Chondroitin Sulfate/Dermatan Sulfate Hybrid Chains in the Development of Cerebellum

Chondroitin sulfate/dermatan sulfate (CS/DS) chains regulate the development of the central nervous system in vertebrates. Previously, we demonstrated that CS/DS hybrid chains from embryonic pig brain exhibit neuritogenic and growth factor binding activities, which depended on their IdoUA content defining the DS-like structure. To elucidate the distribution of such functional sugar chains during the development of the brain, in situ hybridization was performed to examine expression of three CS/DS GalNAc 4-O-sulfotransferases, D4ST-1, C4ST-1, and C4ST-2, and a single uronyl 2-O-sulfotransferase (UST) involved in the biosynthesis of DS in addition to CS intermediates. C4ST-1 and C4ST-2 were ubiquitously expressed in the postnatal mouse brain, whereas the expression of D4ST-1 and UST was restricted in the developing cerebellum and culminated at postnatal day 14 as shown by reverse transcriptase-PCR analysis. In situ analysis of the disaccharides of CS/DS in brain sections revealed that the concentration of CS/DS increases 2-fold during development (postnatal day 7 to 7 weeks). The proportions of DS-specific, principal disaccharides, IdoUA-Gal-NAc(4-O-sulfate) (iA) and IdoUA(2-O-sulfate)-GalNAc(4-O-sulfate) (iB), produced by the sequential actions of D4ST-1 and UST, were higher in the CS/DS chains from cerebellum than those from whole brain sections. A dramatic increase (10-fold) in the proportion of iB during development was noteworthy. In contrast, GlcUA/IdoUA(2-O-sulfate)-GalNAc(6-O-sulfate) (D/iD) and GlcUA/IdoUA-GalNAc(4, 6-O-disulfate) (E/iE) decreased to 50 and 30%, respectively, in the developing cerebellum. These results suggest that the IdoUA-containing iA and iB units along with D/iD and E/iE units in the CS/DS hybrid play important roles in the formation of the cerebellar neural network during postnatal brain development.

of neurites in primary hippocampal neurons of the mouse, suggesting the involvement of DS-type structures, in addition to their sulfation profiles, in the neuritogenesis (21)(22)(23). This notion was strongly supported by the demonstration of significant neuritogenic and growth factor binding activities of CS/DS hybrid chains isolated from embryonic pig brain, which contain a higher proportion of IdoUA than those prepared from adult pig brain (24,25). Therefore, characteristic, functional CS/DS hybrid chains may be distributed in the brain parenchyma, where the neural network forms during development. However, the spatiotemporal distribution of such IdoUA-containing CS/DS hybrid structures in the brain remains largely unknown because of analytical difficulties.
The structural variability of CS/DS chains is biosynthetically generated under the control of multiple sulfotransferases and glucuronyl C5 epimerase that converts GlcUA to IdoUA at the polymer level (for reviews see Refs. 18,26,27). 4-O-Sulfation of GalNAc residues is a high frequency modification of CS/DS, and such a sulfation has been postulated to be involved in the formation of IdoUA by C5-epimerization of GlcUA. Previously, we have demonstrated the substrate specificities of three CS/DS 4-O-sulfotransferases, dermatan 4-O-sulfotransferase-1 (D4ST-1) and chondroitin 4-O-sulfotransferases-1 and -2 (C4ST-1 and C4ST-2), toward partially desulfated DS (28). The results suggest a substantial contribution of D4ST-1 to the formation of IdoUA-rich regions consisting largely of a disaccharide iA unit (IdoUA-GalNAc(4S)) characteristic of DS, and the potential involvement of C4STs in the 4-O-sulfation of GalNAc residues next to IdoUA. In addition, a single uronyl 2-O-sulfotransferase (UST) catalyzes preferentially the 2-O-sulfation of IdoUA of iA units (29), forming another DS-specific iB unit IdoUA(2S)-GalNAc(4S), where 2S and 4S represent 2-O-and 4-O-sulfate groups, respectively. In view of their substrate preferences for DS precursor chains, the gene expression patterns of these sulfotransferases would provide clues for investigating the regional distribution pattern of IdoUA-containing CS/DS hybrid chains in tissues.
