Distinctive Expression Patterns of Heparan Sulfate O-Sulfotransferases and Regional Differences in Heparan Sulfate Structure in Chick Limb Buds*

The skeletal tissue development and patterning in chick limb buds are known to be under the spacio-temporal control of various heparin-binding cell growth factors such as fibroblast growth factors and bone morphogenetic proteins. Different structural regions on heparan sulfate (HS) chains of proteoglycans could be implicated in regional differences in the binding capacities of these cell growth factors, by which they could selectively interact with targeted cells and regulate their signaling in those processes. In this study we first demonstrated by cDNA cloning that one heparan sulfate 2-O-sulfotransferase (HS2ST) and two isoforms of heparan sulfate 6-O-sulfotransferase (HS6ST-1 and -2) occurred in chick embryos and had different substrate specificities each other. We next showed by whole mount in situ hybridization that the HS6ST-1 and HS6ST-2 transcripts were preferentially localized to the anterior proximal region and at the posterior proximal region of the limb bud, respectively, whereas the HS2ST transcript was distributed rather uniformly throughout the bud. Analyses of the structures of HS from different regions of the wing buds have shown variation in that 6-O-sulfated residues are more abundant in the proximal than distal region, whereas iduronosyl 6-O-sulfated residues are abundant in the anterior proximal region and glucuronosyl 6-O-sulfated residues in the posterior proximal region. These results suggest that HS with different sulfation patterns created with multiple sulfotransferase activities provides an appropriate extracellular environment for morphogenetic signal transduction.

The skeletal tissue development and patterning in chick limb buds are known to be under the spacio-temporal control of various heparin-binding cell growth factors such as fibroblast growth factors and bone morphogenetic proteins. Different structural regions on heparan sulfate (HS) chains of proteoglycans could be implicated in regional differences in the binding capacities of these cell growth factors, by which they could selectively interact with targeted cells and regulate their signaling in those processes. In this study we first demonstrated by cDNA cloning that one heparan sulfate 2-O-sulfotransferase (HS2ST) and two isoforms of heparan sulfate 6-Osulfotransferase (HS6ST-1 and -2) occurred in chick embryos and had different substrate specificities each other. We next showed by whole mount in situ hybridization that the HS6ST-1 and HS6ST-2 transcripts were preferentially localized to the anterior proximal region and at the posterior proximal region of the limb bud, respectively, whereas the HS2ST transcript was distributed rather uniformly throughout the bud. Analyses of the structures of HS from different regions of the wing buds have shown variation in that 6-O-sulfated residues are more abundant in the proximal than distal region, whereas iduronosyl 6-O-sulfated residues are abundant in the anterior proximal region and glucuronosyl 6-O-sulfated residues in the posterior proximal region. These results suggest that HS with different sulfation patterns created with multiple sulfotransferase activities provides an appropriate extracellular environment for morphogenetic signal transduction.
In developing limb buds, reciprocal interactions between a specialized ectodermal structure called the apical ectodermal ridge (AER), 1 and the underlying mesoderm are essential for outgrowth and the patterning of skeletal tissues (1). These functions of the AER and mesoderm appear to be controlled at least in part by members of the FGF family, FGF-4, FGF-8, and FGF-10, which are expressed by either tissue (2)(3)(4)(5)(6)(7)(8). For example, mesodermal fibroblast growth factor-10 (FGF-10) is essential for limb bud induction and the maintenance of the AER. The AER promotes proliferation and outgrowth of the sub-ridge mesoderm and maintains the activity of the polarizing zone at the posterior margin of the limb bud (2,3,9). Sonic hedgehog (SHH) signaling by the polarizing region modulates fibroblast growth factor-4 (FGF-4) signaling by the posterior AER, which in turn maintains the polarizing region (SHH/FGF-4 feedback loop) (10).
Signaling of FGFs has recently been shown to involve heparan sulfate (HS) chains attached to the core protein of heparan sulfate proteoglycans (HSPGs). The ternary complex composed of FGF family molecules, FGF receptors (FGFRs), and HS chains is formed on the cell surface (11)(12)(13)(14)(15)(16). In addition, HSPGs may bind growth factors in the extracellular matrix and thereby alter their stability and modulate their effective concentrations (17)(18)(19). Syndecan-3 (integral membrane HSPG) plays such a role in chick limb outgrowth by controlling the FGF signaling between the AER and mesoderm (20). Glypican-3 (glycosylphosphatidylinositol-anchored cell surface HSPG) controls cellular responses to bone morphogenetic protein-4 (BMP-4) in limb patterning (21). Recent analyses in mice of genetic defects in enzymes for HS synthesis further suggest important roles for the interactions between HS and several growth factors in a variety of morphogenetic events, including limb development (22). For example, disruption of the gene for perlecan, an HSPG abundant in basement membranes and connective tissues, caused a severe abnormality of skeletal development by affecting the FGFR-3 signaling pathway (23). In addition, graft experiments of extracellular matrix from the Mouse Posterior Limb Bud-4 cell line suggest that the inhibition of polarizing activity in the anterior limb is regulated by extracellular factors, which are bound to the moieties of extracellular matrix considered as heparan sulfate proteoglycans (24). * This work was supported by Grant-in-aid for Scientific Research on Priority Areas (14082206) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; by a Grant-in-Aid for Scientific Research (B) from the Japan Society for the Promotion of Science; by the preparatory grant for the research at the Matrix Glycoconjugate Group, Research Center for Infectious Disease, Aichi Medical University; and by a special research fund from Seikagaku Corporation. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  Specificities of the interactions between HSPG and ligand molecules are thought to reside, at least in part, in the fine structures of HS chains with specific sequences consisting of highly sulfated (N-, 2-O-, and/or 6-O-sulfated) monosaccharides, such as binding sites for acidic or basic FGF and hepatocyte growth factor (18,(25)(26)(27)(28)(29)(30)(31)(32). The fine structures of HS are also essential for interactions between cells and/or between the cell and extracellular matrix during development and morphogenesis (33). Such structures are generated by complex but strictly regulated modification reactions during the biosynthesis of HS. Therefore, the modification of HS in developing tissues is worth studying to understand the structure-function relationships of HS. It has been shown that modification in the biosynthesis of HS entails a series of reactions: N-deacetylation and subsequent N-sulfation of N-acetylglucosamine (GlcNAc), conversion of glucuronic acid (GlcA) into iduronic acid (IdoA), 2-O-sulfation of IdoA, 6-O-sulfation of N-sulfated glucosamine (GlcNS), and 3-O-sulfation of GlcNS/glucosamine (33). In mammals, each reaction, except for GlcA C-5 epimerase (34,35) and HS 2-O-sulfotransferase (HS2ST) (36), is catalyzed by multiple isoforms that are derived from different but related genes, such as 6-O-sulfation of the GlcNS residue by three isoforms (HS6ST-1, -2, and -3) (37,38) and an alternative spliced form of HS6ST-2 (HS6ST-2S) (39). Since each isoform had a different substrate specificity and restricted tissue expression pattern, the occurrence of multiple isoforms in the biosynthesis of HS could be any important determinative factor for the fine structure of HS.
