reduced values of the maximal response to that of no incubation with GAGs.
Disaccharide Composition Analysis-Aliquots of E-CS/DS affinity fractions corresponding to 100 -500 pmol of disaccharide were exhaustively digested with chondroitinases ABC (5 mIU) or AC-I (2 mIU) in appropriate buffers at 37°C for 2 h (38). The digests were derivatized with a fluorophore 2-aminobenzamide (2AB) (39), and then the excess 2AB was removed by washing with CHCl 3 (40). The 2AB derivatives were subjected to anion-exchange high performance liquid chromatography on an aminobound silica PA-03 column (4.6 ϫ 250 mm, YMC Co., Kyoto, Japan) as described (41). Identification and quantification of the resulting disaccharides were carried out by comparison with authentic unsaturated disaccharides generated by bacterial chondroitinases from the corresponding disaccharide units in the CS and DS chains as follows:  (3,42).
Analysis of the Distribution of IdoUA along the CS/DS Chain-Aliquots (1 nmol as disaccharides) of E-CS/DS affinity fractions were incubated with chondroitinase B (2 mIU) in 100 mM Tris-HCl buffer, pH 8.0, at 30°C for 4 h (43). The digests were labeled with 2AB as described above, and the derivatives were subjected to gel filtration on a column of Superdex Peptide using 0.2 M NH 4 HCO 3 containing 7% 1-propanol as an effluent at a flow rate of 0.4 ml/min (7). Size determination of the peaks dissolved by the chromatography was achieved by delayed extraction matrix-assisted laser desorption ionization time-of-flight mass spectrometry (data not shown), and quantification was carried out by comparison with 2AB-derivatized unsaturated CS disaccharides.

Identification of an E-CS/DS-binding Protein-CS/DS
chains are present at cell surfaces and in extracellular matrices in the mammalian brain. Our previous study showed that the CS/DS chains from PBS-soluble and -insoluble PGs in embryonic pig brains were functionally similar in promoting neurite outgrowth toward E16.5 mouse hippocampal neurons and structurally indistinguishable in disaccharide composition, molecular size, proportion of IdoUA to GlcUA residues, and distribution along the CS/DS chains (7). Hence, we prepared CS/DS chains only from the PBS-soluble PG fraction and used them to isolate their binding proteins from a CHAPS extract of neonatal rat brain. From 387 g of wet embryonic pig brain, 17.6 mg of purified CS/DS chains (E-CS/DS) was obtained (see "Experimental Procedures"), and an analysis of disaccharide composition showed small yet significant proportions of oversulfated ⌬D and ⌬E disaccharide units in addition to large proportions of ⌬O, ⌬C, and ⌬A units ( Table I). The recovery of the disaccharides from a chondroitinase AC-I digestion of E-CS/DS was only 90% with a lower proportion of the ⌬A unit compared with a chondroitinase ABC digestion, indicating the presence of an IdoUA-containing iA unit consistent with our previous report (7).
To isolate E-CS/DS-binding proteins, an E-CS/DS-immobilized gel was prepared and mixed with the membrane-associated protein fraction from whole brains of postnatal day-1 rats. The mixture was rotated overnight and then poured into an open column. After the washing of the column with a buffer containing 0.2 M NaCl, the bound proteins were eluted stepwise with 10 mM EDTA-and 1 M NaCl-containing buffers and then 4 M urea (Fig.  1A). No significant band was detected in the EDTA-or 4 M urea-eluted fractions. Two significant protein bands (17 and 32 kDa) and one faint band at 28 kDa were detected in the fraction eluted from the E-CS/DS affinity column with 1 M NaCl. Neither the band at 17 nor 32 kDa was detected in the 1 M NaCl-eluted fractions obtained from chondroitin-or CS-A-immobilized columns, indicating a specific interaction between these proteins and E-CS/DS. The 17-kDa protein was also found in the 1 M NaCl-eluted fraction from a CS-B-coupled affinity column, suggesting that the binding of this protein to E-CS/DS requires an IdoUA-containing structure.
Liquid phase amino acid sequencing revealed that the NH 2 terminus of the 32-kDa protein was blocked, and the NH 2terminal sequence obtained from the 28-kDa protein did not match any sequences reported to date (data not shown). However, the NH 2 -terminal amino acid sequence of the 17-kDa protein was GKKEKPEKKVKKSDXGEXQXXV, which faithfully matched that of PTN, also known as a heparin-binding growth-associate molecule, except for the four unidentified amino acids denoted by X. This is consistent with a report that PTN is a binding partner for phosphacan, a brain extracellular matrix CS-PG in a CHAPS extract of the mouse brain (44). Reportedly, it binds the protein core of this PG and the CS side chains also. Here we demonstrated that PTN could bind to brain-derived CS/DS chains independently of the core protein and that the interaction appears to require the IdoUA-containing structures, because PTN interacted with E-CS/DS and CS-B but not CS-A.
