Biosynthesis of Heparan Sulfate on P-D-Xylosides Depends on Aglycone Structure*

We have reported that 3-estradiol-P-~-xyloside primes heparan sulfate synthesis in Chinese hamster ovary cells and that the proportion of heparan sulfate made rises with increasing concentration of xyloside (Lugemwa, F. N. and Esko, J. D. (1991) J. Biol. Chern. 266, 6674-6677). Using estradiol as a guide, we varied the structure of the aglycone and showed that P-D-xylosides containing two fused aromatic rings efficiently prime heparan sulfate. Thus, 2-naphthol-P-~-xyloside primed heparan sulfate at low dose (510 p ~ ) and the proportion of heparan sulfate increased with concentration (up to 60% of total glycosaminoglycan). Various ring additions and heterocyclic ring substitutions altered the effi- ciency of heparan sulfate priming, but had no effect on the overall level of glycosaminoglycan synthesis. Re- placement of the bridging oxygen with sulfur (2-naph-thalenethiol-P-D-xyloside) increased the efficiency of heparan sulfate priming. Priming of heparan sulfate correlated with hydrophobicity of the xyloside, but sev- eral exceptions suggested that the chemical structure of the aglycone played an equally important role. Interest- ingly, the heparan sulfate chains generated on 2-naph-thol-pD-xyloside showed a 2-fold decrease in

We have reported that 3-estradiol-P-~-xyloside primes heparan sulfate synthesis in Chinese hamster ovary cells and that the proportion of heparan sulfate made rises with increasing concentration of xyloside (Lugemwa, F. N. and Esko, J. D. (1991) J. Biol. Chern. 266, 6674-6677). Using estradiol as a guide, we varied the structure of the aglycone and showed that P-D-xylosides containing two fused aromatic rings efficiently prime heparan sulfate. Thus, 2-naphthol-P-~-xyloside primed heparan sulfate at low dose (510 p~) and the proportion of heparan sulfate increased with concentration (up to 60% of total glycosaminoglycan). Various ring additions and heterocyclic ring substitutions altered the efficiency of heparan sulfate priming, but had no effect on the overall level of glycosaminoglycan synthesis. Replacement of the bridging oxygen with sulfur (2-naphthalenethiol-P-D-xyloside) increased the efficiency of heparan sulfate priming. Priming of heparan sulfate correlated with hydrophobicity of the xyloside, but several exceptions suggested that the chemical structure of the aglycone played an equally important role. Interestingly, the heparan sulfate chains generated on 2-naphthol-pD-xyloside showed a 2-fold decrease in the proportion of disaccharides containing 6-0-sulfate groups and a striking diminution in non-sulfated iduronic acid containing disaccharides compared to the chains attached to cellular proteoglycans. Thus, both the type of glycosaminoglycan made on a xyloside and its fine structure depends on the aglycone.
Twenty years ago, Okayama et al. (1973) reported that adding p-D-xylosides (D-xylopyranose attached to an aglycone) to cartilage slices results in glycosaminoglycan (GAG)l synthesis.
These compounds were subsequently popularized as inhibitors of proteoglycan synthesis since priming of GAGs resulted in inhibition of chondroitin sulfate proteoglycan biosynthesis *This work was supported by Grant CA46462 from the National Institutes of Health and a grant from Glycomed, Inc. (to J. D. E.) and by a National Defense Science and Engineering Graduate Fellowship (to T. A. F.). The NMR core facility of the Comprehensive Cancer Center was supported by Grant CA13148 from the National Cancer Institute. 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. $ To whom correspondence and reprint requests should be addressed.