Hence, we examined the gene expression of the four CS/DS sulfotransferases described above during postnatal development of the mouse brain by in situ hybridization to probe the distribution of such hybrid chains in the brain parenchyma. Furthermore, we demonstrated a good correlation between the gene expression profiles and the distribution pattern of IdoUAcontaining CS/DS structures produced by the gene products based on an in situ analysis of the disaccharide composition of the CS/DS chains in brain sections. Cloning of Mouse cDNAs Encoding Sulfotransferases Involved in CS/DS Biosynthesis-cDNAs of four mouse sulfotransferases (D4ST-1, C4ST-1, C4ST-2, and UST) were amplified from mouse heart Marathon-Ready cDNA (Clontech) by two-round PCRs using respective primer sets corresponding to the 5Ј-and 3Ј-noncoding regions. For a cDNA fragment (ϳ1.3 kbp) of mouse D4ST-1 (UniGene Cluster Mm. 278349), the first PCR was performed with a forward primer 5Ј-TCG AGG TCG GCG AGG TCT-3Ј and a reverse primer 5Ј-GTG GAC AAG GAC CTC CAA GG-3Ј, followed by nested PCR with nested primers as follows: a forward primer 5Ј-GAG GTC TGG CCG CA-3Ј and a reverse primer 5Ј-GGA CCT CCA AGG ACA GTC CT-3Ј. A cDNA fragment (ϳ1.3 kbp) of mouse C4ST-1 (Uni-Gene Cluster Mm. 339052) was amplified with a forward primer 5Ј-GCA CAC CAA CTC CTG CG-3Ј and a reverse primer 5Ј-TGC CGC TCG CAG CAT CT-3Ј, followed by nested PCR with nested primers as follows: a forward primer 5Ј-AGA GCC TCG GTG AAG CTA-3Ј and a reverse primer 5Ј-ATA ACC CAG TCT CCA TAG AAT TC-3Ј. A cDNA fragment (ϳ1.4 kbp) of mouse C4ST-2 (UniGene Cluster Mm. 28934) was amplified with a forward primer 5Ј-CCA GCT GTG CAC AAG GCT GA-3Ј and a reverse primer 5Ј-TGC CTG TCA CAC CAG GAA GC-3Ј, followed by nested PCR with nested primers as follows: a forward primer 5Ј-GCA CAA GGC TGA AGT GAA GG-3Ј and a reverse primer 5Ј-AAG GAA GCC AGG AGA GAA CC-3Ј. In the case of mouse UST (Uni-Gene Cluster Mm. 41163) cDNA, the cDNA fragment (ϳ1.3 kbp) was amplified with a forward primer 5Ј-CTC TCC ATG TGC AGA CAG CC-3Ј and a reverse primer 5Ј-GGA TGA GGC CAC AGA AGC CA-3Ј, followed by nested PCR with nested primers as follows: a forward primer 5Ј-GAT GGG TGA CCT CTT CCT GG-3Ј and a reverse primer 5Ј-GCC ACA ATC TGG CGC AGG TA-3Ј. All the reactions were performed using KOD-Plus (Toyobo, Tokyo, Japan) in the presence of 5% (v/v) dimethyl sulfoxide. Each amplified cDNA fragment was subcloned into a pGEM-T Easy vector (Promega, Tokyo, Japan) and sequenced in a 377 DNA sequencer (Applied Biosystem, Tokyo, Japan).

Materials
In Situ Hybridization-Brains were quickly removed from postnatal day (P) 7, P14, and P21 ddY mice and 7-week-old (W) male ddY mice and frozen in powdered dry ice. Consecutive brain sections were cut at a thickness of 16 m with a cryostat (Leica Microsystems, Tokyo, Japan), thaw-mounted onto 3-aminopropyltriethoxysilane-precoated glass slides, and stored at Ϫ80°C prior to use. 35 S-Labeled riboprobes were transcribed using a MAXIscript TM T7 kit (Ambion, Austin, TX) with 5Ј-␣ [ 35 S]thiotriphosphate (ϳ30 TBq/mmol) (Amersham Biosciences). Before being loaded on the slides, the radiolabeled probes were mixed individually into a hybridization buffer (50% (v/v) formamide, 10% (v/v) dextran sulfate, 2.5ϫ Denhardt's solution, 4ϫ saline-sodium citrate (SSC), 5 mM EDTA, pH 8.0, 0.5 mg/ml tRNA from brewers' yeast (Roche Diagnostics), 0.1 mg/ml DNA from herring sperm (Wako Pure Chemical Industries, Tokyo, Japan), and 20 mM dithiothreitol) at a concentration of 10 7 cpm/ml. In situ hybridization with 35 S-labeled riboprobes was performed as described previously (30). Briefly, the tissue sections were fixed in 4% (v/v) formaldehyde in phosphate-buffered saline at room temperature for 15 min, followed by digestion with proteinase K (2 mg/ml) at 37°C for 10 min, acetylation in 0.25% (v/v) acetic anhydride in 0.1 M triethanolamine at room temperature for 10 min, and dehydration through graded concentrations of ethanol. After drying, the tissue sections were pretreated with the hybridization buffer without 10% (w/v) dextran sulfate at 55°C for 1 h and then with the hybridization buffer containing the respective probes prepared as above, and incubated in a moisture chamber at 55°C for 18 h. The slides were rinsed twice in 2ϫ SSC containing 10 mM 2-mercaptoethanol at room temperature for 5 min, digested with RNase A (50 g/ml) at 37°C for 30 min, rinsed twice in 50% (v/v) formamide, 2ϫ SSC, 10 mM 2-mercaptoethanol at 55°C for 30 min and subsequently in 2ϫ SSC, 10 mM 2-mercaptoethanol at room temperature for 10 min, and dehydrated with a series of concentrations of ethanol. For signal detection, the processed sections were exposed to x-ray films (BioMax MR; Eastman Kodak Co.) for 1 week. At least three independent experiments were carried out for each probe.