Demonstration of Endogenous PTN in the Hippocampal Cell Culture-PTN is a mitogenic and neuritogenic growth factor originally isolated from the brain (45). To investigate whether PTN is involved in the neuritogenic activity of E-CS/DS, the possible production of PTN in the E16.5 mouse hippocampal cell culture was first examined by immunoprecipitation. After a 24-h culture, PTN was detected in the pooled fraction of the conditioned medium and cell lysate, and culturing hippocampal cells on the E-CS/DS-containing substratum did not significantly influence the expression level of PTN (Fig. 2, left panel). The ϳ180-kDa band was observed probably because of nonspecific binding, because it was also found in the control sample    obtained without the anti-PTN antibody (right lane of the left panel). In addition, the expression of midkine (MK), which is another neurite-promoting growth factor and has 45% homology to PTN (46), was not detectable under the same conditions (Fig. 2, right panel).

Involvement of PTN in the E-CS/DS-induced Neurite
Outgrowth-A previous study (7) showed that PTN specifically bound to immobilized E-CS/DS with a K D of 22 nM in a BIAcore system. In view of the fact that E-CS/DS is a mixture of structurally different CS/DS chains, we hypothesized that PTNbound and -unbound fractions of E-CS/DS might exhibit distinct characteristics in terms of neuritogenic activity, given that PTN is involved in the neuritogenic activity of E-CS/DS. Therefore, we fractionated E-CS/DS on a PTN-immobilized affinity column into unbound, low affinity, and high affinity subfractions (E-CS/DS-U, E-CS/DS-L, and E-CS/DS-H, respectively), which were eluted with the equilibrating buffer followed by 0.4 M NaCl and 1.0 M NaCl-containing buffers. The unbound fraction contained most (88%) of the parent E-CS/DS chains, whereas E-CS/DS-L and E-CS/DS-H accounted for 10 and 2%, respectively (Fig. 3A).
Neurite outgrowth assays using embryonic mouse hippocampal cell cultures showed that the neurons cultured on an E-CS/ DS-L-containing substratum exhibited an elaborate morphology with multiple neurites, whereas those cultured on an E-CS/ DS-H-containing substratum bore one or two long neurites (Fig. 3B). However, no significant change in morphology was observed for the neurons cultured on an E-CS/DS-U-containing substratum. This observation was confirmed by the systematic morphometric analysis, which demonstrated that both E-CS/ DS-L and E-CS/DS-H significantly increased the neurite length at a dose of 1.0 g/cm 2 , but E-CS/DS-U failed to do so even at a dose of 3.0 g/cm 2 (Fig. 3C). E-CS/DS-L showed a similar neurite extension-promoting activity at doses of 1.0 and 3.0 g/cm 2 , whereas E-CS/DS-H displayed stronger activity at 3.0 g/cm 2 . Most interestingly, E-CS/DS-L promoted the formation of more than two neurites like E-CS/DS, whereas E-CS/DS-H caused one or two long neurites to form on each cell (Fig. 3D).
These results indicated that the neurite outgrowth-promoting materials in E-CS/DS were enriched in its PTN-bound fractions, suggesting a possible involvement of endogenous PTN in the E-CS/DS-induced neurite outgrowth. In addition, the promoting effects of the low and high affinity fractions might involve distinct molecular mechanisms, because they gave a different cell morphology.
To further examine whether endogenous PTN was involved in the E-CS/DS-induced neurite outgrowth in the hippocampal cell culture, an anti-PTN antibody was added to the culture medium. The antibody significantly suppressed the E-CS/DS-L-induced neurite outgrowth at a concentration of 1.0 g/ml, and exhibited a stronger inhibition at a higher concentration (10 g/ml) (Fig. 4). On the other hand, the antibody only slightly interfered with the E-CS/DS-H-induced neurite outgrowth (Fig. 4). Control IgG at 10 g/ml did not influence the promoting effects of either E-CS/DS-L or E-CS/DS-H. Thus, it appears that endogenous PTN mediates the neurite outgrowth In the case of E-CS/DS, 2 g was coated on the coverslip for neurite outgrowth assay. After a 24-h culture, the cells were immunostained, and the cell morphology was analyzed as described under "Experimental Procedures." Representative images (B) are shown, and the neurite outgrowth was determined as the mean length of total neurites per cell (C) and the mean number of primary neurites longer than the cell body per cell (D). The values represent the mean Ϯ S.E. from three independent experiments. *, 0.01 Ͻ p Ͻ 0.05, and **, 0.001 Ͻ p Ͻ 0.01, significant difference from the control. Scale bar, 50 m.