Thus, under normal conditions core protein expression, the extent of xylosylation, or the flux of core protein substrates passing from the endoplasmic reticulum through the Golgi may limit GAG synthesis. Flooding cells with p-D-xylosides apparently bypasses the restriction and provides a large number of primers for chain elongation to occur.
could not be assigned.
agnostic protons are included in Table I. The migration of the compounds on thin layer C,, reverse-phase plates (E. Merck) (methanol/ H,O, 7:3, v/v) or on Silica Gel-GOA plates (Whatman) (ethyl acetate/ methanoVacetic acid, 90:10:1, v/v) is also presented. Synthesis ofAglycones-2-(2-Naphthoxy)ethanol was made by stirring 2-naphthol (7.2 g, 50 mmol) and K&03 (6.9 g, 50 mmol) in 10 ml of acetone for 30 min a t room temperature. 2-Bromoethanol (6.2 g, 50 mmol) was added, and the mixture was gently refluxed for 7 h. The reaction was diluted with water, and the organic layer was repeatedly extracted with 2 M NaOH. The NaOH washings were extracted with diethyl ether, and the ether extracts were added to the original organic layer. The organic layer was dried with CaCl,, filtered through diatomaceous earth, and concentrated to a solid. The solid was dissolved in diethyl ether, and the precipitate that formed was removed by centrifugation. The supernatant was concentrated, dissolved in acetone, and placed a t -20 "C. The amorphous product that formed was collected and dried under vacuum.
6-Ethoxy-2-naphthol was prepared by stirring 2,6-dihydroxynaphthalene (3.2 g, 20 mmol) for 1 h a t room temperature in 30 ml of dimethylformamide containing NaH (0.48 g, 20 mmol). Bromoethane (2.2 g, 20 mmol) was added, and the mixture was stirred a t 50-55 "C for 18 h. Dimethylformamide was removed by coevaporation with toluene, and the resulting solution was diluted with ethyl acetate. Several volumes of 0 . 1 4 . 2 M NaOH were added, and the ethyl acetate layer was removed. The aqueous layer was extracted with ethyl acetate, and the organic extacts were combined with the original ethyl acetate phase, dried with Na2S04, and concentrated. The sample was diluted with acetone, and the precipitate that formed was filtered. The filtrate was concentrated to dryness, dissolved in acetone, and insoluble material was again removed by filtration. The filtrate was concentrated, diluted with a minimal volume of toluene and product was purified by silica gel chromatography (20 g) by washing with hexanes and increasing the proportion of ethyl acetate.
6-Ethoxy-2-naphthol-~-~-xylopyranoside was prepared by adding under argon 6-ethoxy-2-naphthol (0.75 g, 4 mmol) to 5 ml of acetonitrile containing 1 g of 4-A molecular sieves and NaH (0.12 g, 5 mmol). The solution was warmed to 40 "C for 15 min. In a separate container, acetobromo-a-D-xylopyranoside (1.8 g, 5.3 mmol) was dissolved in 5 ml of acetonitrile containing 1 g of 4-A molecular sieves. The first solution was added dropwise to the second solution with stirring. After 30 min at 23 "C, the solution was diluted with chloroform, washed with 5% NaHC03(,,, (w/v), and the organic layer was dried with NazS04. After concentration, the material was deacetylated as described above, and product was purified on a silica gel column (15 g) by washing with hexanes followed by ethyl acetate/toluene/acetic acid (5:5:0.1, v/v) and eluting by increasing the proportion of ethyl acetate.
4-Phenylphenol and 9-phenanthrol-P-o-xylopyranosides were similarly prepared but the product precipitated during the reaction. The precipitate was removed by centrifugation, and the supernatants were extracted with ethyl ether. The combined sediment and ethyl ether extracts were dried and then deacetylated as described above. Xylosides were purified by silica gel chromatography (15 g) in solvent A.
3-d-Equilenin-~-~-xylopyranoside was also prepared as described above, but an unknown precipitate that formed during the coupling reaction was removed by filtration. The filtrate was extracted with ethyl acetate and then concentrated by evaporation. The sample was then deacetylated as described above, and 3-~-equi~enin-~-~-xy~opyranoside was purified by silica gel chromatography (20 g) in solvent A.