Quantitative Real Time RT-PCR-Total RNA was extracted from entire brains or cerebella of P7, P14, and P21 ddY mice and 7W male ddY mice using a QuickPrep total RNA extraction kit (Amersham Biosciences). The cDNA was synthesized from ϳ1 g of the total RNA using Moloney murine leukemia virusreverse transcriptase (Promega) and an oligo(dT) 20  Determination of the Disaccharide Composition of CS/DS Chains in the Brain Sections on Glass Slides-The digestion of CS/DS chains in the brain sections on glass slides with CS/DSdegrading enzymes was carried out as described previously (31) with slight modifications. The tissue sections prepared as above were fixed with acetone/methanol (1:1, v/v) at room temperature for 3 min and dried. For the analysis of CS/DS chains in the cerebella, unnecessary portions of the brain sections were scratched off with a fine surgical knife under a dark field microscopic view. The photographs of the clipped tissue sections were scanned under a dark field microscope (SZX12; Olympus, Tokyo, Japan) equipped with a digital camera (HC-300Z/OL; Olympus) for calculating the area using morphological analysis software (Mac Scope; Mitani Corp., Tokyo, Japan). The sections were rehydrated in distilled water at room temperature for 10 min. Depending upon the relative size of each section, an appropriate volume (25-150 l) of an enzyme solution mixture containing chondroitinase ABC (25 mIU/ ml) or AC-I (25 mIU/ml) in 50 mM Tris-HCl buffer, pH 7.3 (32), was applied on the sections. It should be noted that the enzymatic digestion with chondroitinase ABC was also carried out in the above-mentioned buffer instead of the conventional buffer (50 mM Tris-HCl buffer, pH 8.0, containing 60 mM sodium acetate) (32), based on the prior selection of an optimal buffer for digestion with chondroitinase ABC and subsequent efficient labeling with a fluorophore, 2-aminobenzamide , and C4ST-2 (I-L), and exposed to an x-ray film for 1 week. No labeling was observed in their respective neighboring sections hybridized with the corresponding cRNA sense probes (data not shown). Both C4ST-1 and C4ST-2 transcripts showed a widespread expression throughout the postnatal development, whereas the D4ST-1 transcript was detected at a low yet significant level in the cerebellum. OB, olfactory bulb; CPu, caudate putamen; Cx, cerebral cortex; Hi, hippocampus; Th, thalamus; Md, midbrain; Cb, cerebellum; Bs, brainstem. Scale bars, 5 mm.
(2AB). 4 The sections were incubated in a moisture chamber at 37°C for 2 h. The enzyme solutions were then collected, dried, and subjected to derivatization with 2AB as described previously (33). After the removal of excess 2AB reagent by paper chromatography, the resultant 2AB derivatives of the unsaturated disaccharides listed below were identified and quantified by anion-exchange HPLC on an amine-bound silica PA03 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 (33) or by gel filtration on a Superdex TM Peptide HR10/30 column (Amersham Biosciences) equilibrated with 0.2 M NH 4 HCO 3 containing 7% (v/v) 1-propanol as described previously (24): Alternatively, a GAG fraction was prepared from brain sections by the conventional method (24,34) with slight modifications. Several brain sections were directly homogenized with acetone, and the acetone-insoluble materials were collected by centrifugation. After drying, the CS/DS chain-containing precipitates were treated with 1 M NaBH 4 , 0.05 M NaOH at room temperature for 2 h to liberate O-linked saccharides including GAG chains from the core proteins. After neutralization with 1 M acetic acid and subsequent treatment with 5% (w/v) trichloroacetic acid, a GAG-containing fraction was recovered by ethanol precipitation. An aliquot of the resultant GAG fraction was used for the analysis of the disaccharide composition described above.
Brain sections on glass slides were immunostained with the monoclonal anti-CS antibody 473HD, which has been originally reported to recognize the GAG moiety of a brain-derived CS-PG, DSD-1-PG/phosphacan, and has long been postulated to react with unique CS/DS hybrid structures (20), before and after treatment with the enzyme solution containing chondroitinase ABC to confirm the efficient removal of CS/DS chains from the sections.  (28,35), mainly D4ST-1 has been considered to be involved in the formation of IdoUA in the biosynthesis of CS/DS hybrid chains (28,35). Thus, to gain insights into the distribution of such IdoUA-containing CS/DS hybrid chains in the mouse brain during postnatal development, expression of the mouse counterparts of the three human family members of CS/DS GalNAc 4-O-sulfotransferases, D4ST-1, C4ST-1, and C4ST-2, was examined by in situ hybridization with their respective 35 S-labeled antisense probes. Fig. 1 shows autoradiograms of the sagittal sections from the brains of P7, P14, P21, and 7W mice. Distinct patterns of labeling were observed for C4ST-1 and C4ST-2 in widespread regions of the developing brain, including the olfactory bulb, caudate putamen, cerebral cortex, hippocampus, thalamus, midbrain, and cerebellum ( Fig. 1, E-L). In the negative control experiments with the respective 35 S-labeled sense probes, essentially no label was observed in any region (data not shown), confirming the specific hybridization of the antisense probes. In contrast, the expression level of D4ST-1 was low compared with levels of C4STs. Interestingly, however, the transcript of D4ST-1 was exclusively expressed in the cerebellum throughout the postnatal developmental stages examined ( Fig. 1, A-D). Similar results were obtained in the autoradiograms of consecutive coronal sections of the brain, including the cerebellum (Fig. 2). Consistent with these findings, quantitative real time RT-PCR revealed that the expression level of D4ST-1 in the cerebellum was relatively higher than that in the whole brain and culminated at P14 during postnatal development (Fig. 3A), whereas C4ST-1 was expressed at a similar level in both the cerebellum and the whole brain at all stages exam-4 C. Mitsunaga, T. Mikami, and K. Sugahara, unpublished results. ined (Fig. 3B), indicating its ubiquitous expression in the postnatal brain. These gene expression patterns suggest that IdoUA-containing CS/DS hybrid chains are relatively abundant in the cerebellum during postnatal brain development.