The aim of this study is to investigate the relation between expression patterns of HS O-sulfotransferases and differences of HS structure in developing tissues, especially with regard to roles of HS in regulating the activities of heparin-binding morphogenetic and cell growth factors. We used chick limb buds as a model of developing tissue, because many heparin-binding factors (FGFs, WNTs, BMPs, hepatocyte growth factors, and SHH) are known to be expressed in the limb buds and play pivotal roles in the development and subsequent patterning of skeletal tissues (1-10). We first cloned cDNAs for chicken homologues of mammalian HS2ST and HS6STs but only detected those for HS2ST, HS6ST-1, and HS6ST-2 at least in the chicken cDNA libraries examined. We next examined the expression of these transcripts in chick embryos by whole mount in situ hybridization. We then characterized the structures of heparan sulfate samples obtained from different regions of the wing bud and confirmed that the enzyme expression patterns reflected the regional differences in structure of heparan sulfate chains. We hypothesize that HS with different sulfation patterns created by multiple sulfotransferases provides regional extracellular environments appropriate for morphogenetic signal transduction.  Table I.
Screening of gt 11 cDNA Library-A cDNA library of limb buds of chick embryos at stage 21 was constructed in the lambda gt11 vector (Clontech, Palo Alto, CA) as described previously (41). Approximately 2 ϫ 10 6 plaques were screened using the following 32 (38). Hybond N ϩ nylon membrane replicas of the plaques from the gt11 cDNA library were subjected to alkali fixation by the manufacturer's recommended method, and prehybridized in a solution containing 5ϫ SSC, 20% formamide, 5ϫ Denhardt's, 0.5% SDS, 50 g/ml salmon sperm DNA overnight at 42°C. Hybridization was carried out at 42°C in the same buffer containing 32 P-labeled probe for 16 h, and the membranes were washed in 0.5 ϫ SSC, 0.1% SDS for 6 h at 42°C several times. Another cDNA library from chick retina at stage 23, which was constructed in the lambda gt11 vector (42), was also screened by the same method. Positive clones were subcloned into pBluescript II KS(Ϫ) (Stratagene, La Jolla, CA).
DNA Sequence Analysis-The subcloned cDNAs were sequenced on both strands by the dideoxy chain termination method using Taq polymerase (dye terminator cycle sequencing; PerkinElmer Life Sciences (Norwalk, CT) with a DNA sequencer (Applied Biosystems, PRISM 310). The obtained DNA sequences were compiled and analyzed using GENETYX-MAX computer programs (Software Development Co., Ltd., Tokyo, Japan). The deduced nucleic acid and protein sequences were compared with those of mouse and human HS2ST and HS6STs.