2-Naphthalenethiol-P-~-xylopyranoside was synthesized by dissolving 2-naphthalenethiol(O.8 g, 5 mmol) in 4.2 ml of 1.0 M NaOH and 40 ml of acetone. Acetobromo-a-o-xylopyranoside (0.68 g, 2 mmol) was dissolved in 1 ml of acetone and added to the first solution. After stirring the mixture at 23 "C for 6 h, the sample was concentrated and deacetylated as described above. A precipitate formed during concentration of the ethyl acetate and was found to be 2-naphthalenethiol-P-~-xylopyranoside. This precipitate was washed with ethyl acetate, and the product was crystallized from methanol. cis /trans-Decahydro-2-naphthol-~-~-xylopyranoside was synthesized by the Koenigs-Knorr reaction as described by VanAken et al. (1986). Cis/trans-decahydro-2-naphthol (0.4 g, 2.6 mmol) was dried overnight and dissolved in 15 ml of dichloromethane containing 5 g of Drierite. Silver silicate (1.58 g, 5.4 mmol) was added, and the solution was stirred for 30 min at 23 "C. Acetobromo-a-o-xylopyranoside (1.2 g, 3.5 mmol) was added, and after 5 min at 23 "C the reaction was filtered through diatomaceous earth and concentrated to a syrup. The syrup was deacetylated as described above. Product was purified by silica gel chromatography (12 g) in solvent A.
One day prior to radioactive labeling experiments, cells were harvested with trypsin, centrifuged, and washed with sulfate-free Ham's F-12 medium supplemented with 10% dialyzed fetal bovine serum and 100 units/ml penicillin G (labeling medium). Approximately 0.5-1 x lo5 cells were transferred in 1 ml to separate wells of 24-well tissue culture plates (Corning) and incubated at 37 "C in labeling medium. Xylosides were prepared as 100 x stocks in Me2SO/water (l:l, v/v).
[35S]H,S0, (40 Ci/mg, DuPont-New England Nuclear) was added to fresh labeling medium (160 pCi/ml), and aliquots were adjusted to different xyloside concentrations. The final concentration of MezSO was always 0.5% (v/ v). Experiments were initiated by replacing the medium in the wells with 0.3 ml of radioactive medium containing xyloside. The cells were incubated for 4-5 h at 37 "C. Background 35S0, incorporation into GAGs was determined from cells incubated in the absence of xyloside.
GAG Preparation and Analysi~-[~~S]GAGs were isolated by anionexchange chromatography using a modification of previously described methods (Bame and Esko, 1989). Cells and media were adjusted to 0.1 M NaOH, and after 15 min acetic acid (1 molar equivalent) was added to neutralize the sample. Chondroitin sulfate A (2 mg, Sigma) and 1/6 volume of a solution containing 1 mg/ml Pronase, 1.92 M NaCl, 0.24 M sodium acetate (pH 6.5) were added. After incubating the samples overnight at 40 "C, they were diluted 3-fold with water and loaded onto 0.5-ml columns of DEAE-Sephacel (Pharmacia LKJ3 Biotechnologies Inc.) pre-equilibrated with 0.25 M NaCl in 20 m~ sodium acetate buffer (pH 6.0). The columns were washed with 15 ml of buffer, and GAGs were eluted with 2.5 ml of 1.0 M NaCl in 20 rn sodium acetate buffer (pH 6.0). GAGs were precipitated by adding 10 ml of cold 95% ethanol and incubating the samples at 4 "C for 2 2 h. The samples were centrifuged for 10 min, the supernatant was aspirated, and the pellets were resuspended in 1 ml of 0.5 M sodium acetate in ethanoVwater (1/9, v/v) and reprecipitated with 4 ml of ethanol. The final pellets were dried under vacuum and resuspended in 0.2 ml of 20 m~ sodium acetate (pH 6.0).
Aliquots (45 p1) of the resuspended pellets were treated with low pH nitrous acid (Shively and Conrad, 1976), which cleaves heparan sulfate at N-sulfated glucosamine residues and generates a series of short oligosaccharides. The remaining chondroitin sulfate was isolated by anion-exchange chromatography as described above, and the amount of [35Slheparan sulfate was determined as the difference between the total P5S1GAG and the remaining [35Slchondroitin sulfate.