Expression of CS/DS GalNAc 4-O-Sulfotransferases in the
Quantification of CS/DS Chains in Brain Sections on Glass Slides-In order to study the regional distribution pattern of IdoUA-containing CS/DS hybrid structures in the postnatal mouse brain, the disaccharide composition of the CS/DS chains was determined using the in situ digestion method in histological sections on glass slides with CS/DS-degrading enzymes (31), because this method has been established for rapid determination of the concentration of CS/DS chains in the targeted local area with a defined thickness on the tissue section. Because the DS-specific D4ST-1 was abundantly expressed in the cerebellum, the analysis was focused on the structural difference in CS/DS chains between the cerebellar region and the entire brain. For the analysis of the cerebellar CS/DS, brain sections on glass slides were processed to analyze the cerebellar region by removing the other parts with a small surgical knife. The enzyme solution containing chondroitinase ABC was mounted on the respective sections to digest the CS/DS chains directly on the slides. Chondroitinase ABC catalyzes the eliminative cleavage of almost all the galactosaminidic linkages in CS/DS chains (36,37). Exhaustive treatment with chondroitinase ABC efficiently removed CS/DS chains from the sections, as confirmed by the elimination of the immunoreactivity toward anti-CS mAb 473HD in the sections (data not shown). The enzyme reaction mixture was recovered and derivatized with a fluorophore, 2AB. The 2AB derivatives of the CS/DS disaccharide components were subjected to anion-exchange HPLC, and their fluorescence intensity was monitored. Representative chromatograms obtained using enzyme digests of the whole brain sections and cerebellar sections are shown in Fig. 4. The unsaturated CS/DS disaccharides observed in the digests were identified by comparison with the elution positions of the authentic 2AB-labeled disaccharide standards. To verify the validity of the in situ method, the conventional method was also used, where CS/DS chains were first purified from the sections, then digested with chondroitinase ABC, and subjected to anion-exchange HPLC after 2AB-labeling (as described under "Experimental Procedures"). A similar chromatographic profile of the fluorescently labeled disaccharides was observed (data not shown), except that the recovery of total disaccharide obtained with the conventional method was ϳ70% less than that obtained using the in situ method. Thus, the in situ method was more efficient for the highly sensitive detection and quantitative estimation of CS/DS disaccharides derived from the tissue sections. The lower recovery with the conventional method  was assumed to be due to the low efficiency of the purification of minute amounts of the materials.
The data obtained from the direct digestion of CS/DS chains in the brain sections are summarized in Table 1. Accurate values for the nonsulfated disaccharide ⌬O (⌬Di-0S) could not be obtained because of the methodological imperative of its loss during the step of purification to remove the excess reagents for 2AB derivatization and impurities derived from tissue sections. Nevertheless, the approximate proportion of ⌬O was roughly estimated for the CS/DS chains derived from tissue sections of whole brain and the cerebellum at P14 and 7W after the removal of excess reagents by paper chromatography before HPLC analysis. The results indicated that ϳ5% of all the disaccharides were ⌬O in both samples and that the disaccharide composition of CS/DS chains, including ⌬O, was essentially unchanged. Therefore, ⌬O was not pursued in this study. In both the whole brain and cerebellar sections, the proportion of ⌬A increased gradually, whereas that of ⌬C decreased with development, which was consistent with the previously reported developmental change in the sulfation pattern of the CS chains from embryonic chick brain (10). The temporal change in composition in the cerebellar sections was more dynamic than that in the whole brain sections (Table 1), which may reflect the dramatic development of the cerebellar system after birth. Interestingly, a marked (2-fold) increase in the concentration of total CS/DS in the cerebellum was observed, whereas that in the entire brain was moderate (20%) ( Table 1).