promoted by E-CS/DS-L but not by E-CS/DS-H, supporting that E-CS/DS-L and E-CS/DS-H induced neurite outgrowth through distinct molecular mechanisms.
PTN Synthesized by Adherent Cells Mediates Dendrite-like Neurite Outgrowth-PTN is highly expressed in the embryonic brain, especially in the CA1 subregion of the hippocampus (47). To clarify the origin of the endogenous PTN in the hippocampal cell culture, the hippocampal cells were fractionated into nonadherent neuronal cells and adherent cells, which account for 73-86 and 14 -27% of the total cell population, respectively, by preculturing in a serum-containing medium. Immunoprecipitation revealed PTN in the adherent cell fraction, but not in the neuronal cell fraction (Fig. 5A), indicating that endogenous PTN in the hippocampal cell culture was derived from the adherent cells, which mainly contained glia cells but might also include neuronal stem cells and precursor cells for neurons and glia cells.
The purified hippocampal neurons were then used for the neurite outgrowth assay. E-CS/DS-L showed no significant activity, whereas E-CS/DS-H exhibited marked promoting activity (Fig. 5, B and C). Most notably, addition of exogenous PTN (0.1 and 1.0 g/ml) to the culture medium induced significant neurite outgrowth of the cells cultured on an E-CS/DS-L-containing substratum but had only a slight effect on the neurite outgrowth induced by E-CS/DS-H alone (Fig. 5, B and C). Exogenous PTN did not exhibit significant neuritogenic activity toward the cells cultured on P-ORN in the negative control experiment (Fig. 5C).
It has been demonstrated that whale cartilage CS-A has no effect, whereas shark cartilage CS-D and squid cartilage CS-E promote the outgrowth of dendrite-like and axon-like neurites, respectively, toward neuronal cells in hippocampal cell cultures (23,25). Hence, the effects of these CS variants on neurite outgrowth were examined in a culture of purified hippocampal neurons. CS-A showed no significant activity with or without exogenous PTN at 0.1 g/ml. CS-D promoted neurite outgrowth slightly, whereas CS-E promoted it markedly. Most interestingly, exogenous PTN (0.1 g/ml) significantly increased the CS-D-induced neurite outgrowth (Fig. 5D) but had little effect on the CS-E-induced neurite extension (Fig. 5D). These results suggested that a subpopulation of CS/DS chains such as E-CS/ DS-L and CS-D can cooperate with PTN synthesized mainly by glia cells to promote dendrite-like neurite outgrowth in the hippocampal cell culture, and that another subpopulation of CS/DS chains such as E-CS/DS-H and CS-E exhibit activity to promote the formation of axon-like neurites in a PTN-independent manner.

PTN Interacts with E-CS/DS-L and E-CS/DS-H with Distinct Kinetics-The different functions of PTN in the E-CS/DS-Land E-CS/DS-H-induced neurite outgrowth suggested that E-CS/DS-L and E-CS/DS-H interact with PTN in distinct ways.
To examine this possibility, a quantitative kinetic analysis of the interaction of PTN with immobilized E-CS/DS-L and E-CS/ DS-H was carried out using a BIAcore system. The sensorgrams are shown in Fig. 6, and the kinetic parameters obtained are summarized in Table II (27), whereas the interaction of PTN with E-CS/DS-H is marked by quick binding and an extremely slow dissociation, reflecting the structural difference between the PTN-binding domains of these subfractions, which share a common structural element. A similar slow dissociation in the BIAcore system was observed for the interactions of PTN with CS-E (data not shown) and CS-H (hagfish notochord CS/DS chains) (26), both of which promote the outgrowth of axon-like neurites in vitro (23,25).