Transfection of cDNA and Transient Expression of Chick HS2ST, HS6ST-1, and HS6ST-2 in COS-7 Cells-For the construction of the expression vectors, the polymerase chain reaction products containing the open reading frames of those enzymes were ligated into the EcoRI/ EcoRV site of the pFLAG-CMV2 expression vector (Eastman Kodak Co., Rochester, NY). The inserted sequences were confirmed on a single strand as described above. COS-7 cells (5.5 ϫ 10 5 ) precultured for 48 h in a 60-mm culture dish were transfected with 5 g of each expression vector. The transfection was performed using LipofectAMINE according to the manufacturer's recommended protocol (TransFast, Promega, Madison, WI). After being cultured in Dulbecco's modified Eagle's medium containing 10% (v/v) fetal calf serum and antibiotics for 72 h, the cell layers were washed with Dulbecco's modified Eagle's medium alone, scraped, and homogenized in 1 ml of 10 mM Tris-HCl, pH 7.2, 0.5% (w/v) Triton X-100, 0.15 M NaCl, 20% glycerol, 10 mM MgCl 2 , and 2 mM CaCl 2 . The homogenates were subjected to stirring for 1 h and then centrifuged at 10,000 ϫ g for 30 min. Each FLAG fusion protein in the supernatant (cell extracts) was isolated by anti-FLAG M2 (Eastman Kodak Co.) affinity chromatography according to the method described by the manufacturer. Assay for Sulfotransferase Activities-Activities of HS2ST and HS6STs expressed in the FLAG-bound fractions were measured by monitoring the incorporation of [ 35 S]sulfate into various glycosaminoglycans as described previously (37). Briefly, the standard reaction mixture (50 l) contained 2.5 mol of imidazole-HCl, pH 6.8, 3.75 g of protamine chloride, 25 nmol (as hexosamine) (500 M) of acceptor glycosaminoglycans, 50 pmol of [ 35 S]PAPS (about 5 ϫ 10 5 cpm, 1 M), and enzyme. After incubation for 20 min at 37°C, the reaction was stopped by heating at 100°C at 1 min. Carrier chondroitin sulfate A (0.1 mol as glucuronic acid) was added to the reaction mixture, and the 35 S-labeled polysaccharides were isolated by precipitation with ethanol containing 1.3% potassium acetate and 0.5 mM EDTA, followed by gel chromatography on a Fast desalting column HR 10/10 (Amersham Biosciences) to remove [ 35 S]PAPS and its degradation products. The amounts of enzymes added to the reaction mixture were chosen to obtain a linear incorporation of [ 35 S]sulfate. One unit of enzyme activity was defined as the amount required to transfer 1 pmol of sulfate to CDSNS-heparin per minute.
Analysis of Enzymatic Reaction Products-The reaction products of enzymes were analyzed by high performance liquid chromatography (HPLC) as described previously with some modifications (18). Briefly, 35 S-labeled products were digested with a mixture of 10 milliunits of heparitinase I, 5 milliunits of heparitinase II, and 10 milliunits of heparitinase III in 40 l of 50 mM Tris-HCl, pH 7.2, 1 mM CaCl 2 , and 4 g of bovine serum albumin at 37°C for 2 h. The digests were subjected to gel chromatography on HiLoad 16/60 Superdex 30 pg, equilibrated with 0.2 M NH 4 HCO 3 . The 35 S-labeled fractions containing disaccharides and monosaccharides were injected into a PAMN column together with standard unsaturated disaccharides of the kit (Seikagaku Corp.). Fractions of 0.6 ml were collected, and radioactivity was measured.
Northern Blot Analysis-Poly (A) ϩ RNA samples from chick embryo tissues at various stages of development (stages 18 -19, 20 -21, 23-24, 26, and 31) were prepared, and 2 g of each sample was blotted onto the membranes. The membranes were prehybridized in ExpressHyb Hybridization Solution TM (Clontech) at 68°C for 30 min and then hybridized in the same solution containing 32 P-labeled probes (1 ϫ 10 6 cpm/ ml) at 68°C for 1 h. The probes used were a 2271-base fragment containing the coding region of chick HS2ST cDNA, a 1649-base fragment containing the coding region of chick HS6ST-1 cDNA, and a 1960-base fragment containing the coding region of chick HS6ST-2 cDNA, respectively. These probes were labeled with [␣-32 P]dCTP by random oligonucleotide priming (Ready-to-Go DNA labeling kit, Amersham Biosciences). The membranes were washed several times with 2ϫ SSC (w/v) and 0.05% (w/v) SDS at room temperature and subsequently with 0.1ϫ SSC (w/v) and 0.1% (w/v) SDS at 50°C two times. The membranes were exposed to x-ray film with an intensifying screen at Ϫ80°C.
Whole Mount in Situ Hybridizations-Whole mount in situ hybridization was carried out as described by Wilkinson and Nieto (43). Embryos were fixed in 4% paraformaldehyde/PBS overnight and then digested with 10 g/ml of proteinase K in PBS containing 0.1% Tween 20 at 25°C for 5-10 min. Hybridization was performed at 55°C in 5ϫ SSC, 50% (v/v) formamide, 1% (w/v) SDS, 50 g/ml heparin, and 50 g/ml yeast tRNA using digoxigenin-labeled RNA as probes. Subsequent washing was carried out in 2ϫ SSC, 50% (v/v) formamide, and 1% (w/v) SDS at 55°C. Digoxigenin-labeled RNA probes were prepared by generating the large strand with a T3 RNA polymerase or T7 RNA polymerase reaction using chick HS2ST cDNA (nucleotide numbers 1259 -2148 bp), chick HS6ST-1 cDNA (678 -1382 bp), or chick HS6ST-2 cDNA (1-1505 bp), and subsequently by subjecting it to mild alkali (pH 10) treatment for 10 min at 60°C. The development was carried out for 2-5 h at 25°C using nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate stock solution (Roche Molecular Biochemicals GmbH, Germany). For some embryos, the background was diminished with graded ethanol. Photographs of purple signals in embryos positioned on a 1% agarose bed were taken in 50% (v/v) glycerol PBS.