The proportions of disulfated disaccharide components, ⌬D, ⌬E, and ⌬B, in the CS/DS chains obtained from both the whole brain and cerebellar sections were relatively small throughout the postnatal development (Table 1). However, growing evidence suggests the biological significance of even such small proportions of disaccharide units in CS/DS chains (3). Therefore, the temporal change in their proportion was also examined. In the whole brain sections, the proportion of ⌬B and ⌬E showed a gradual increase and concomitant decrease with development, respectively, whereas that of ⌬D appeared unaltered. In contrast, dramatic changes in the proportion of the three disulfated disaccharide components in the cerebellar CS/DS chains were observed; the proportion of ⌬D and ⌬E decreased to 50 and 30%, respectively, whereas that of ⌬B increased as much as 10-fold with development. The marked sharp increase in the proportion of ⌬B in cerebellar CS/DS is especially noteworthy and may provide clues for understanding the distribution of IdoUA-containing CS/DS hybrid structures in the postnatal brain as described below.

A Significant Proportion of the IdoUA-containing Disaccharide Unit iA Exists in the Cerebellar CS/DS Chains-Recently, it
was demonstrated that CS/DS chains isolated from a soluble PG fraction of embryonic pig brains contained a significantly higher proportion (8 -9%) of iA than the corresponding fraction obtained from adult pig brains (24). Therefore, iA units were compared between whole brain and the cerebellum. Unsaturated disaccharide ⌬A, released by digestion with chondroitinase ABC, is derived from both disaccharide A (GlcUA-GalNAc(4S) and iA (IdoUA-GalNAc(4S)) units embedded in the CS/DS chains, where "i" stands for IdoUA. Because chondroitinase AC-I cleaves the galactosaminidic linkages bound to GlcUA but not to IdoUA or 2-Osulfated GlcUA (36,37), the linkages sensitive to chondroitinase ABC but resistant to chondroitinase AC-I can be used as a criterion for estimating the IdoUA-containing structures in CS/DS chains. As expected, the amount of ⌬A obtained from the brain sections after treatment with chondroitinase AC-I was always slightly less than that obtained from the corresponding sections treated with chondroitinase ABC. Interestingly, the ratio of ⌬A in the chondroitinase AC-I digest to that in the chondroitinase ABC digest was 0.95-0.99 for the whole brain sections and 0.89 -0.93 for the cerebellar sections (Table 2). These results suggest that the proportion of iA in the total 4-O-sulfated disaccharide units in the whole brain sections and cerebellar sections was 1-5 and 7-11%, respectively (Table 2), although the proportion tend to be slightly overestimated in view of the resistance of D units to the action of chondroitinase AC-I. The relatively larger proportion of iA in the CS/DS chains of the cerebellum rather than whole brain suggested the significant expression of the DS-type structure in the developing cerebellum. These results confirm the notion that embryonic brain contains a significant proportion of iA besides the predominant A unit (25), and verify the validity of the above-mentioned proposal that the gene expression profiles of CS/DS GalNAc 4-Osulfotransferases, especially that of D4ST-1, reflect the distribution

composition of CS/DS chains obtained by digestion with chondroitinase ABC of whole brain sections and cerebellar sections of mice on glass slides
Unsaturated disaccharides generated by digestion of CS/DS chains with chondroitinase ABC were analyzed by anion-exchange HPLC after labeling with the fluorophore 2AB as described in the legend to Fig. 4. The values obtained from several independent experiments were used to calculate picomoles (mol %) of unsaturated disaccharides per mm 3 of each section, which is expressed as the means Ϯ S.E. Note that the nonsulfated disaccharide ⌬O was not obtained due to a methodological restriction, its loss during the purification step for removing the excess 2AB-labeling reagents.   JULY 14, 2006 • VOLUME 281 • NUMBER 28

CS/DS Hybrid Chains in the Brain
pattern of IdoUA-containing CS/DS hybrid chains in the postnatal brain.

Expression of D4ST-1 and UST Involved in the Biosynthesis of Disulfated Disaccharide iB Units in the
Cerebellar CS/DS Chains-As mentioned above, the proportion of ⌬B in the cerebellar CS/DS chains was dramatically increased by ϳ10-fold from P7 to 7W, despite its relatively small increase (ϳ2-fold) in CS/DS chains derived from whole brain during the same time period. ⌬B is formed by chondroitinase digestion from a disulfated B unit (GlcUA(2S-)-GalNAc(4S)) and iB unit (IdoUA-(2S)-GalNAc(4S)). However, it should be noted that the B unit has never been reported in mammalian tissues, although it was recently detected for the first time in shark skin (23). Therefore, it is most likely that ⌬B was derived from iB rather than B units. Indeed, ⌬B was not produced by chondroitinase AC-I specific for CS. Because the iB unit is a characteristic component of typical DS chains, comparison of the proportion of iB in the total CS/DS disaccharides in a particular brain region is useful for a preliminary survey of the distribution of CS/DS hybrid structures.
In the biosynthesis of DS chains containing the iB unit, it has been postulated that a biosynthetic intermediate (IdoUA-GalNAc(4S); iA unit) is first produced from a disaccharide unit (GlcUA-GalNAc; O unit) by the cooperative actions of C5 epimerase and D4ST-1 (28,35,38). UST catalyzes thereafter 2-Osulfation of uronyl residues, particularly IdoUA residues in iA units in DS chains (29). As expected, in situ hybridization revealed that UST, as well as D4ST-1, was highly expressed in cerebellum during postnatal development (Fig. 5). Real time RT-PCR also revealed that the expression level of UST was relatively high in the cerebellum throughout development and culminated at P14 (Fig. 3C). Thus, the iA units embedded in cerebellar CS/DS chains may be efficiently sulfated by the action of UST, leading to the high production of the DS-specific iB unit in the postnatal cerebellum.
