Oversulfated Dermatan Sulfate Exhibits Neurite Outgrowth-promoting Activity toward Embryonic Mouse Hippocampal Neurons

Brain-specific chondroitin sulfate (CS) proteoglycan (PG) DSD-1-PG/6B4-PG/phosphacan isolated from neonatal mouse brains exhibits neurite outgrowth-promoting activity toward embryonic rat and mouse hippocampal neurons in vitro through the so-called DSD-1 epitope embedded in its glycosaminoglycan side chains. Oversulfated CS variants, CS-D from shark cartilage and CS-E from squid cartilage, also possess similar activities. We have proposed that the neuritogenic property of the DSD-1 epitope may be attributable to a distinct CS structure characterized by the disulfated D disaccharide unit [GlcUA(2S)-GalNAc(6S)]. In this study, we assessed neuritogenic potencies of various oversulfated dermatan sulfate (DS) preparations purified from hagfish notochord, the bodies of two kinds of ascidians and embryonic sea urchin, which are characterized by the predominant disulfated disaccharide units of [IdoUA-GalNAc(4S,6S)] (68%), [IdoUA(2S)-GalNAc(4S)] (66%) plus [IdoUA(2S)-GalNAc(6S)] (5%), [IdoUA(2S)-GalNAc (6S)] (>90%), and [IdoUA-GalNAc(4S,6S)] (74%), respectively. They exerted marked neurite outgrowth-promoting activities, resulting in distinct morphological features depending on the individual structural features. Such activities were not observed for a less sulfated DS preparation derived from porcine skin, which has a monosulfated disaccharide unit [IdoUA-Gal-NAc(4S)] as a predominant unit. The neurite outgrowth-promoting activities of these oversulfated DS preparations and DSD-1-PG were eliminated by the specific enzymatic cleavage of GalNAc-IdoUA linkages characteristic of DS using chondroitinase B. In addition, chemical analysis of the glycosaminoglycan side chains of DSD-1-PG revealed the DS-type structures. These observations suggest potential novel neurobiological functions of oversulfated DS structures and may reflect the physiological neuritogenesis during brain development by mammalian oversulfated DS structures exemplified by the DSD-1 epitope.

and 6S represent 2-O-, 4-O-, and 6-O-sulfate group, respectively). In addition, we have demonstrated that particular neurotrophic CS-PGs such as DSD-1-PG/6B4-PG/phosphacan purified from neonatal mouse brains (12,13) and appican (14), the PG form of amyloid precursor protein expressed by rat C6 glioma cells, contained significant proportions of D and E units, respectively (15)(16)(17). The proportions of these units developmentally change in the chick and mouse brain (16,18), suggesting that CS chains differing in the degree and profile of the sulfation may possess distinct functions in development. In contrast to CS chains, the presence and the possible functions of DS-type GAG chains containing IdoUA in the mammalian brain are poorly understood.
CS-PGs and CS GAGs have been reported to inhibit neurite outgrowth in vitro (5, 19 -23). This widely accepted concept is consistent with the recent findings that enzymatic degradation of CS chains permitted the axonal regeneration after a spinal cord injury (24) and the reactivation of the ocular dominance plasticity in the adult visual cortex (25). In strong contrast, DSD-1-PG stimulates neurite outgrowth of cultured rat hippocampal neurons through its CS side chains containing the unique structure (referred to as DSD-1 epitope), recognized by the monoclonal antibody (mAb) 473HD (12). The 473HD-reactive epitope is also found in CS-C and oversulfated CS-D from shark cartilage, and the CS-D preparation itself exhibits neurite outgrowth-promoting activity (15,26). Interestingly, another oversulfated CS variant, CS-E derived from squid cartilage, also enhances neurite extension but in a DSD-1 epitopeindependent manner as revealed by the mAb 473HD-resistant property (27). Because CS-E interacts specifically with several heparin (Hep)-binding growth factors known as neurotrophic factors (28), a possibility exists that the neurite outgrowthpromoting activities of such oversulfated CS variants may be elicited through their binding of Hep-binding growth factors.