Preparation and Structural Analysis of HS in the Chick Wing Bud-We obtained wing buds from 144 chick embryos at stage 24 in total through the four independent experiments (42, 31, 33, and 38 embryos in the 1st, 2nd, 3rd, and 4th experiments), and each wing bud was separated into three regions: anterior proximal, posterior proximal, and distal (see Fig. 5). The distal region contained the AER and mesenchyme adjacent to the AER. These samples (four different batches for each region and 12 samples in total) were lyophilized and weighed. Each was suspended in 700 l of 0.2 M NaOH, incubated for 16 h at room temperature, neutralized with 4 M acetic acid, and then treated with DNase (100 g) and RNase (100 g) at 37°C for 2 h. Proteinase K (100 g) was added, and incubation continued for 16 h at 37°C. The reaction was stopped by heating at 100°C for 2 min. The samples were centrifuged at 13,000 rpm for 10 min to remove insoluble materials. The supernatants were applied to a DEAE-Sephacel column equilibrated with 50 mM Tris-HCl buffer (pH 7.2) containing 0.2 M NaCl and subsequently washed with 10 column volumes of 0.2 M NaCl in 50 mM Tris-HCl buffer (pH 7.2). The fractions eluted with 2.0 M NaCl in the same buffer were collected. Two and one-half volumes of cold ethanol containing 1.3% (w/v) potassium acetate were added to these fractions, and the glycosaminoglycans were recovered as precipitates by centrifugation at 13,000 rpm for 30 min at 4°C. The precipitates were dissolved in water, and portions of the solutions were treated with a mixture of 10 milliunits of heparitinase I, 5 milliunits of heparitinase II, and 10 milliunits of heparitinase III in 40 l of 50 mM Tris-HCl buffer (pH 7.2, containing 1 mM CaCl 2 , and 4 g of bovine serum albumin) at 37°C for 2 h. After the digests were filtered with Ultrafree-MC (5,000 molecular weight limit, Millipore Corp., Bedford, MA), unsaturated disaccharides in the filtrates were analyzed by a reversed-phase ionpair chromatography using Senshu Pak column Docosil with a fluorescence detector according to Toyoda's method (44) with a slight modification of elution conditions. The other portions were subjected to nitrous acid degradation at pH 1.5. The products degraded with nitrous acid were first labeled with [ 3 H]sodium borohydride, then subjected to gel filtration, and the labeled O-sulfated disaccharide fractions were then subjected to HPLC on a column of Partisil-10 SAX. The nitrous acid degradation and the gel filtration and subsequent HPLC were performed as described previously (38). The amounts of the disaccharides and total HS were estimated from their absorbancy on the chromatographic patterns using standard disaccharides as described previously (38).

Cloning of cDNAs for Chick 2-O-and 6-O-Sulfotrans-
ferases-Considering the regulatory roles of heparan sulfate 2-O-and 6-O-sulfations in the signaling of cell growth factors, which have only recently been elucidated, we first performed cDNA cloning of chicken homologues of mammalian HS2ST, HS6ST-1, HS6ST-2, and HS6ST-3 using two different chicken cDNA libraries (limb buds and retina from stage 21 and 23 chick embryos). Despite several attempts, cross-hybridization with mouse cDNAs as probes only yielded clones of chick HS2ST, chick HS6ST-1, and chick HS6ST-2 from both the libraries. The nucleotide sequences of putative chick HS2ST, chick HS6ST-1, and chick HS6ST-2 contained single open reading frames and encoded type II transmembrane proteins, composed of 356, 401, and 403 amino acid residues, respectively ( Fig. 1, A-C). There were six, two, and four potential N-linked glycosylation sites in the chick HS2ST, HS6ST-1, and HS6ST-2 proteins, respectively. The predicted amino acid sequences of chick HS2ST and chick HS6ST-1 were 97 and 90% identical to those of mouse HS2ST and mouse HS6ST-1, respectively (Fig.  1, A and B). A comparison of predicted amino acid sequence between chick HS6ST-2 and mouse HS6ST-2 revealed that the central region and the region near the C-terminal of mouse HS6ST-2 were missing in chick HS6ST-2, but the other regions of mouse HS6ST-2 were well conserved in chick HS6ST-2 with 85% identity (Fig. 1C). We have recently found that mammalian HS6ST-2 has two forms (short and original long forms) due to alternative splicing of exons 2 and 3 corresponding to the central region and designated the short form HS6ST-2S (39). Characteristics of chick HS6ST-2 described above suggest that it corresponds to the spliced form of HS6ST in mouse, HS6ST-2S. A comparison of the predicted amino acid sequence between chick HS6ST-1 and chick HS6ST-2 revealed only 63% identity overall, but that the sequence of the center region, especially at the putative PAPS binding sites (5Ј-PSB and 3Ј-PB), was well conserved (45,46). Little identity was observed in the aminoterminal and carboxyl-terminal sequences between these two isoforms. Because we found in mice (38) and recently in humans (39) that the original long form of HS6ST-2 was highly  FIG. 1. Predicted amino acid sequences of chick HS2ST, HS6ST-1, and HS6ST-2 and their identity to mouse counterparts. The predicted amino acid sequences of chick HS2ST (A), HS6ST-1 (B), and HS6ST-2 (C) were aligned with mouse counterparts using the Genetyx computer program. The dashes indicate the positions skipped for the alignment. The boxes indicate that the predicted amino acid in the alignment expressed in brain, we examined whether the original long form of HS6ST is expressed in chick brain. PCR was performed using poly (A) ϩ RNA isolated from heads of stage 31 chick embryos as a template and primer oligonucleotides with the sequences 5Ј-CTGCACATCCAGAAGACGGGCG-3Ј (the sequence for nucleotides 259 -280 of chick HS6ST-2) and 5Ј-GTGTAGCAGCTGGGCAGCTCCT-3Ј (the complementary sequence for nucleotides 641-662 of chick HS6ST-2). We also amplified human adult brain cDNA (Clontech) by PCR using the following primers as a positive control. The nucleotide sequence between the two primers corresponded to the sequence of nucleotides 270 -793 of human HS6ST-2, in which the sequence absent in human HS6ST-2S was contained. The chick brain cDNA template gave only one band corresponding to the shorter product from the human cDNA on agarose gel, whereas the human brain cDNA template gave two bands. The results indicate no expression of the original long form of HS6ST-2 at least at embryonic stages. We also tried to detect a chick homologue of HS6ST-3 or possibly other HS6ST homologues if present by RT-PCR using poly (A) ϩ RNA isolated from the stage 31 chick embryo head. The sequences of primers were chosen from the sequence conserved among three isoforms of mouse and human HS6STs and have the sequence 5Ј-TTCCT-GCACATCCAGAAGAC-3Ј (nucleotides 445-464 of mouse HS6ST-3) and 5Ј-ATGGCCTTCTTCGGCCT-3Ј (a complementary sequence for nucleotides 1039 -1055 of mouse HS6ST-3). However, there was no positive signal to suggest the presence of HS6ST-3 or other HS6ST homologues in chick embryos. In addition, recent publications describe the presence of only one HS6ST in Drosophila (47) or difficulty in detecting HS6ST in Xenopus (48). Considering these results, we concluded that chickens have one HS2ST and only two isoforms of HS6ST (HS6ST-1 and -2).