The expression levels of another two sulfotransferases, chondroitin 6-O-sulfotransferase-1 (C6ST-1) and GalNAc 4-sulfate 6-O-sulfotransferase (GalNAc4S-6ST), were also analyzed by real time RT-PCR. The former catalyzes the 6-O-sulfation of GalNAc residues and contributes substantially to the formation of the C and D units (39,40), whereas the latter catalyzes the 6-O-sulfation of GalNAc(4S) in A units and is responsible for the formation of the E unit (41). In contrast to D4ST-1 and UST, although the expression levels of C6ST-1 and GalNAc4S 6ST in the cerebellum were higher than those in the whole brain, they gradually decreased with development (Fig. 3, D and E). Their temporal expression profiles in the cerebellum may support the gradual decrease in the proportions of ⌬C, ⌬D, and ⌬E in the cerebellar CS/DS chains with development ( Table 1).
Distribution of IdoUA Residues along Mouse Brain CS/DS Chains-To further characterize the structural properties of the IdoUA-containing CS/DS hybrid chains, the distribution of IdoUA residues along the chains was compared between CS/DS chains derived from the entire brain and cerebellum at 7W. Each section was treated with chondroitinase AC-I, and the

TABLE 2 Determination of iA units in the cerebellum during postnatal development
The CS/DS chains existing in the brain sections were exhaustively digested with chondroitinase ABC or AC-I (abbreviated as ABC and AC-I, respectively), and the resultant unsaturated disaccharide ⌬A in addition to other units was identified and quantified by anion-exchange HPLC several times as described under "Experimental Procedures." Values are expressed as the mean Ϯ S.E. Note that ⌬A is generated by chondroitinase digestion from both A and iA units in CS/DS chains. The average ⌬A values obtained after digestion with AC-I and ABC from several independent experiments (N) were used to calculate mol % of iA units in the total of iA and A.

Stages Enzymes
Unsaturated disaccharide ⌬Di-4S digest was labeled with 2AB and analyzed by gel filtration. As shown in Fig. 6, fluorescent peaks derived from each digest were detected at positions corresponding to tetra-, hexa-, and octasaccharides, in addition to monosulfated and disulfated disaccharides, which are designated mono-S and di-S fractions, respectively. Considering the substrate specificity of chondroitinase AC-I (36,37), the oligosaccharides resistant to chondroitinase AC-I should be derived from sequences containing iA, iB, iC, iD, iE, and/or D units (GlcUA(2S)-GalNAc(6S)). Each broad peak close to V 0 in both chromatograms (Fig. 6) was composed of unidentified impurities, as judged by the resistance to treatment with chondroitinase ABC followed by 2AB labeling, which gave no fluorescent peaks corresponding to dis-accharides (data not shown). In addition, anion-exchange HPLC analysis showed that the di-S fraction contained a small proportion of ⌬E and a larger proportion of impurities, which were not eluted from the column (data not shown). The molar ratio (oligosaccharides/disaccharides) of the oligosaccharides (the sum of octa to tetra) to those of the disaccharides was calculated based on the peak area, excluding the impurities in the di-S fraction ( Table 3). The oligosaccharides/disaccharides ratio in the cerebellar CS/DS chains was ϳ3 times that in the whole brain CS/DS chains. These results suggest that characteristic domain structures constituted of IdoUA-containing iA, iB, iC, iD, and iE units or GlcUA-containing rare D units are embedded ϳ3 times more frequently in the cerebellar CS/DS chains compared with the CS/DS chains derived from other areas of the brain during postnatal development.

DISCUSSION
In this study, the characteristic spatiotemporal distribution of IdoUA-containing CS/DS hybrid chains in the postnatal mouse brain was demonstrated by in situ hybridization for sulfotransferases involved in the biosynthesis of DS and by chemical in situ analysis of CS/DS chains of the brain sections. Both approaches showed that IdoUA-containing CS/DS hybrid structures are more concentrated in the cerebellum when compared with whole brain, providing insight into the important biological functions of IdoUA-containing CS/DS hybrid chains during postnatal cerebellar development as discussed below.