In each experiment a pair of wells (unlabeled) was washed three times with 1 ml of cold phosphate-buffered saline and solubilized in 0.1 M NaOH. An aliquot was assayed for protein using the method of Lowry et al. (1951) with bovine serum albumin as standard.
Disaccharide Analysis-Approximately 5 x lo6 wild-type CHO and mutant pgsA-745 cells were labeled for 36 h in 15 ml of low glucose (1 rn) Ham's F-12 medium containing 0.1 mCi/ml ~-[6-~H]ghcosamine (100 Ci/mmol, DuPont-New England Nuclear). pgsA-745 cells were supplemented with 50 p~ 2-naphthol-P-~-xyloside. Radioactive proteoglycans and GAGs were isolated by anion-exchange chromatography as described above. Samples were p-eliminated by incubation for 24 h at 4 "C in 0.5 ml of 0.5 M NaOH containing 1.0 M NaBH,. After neutralization with acetic acid, the samples were diluted with water (lO-fold), and L3H]GAGs were reisolated by anion-exchange chromatography. Chondroitin sulfate was depolymerized by treatment with chondroitinase ABC (Seikagaku), and heparan sulfate was isolated by another round of anion-exchange chromatography. The remaining PHIheparan sulfate was completely deacetylated in 200 pl of 70% aqueous hydrazine containing 1% hydrazine sulfate (96 "C, 6 h) as described by GUO and Conrad (1989). Excess hydrazine was removed under a stream of Nz gas, water was added, and the samples were lyophilized.
[3H]Heparan sulfate chains were cleaved to disaccharides essentially as described by Shively and Conrad (1976). The pellets were dissolved in 45 pl of 1.5 M HNOz (pH 4.0), and after 15 min at 23 "C, the pH was adjusted to 1.5 by adding 45 p1 of 3 M HzS04 and 60 pl of 1.5 M HNOz (pH 1.5). After 10 min at 23 "C, the reaction was stopped by adding 360 pl of 1 M Na2C03. Disaccharides were reduced by adding 160 pl of 3 M NaBH, in 0.1 M NaOH (50 "C, 30 min). The samples were neutralized with acetic acid and dried under a stream of Nz gas. Methanol was added to the samples, and they were dried again. The process was repeated, and finally the samples were dissolved in water and lyophilized. Disaccharides were dissolved in 0.5 ml of 0.5 M pyridinium acetate (pH 5.0) and purified by gel filtration chromatography (1 x 100-cm Bio-Gel p2 column, Bio-Rad) in the same buffer. Over 90% of the material eluted as disaccharides. The fractions were pooled, lyophilized, dissolved in water, and desalted by descending paper chromatography on Whatman 3MM paper in butanol/ethanol/water (52:32:16, v/v). Disaccharides were eluted from the paper with watedethanol (9:1, v/v).
The disaccharides were separated by reverse-phase ion-pairing chromatography essentially as described by Guo and Conrad (1988) using a Hi-Chrom S5 ODS C,, column (4.6 x 50 mm, Regis Chemical Co.) at a flow rate of 1 ml/min. Starting buffer contained 38 mM NH4H2P04, 2 mM H,PO,, and 1 m~ tetrabutylammonium phosphate in water (pH 3.6). Disaccharides were eluted with a gradient of water/acetonitrile (70:30, v/v) containing 38 mM NH,H,PO,, 2 mM H3P0,, and 1 mM tetrabutylammonium phosphate (pH 4.2). Their elution was monitored with a Radiomatic FLO-ONE\\Beta in-line detector. Radioactivity was measured at 6-s intervals, and the data were averaged over 0.5-min time frames. Individual peaks were identified by comparison with heparin disaccharides and published results (Guo and Conrad, 1988). Heparin disaccharides were prepared as described above and labeled by reduction with NaBL3H1, (500 CUmmol, Amersham Corp.) as described by Guo and Conrad (1988). Non-sulfated disaccharides from the reversephase ion-pairing column were pooled and analyzed by descending paper chromatography on Whatman 1MM paper in ethyl acetate/acetic acid/H20 (3:1:1, v/v) a s described by Bame et al. (1991). Radiolabeled GlcA-AmanR used as a standard for paper chromatography was prepared from Escherichia coli K-12 N-acetylheparosan by hydrazinolysis and pH 4.5 nitrous acid cleavage as described by Lindahl et al. (1973). The disaccharides were radiolabeled by reduction with NaBc3H1, as described by Guo and Conrad (1988).