Characterization of Chick HS 6ST-2-The predicted amino acid sequences of cloned chick HS2ST and HS6ST-1 cDNAs showed very high identity to those of mouse HS2ST and mouse HS6ST-1, respectively. Therefore, it is very likely that chick HS2ST and HS6ST-1 have enzyme properties and substrate specificities similar to those of mouse isologues. However, chick HS6ST-2 was devoid of 40 amino acid residues and 30 amino acid residues found in the central region and the carboxylterminal region of mouse HS6ST-2, respectively (Fig. 1C). Therefore, it should be examined if the protein encoded by chick HS6ST-2 has the HS6ST activity and, if so, whether or not its substrate specificity is similar to that of mouse HS6ST-2 (original form). The cDNA for chick HS6ST-2, as well as for chick HS6ST-1 and mouse HS6ST-1 and -2 for comparison, was inserted into a mammalian expression vector, pFLAG-CMV2, and introduced into COS-7 cells. From the cellular extracts of the transfected cells, the FLAG-tagged proteins were purified with an anti-FLAG antibody-conjugated affinity column as described under "Experimental Procedures." The sulfotransferase activities of the purified proteins were examined toward heparin derivatives, NS-heparosan, and heparan sulfates from different sources (Table II), among which disaccharide compositions differed widely (Table I)

TABLE II
Acceptor substrate specificities of the recombinant sulfotransferases purified by anti-FLAG antibody affinity column chromatography Sulfotransferase activities were assayed using various glycosaminoglycans as acceptors. Sulfotransferase fractions were prepared from COS-7 cells transfected with pFLAG-CMV2-chick HS6STs (-1 and -2), and pFLAG-CMV2-mouse HS6STs (-1, -2, and -3), respectively, as described under ''Experimental Procedures.'' Recombinant proteins were purified on the anti-FLAG antibody column. a Values are means of triplicate independent experiments with the standard deviation less than 5% of each value and indicate the relative rate of the incorporation into various substrates to that into CDSNS-heparin. The amount of each enzyme used was designed to give the same rate of 35 S incorporation into CDSNS-heparin. was able to transfer sulfate to all the glycosaminoglycans examined, and NS-heparosan was a better acceptor than CDSNSheparin under the standard reaction conditions used in this study, suggesting that chick HS6ST-2 prefers an N-sulfoglucosamine residue in glucuronosyl N-sulfoglucosamine unit to the one in iduronosyl N-sulfoglucosamine unit at least at the used substrate concentration (500 M). The preference of chick HS6ST-2 for these units varied depending on the substrate concentrations, and chick HS6ST-2 had an opposite preference at the lower substrate concentrations than 200 M (data not shown). Such a substrate preference is very similar to that of mouse HS6ST-2 (Table II) and of human HS6ST-2 (38, 39). The substrate preference of chick HS6ST-1 was also similar to that of mouse HS6ST-1 (Table II) and human HS6ST-1 (39), and chick HS6ST-1 had a preference to an N-sulfoglucosamine residue in iduronosyl N-sulfoglucosamine unit at all the concentrations (data not shown). To confirm the position of the sulfation and examine the substrate specificity, we analyzed the structure of the 35 S-labeled products obtained by incubation with the purified chick HS6ST-2 and heparan sulfate from pig aorta as described under "Experimental Procedures." The products were digested with a mixture of heparitinase I, II, and III, and the digests were subjected to Superdex 30 pg gel chromatography and subsequently the fractions containing disaccharides and monosaccharides to PAMN column chromatography. The radioactivity was detected in the fraction of GlcNSO 3 (6SO 4 ), ⌬Di-(N,6)diS, and ⌬Di-(N,6,2)triS among which GlcNSO 3 (6SO 4 ) was derived from the non-reducing end of the products (Table III), indicating that chick HS6ST-2 transfer sulfate to position 6 of GlcNSO 3 residues not only at the inside chain whether or not the adjacent hexuronic acid residues are 2-O-sulfated but also at the non-reducing terminal of HS. The activity of chick HS 6ST-2 to transfer sulfate to position 6 of GlcNSO 3 residues at the non-reducing terminal should be of high affinity, because the GlcNSO 3 residues are only minor (less than a few percent) (39). The observed substrate preference was rather similar to that of human HS6ST-2S (39). All the data indicate that chick HS6ST-2 is an isologue of mammalian HS6ST-2, although there were large structural differences in the carboxyl terminus.