Typical mammalian DS chains are characterized by distinct iA and iB disaccharide units as exemplified by DS in skin (42). 4-O-Sulfation of GalNAc is postulated to be responsible for the prior formation of IdoUA by C5-epimerization of GlcUA (38), which is an initial step in the biosynthesis of DS. Our previous study revealed that the 4-O-sulfation is catalyzed by members of the CS/DS 4-O-sulfotransferase family as follows: D4ST-1, C4ST-1, and C4ST-2 (28). Their substrate preferences suggest that D4ST-1 contributes substantially to the formation of IdoUA in vivo, resulting in production of the iA unit along the precursor chains. In addition, Kobayashi et al. (29) has reported that UST catalyzes efficiently the 2-O-sulfation of IdoUA residues of iA units, forming iB units. The preferential expression of D4ST-1 and UST in the developing cerebellum (Figs. 1-3 and 5) was positively correlated with the relatively larger proportions of iA and iB units along the cerebellar CS/DS chains compared with the chains from whole brain (Table 1). These find-FIGURE 6. Analysis of the distribution of IdoUA residues in the CS/DS chains expressed in the postnatal mouse brain. Tissue sections of whole brain (ϳ53 mm 2 ) (A) and the cerebellum (ϳ36 mm 2 ) (B), respectively, of 7W mice were treated with an enzyme solution containing chondroitinase AC-I. The size distribution of the resultant 2AB-labeled oligosaccharides was analyzed by gel filtration as described under "Experimental Procedures" The broad peaks marked by horizontal bars are derived from unidentified impurities as they were not digested by chondroitinase ABC any more. The numbered arrows indicate the elution positions of 2AB-labeled, size-defined DSoligosaccharides derived from porcine skin or a mixture of authentic CS disaccharides: arrow 1, octasaccharides; arrow 2, hexasaccharides; arrow 3, tetrasaccharides; arrow 4, disulfated disaccharides (di-S); and arrow 5, monosulfated disaccharides (mono-S).

TABLE 3 Molar ratios of oligosaccharides to disaccharides produced by digestion with chondroitinase AC-I of postnatal mouse brain CS/DS chains
Whole brain and cerebellar sections of 7W mice were digested with chondroitinase AC-I, and the digests were analyzed by gel filtration as described in the legend to Fig.  6. The yield of each oligosaccharide fraction was calculated based on the peak area in the chromatograms shown in Fig. 6. The mono-S plus di-S fractions were considered the disaccharide fraction, whereas the tetra-and hexasaccharides plus octasaccharides were treated as oligosaccharides. Note that the values derived from unidentified impurities were subtracted from the peak area of the di-S fraction.  JULY 14, 2006 • VOLUME 281 • NUMBER 28

CS/DS Hybrid Chains in the Brain
ings verify the reliability and potential applicability of these gene expression profiles for probing the presence of IdoUAcontaining CS/DS hybrid chains. In contrast, C4ST-1 and C4ST-2 showed a widespread expression in the postnatal brain ( Figs. 1-3). Although their expression cannot be used for surveying the distribution of IdoUA-containing DS-type structures, they might also be partially involved in the formation of IdoUA in CS/DS chains in the postnatal brain. It has been reported that the sulfation profiles of CS chains change markedly with embryonic development in the chick brain (10). This change, which is characterized by an increase in 4-O-sulfation and a decrease in 6-O-sulfation, was also observed in the postnatal mouse brain-derived CS/DS chains, especially in the cerebellar region (Table 1), where neural development proceeds dramatically during the postnatal period. In addition, temporal change in the proportions of iA and iB units was also observed. Despite the up-regulation of D4ST-1 expression in the developing cerebellum, the postnatal maturation of the cerebellum resulted in a decrease in the proportion of iA among A and iA from 11 (at P7 and P14) to 7% (at P21 and 7W) ( Table 2). This apparent discrepancy may be explained by a significant increase in the iB units (Table 1), which are produced by UST efficiently utilizing a fraction of the preformed iA units as intermediates. In view of recent reports about the contribution of IdoUA-containing disaccharides in addition to the sulfation profiles to the neuritogenic activity of embryonic pig brain-derived CS/DS (24,25), the dynamic changes in disaccharide composition, including iA and iB units, in the cerebellar CS/DS chains appear to reflect the neural development and/or maturation of the cerebellum.
Increasing evidence shows that most neuritogenic CS/DS chains possess the ability to bind growth factors (3,20,(22)(23)(24)43), suggesting that these neuroregulatory activities are exerted at least in part through signal transductions mediated by several growth factors. Most interestingly, Bao et al. (25) have reported that embryonic pig brain-derived CS/DS, which contains a significant proportion of IdoUA residues, supports in part pleiotrophin (PTN)-mediated neuritogenesis of primary hippocampal neurons by recruiting endogenous PTN produced by glial cells. PTN, a heparin-binding growth factor, binds not only to heparan sulfate (HS)-PGs but also to CS-PGs through their CS moieties. Typically, such CS-PGs include a cell membranetethered receptor-type tyrosine phosphatase, PTP␦/RPTP␤, and its splicing variant, phosphacan/DSD-1-PG, which does not have the carboxyl-terminal membrane-spanning domain and is a soluble matrix component (44 -46). The PTN-PTP␦ signaling regulates the morphogenesis of Purkinje cell dendrites in the slice cultures of the developing cerebellum, and an exogenous application of CS variants such as CS-C, CS-D, and CS-E into the culture medium causes abnormal morphogenesis (15). In addition, Maeda and co-workers (16,47) reported that the affinity of PTN for phosphacan depends on the structure of its CS chains, especially on the structure containing D units. In view of the fact that UST is responsible for the biosynthesis of the D unit in addition to the DS-specific iB unit (29), the preferential expression of UST in the developing cerebellum (Figs. 3C and 5) is consistent with the notion that CS chains with oversulfated disaccharide D units are involved in the neurito-genesis of cerebellar neurons. The decrease in the proportion of D unit with cerebellar development (Table 1) may be attributed to a decreased production of its precursor disaccharide unit (GlcUA-GalNAc(6S)) ( Table 1), which probably results from a gradual decrease in the expression of C6ST-1 with development (Fig. 3E). In contrast, the temporal change in the proportion of another disulfated disaccharide unit E may be regulated by the expression level of GalNAc4S-6ST (Fig. 3D).