Heparan Sulfate Priming on 2-Naphthol-~-~-xyloside-We
recently showed that estradiol-P-D-xyloside primes over 50% heparan sulfate in CHO cells, whereas other hydrophobic xylosides prime only chondroitin sulfate (Lugemwa and Esko, 1991). To test if smaller, less complex aglycones than estradiol might facilitate biosynthesis of heparan sulfate, a series of P-D-xylosides were synthesized and fed to a mutant CHO cell line that lacks xylosyltransferase (pgsA-745). The absence of GAG synthesis on endogenous proteoglycans in mutant pgsA-745 (Lidholt et al., 1992) makes the cells a convenient system for studying priming of GAGs by exogenous P-o-xylosides.
[35SlHeparan sulfate synthesis occurred at 3 p, increased up to 30 p, and then remained nearly constant. Priming of total [35SlGAG, [35Slchondroitin sulfate, and [35Slheparan sulfate by 3-estradiol-P-D-xyloside behaved in a similar fashion except that the curves were shifted to higher concentrations (Fig. 2 3 ) . However, above 30 p~ a pronounced decrease in both chondroitin sulfate and heparan sulfate occurred, probably due to the detergent properties of estradiol-P-o-xyloside. High concentra- Nearly all of the xylosides tested to date prime GAGs in a dose-dependent manner. Since the midpoint for maximal priming of heparan sulfate for 2-naphthol-p-~-xyloside was about 10 PM (Fig. 2), we used this concentration to evaluate the relative efficacy of other xylosides. At this concentration, 2-naphthol-Po-xyloside primed heparan sulfate better than 3-estradiol-P-oxyloside (32% of total GAG uemus lo%, respectively).
Ring Modifications and Additions-We examined whether adding alkoxy1 "spacers" that increase the distance between naphthol and xylose would affect priming (Fig. 3). Since the addition of the spacers would also change the hydrophobicity of the aglycone, we tested a set of related compounds containing similar "tails" attached to the 6-position of 2-naphthol-P-D-xyloside. The addition of a 2-carbon spacer (2-(2-naphthoxy)-lethyl-P-D-xyloside) had no effect on priming, nor did the addition of a 2-carbon tail (6-ethoxy-2-naphthol-/3-~-xyloside). In contrast, the addition of a 4-carbon spacer (4-(2-naphthoxy)-lbutyl-a-D-xyloside) or a 4-carbon tail (6-butoxy-2-naphthol-P-oxyloside) decreased priming of heparan sulfate. Neither modification affected to a significant degree the overall extent of GAG priming. The corresponding 8-carbon adducts also were tested, but the compounds exhibited strong detergent properties and solubilized the cells (data not shown).
A variety of other ring modifications and additions to the aglycone were tested (Fig. 5). Changing the aglycone to l-naph-tho1 or 9-phenanthrol had little effect on heparan sulfate priming relative to 2-naphthol (20% for both compounds versus 26%, respectively). Separation of the aromatic rings in 4-phenylphenol-P-D-xylose somewhat reduced priming of heparan sulfate (15%), suggesting that the planar structure afforded by the fused aromatic rings in naphthol may be important. The xyloside of 3-d-equilenin was found to be more potent than 2-naphthol-p-o-xyloside (55% heparan sulfate at 10 PM), and the dose response showed a clear shift to lower concentration (data not shown). 3-d-Equilenin resembles 3-estradiol, but the aromatic A and B rings correspond to those of 2-naphthol. These results point to the importance of two fused aromatic rings and suggest that certain adducts might further enhance priming of heparan sulfate.