We also compared the substrate preference of chick HS6ST-2 to that of human HS6ST-2 and of human HS6ST-2S (short form). The results indicated that, in this regard, chick HS6ST-2 appeared to be more similar to human HS6ST-2S than human HS6ST-2 (data not shown). HS6ST-2S is a spliced form of HS6ST-2. However, the gene portions corresponding to the spliced exons, exon 2 and exon 3, of mammalian HS6ST-2 can not be detected in any HS6STs of lower animals such as chicken and Drosophila (47). These facts indicate that the long form of HS6ST-2 may be a mammalian-specific form.
Northern Blot Analysis-We then examined tissue-and stagedependent differences in expression patterns of transcripts among the genes for HS2ST, HS6ST-1, and HS6ST-2 using Northern blots of poly (A) ϩ RNAs obtained from four typical tissues (head, heart, trunk, and limb) dissected from chick embryos at various stages. We first performed hybridization in samples at stage 31 with 32 P-labeled probes prepared from each cDNA (Fig. 2). The expression patterns of the two HS6ST isoforms differed. The major transcript of 2.1 kb for chick HS6ST-1 was detected in all the tissues but with an apparently lower staining intensity in the heart (Fig. 2B), whereas only one transcript of 3.2 kb for chick HS6ST-2 was detected in all the tissues with a higher intensity in the limb and head samples than in the heart and trunk samples (Fig. 2C). Multiple transcripts for chick HS2ST (2.3, 3.2, and 7.5 kb) were detected in all the tissue samples but there were significant differences in their intensities, especially those of 3.2 and 7.5 kb among the different tissues ( Fig. 2A). In limb bud, all the transcripts were highly expressed at stage 31. We also examined expression patterns of HS2ST, HS6ST-1, and HS6ST-2 in the above four tissues at different stages (22)(23)(24)26) but found no distinctive differences (data not shown).
Whole Mount in Situ Hybridizations-We next examined the precise expression sites of these transcripts at different developmental stages by whole mount in situ hybridization. The experiments were performed on chick embryos at stages 17-24 using probes for chick HS2ST, chick HS6ST-1, and chick HS6ST-2.
The staining patterns demonstrated not only tissue-dependent but also stage-dependent differences in the expression. The patterns at stage 24 (Fig. 3, A, B, and C) and at stage 20 (Fig.  3, D, E, and F)  the expression of HS2ST was detected in the most regions of the first and second pharyngeal arches (Fig. 3, A and D). HS6ST-1 and -2 transcripts were found in forebrain, midbrain, hindbrain, tail bud, mesonephros, trigeminal nerve, facial nerve, glossopharyngeal nerve, vagus nerve, neural crest, dorsal aspects of the neural tube, and dorsal root ganglia, which are all composed of cells from the neural crest and neurogenic epithelial placodes (Fig. 3, B and E, for HS6ST-1; Fig. 3, C and  F, for HS6ST-2). Interestingly, the patterns exhibited regionand stage-dependent differences within limb bud tissues, which confirmed that limb buds are a good systems for testing our hypothesis. In both wing and leg buds, significant levels of HS2ST transcripts were detected at the overlying ectoderm and mesenchyme throughout stages 21, 23, and 24 (Fig. 4). HS6ST-1 and -2 transcripts were clearly detectable in the anterior proximal and posterior proximal regions of the limb mesenchyme and ectoderm, and their expression varied in intensity, depending on the stage of differentiation (Fig. 4, B and C). It should be noted that apical ectodermal ridge (AER) did not express either of the two and in limb buds there were distinctive differences in the expression patterns of the two HS6STs. The HS6ST-1 transcript was mostly detected in the anterior proximal region, whereas the HS6ST-2 transcript was detected in the posterior proximal region at all the stages examined from 21 to 24 (Fig. 4, B and C, respectively).
Structural Analysis of HS in the Chick Wing Bud-Because HS6ST-1 and -2 in the limb buds were strongly expressed in the proximal region with preference for the former in the anterior half and the latter in the posterior half, and also considering their difference in the acceptor substrate specificity, we expected that HS in the wing bud differs in the sulfation pattern and degree, depending upon the regions; i.e. HS in the bud is higher in 6-O-sulfation degree in the proximal region than in the distal region and different in 6-O-sulfation pattern between the anterior and posterior halves of the proximal region. To examine this possibility, we chose the wing buds at stage 24 when the spacio-temporal control of various heparinbinding cell growth factors are essential for the subsequent skeletal tissue development and patterning in the limbs. Based upon our preliminary observations, 6-O-sulfated residues appeared to be minor in HS. Therefore, to analyze such structural differences in HS with statistically significances, we obtained the wing bud samples in the four independent experiments. We dissected wing buds from 42, 31, 33, and 38 embryos in the 1st, 2nd, 3rd, and 4th experiments, respectively, and separated them into three regions; the distal area that contained the AER and underlying mesenchyme (progress zone), the anterior half of the proximal region, and the posterior half of the proximal region (Fig. 5A) in the respective experiments. HS chains were prepared from those three regions, which were from 12 samples in total (the three regions from the four different batches). Their structures were individually analyzed, and the values thus obtained for the respective regions were all given with the standard deviations among the four batches of the wing buds (see Fig. 5). First, their HS compositions and amounts were analyzed by HPLC after digestion with a mixture of heparitinases as described under "Experimental Procedures." Relative amounts of HS (a ratio of hexuronic acid, picomoles to wing bud dry weight, nanograms) were similar among the three regions (Fig. 5A). The percent compositions (Fig. 5B) (Fig. 5C) demonstrated that the posterior region is richer in GlcA-AMan(6S) derived from the disaccharide unit of GlcA-GlcNS(6S) adjacent to the GlcNSO 3 residue at the non-reducing terminal than the anterior region (18% to 11% of the total O-sulfated disaccharide derivatives), whereas the anterior region is significantly richer in IdoA-Aman(6S) derived from the IdoA-GlcNS(6S) unit adjacent to the GlcNSO 3 residue at the non-reducing terminal than the posterior region (13% to 7% of the total O-sulfated disaccharide derivatives). It is also of note that, although the difference is of no statistical significance, the anterior region is slightly richer in IdoA(2S)-Aman(6S) derived from the IdoA(2S)-GlcNS(6S)  HS2ST (A), HS6ST-1  (B), and HS6ST-2 (C) (dorsal view, top is anterior). At all stages, HS2ST transcripts were detected throughout both mesenchyme and ectoderm in wing and leg limb buds apparently with lower levels in the ectoderm. At all stages, HS6ST-1 transcript was detected in limb bud mesenchyme and restricted mainly to the anterior proximal regions of the wing bud and leg bud, and to some extent to the posterior proximal region. In contrast to HS6ST-1, HS6ST-2 transcript was detectable mostly in the posterior region of mesenchyme at all stages. At stage 24, the transcripts of HS6ST-2 and HS6ST-1 were detected in the wing bud and leg bud mesenchymes surrounding chondrogenic regions except for the mesenchymes underlying the AER. unit adjacent to the GlcNSO 3 residue at the non-reducing terminal than the posterior region (35% to 33% of the total Osulfated disaccharide derivatives).