Interestingly, CS-D from shark cartilage containing D units and an oversulfated DS preparation derived from ascidian Styela plicata, which is characterized by a prominence of the disaccharide unit iB, promote the outgrowth of neurites in rodent hippocampal neurons in vitro, resulting in the induction of multiple shorter neurites. In contrast, they do not produce an axonic longer neurite, which is typically observed in neurons cultured on defined substrates such as other marine organismderived CS/DS hybrid chains containing mainly E/iE units (13,21). The apparently similar morphogenetic potential of D and iB units is consistent with the finding that the cerebellar CS/DS chains with significant proportions of iB, in addition to D, function at least in part in the neuritogenesis of cerebellar neurons during postnatal development. In support of this notion, we recently observed that DS preparation from ascidian S. plicata promoted neurite outgrowth not only of hippocampal neurons (21) but also of cerebellar neurons when used as a substrate for dissociated culture. 5 However, several studies have suggested that CS/DS chains, which can promote neurite outgrowth of a particular cell type, do not always guarantee neuritogenesis of another. For example, CS-E, which is characterized by a predominance of the E unit, promotes neurite outgrowth of embryonic mouse hippocampal neurons in vitro (13,21), although it is potently inhibitory toward dorsal root ganglion explants from chick embryos (48). DSD-1-PG has been also reported to show similar opposing effects on neuritogenesis dependent on neuronal lineages (49). In view of these findings, further investigation is needed for clarification of the biological importance of CS/DS hybrid chains containing iA and/or iB units in the primary cerebellar neurons.
Notably, recent studies demonstrated that oligosaccharide structures rich in iA units present in DS chains are capable of promoting the signal transduction mediated by basic fibroblast growth factor (bFGF) (50). bFGF has various neuroregulatory functions influencing neuronal proliferation, differentiation, and neuroprotection (51)(52)(53). Although it is well established that the bFGF-mediated events are supported by HS chains as co-receptors (54), the CS/DS hybrid chains expressed in the postnatal cerebellum may also support bFGF-mediated neurotrophic effects on cerebellar neurons such as Purkinje cells and granular cells.
Gel filtration analysis of the enzymatic digest of the brain sections obtained with chondroitinase AC-I revealed that the clusters consisting of IdoUA-containing iA and iB units, and GlcUA-containing rare D units were distributed ϳ3 times more frequently in the cerebellar CS/DS chains than in the whole brain CS/DS chains (Fig. 6). Because the digestion of mouse brain CS/DS 5 T. Mikami and K. Sugahara, unpublished results. chains with chondroitinase B, which cleaves the galactosaminidic linkages bound to IdoUA residues (37,55), gave no detectable disaccharide units (data not shown), the above-mentioned oligosaccharide clusters may not be composed of consecutive IdoUA-containing disaccharides, but rather of IdoUA-rich disaccharide sequences containing other units such as D unit(s). These characteristic domain structures have also been observed in the embryonic pig brain CS/DS subfractions, which promote a dendrite-like neurite outgrowth (25), supporting the possible requirement of such distinct CS/DS hybrid domain structures for the neuritogenic activities toward cerebellar neurons.
Interestingly, the abundant distribution of IdoUA-containing CS/DS hybrid chains in the postnatal cerebellum is probably correlated with the epitope of the monoclonal antibody (mAb) 473HD (12). 473HD has long been postulated to react with CS/DS hybrid structures, and its epitope was preferentially distributed in the cerebellum among adult brain subregions examined (12,20). Several studies have shown that the 473HD epitope was also distributed in the embryonic brain (49,57), where dynamic neurogenesis occurs before onset of cerebellar formation. 473HD can neutralize the neurite outgrowth-promoting activities of CS-D and embryonic pig brain CS/DS (13,25). Most recently, Ito et al. (56) showed that the 473HD epitope might be associated with structure(s), which include IdoUA-containing units, in addition to A-D and/or D-A tetrasaccharide sequences. These findings may imply that functional IdoUA-containing CS/DS hybrid chains are prominent in the brain subregions where the drastic formation of neural network is observed. Therefore, application of the systematic analyses performed in the present study to the embryonic central nervous system will facilitate a more comprehensive understanding of the biological functions of IdoUA-containing CS/DS hybrid chains in the developing central nervous system, including a hippocampus, whose neurons extend neurites in response to the stimulation with brain CS/DS chains from the embryonic stage (24,25).