The various modifications to 2-naphthol-p-~-xyloside shown in Figs. 3-5 would affect aqueous solubility of the xylosides and possibly their transfer across cell membranes. As shown in Table 11, the extent of heparan sulfate priming seemed to cor- ~~ relate with the partitioning of xyloside into the organic phase of an octanol-water mixture. However, the correlation did not hold for all compounds since 9-phenanthrol-P-D-xyloside primed heparan sulfate to a similar extent as 2-naphthol-p-~xyloside, but was more soluble in the octanol phase. 4-n-Butylphenyl-p-D-xyloside also preferentially partitioned into octanol but primed heparan sulfate poorly.
The Fine Structure of Heparan Sulfate Primed by 2-Naphthol-P-D-xyEoside-Priming of chondroitin sulfate and dermatan sulfate on xylosides can result in undersulfation of the chains relative to those produced on endogenous core proteins (Gibson et al., 1977;Coster et al., 1991). To test if the heparan sulfate chains built on xylosides might be undersulfated, cells were incubated with 10 p~ 2-naphthol-P-~-xyloside and various concentrations of 35S04 at constant radiospecific activity (Fig.   6A). Incorporation of 35s04 increased as the concentration of 35S04 in the medium was increased and remained nearly constant above 75 p~. This finding suggested that adding 275 p~ sulfate would ensure full sulfation of GAGs.
To test if the extent of sulfation would depend on the dose of p-o-xyloside, cells were labeled with ~-[6-~H]glucosamine and 35S04 in the presence or absence of inorganic sulfate and increasing concentrations of 2-naphthol-P-~-xyloside (Fig. 6B). In the absence of added inorganic sulfate, the ratio of incorporated 35S and 3H counts decreased when 2-naphthol-P-~-xyloside was increased from 1 to 30 p~ and then remained constant at higher concentrations. When 100 PM inorganic sulfate was included, the ratio of incorporated 35S to 3H remained essentially constant at all concentrations of 2-naphthol-P-D-xyloside up to 300 p~, the highest concentration tested. Thus, the inclusion of 100 Partitioning of xylosides in octanol-water mixtures and heparan sulfate synthesis Xylosides were partitioned between n-octanol and water as described by Chiou et al. (1977). Xylosides (0.1 M stocks in Me2SO) were added to water saturated with n-butanol and the absorbance at 280 or 266 nm was measured by spectrophotometry. Aliquots (3 ml) were added to 3 ml end overnight at 23 "C. After centrifuging the samples for 20 min on the of n-octanol saturated with water, and the tubes were mixed end-overhigh setting of a clinical centrifuge (IEC model HN-SII), the n-octanol phase was removed and the absorbance at 280 or 266 nm of the aqueous phase was measured. Octanol-water partition coefficients (KO,,. =   C,,,,,/C,,,) were determined in the following way. The absorbance of the aqueous phase before and after partitioning was used to calculate the amount of material that partitioned into the octanol phase (Coetsnol). This value was divided by the absorbance of the sample in the water phase aRer partitioning (Cwa-r). Data are presented as the mean of duplicate determinations. Individual values did not vary by more than 12% from the mean. The proportion of heparan sulfate synthesis was determined in separate experiments in which the xylosides were tested at 10 p~. GAGs were analyzed as described under "Experimental Procedures."

Xyloside
Data are expressed as 35S cpdpg cell protein and are the mean of duplicate determinations which did not vary by more than 5% from the mean. and heparan sulfate from proteoglycans of wild-type CHO cells was qualitatively similar (Fig. 7). However, the amount of 6-0sulfated disaccharides declined about 2-fold (ISMS, derived mostly from IdoA20S03-GlcNS0360S03 units, IMS, derived mostly from IdoA-GlcNS0360S03 units, and GMS, derived mostly from GlcA-GlcNS0360S03 units).2 The amount of ISM (derived from IdoA20S03-GlcNS03 units) did not change. The decline in the 6-0-sulfated residues was accompanied by a corresponding increase in non-sulfated disaccharides. When the latter were resolved by paper chromatography, a striking decrease in IdoA-aManR was found (Table 111). The changes in disaccharide composition were also seen in a separate experiment in which the cells were incubated with only 10 p 2-naphthol-&o-xyloside.