Recent studies have demonstrated that embryonic sea urchin Strongylocentrotus purpuratus produces oversulfated DS chains composed primarily of the iE disaccharide unit [IdoUA-GalNAc(4S,6S)] (74%), and the degree of the 4-O-sulfation is decreased markedly in the adult sea urchin (33) (see Table I). Although some of these DS preparations exert anticoagulant activity (31,32,34), we have shown that DS-E/CS-H derived from hagfish notochord interacts significantly with a Hep-binding neuroregulatory factor midkine (MK) in vitro and inhibits MK-mediated neural adhesion similarly to CS-E (11). Thus, various above-mentioned oversulfated DS preparations may exert neurite outgrowth-promoting activity as in the case of oversulfated CS variants. Here, we assessed neurite outgrowth-promoting activities of these DS variants toward cultured E16 mouse hippocampal neurons and investigated the involvement of IdoUA-containing DS-type structures in the neuritogenic properties. The preliminary findings were reported in abstract form (35). CS-H (3.0 M NaCl-eluted fraction) isolated from notochord from hagfish (Eptatretus burgeri) (29) was from the late Prof. Nobuko Seno (Ochanomizu University, Tokyo, Japan). Oversulfated DS preparations from ascidians A. nigra and S. plicata were isolated as reported previously (31,32). DS preparations from sea urchin S. purpuratus were isolated from the embryos and adult body walls as described previously (33). The purified DSD-1-PG was prepared from postnatal day 1 and postnatal day 15 mouse brains as described previously (12).
Enzymatic Treatment-Enzymatic digestion with chondroitinases ABC, AC-I, AC-II, or B was carried out using 5 g of each CS/DS polymer or 5 g (as GAG polysaccharide) of DSD-1-PG and 7 mIU of each chondroitinase in a total volume of 50 l of the appropriate buffer at 37°C (or 30°C for digestion with chondroitinase B) for 60 min as described previously (36). A highly purified preparation of chondroitinase ABC was used only for a digestion of DSD-1-PG in place of the conventional preparation. After incubation, the reaction mixture was boiled at 100°C for 1 min. The two-fifths volume (corresponding to 2 g of GAG) of each mixture was used for a cell culture substrate described below.
Cell Culture-Primary cultures of hippocampal neurons were established from E16 mouse brains as described previously (15) with slight modifications. The hippocampi were dissected from E16 mouse embryos, and the hippocampal blocks were washed 10 times with Hanks' balanced salt solution and dissociated with 0.25% (w/v) trypsin in the same Hanks' solution for 10 min at 37°C followed by a series of gentle triturations. The single cells were resuspended with the culture medium, Eagle's minimum essential medium containing N2 supplements (Invitrogen, Tokyo, Japan), 0.1 mM pyruvate, 0.1% (w/v) ovalbumin, 0.029% L-glutamine, 0.2% sodium hydrogen carbonate, and 5 mM HEPES and plated at a cell density of 10,000 cells/cm 2 on coverslips precoated with a defined substrate. Such cultures were maintained at 37°C with 5% CO 2 in a tissue culture incubator.
After 24 h in culture, the cells were fixed in 4% (w/v) paraformaldehyde for 30 min, washed three times with PBS, and permeabilized with 0.2% (w/v) Triton X-100 in PBS for 30 min at room temperature. The cells were immunostained with anti-microtuble-associated protein 2 after 100-fold dilution (Leico Technologies Inc., St. Louis, MO) (37) and anti-neurofilament after 250-fold dilution (Sigma) (38), which specifically reacts with phosphorylated and nonphosphorylated forms of neurofilament H subunit, antibodies in PBS containing 3% (w/v) bovine serum albumin followed by development using Vectastain ABC kit (Vector Laboratories Inc., Burlingame, CA) with 3,3Ј-diaminobenzidine as a chromogen. At least three independent experiments were carried out for each culture condition.
Analysis of Neuronal Morphology-The immunostained cells on each coverslip were scanned and digitalized with a ϫ20 objective on an optical microscope (BH-2; Olympus, Tokyo, Japan) equipped with a digital camera (HC-300Z/OL; Olympus). The photographs were analyzed using morphological analysis software (Mac SCOPE; Mitani Corp., Tokyo, Japan). In morphometric analysis, only the clearly isolated neurons with at least one process being longer than a cell body diameter were chosen at random. The length of the longest neurite and the number of the primary neurites were determined by drawing and counting the corresponding neurite(s), respectively, of at least 100 neurons on duplicate coverslips of each substrate condition per experiment.
The surface area of neuronal soma (m 2 ) was also measured as the criterion of cell adhesion.