Because most of the observed differences in the HS compositions were statistically significant, the structural differences shown above were consistent with the ones expected given the expression patterns of HS6STs and the substrate specificities of the expressed HS6STs if the acceptor substrate concentrations in vivo were at the levels similar to those under the used standard reaction conditions in vitro.
The HS contents of the distal, anterior-proximal, and posterio-proximal regions in the wing buds at stage 24 were almost equal, but the HS compositions varied, depending on the regions. Considering the generally accepted fact that O-sulfated residues are rather minor, compared with non-sulfated and N-sulfated residues in normal HS, it is likely that even the subtle differences in the compositions of the 6-O-sulfated disaccharide units among the three regions in chick wing buds may be enough to generate HS chains with region-dependent different functions. Although this study only analyzed HS in the wing buds at stage 24, the stage-dependent different expression patterns of heparan sulfate O-sulfotransferases ( Fig.  4) strongly suggest that the O-sulfotransferases are also involved in generating HS chains with stage-dependent different structures.

Heparan Sulfate O-Sulfotransferases in Chick Limb
Buds-In the FGF-signaling pathway, HS on the cell surface may serve as a low affinity receptor for FGFs and facilitate the interaction of ligands with the high affinity receptors of tyrosine kinases (11, 14 -16). In addition, interactions of FGFs with HS in the extracellular matrix as well as on the cell surface may serve to regulate the diffusion and/or concentration of FGF (49). It has been shown that specific O-sulfated saccharide sequences in HS are required for such interactions (15,18,30,50). Therefore, one can speculate that a tissue-or cell-specific HS biosynthesis, especially the O-sulfation reaction, regulates FGF signaling. Actually, immunostaining of FGFs bound to HS chains (49,51) and of FGF receptors bound to FGF⅐HS complexes (16) revealed that specific HS sulfation patterns are distinct, depending upon the tissues and tissue regions. However, there has been no in vivo evidence for relationships between specific HS structures and expression patterns of enzymes of the HS biosynthetic pathway in terms of the structures involved in tissue-or cell-specific FGF signaling, although recent publications on the null mutation in mice and Drosophila of genes involved in the biosynthesis of heparan sulfate proteoglycans have provided some clues (23, 54 -62).
In the present study, we observed that the expression pat-terns of O-sulfotransferases are regulated in a tissue-and stage-dependent manner in developing chick limb buds. We also analyzed HS structures in three different regions of the developing wing bud and have found region-specific HS structural alterations in which the observed differences in the expression patterns of O-sulfotransferases may be greatly involved.
In the development of vertebrate limbs, functions of some HSPGs have been reported. For example, chick syndecan-3 expressed in the distal mesenchymal cells underlying the AER plays an important role during limb outgrowth by regulating the interaction of FGFs with the underlying mesoderm (20). On the other hand, our results indicate that 6-O-sulfated structures of HS are preferentially found in the proximal regions of the limb buds. These findings suggest that the function of syndecan-3 in the distal mesenchyme may not require high 6-O-sulfated HS. We have recently examined the affinity of a variety of chemically and enzymatically modified heparin preparations with FGF-10, 2 a mesenchymal FGF that is essential for limb induction and AER maintenance by activating the FGFR-2b isoform in the AER (3,63), and found that FGF-10 needs the 6-O-sulfated structures at least for the binding to HS. Taken together, it is possible that the low 6-O-sulfated HS in the distal region has a lower affinity for FGF-10 than the high 6-O-sulfated HS in the proximal region, thus facilitating the diffusion of FGF-10 from the distal mesenchymal cells to the AER, which is essential for outgrowth and the patterning of skeletal tissues of developing limb buds. In relation to this, it is interesting that the FGFR-2b isoform, a signal receptor for FGF-10, is preferentially expressed throughout the ectoderm, including the AER, whereas the FGFR-2c isoform, a signal receptor for FGF-8 produced by the AER, is expressed by both mesoderm and ectoderm (63).