DISCUSSION
Xylosides containing D-xylopyranose in P-linkage to hydrophobic aglycones efficiently prime GAGs. In general, P-D-XYlOsides prime chondroitin sulfate well and only weakly prime heparan sulfate (Lugemwa and Esko, 1991). This selective priming of chondroitin sulfate on P-D-xylosides is somewhat surprising since the biosynthesis of both types of GAGs occurs on the common intermediate, -GlcA~l-3Gal~l-3Gal~l-4Xylpl-aglycone. "ransfer of GalNAcpl-4 to the terminal GlcA residue initiates the assembly of chondroitin sulfate, whereas transfer of GlcNAcal-4 initiates heparan sulfate synthesis. Our results show that the addition of GalNAc01-4 or GlcNAcal-4 apparently depends on the structure of the aglycone attached to the Xyl residue. Xylosides containing simple aglycones, such as methanol, benzyl alcohol (Robinson et al., 1975;Sobue et al., 1987), or p-nitrophenol (Okayama et al., 1973 allow attachment of GalNAc, but not GlcNAc. In this report we showed that xylosides containing simple aglycones, such as naphthol, allow the transfer of both GlcNAc and Gal-NAc. in the presence of 10 p~ of the indicated xylosides. GAGs were analyzed as described under "Experimental Procedures." Data are expressed as 36S cpdpg cell protein and are the mean of duplicate determinations which did not vary by more than 5% from the mean.
p inorganic sulfate was sufficient to permit full sulfation of GAGs. We next examined the fine structure of the heparan sulfate primed on 2-naphthol-p-~-xyloside by analyzing the disaccharide composition of the chains. The composition of heparan sulfate primed on 2-naphthol-P-~-xyloside in pgsA-745 cells Some ISMS, IMS, and ISM may have originated from the corresponding N-acetylated disaccharide units. However, the modified disaccharides tend to predominate in sections of the chain that have undergone extensive GlcN N-sulfation.

Xylosides That Prime Heparan Sulfate
What is the mechanism that controls the addition of GlcNAc and GalNAc and therefore the type of GAG chain assembled on particular xylosides? One possibility is that chondroitin sulfate and heparan sulfate synthesis occurs in different subcellular compartments. Selective entry of xylosides according to the structure or hydrophobicity of the aglycone could explain the observed differences in heparan sulfate priming. Although some work has been done to localize various enzymes involved in chondroitin sulfate (Sugumaran et  of 36S0, at constant radiospecific activity. E , one set of cells were incubated in medium containing 1 ~l l~ glucose, 100 1.1~ added Na2S04, 100 pCi/ml [6-3H]-~-glucosamine, and 50 pCi/ml %04 (0). A second set of cells were incubated in medium containing 100 pCi/ml 8 pCi/ml [35SlH,S04,and no added Na2S04 (0). The concentration of 2-naphthol-p-D-xyloside was varied, and the radioactive GAG chains were counted by liquid scintillation spectrometry using dual channels to determine the amount of 36S cpm and 3H cpm. The data were corrected for about 25% spillover of 35S counts in the 3H channel and about 5% spillover of 3H counts in the 35S window. Data in panel A are presented as total incorporation of 35S cpdwell whereas data in panel B are presented as the ratio 36SPH in the GAG chains. All data are the mean of duplicate determinations which varied by less than 10% from the mean. Yanagishita, 1993), definitive evidence for their location is lacking (Kjellen and Lindahl, 1991). The endoplasmic reticulum, Golgi, and plasma membranes differ in protein and lipid composition (Daum, 19851, suggesting the possibility that xylosides of different hydrophobicity or chemical properties might partition preferentially into particular compartments. Molecules such as benzo(a)pyrene have been shown to localize to distinct sites within cells (Plant et al., 1985) and exogenous membrane lipids sort to different subcellular locations (Pagano, 1990). Perhaps xylosides with different aglycones follow distinct pathways after they enter cells as well.