Analysis of the 2AB Derivatives of Chondroitinase Digests Prepared from the CS Chains of DSD-1-PG-DSD-1-PG (2 g as GlcUA) was incubated with 6 mIU of chondroitinase B (a native or heat-inactivated preparation) in a total volume of 60 l at 30°C for 60 min as described above. The samples were dried in a vacuum concentrator and derivatized with 2AB, and the excess 2AB reagent was removed by paper chromatography (39). The three-fifths volume of the 2AB derivative was subsequently digested with 5 mIU of chondroitinase AC-I in a total volume of 20 l at 37°C for 60 min. Each 2AB derivative corresponding to a one-eighth volume of the starting material was analyzed using anion exchange HPLC on an amine-bound silica PA-03 column (4.6 ϫ 250 mm; YMC Co., Kyoto, Japan) with a linear gradient of NaH 2 PO 4 from 16 to 530 mM over 60 min at a flow rate of 1 ml/min (39) or by gel filtration HPLC on a Superdex™ Peptide HR10/30 column (Amersham Biosciences) as described previously (40).

Oversulfated DS-E/CS-H Stimulates Neurite Outgrowth of Cultured Mouse Hippocampal Neurons-Previous observations
suggested that oversulfated CS variants such as CS-D and CS-E can promote neurite outgrowth of E18 rat hippocampal neurons (15,26,27). To clarify whether oversulfated DS also possesses such activities, we assessed the oversulfated DS-E, originally isolated from hagfish notochord and hence termed as CS-H after hagfish (29), for its neurite extension activity toward mouse E16 hippocampal neurons using the previously established assay system (15,26,27). DS-E/CS-H contains a high content (68%) of the H or iE disaccharide unit [IdoUA-GalNAc(4S,6S)] with small proportions of [IdoUA-GalNAc(4S)] (iA unit) and [IdoUA-GalNAc(6S)] (iC unit) (29). The average number of sulfate groups/disaccharide unit (S/Unit) was 1.68, whereas the S/Unit values for CS-D and CS-E were 1.21 and 1.53, respectively (Table I). Mouse E16 hippocampal neurons were cultured at a low cell density on plastic coverslips precoated with P-ORN and subsequently with CS-D, CS-E, or DS-E/CS-H. After 24 h of incubation, the neurons were fixed and immunostained with anti-microtuble-associated protein 2 (37) and anti-neurofilament (38) antibodies. To evaluate neurite outgrowth-promoting activity of the substrate coated with CS variants, the length of the longest neurite of each of 100 randomly selected cells was first measured. Approximately 80% of the neurons cultured on the P-ORN-coated control coverslip had some short neurites (Ͻ30 m) (Figs. 1A and 2). CS-D and CS-E displayed marked neurite outgrowth-promoting effects compared with P-ORN-control (Figs. 1, A-C, and 2), being consistent with previous observations (15,27). The neurons cultured on the CS-D-coated substrate showed a flattened morphology and extended multiple neurites (Fig. 1B). On the other hand, the neurons cultured on the CS-E-coated substrate exhibited a round-shaped morphology, and ϳ40% of the neurons possessed a prominent long neurite longer than 60 m in length (Figs. 1C and 2). The DS-E/CS-H-coated substrate also exerted neurite outgrowth-promoting activity comparable with that of the CS-E-coated substrate (Fig. 2). Notably, DS-E/CS-H-induced morphological properties of the cultured hippocampal neurons were also very similar to those observed with CS-E ( Fig. 1, C and D).
To confirm the activities of CS-D, CS-E, and DS-E/CS-H, these CS/DS variants were pretreated with chondroitinase AC-I or chondroitinase B, which degrade the GalNAc-GlcUA linkages in CS or the GalNAc-IdoUA linkages in DS, respectively, before coating the P-ORN surface. As expected, the neurite outgrowth-promoting activities of CS-D and CS-E were eliminated by digestion with chondrotinase AC-I and were only slightly decreased by digestion with chondroitinase B (Fig. 3, A and B). In contrast, the activity of DS-E/CS-H was abolished by digestion with chondroitinase B but not with AC-I (Fig. 3C). Treatment with heat-inactivated chondroitinases and the enzyme incubation buffers did not affect the neurite extension (data not shown). These results suggest that it is indeed these oversulfated polymeric structures themselves that exhibited neurite outgrowth-promoting activities but not contaminants and that the predominant sulfated structures, GalNAc(4S,6S)containing disaccharide units, in CS-E and DS-E/CS-H are possibly responsible for the neurite outgrowth-promoting activities.