Effects of mAb 473HD on the E-CS/DS-induced Neurite
Outgrowth-The monoclonal antibody 473HD recognizes an epitope structure embedded in the CS side chains of DSD-1-PG/phosphacan isolated from neonatal mouse brains (23). DSD-1-PG/phosphacan promotes neurite outgrowth of hippocampal neurons from E18 rats (23) and E 16. To further investigate the relation between the 473HD epitope(s) and the PTN-binding domains in the E-CS/DS chains, inhibition assays were carried out using a BIAcore system. To examine the effects of 473HD on the binding of PTN to E-CS/DS, the surface of an E-CS/DS-immobilized sensor chip was first saturated with 473HD by repeated injections followed by an injection of 100 ng of PTN, and the response was recorded. The obtained maximal response accounted for only 5.7% of the value obtained without prior injection of 473HD (Fig. 8C), suggesting that the saturation of E-CS/DS with 473HD almost completely inhibited the binding of PTN to the E-CS/DS chains. Hence, the sites in E-CS/DS for the recognition of 473HD and for the binding of PTN largely overlap. Similarly, a reciprocal experiment was carried out, where the inhibition by PTN of the interaction of 473HD with E-CS/DS was tested by saturating the E-CS/DS-immobilized sensor chip with repeated injections of PTN followed by an injection of 50-fold diluted 473HD. Fig. 8D demonstrated that the saturation by PTN inhibited 34.5% of the binding of 473HD to E-CS/ DS. The partial inhibition may be because PTN with a small molecular mass (17 kDa) could not block the 473HD epitope(s), whereas the bulky IgM antibody 473HD covered almost completely the PTN-binding sites. Another possibility is that the putative multiple PTN-binding sites with different affinities only partially overlap the 473HD epitope(s) so that PTN could not completely block the epitope(s). Alternatively, the interaction between PTN and E-CS/DS might be weaker than that between the 473HD antibody and E-CS/DS, which would result in an incomplete saturation of the E-CS/DS chain with PTN. In fact, the response of PTN to the E-CS/DS-immobilized sensor chip decreased by 1/3 to 1/2 within 30 s after reaching the maximum (data not shown). Therefore, it was concluded that the 473HD epitope(s) and the PTN-binding domain(s) along E-CS/DS chains largely overlap. To our knowledge, this is the first demonstration of the close relation between the epitope(s) for an anti-CS/DS antibody and the functional binding domain(s) of a growth factor in the CS/DS chains.
Effects of Various GAGs on the Binding of PTN to E-CS/ DS-In view of the functional importance of the PTN binding for the neurite outgrowth-promoting activity, the structural characteristics of such binding domains were investigated using various CS and HS variants in inhibition assays using a BIAcore system. As shown in Fig. 9A, heparin and HS potently inhibited the binding of PTN to immobilized E-CS/DS (IC 50 ϭ 0.09 and 0.16 g/ml, respectively), reflecting the heparin-binding property of PTN. Oversulfated CS-E and CS-D also strongly suppressed the binding (IC 50 ϭ 0.13 and 0.24 g/ml, respectively), and CS-C had a moderate effect (IC 50 ϭ 1.47 g/ml). However, CS-A only slightly influenced the binding of PTN to E-CS/DS (IC 50 Ͼ 100 g/ml). Most interestingly, CS-B, which has a comparable degree of sulfation (ϳ1.05 sulfate/disaccharide) with CS-C, exhibited strong inhibition (IC 50 ϭ 0.22 g/ml) in this interaction. The overlaid sensorgrams obtained by inhibition assays using 1.0 g/ml of each CS or HS variant are shown in Fig. 9B as examples. The results suggested that oversulfated disaccharides and/or IdoUA-containing structures in the E-CS/DS chains might be involved in the binding to PTN.   E-CS/DS contained DS-like structures, which seemed to be important for the binding of PTN to E-CS/DS (Fig. 1B and Fig.  9A). Previous studies (7) also showed the critical role of the iA unit as demonstrated by the abolishment of the neuritogenic activity of E-CS/DS by digestion with chondroitinase B, which specifically cleaves the galactosaminidic linkages bound to Ido-UA in DS or CS/DS hybrid chains (50). Hence, the distribution of IdoUA along the CS/DS chains in these three affinity fractions was examined. Each fraction was extensively digested with chondroitinase B, labeled with the fluorophore 2AB, and then separated by gel filtration. The digestibility of the fractions with chondroitinase B was comparable (11.4, 9.6, and 8.8% for E-CS/DS-U, E-CS/DS-L, and E-CS/DS-H, respectively), indicating that they shared a similar GlcUA/IdoUA ratio. However, the gel filtration patterns for the fluorescencelabeled digests of these three fractions differed. More than half of the components (67.7% in mol percentage) of the digests of E-CS/DS-U were small oligosaccharides ranging from di-to tetrasaccharides, which were derived from IdoUA-clustered regions. In contrast, these small oligosaccharides accounted for only 44.7 and 24.3% of all the components of the digests of E-CS/DS-L and E-CS/DS-H, respectively. More significantly, the disaccharides generated by chondroitinase B digestion of E-CS/DS-U, E-CS/DS-L, and E-CS/DS-H accounted for 10.2, 6.7, and 0.9%, respectively, and were released from consecutive IdoUA-containing domains in the