Another explanation for the differential priming ability of various xylosides should be considered. Free xylose and xylosides containing simple aglycones (e.g. methanol) prime chondroitin sulfate readily, suggesting that chondroitin sulfate assembly probably occurs by default. In contrast, heparan sulfate synthesis on xylosides requires a more complex aglycone. One can speculate that the aromatic rings

TABLE I11
Disaccharide composition of heparan sulfate chains and nitrous acid cleavage. The disaccharides were reduced and analyzed by reverse phase ion pairing chromatography as described under Heparan sulfate chains produced on 2-naphthol-P-o-xyloside or endogenous core proteins were depolymerized to disaccharides by hydrazinolysis "Experimental Procedures." The amount of radioactivity in each of the disaccharides was used to determine the relative abundance of each unit. The abbreviations are the same as in Fig. 7 hydrophobic determinant on proteoglycan core proteins that interacts with the transferase catalyzing the addition of the GlcNAca1,4 to the linkage tetrasaccharide. In this model, core proteins that do not normally carry heparan sulfate chains would lack this determinant. If the latter hypothesis is correct, tetrasaccharide intermediates containing naphthol may preferentially bind the transferase that catalyzes the addition of GlcNAcal-4 to the terminal GlcA unit. This enzyme may differ from the GlcNAcal-4 transferase involved in chain polymerization. A previous study of the substrate specificity of GalNAcp1,4 addition suggests that two different enzymes catalyze chondroitin sulfate initiation and polymerization (Rohrmann et al., 1985).
The model predicts that GlcNAca1-4 transferase recognizes the aglycone as well as the terminal GlcA of the linkage tetrasaccharide. Freeze et al. (1993) showed that Gal transferase 11, which transfers Galp1,3 to the intermediate Galp1,4Xylplduring GAG synthesis, recognizes both the anomeric configuration at the reducing end of this disaccharide as well as the terminal galactose. Gal addition to xylose catalyzed by Gal transferase I occurred to both p-nitrophenyl-a-and p-nitrophenyl-p-o-xylose in cells. The addition of the second Gal residue, however, did not occur to the a-linked compound, suggesting that Gal transferase I1 distinguishes between a and p configuration of the terminal xylose. The a2-3 sialyltransferase that generates NeuAca2-3Gal~l-4xylosides exhibits similar anomeric selectivity (Freeze et al., 1993).
Examination of disaccharides generated from heparan sulfate chains made on 2-naphthol-p-D-xyloside and those derived from CHO cell heparan sulfate proteoglycans showed striking differences. First, a reduction in the proportion of all 6-0-sulfated disaccharides occurred, suggesting that the action of 6-O-sulfotransferase(s) may depend on the aglycone attached to the end of the chain. The differences also might reflect the enhanced flux of chains through the cells, since the addition of xylosides causes a 34-fold increase in heparan sulfate synthesis over that occurring on endogenous core proteins. When this stimulation is taken into account, the number of 6-0-sulfated residues produced on GAGS primed by 2-naphthol-p-~-xyloside in pgsA-745 cells was actually greater than that produced on proteoglycans in wild-type cells. Previous reports have shown that the structure of dermatan sulfate primed by P-o-xylosides can differ from that synthesized on core proteins (Coster et al., 1990;Fransson et al., 1991). Chain length, extent of epimerization of GlcA to IdoA, degree of sulfation, and the frequency of periodic repeats of clustered GlcA-GalNAc declined, depending on the concentration of xyloside (Fransson et al., 1992). In our study, the amount of non-sulfated IdoA-containing disaccharides in heparan sulfate declined dramatically (Table 1111, but when the enhanced rate of synthesis of GAG chains was taken into account, the total number of IdoA units actually increased.