The Properties of DS-E/CS-H-induced Neurite Outgrowth-The amounts of immobilized GAG influence the morphologies  41 and 42). Hence, the disaccharide composition here was calculated by multiplying the total disaccharides (mol %) detected by digestion with chondrotinase ABC by a factor of 0.9, and the S/Unit value may be slightly underestimated.
f Note that the CS chains attached to the purified DSD-1-PG preparation derived from postnatal mouse brain contains an unidentified compound at 2.0 mol %, which elutes at the same position unsaturated disaccharide unit [⌬HexUA-GalNAc(4S,6S)] and are resistant to chondro-6-sulfatase (15). It also contains a small proportion of IdoUA as evident from the digestibility to chondroitinase B (see "Results"). of cultured neurons. For example, cell adhesion is progressively inhibited by an increase in the concentration of substratebound CS preparation from shark cartilage (43). Hence, effects of varying amounts of DS-E/CS-H on its neurite outgrowthpromoting activity were next examined. Compared with neurons cultured on the DS-E/CS-H substratum at a standard concentration (2 g/coverslip), those cultured on the substrata coated with the two higher dosages (4 and 10 g/coverslip) of DS-E/CS-H also showed neurite outgrowth promotion. The length of the longest neurite appeared to increase in a dose-dependent manner (Fig. 4). Furthermore, the apparent neuronal cell adhesion on all of these DS-E/CS-H substrata remained essentially unchanged (data not shown). No higher amounts were tested because of the limited availability of DS-E/CS-H.
To further evaluate the effects of DS-E/CS-H, we also assessed laminin-induced neurite extension of E16 mouse hippocampal neurons as control experiments. This molecule is a representative ECM component, which promotes "neuronal polarity" (i.e. an asymmetric axonal growth) of cultured neurons in vitro (44,45), resembling the morphology of hippocampal neurons with prominent long neurites cultured on the substratum coated with CS-E or DS-E/CS-H (Fig. 1, C and D). As expected, cells grown on the laminin-coated substratum extended the polarized neurites (data not shown). The average length of the longest neurites sprouted from the individual neurons cultured on the laminin substratum increased in a dose-dependent manner (Fig. 4). It is noteworthy that DS-E/ CS-H showed a significantly higher activity in neurite out-growth promotion compared with that of laminin at the three dosages (Fig. 4). CS-E also exhibited the same dose-dependent profiles of DS-E/CS-H (data not shown). Thus, these oversulfated CS/DS variants surely exerted significant neurite outgrowth-promoting activities comparable with that of laminin at least in our in vitro assay system.

Various Oversulfated DS Variants Also Possess Neurite Outgrowth-promoting Potencies in a Sulfation Pattern-dependent
Manner-The neurite outgrowth promotion by substrates coated with DS-E/CS-H or various individual oversulfated CS variants suggested that other oversulfated DS preparations, which differ in the positions of sulfate groups, may also possess similar activities. To test this idea, we assessed the neurite outgrowth-promoting activities of previously characterized various oversulfated DS preparations with different structures (31)(32)(33). The oversulfated DS preparations along with oversulfated CS-D and CS-E used in this study are summarized in Table I. All of these preparations are derived from lower marine organisms; the two preparations are derived from bodies of ascidians S. plicata and A. nigra, which are characterized by the predominant disulfated disaccharide units, [IdoUA(2S)-GalNAc(4S)] (B unit) (32) and [IdoUA(2S)-GalNAc(6S)] (iD unit) (31), which account for 66% and greater than 90% of the disaccharides of the parent DS chains, respectively. It should be noted that the S. plicata DS preparation also contains the iD disaccharide unit [IdoUA(2S)-GalNAc(6S)] at 5% (Table I). Two other DS preparations were purified from embryo and adult tissues of sea urchin S. purpuratus. The embryonic preparation was composed primarily of the iE disaccharide unit [IdoUA- GalNAc(4S,6S)] (74%), and the S/Unit was 1.74 (33). Although the adult preparation contained this disulfated disaccharide unit at 25%, it also contained the B disaccharide unit [IdoUA(2S)-GalNAc(4S)] at 16% (33), and the S/Unit value was 1.41, which was significantly lower than that of the embryonic preparation but significantly higher than that (1.06) of porcine skin DS (Table I). As shown in Fig. 5, these oversulfated DS preparations exhibited neurite outgrowth-promoting activities. These activities were eliminated by digestion with chondroitinase ABC or B but not with chondroitinase AC-I (data not shown), suggesting that the observed activities were attributable to the DS chains or DS-like structure in the CS/DS hybrid chains and ruling out the possibility that the activities were due to contaminants in these DS preparations.