parent polymers. These results suggest that IdoUA residues are more scattered along the CS/DS chains in the PTN-bound fractions than in the PTNunbound fraction and that a long consecutive IdoUA-containing sequence is not required, but hybrid sequences composed of mixed GlcUA and IdoUA residues are critical for the binding of PTN to E-CS/DS. DISCUSSION In this study, endogenous PTN mainly synthesized by glia cells was identified as the predominant ligand for E-CS/DS in the neonatal rat brain and mediated the CS/DS chain-induced outgrowth of dendrite-like neurites toward mouse hippocampal neurons in culture. The results are consistent with the observation that PTN is often located in the glia-rich regions of the developing brain, such as the radial glial processes in the cerebral cortex of embryonic rats (46). The PTN-binding domains largely overlap the 473HD epitopes in the E-CS/DS chains and form the functional neuritogenic domains of the CS/DS chains. Both oversulfated disaccharides and scattered IdoUA-containing disaccharides are critical for the PTN-binding neuritogenic domain.

Comparison of Disaccharide Composition and Distribution of IdoUA along the E-CS/DS Chains in the Affinity Fractions-To
CS/DS-PGs are mainly located at extracellular matrices and cell surfaces. The E-CS/DS chains, purified from the PBSsoluble PG fractions of embryonic pig brains, mainly represent the CS/DS side chains of secreted CS/DS-PGs, such as neurocan, phosphacan, brevican, and versican, and may also include CS/DS chains of shedded ectodomains of the membrane-associated CS/DS-PGs (NG2 and neuroglycan C, etc.) and hybrid type CS/HS-PGs such as syndecans (1). PTN, a neuritogenic growth factor, binds HS chains of syndecan-3 (51) and also interacts with the core protein and CS chains of phosphacan (44), both of which are involved in the PTN-mediated neurite extension and neuronal migration (44,52,53). In this study, PTN specifically bound to E-CS/DS-and CS-B-immobilized gels, but not to CS-A-immobilized gels, suggesting a possible involvement of the IdoUA-containing structures in the E-CS/DS chains in the binding to PTN. This notion was supported by the finding that CS-B strongly inhibited the binding of PTN to immobilized E-CS/DS, whereas CS-C with a comparable degree of sulfation to CS-B exhibited 6.7-fold less inhibitory activity. Oversulfated CS-E and CS-D strongly inhibited the binding of PTN to E-CS/DS, which is consistent with a recent report (8) that PTN bound to CS-E and CS-D with high affinity; a small proportion of D unit (1.3%) in the CS chains of phosphacan mediated the binding of PTN to phosphacan. Thus, both IdoUA-containing disaccharides and oversulfated disaccharides (see below) in the E-CS/DS chains appear to be required for the interaction of PTN with E-CS/DS.
The affinity chromatography on a PTN-coupled column enriched the neurite outgrowth-promoting E-CS/DS chains in the bound subfractions. The endogenous PTN mediated the E-CS/ DS-L-induced neurite outgrowth as shown by the inhibition with the anti-PTN antibody but was not required for such activity of E-CS/DS-H, suggesting that the neuritogenic activities of E-CS/DS-L and E-CS/DS-H are expressed via distinct mechanisms. This speculation was supported by the following observations. 1) E-CS/DS-L and E-CS/DS-H induced different cell morphologies. The cells cultured on the substratum coated with E-CS/DS-L exhibited a dendrite-like cell morphology, whereas those cultured on the E-CS/DS-H-coated substratum formed axon-like neurites. It should be noted, however, that identification of dendrite or axon requires longer incubation periods. 2) After the removal of non-neuronal cells from the hippocampal cell culture, the neuritogenic activity of E-CS/ DS-L toward purified neurons largely diminished but was restored with the addition of exogenous PTN to the medium. In contrast, the same treatments had little effect on the neuritogenic activity of E-CS/DS-H. 3) The interactions of PTN with E-CS/DS-L and E-CS/DS-H were kinetically different. Although PTN bound to E-CS/DS-L and E-CS/DS-H at comparable rates, it was released from the former at a 60-fold higher rate than from the latter, reflecting the tight association with the latter, which suggests that the PTN-binding domains for E-CS/DS-L and -H share a common structure, but either one may contain an additional structural element. Thus, we propose a model for the relation of PTN to the neuritogenic activity of the E-CS/DS chains (Fig. 10). It is likely that E-CS/DS-L efficiently captures and presents PTN produced mainly by adherent glia cells to a high affinity receptor (51, 55) on the neuronal cell surface to promote neurite outgrowth acting as a co-receptor for the PTN signaling. On the other hand, a portion of the E-CS/DS-H chains, which are close to the adherent cells in the culture, would tightly seize the adherent cell-derived PTN and prevent it from being exposed to the neuronal cell surface. At present, the molecular mechanism, through which the E-CS/DS-H chains promote axon-like neurite outgrowth, remains to be investigated. The hippocampal neurons may express a specific receptor for E-CS/DS-H, or a growth factor other than PTN may mediate the neuritogenic activity of E-CS/DS-H.