In mouse limb development, the other HSPGs are also required. At early stages in the development, glypican-3 (Gpc-3) is expressed in the anterior-and posterior-proximal regions as well as in distal mesenchymal cells of the limb bud. However, it is not expressed in the AER (21). From a genetic analysis of Gpc-3 deficiency and haploinsufficiency for BMP-4, it was concluded that the function of glypican-3 was to control cellular responses to BMP-4 positively in limb patterning and skeletal development (21). The present results of whole mount in situ hybridization suggest that the regions expressing HS6STs in the limb bud are overlapped with the regions where Gpc-3 is expressed. In the AER adjacent to those regions, regression takes place after the most distal skeletal progenitors are specexperiments), and wing buds in each batch were separated into the three regions shown in A. These samples (the three regions from the four batches, and 12 samples in total) were lyophilized and weighed, and HS samples were prepared from those tissue samples as described under "Experimental Procedures." The HS amount in each region per one wing bud (picomoles/wing bud), dry weight of each region per one wing bud (nanograms/wing bud), and the relative HS content in each region (picomoles of HS per ng dry weight of wing bud) in the three regions were mean values of the four experiments described above. The relative contents of HS were similar to each other among the distal, anterior-proximal, and posterio-proximal regions of the wing buds at stage 24. B, HS samples were digested with a mixture of heparitinase I, II, and III as described under "Experimental Procedures." Digests were subjected to reversed-phase ion-pair chromatography with postcolumn fluorescence labeling as described under "Experimental Procedures." The histogram shows the percent unsaturated disaccharide compositions of the HSs in the three regions of the wing buds. The height of each column represents mean of the values obtained by the four independent experiments, and each error bar on the top represents its standard deviation. Compared with HS samples from both proximal regions, the HS sample from the distal region had less 6-O-sulfated disaccharides with statistical significance. C, HS samples from the anterior and posterior proximal regions were subjected to nitric acid degradation at pH 1. ified and is important to limit limb outgrowth (64 -67). BMPs are considered strong candidates for factors that promote the regression and necrotic zone induction in the proximal regions of AER (68). Interestingly, BMPs have also been shown to be factors to maintain the AER in the distal region (69). Therefore, it is likely that 6-O-sulfated HS directly or indirectly influences such regional differences in BMP signaling and the difference in HS sulfation may involve the opposite functions of BMPs.
An analysis of the region-dependent difference in HS structure has demonstrated that the posterior region is richer in GlcA-GlcNS(6S) than the anterior region, which is a disaccharide rarely found in the sequences for cell growth factors and morphogens to bind to. In this posterior region, there is a zone of polarizing activity (ZPA) that regulates the anterior-posterior growth of limb bud. Sonic hedgehog (SHH), a member of the hedgehog family, is produced by ZPA cells and is the key mediator of this polarizing activity (70). The diffusion of hedgehog is important for determination of the anterior-posterior axis of developing limb buds and is known to be regulated by HS (54,71). Recently it was reported that a graft of porcine intestinal mucosa HS into the anterior of chick limb bud induced ectopic ZPA from anterior limb mesenchyme (24). Based on these results, it is possible that structural differences of HS between the anterior and posterior regions are important for determination of the anterior-posterior axis of developing limb buds.
The structural differences of HS between the anterior and posterior proximal regions were consistent with the ones expected from the expression patterns of HS6STs and their acceptor substrate specificities if the substrate concentrations in vivo were at the levels similar to those under the used standard reaction conditions in vitro for Table II (500 M as hexosamine). Although the actual concentrations of acceptor substrates in the sulfation sites of cells expressing HS6STs are hardly known, the observed differences have enabled us to speculate that the sulfation sites in the Golgi apparatus may form the following environment. The concentrations of HS6STs and their acceptor substrates are kept high enough by direct or indirect interactions with itself such as oligomerization of glycosyl transferases themselves in the Golgi (72,73) or with other proteins such as distinct dependence of Golgi localization of Pipe, a Drosophila enzyme with structural similarity to the mammalian HS2ST, on the Windbeutel protein (74).
Heparan Sulfate O-Sulfotransferases in Other Tissues-HS6ST-1 and -2 transcripts were detected in trigeminal nerve, facial nerve, glossopharyngeal nerve, vagus nerve, and dorsal root ganglia with different intensities, and HS2ST transcript was detected mainly in the first and second pharyngeal arches (Fig. 3). These nerves and ganglia are composed of cells from the neural crest and neurogenic epithelial placodes. Recent evidences have demonstrated that heparan sulfate proteoglycans are implicated in controlling neural crest migration by modifying the FGF signaling. A null mutation of the gene for perlecan, a heparan sulfate proteoglycan abundant in basement membranes and some pericellular matrices, causes a high incidence of malformations of the cardiac outflow tract, due to the abnormal abundance of mesenchymal cells in the tract, which were derived from an uncontrolled migration of neural crest cells, suggesting that perlecan is involved in control of those events (52). Kubota and Ito (53) examined the roles of FGF-2 and FGF-8 in the migration of mesencephalic mouse neural crest cells and have found that FGF binding to heparan sulfate proteoglycans is needed for the migration, and the expression of FGF-8 in the mandibular arch epithelium is a prerequisite for the differential localization of FGF-2. Therefore, the observed tissue-dependent differences in the expres-sions among the heparan sulfate O-sulfotransferase genes in the neural crest cell-containing tissues suggest close relationships between the O-sulfation patterns of heparan sulfate proteoglycans and the tissue-specific differences of neural crest cell differentiation and migration.
Conclusions-The present results suggest that differential sulfation patterns of HS produced by multiple sulfotransferases provide appropriate cellular environments for respective morphogenic signal transductions.