To quantify the neurite outgrowth-promoting activities, the average length of the longest processes was compared among the neurons cultured on the substratum coated with each one of the different CS/DS variants (Fig. 6). In contrast to various DS preparations from marine organisms, no neurite outgrowthpromoting activity was observed for the conventional DS preparation (CS-B) from porcine skin (Figs. 5B and 6), which is characterized by the monosulfated iA disaccharide unit [Ido-UA-GalNAc(4S)] (89%). When cultured on the substratum coated with the DS preparation from embryonic sea urchin, the iE content (74%) of which was higher than that (68%) in hagfish DS-E/CS-H, neurons developed one to three long neurites, as in the case of hagfish DS-E/CS-H (Figs. 1D and 5E). The mean length of the longest neurites observed for neurons cultured on the substrates coated with the embryonic sea urchin DS or hagfish DS-E/CS-H preparations was similar, showing the highest values among various DS preparations examined (Fig. 6). On the other hand, the extent of the neurite outgrowth of neurons cultured on the adult sea urchin DS preparation (S/Unit ϭ 1.41) was significant but clearly weaker compared with that observed for neurons cultured on the substratum coated with the embryonic sea urchin DS preparation (Figs. 5,  E and F, and 6). Although the neurite outgrowth promoted by the two ascidian DS preparations (S/Unit ϭ 1.71 or 1.90) were inferior to that observed for neurons stimulated by the GalNAc(4S,6S)-containing CS/DS variants (Fig. 6), the neurons grown on the substratum coated with either one of both ascidian DS preparations extended multiple dendrite-like neurites, exhibiting similar morphological features to neurons cultured on the shark cartilage CS-D-coated substrate (Figs. 1B and 5, B  and C). These findings strongly support the crucial importance Differential Neuritogenic Properties of Oversulfated DS Variants-It has been suggested that axonal growth is facilitated by low cell-to-substrate adhesion conditions, whereas dendritic arborizations are elaborated under high adhesion conditions, which leads to the increased spreading of a neuronal soma (44 -48). To investigate the mechanism of the differential effects of the oversulfated DS variants on the neurite outgrowth enhancement, we analyzed the surface area of the cell soma by calculating the mean surface area of 100 cells as an index of cellular adhesion and spreading in addition to the mean num-ber of primary neurites/cell cultured on the substrata coated with individual preparations of various DS and oversulfated CS variants, CS-D and CS-E. These two indices (44) were confirmed useful to characterize the neuritogenic activities of CS/DS chains for the first time in this study.
The apparent cell surface area was decreased when the neurons were cultured on the substrate coated with any of the oversulfated CS/DS variants tested compared with the surface area of the neurons cultured on the P-ORN-coated control substrate (Fig. 7A). A decrease in the cell surface was particularly striking for the neurons cultured on the CS-E and DS-E/ CS-H-coated substrata, suggesting that these substrata provide relatively lower adhesive conditions for neurons, leading to the elongation of a prominent long neurite, which resembles the premature axon morphology. However, the numbers of primary neurites observed for neurons cultured on these two substrata were significantly different. Although only a single long neurite per neuron was frequently observed for neurons cultured on the CS-E substratum, most neurons grown on the DS-E/CS-H-coated substratum had on average 3.0 neurites (Fig. 7B), i.e. at least a single long neurite and two short neurites/cell. Both parameters, the cell surface area and the number of neurites, obtained for neurons cultured on the substratum coated with the embryonic sea urchin DS preparation were greater than those obtained for the neurons cultured on the substratum coated with CS-E or DS-E/CS-H (Fig. 7). Thus, the effects of these three CS/DS variants, which contain the E or iE units [GlcUA/IdoUA-GalNAc(4S,6S)] as major components, on neurons appear to be attributable to their low adhesive properties and permissive for longer neurite outgrowth. In addition, the IdoUA-containing DS-type structures, which are present in the DS-E/CS-H and the embryonic sea urchin DS preparation but not present in CS-E, may have additional effects to enhance the formation of multiple neurites.
The surface area of the cell soma of the neurons cultured on the substratum coated with shark cartilage CS-D or the DS preparations from two kinds of acidians, which stimulated the dendrite-like neurite formation, was larger than the surface areas observed for the neurons cultured on the substratum coated with CS-E or DS-E but smaller than that observed for the neurons cultured on the P-ORN-coated control substratum (Fig. 7). Thus, the surface area of neurons also appears to be correlated with the adhesiveness and morphological traits of the neurons cultured on the substrata coated with these CS/DS variants (Fig. 7).