CS-D and CS-E promote the outgrowth of dendrite-and axon-like neurites, respectively, toward neurons in hippocampal cell cultures (24,55). In this study, CS-D did not significantly show activity once the non-neuronal cells were removed from the culture, and the addition of exogenous PTN markedly stimulated the activity of CS-D. In contrast, CS-E exhibited neuritogenic activity toward purified hippocampal neurons independently of endogenous or exogenous PTN. Thus, the neuritogenic activities of CS-D and CS-E resemble those of E-CS/ DS-L and E-CS/DS-H, respectively. Most likely, the neuritogenic activity of CS-D involves PTN, whereas that of CS-E does not. In fact, the affinity for the binding of CS-E to immobilized PTN is 3.5-fold higher than that for the binding of CS-D to immobilized PTN (8). mAb 473HD recognizes an epitope embedded in the CS/DS side chains of DSD-1-PG/phosphacan and inhibits the neuritogenic activity of DSD-1-PG/phosphacan (23), suggesting that this epitope is the functional domain of this PG. In the present study, 473HD suppressed the neuritogenic activities of both E-CS/DS-L and E-CS/DS-H, suggesting that the 473HD epitope is shared by both types of E-CS/DS chains. Most interestingly, the 473HD epitope is also present in CS-C and CS-D (6, 24), and recognizes structures containing an A-D tetrasaccharide sequence [GlcUA-GalNAc(4S)-GlcUA(2S)-GalNAc(6S)] (54). Such a sequence may be present in the CS/DS side chains of various PGs including phosphacan in the brain. This study revealed that the 473HD epitope and the PTN-binding domains largely overlap, which may explain the inhibitory effects of 473HD on the neuritogenic activity of E-CS/DS-L. In addition, 473HD significantly inhibited the neuritogenic activity of E-CS/DS-H for axon-like neurites, suggesting that the functional domain of the neuritogenic activity of the E-CS/DS-H chains also includes the 473HD epitope (Fig. 10).
D/iD and E/iE disaccharides are the structural elements for the neuritogenic activities of various oversulfated CS and DS chains derived from marine organisms (see Refs. 24 -27 and 55 and for review see Refs. 3 and 42). Although IdoUA-containing structures are essential for the neuritogenic activity of E-CS/DS (7), the possible contribution of small proportions (2.5-3.6%) of D and E units in E-CS/DS chains to this activity remains to be firmly established. The disaccharide analysis in this study revealed that E-CS/DS-U, E-CS/DS-L, and E-CS/ DS-H differed considerably. The enrichment of oversulfated units including D/iD, E/iE, B/iB, and T/iT in E-CS/DS-L and E-CS/DS-H indicates that these units are most likely involved in their neuritogenic activities, being consistent with our previous finding that D/iD-or E/iE-containing oversulfated CS/DS chains tend to form multiple neurites or a long neurite, respectively (24,25,56). Analysis of the chondroitinase B digests of these three CS/DS preparations showed that critical IdoUA residues in iA, iD, iE, or iB units were more sparsely distributed along the CS/DS chains in E-CS/DS-H than in E-CS/DS-U, suggesting that a long sequence of consecutive IdoUA-contain-ing disaccharides may not be necessary for the PTN binding. Further studies are required to elucidate the functional sequences in the E-CS/DS chains responsible for these activities, which would provide valuable information not only for a better understanding of CS/DS-dependent neuritogenesis but also for drug designs for regenerating neurons and treating dementia.