These CS and DS preparations had different effects on the number of primary neurites. Notably, a larger number (5.0 versus 3.0) of primary neurites were observed for the neurons cultured on the substratum coated with the DS preparation purified from ascidian A. nigra compared with those cultured on the CS-D-coated substratum (Fig. 7B) ]. This appears to correlate with the findings that shark cartilage CS-D contains only GlcUA as uronic acid, whereas the A. nigra DS contains mainly IdoUA. In contrast to the DS from A. nigra, the DS preparation derived from S. plicata, which contains B unit [IdoUA(2S)-GalNAc(4S)] as a major component, developed only 2.5 neurites/neuron, similarly to P-ORN, suggesting that the B unit does not appear to have stimulatory effects to increase the number of primary neurites despite its IdoUA-containing nature. The observed neurite outgrowth-promoting activity of the S. plicata DS preparation may be attributable to iD units contained at 5% rather than B units contained at 66% as discussed below. Taken together, in vitro dendritic neurite outgrowth, promoted by CS/DS-coated sub- chondroitinase AC-II, implying the existence of an IdoUA-containing structure (12). Therefore, we investigated whether the neuritogenicity of DSD-1-PG is associated with IdoUA residues of the CS chains of DSD-1-PG. The DSD-1-PG preparation was digested with various chondroitinases and used for coating P-ORN-precoated coverslips. Hippocampal neurons were cultured on such substrata, and the neurite outgrowth was evaluated. Neurite outgrowth-promoting activity of DSD-1-PG was abolished by digestion with highly purified chondroitinase ABC but not chondroitinase AC-II as described previously (Fig. 8 and Ref. 12). In contrast, chondroitinase B eliminated the neurite outgrowth markedly (Fig. 8), suggesting that the biologically active GAG side chain(s) of DSD-1-PG contain IdoUA residue(s) and are most likely CS/DS hybrid chains. Digestion with chondrotinase AC-I had much weaker effects on the neuritogenic activity of DSD-1-PG compared with those of chondroitinases ABC and B (Fig. 8), consistent with the specificity of chondroitinase AC-I, which does not cleave a GalNAc linkage that is bound to 2-O-sulfated GlcUA (49). These results are in good agreement with our hypothesis that the neuritogenic DSD-1 epitope contains a D unit (15,26) and an IdoUA residue, as will be described under "Discussion." To confirm the presence of IdoUA residue(s) in the CS/DS moiety of DSD-1-PG, the chondroitinase B digest of DSD-1-PG, which should contain chondroitinase B cleavable sites bearing IdoUA, was derivatized with a fluorophore 2AB, and the 2AB derivative was purified, digested with chondroitinase AC-I to generate oligosaccharides, and subjected to anion exchange or gel filtration HPLC to monitor fluorescence intensity. Fluores-cent peaks were observed at the positions of the authentic 2AB-labeled unsaturated CS disaccharide units, ⌬Di-6S and ⌬Di-4S, by anion exchange HPLC (Fig. 9), suggesting the presence of GlcUA-GalNAc(6S)-IdoUA/IdoUA(2S) and GlcUA-GalNAc(4S)-IdoUA/IdoUA(2S) trisaccharide sequences, respectively. Another peak eluted after ⌬Di-4S-2AB was also observed and is likely a 2AB-derivatized oligosaccharide. The two peaks eluted before ⌬Di-6S-2AB are derived from the chondroitinase AC-I preparation as examined by control experiments. No other appreciable peaks were observed for the control digest, which was prepared by treatment of DSD-1-PG with heat-inactivated chondroitinase B (data not shown). These results together suggest that the 2AB-labeled di-and oligosaccharides are derived from the IdoUA-containing chondroitinase B-cleavage sites and that the alleged CS chains containing the DSD-1 epitope of DSD-1-PG contain at least a small number of IdoUA residues and possess a DS-like structural characteristic that is involved in the neuritogenic activity of DSD-1-PG. Thus, the GAG side chains of DSD-1-PG should be regarded as CS/DS hybrid chains.

DISCUSSION
In this study, we demonstrated that structurally different oversulfated DS preparations derived from lower marine organisms, in addition to the oversulfated CS variants reported previously (15,26,27), possessed neurite outgrowth-promoting activities. Morphometric analyses of primary neurons grown on the CS/DS-coated substrata showed good correlations between their characteristic sulfation patterns and neuritogenic properties, providing biochemical and cell biological bases for elucidating the underlying mechanisms. The morphometric traits of neurite developing cells cultured on the substrata coated with multiple CS/DS variants can be divided into two major  (Table I) (Table I), exhibited similar morphologies: a relatively small cell soma and a lower neurite sprouting with a prominent long neurite. The different morphologies of hippocampal neurons cultured on substrates coated with structurally different DS variants supports the notion that the neurite outgrowthpromoting activities of the CS/DS variants are dependent on the sulfation pattern but not on the mere charge density.
The adult sea urchin DS preparation exhibited only a modest neurite outgrowth-promoting activity as evaluated by the mean length of the longest neurites (Fig. 6) and the mean surface area of the cell soma (Fig. 7A), whereas the activity expressed by the index of the number of newly formed primary neurites (Fig. 7B) was comparable with that observed for the neurons cultured on the P-ORN control substratum. Such a modest activity of the adult sea urchin DS also appears to be correlated with its structure. Although the DS preparation contains appreciable proportions of disulfated disaccharide units, iE [Ido-UA-GalNAc(4S,6S)] at 25% and B [IdoUA(2S)-GalNAc(4S)] at 16% (Table I) (33), the former rather than the latter is presumably involved in the activity. This is because porcine skin DS (CS-B), which contains B units at 6%, was not neuritogenic, whereas DSD-1-PG/phosphacan, purified from neonatal mouse brains, which contains a small proportion (5.0%) of D or iD units, exhibits a marked neurite outgrowth promoting activity (15). Hence, similar activity observed for the ascidian S. plicata DS preparation, which contains B units (66%) as a predominant component, is presumably due to the small yet significant proportion (5%) of iD units (Table I) (32). Thus, relatively small proportions of iE units in the adult sea urchin DS and iD units in the S. plicata DS are assumed to be involved in the biological activities, although a possibility cannot be excluded that combinations of a B unit and an iD or iE unit may form functional domain structures.
The oversulfated CS/DS-induced neuronal morphologies are in agreement with the notion (50) that permission of either axonal or dendritic neurite formation can be regulated by the neuronal cell-to-substrate adhesion. In addition, in this study the number of primary neurites/cell was found to be another index that can be used for evaluation of the capacity to generate multiple dendritic neurites and revealed the importance of IdoUA residues in DS for such neuritogenic activities. Although the functional importance of IdoUA in Hep/heparan sulfate has been established for their specific bindings to matrix proteins, growth factors, and chemokines (51)(52)(53)(54), no neuritogenic activity was observed for the bovine intestinal Hepimmobilized substrate in our assay conditions, 2 despite the high IdoUA content and the high charge density of Hep comparable with the oversulfated DS variants used in this study. Oohira et al. (43) also reported that Hep coupled to dipalmitoylphosphatidylethanolamine did not serve as an effective substrate for neurite elongation. Hence, IdoUA residues in the DS chains but not in Hep or its structural analog heparan sulfate chains appear to have potentially important implications in the neuritogenic potencies in addition to particular sulfation patterns.
The expression and function of DS chains in mammalian brains have not been clearly demonstrated, although DS-specific GalNAc-4-O-sulfotransferase and uronyl 2-O-sulfotransferase, which are required for the synthesis of iE and D/iD units, respectively, have been molecularly cloned, and their messenger RNAs are expressed in the human brain (55,56). Although recent analyses of CS-and DS-containing fractions from the embryonic mouse brain (10,16,57) showed small proportions of oversulfated CS or DS disaccharide units such as B, E or iE, and D or iD, their rigorous identification as IdoUAcontaining units has not been reported. In this study, we showed that chondroitinase B treatment of DSD-1-PG purified from neonatal mouse brains disordered its neuritogenic property and also showed the presence of the chondroitinase Bsensitive structures including at least GlcUA-GalNAc(4S)-Ido-UA/IdoUA(2S) and GlcUA-GalNAc(6S)-IdoUA/IdoUA(2S) trisaccharide sequences by chemical analysis of the CS moiety of DSD-1-PG. These observations provide evidence for the existence of DS-type disaccharide unit(s) in the DSD-1 epitope. Hence, it is suggested that the DSD-1 epitope contains a neuritogenic functional domain with an IdoUA residue in addition to a disulfated D unit reported previously (15).
The neuritogenic DSD-1 epitope, found in the DSD-1-PG/ phosphacan from the mouse brain (12), is one of the most attractive targets for structural analysis of the neuritogenic functional domain(s). Chondroitinase AC-I treatment reduced the neuritogenic activity only weakly, probably because it does not cleave GalNAc-IdoUA or GalNAc-GlcUA(2S) (49), consistent with our hypothesis that the epitope contains a D unit (15,26) and IdoUA demonstrated in this study. Chondroitinase B abolished the neuritogenicity but did not yield ⌬HexUA(2S)-GalNAc(6S), suggesting that IdoUA or IdoUA(2S) is not present as an iD [IdoUA(2S)-GalNAc(6S)] unit but is present together with a D [GlcUA(2S)-GalNAc(6S)] unit forming the neuritogenic epitope or that it is present as an iD unit but in a sequence resistant to the action of chondroitinase B, whose sequence specificity is not well understood. As previously reported (12), chondroitinase AC-II did not affect the neuritogenic activity, probably because it acts in an exolytic fashion (49) and may not act beyond an expected IdoUA-containing point in the CS/DS chains. As reported previously (12), the reactivity of DSD-1-PG to mAb 473HD, which neutralizes the neuritogenic activity of DSD-1-PG, is abolished by digestion with chondroitinase AC-I but not with chondroitinase AC-II, whereas another CS epitope, recognized by mAb CS-56, is lost by treatment with either chondroitinase AC-I or AC-II, suggesting that DSD-1 epitope may be located in the proximity of the core protein (12). Alternatively, or in addition, the chondroitinase AC-I-resistant neuritogenic activity may suggest that the neuritogenic DSD-1 epitope is composed of IdoUA-rich disaccharide clusters with an adequate length for the attachment to the P-ORN surface even after cleavage by chondroitinase AC-I digestion. The mAb 473HD epitope includes the neuritogenic DSD-1 epitope recognized by the mAb but is not identical with the latter, because chondroitinase AC-I eliminates the former but not the latter (12). Isolation and structural elucidation of the neuritogenic DSD-1 epitope and the mAb 473HD epitope remain to be achieved.
Notably, soluble forms of shark cartilage CS-C and bovine mucosa DS have been reported to be rapidly internalized by embryonic rat mesencephalic neurons and enhance neurite outgrowth when applied to the culture medium (47). However, it is not clear whether these observations represent physiological phenomena, because CS and DS usually exist as side chains covalently attached to the core proteins as ECM components, and no mammalian glycosidase, which can liberate poly-mer CS or DS chains, has been reported. Because PGs interact with various ECM molecules through the core proteins and/or GAG moieties (5), PGs would consequently be deposited in the ECM. In addition, because the immobilized forms of CS-C and DS were not active in the present assay system (15,26), it is difficult at present to evaluate the observed activities of the soluble CS and DS chains added to the culture medium. In the present study we chose the assay system where GAG chains are immobilized on the P-ORN surface to mimic ECM, excluding the influence of other ECM molecules, and our findings would represent at least some aspects of the physiological functions of oversulfated CS/DS present in the ECM.
Although the neuritogenic mechanism mediated by oversulfated CS/DS chains has been only poorly understood, Hepbinding growth factors may be involved in the mechanism. Accumulating evidence suggests that CS/DS chains, in addition to Hep/heparan sulfate, may act as binding partners or coreceptors for Hep-binding growth factors and chemokines/cytokines (28,58). For example, MK and pleiotrophin (PTN) are involved in neurite outgrowth and neuronal migration (59 -61) through interactions not only with the heparan sulfate chains of the transmembrane receptor syndecan 3 (62) but also probably with a receptor-type protein tyrosine phosphatase (PTP/ RPTP␤) (63,64), which is a transmembrane CS-PG-type isoform of DSD-1-PG/phosphacan (also known as PTP-S) (65). It is suggested that the PTP-mediated MK and PTN signalings may control cytoskeletal changes including neuritogenesis, through the action of the intracellular Src family kinase (62,66). The bindings of MK and PTN to PTP and their biological activities are strongly inhibited by CS-D, CS-E and CS-C but not CS-A (63,64), suggesting that certain oversulfated and/or IdoUA-containing structures in the CS/DS moiety of PTP may play an important role in the neurite outgrowth promotion by MK and PTN. In addition, phosphacan inhibits strongly the PTN-induced neuronal migration (64), suggesting that DSD-1-PG/phosphacan may also regulate the PTP-mediated signaling by inhibiting competitively the binding of PTN to PTP. Thus, the effects of the immobilized oversulfated CS/DS variants observed in this study may be mimicking the in vivo functions of DSD-1-PG/phosphacan.
It has been reported that cultured astrocytes synthesize and release CS/DS-PG biglycan, which is generally known as an ECM component in a variety of non-neural connective tissues such as bone, cartilage, and skin, and the CS/DS hybrid chains of the biglycan purified from astrocytes support the survival of neocortical neurons (67). The neurotrophic effects of the CS/DS chains are inhibited by nonselective tyrosine kinase inhibitors genistein and erbstatin, suggesting the involvement of the enhancement of signaling in tyrosine kinase pathways by CS/DS chains (68). Another line of evidence suggests that the injection of the biglycan and CS-C preparation into the basal ganglia facilitates learning and long term ventralpalloidal-cortical chorinergic activity (69). In view of these findings and those from the present study, characterization of the minimal structures of neuroactive CS/DS chains such as DSD-1 epitope and oversulfated CS/DS would provide useful information applicable to drug development for regenerating neurons and treating dementia.