INTRODUCTION

Heparan sulfate proteoglycans interact with a large number of proteins involved in inflammation, coagulation, lipid metabolism, and development (1, 2, 3). To a large extent these interactions depend on specific sugar residues in the carbohydrate chains and amino acids of the protein (1, 2). The best studied example to date is the ``lock-and-key'' interaction between heparan sulfate and heparin with antithrombin III (4). More recent studies have focused on sequences that mediate binding to basic fibroblast growth factor (5, 6), hepatocyte growth factor (scatter factor) (7, 8), and lipoprotein lipase (9). These proteins exhibit different, albeit overlapping, sugar specificities for binding heparan sulfate (Table I).

Table I. Heparan sulfate binding sequences Ligand Binding sequence Refs. Antithrombin -GlcNSO3 (6OSO3)-GlcUA-GlcNSO3 (3OSO3) (±6OSO3)-IdUA (2OSO3)-GlcNSO3- 4 Basic fibroblast growth factor -HexUA-GlcNSO3-HexUA-GlcNSO3-IdUA (2OSO3)- 5, 6 Hepatocyte growth factor (scatter factor) [-IdUA-GlcNSO3 (6OSO3)-]n 7, 8 Lipoprotein lipase [-IdUA(2OSO3)-GlcNSO3 (6OSO3)-]n 9

The mechanisms that control the production of specific binding sequences in heparan sulfate are unknown. Presumably, control is exerted at the level of biosynthesis, which in turn is determined by the action of the enzymes that catalyze chain polymerization, GlcNAc N-deacetylation and N-sulfation, glucuronic acid (GlcUA)¹ epimerization, 2-O-sulfation of uronic acid residues, and 3-O- and 6-O-sulfation of glucosaminyl residues. In order to understand how these reactions are orchestrated, we and others have undertaken a genetic analysis of heparan sulfate assembly (10). Mutants altered in chain initiation (11, 12), polymerization (13, 14), and sulfation (15, 16, 17, 18, 19, 20, 21) have been biochemically characterized and used in a variety of systems to establish the importance of heparan sulfate and overall sulfation in binding (10). Recently, Rosenberg and co-workers identified mutants altered in antithrombin III binding and suggested that in addition to the biosynthetic enzymes, other factors may exist that alter the production of antithrombin binding sequences (22, 23, 24). In this report, we describe a new Chinese hamster ovary (CHO) cell mutant altered in sulfation of heparan sulfate. This strain lacks 2-O-sulfotransferase activity in vitro and produces heparan sulfate chains lacking 2-O-sulfated iduronic acid and glucuronic acid.

EXPERIMENTAL PROCEDURES

CHO-K1 cells were obtained from the American Type Culture Collection (CCL-61, Rockville, MD). The cells were grown under an atmosphere of 5% CO₂ in air and 100% relative humidity in Ham's F-12 growth medium (Life Technologies, Inc.) supplemented with 7.5% (v/v) fetal bovine serum (HyClone Laboratories), 100 µg/ml streptomycin sulfate and 100 units/ml penicillin G. Sulfate-free medium was prepared from individual components (25), substituting chloride salts for sulfate and fetal bovine serum that had been dialyzed exhaustively against phosphate-buffered saline (26). Low glucose medium contained 1 mM glucose instead of 10 mM.

bFGF was radioiodinated using the modified chloramine T method described by Olwin and Hauschka (27). Recombinant human bFGF (20 µg, a generous gift from Chiron Corp.) was mixed with 0.5 M sodium phosphate (20 µl, pH 7.0), 0.45 M chloramine T (30 µl), and 2 mCi of Na¹²⁵I (DuPont NEN) in a final volume of 100 µl. After 2 min, 25 mM dithiothreitol was added (100 µl), and the ¹²⁵I-bFGF was purified on a column of heparin-Sepharose CL-6B (Pharmacia Biotech Inc.) equilibrated with 20 mM HEPES (pH 7.4), 0.5 M NaCl, and 0.2% (w/v) bovine serum albumin. The ¹²⁵I-bFGF was eluted with a solution of 3 M NaCl, 20 mM HEPES (pH 7.4), and 0.2% (w/v) bovine serum albumin. The specific activity was 2-5 × 10⁵ cpm/ng.

Wild type CHO cells were mutagenized with ethylmethane sulfonate as described previously (28) and frozen under liquid nitrogen. A portion of cells was thawed, propagated for 3 days, and used to establish replica plates as described in the legend of Fig. 1. Mutants tentatively defective in binding ¹²⁵I-bFGF were picked from the original master plates using cloning cylinders and trypsin as described elsewhere (28).

Cells were labeled for 24 h with 100 µCi/ml Na³⁵SO₄ (25-40 Ci/mg, DuPont NEN) in sulfate-free medium or with 100 µCi/ml D-[6-³H]glucosamine HCl (40 Ci/nmol, DuPont NEN) in low glucose medium. Radiolabeled GAG chains were isolated in the following way. The spent medium was removed and the cell layer was washed three times with cold phosphate-buffered saline without calcium or magnesium (26) and then treated with 1 ml of 0.1 M NaOH. After 15-20 min, an aliquot was removed for protein determination using the Bio-Rad protein assay kit (Bio-Rad Laboratories) and bovine serum albumin as standard. The remaining material was adjusted to pH 7 with 10 M acetic acid and combined with the spent medium. Chondroitin sulfate A (2 mg) was added and 0.166 volume of a protease solution containing 1 mg/ml Pronase (Boehringer Mannheim) in 0.24 M sodium acetate (pH 6.5) and 1.92 M NaCl. After overnight incubation, the reaction mixture was diluted 5-fold with water to reduce the salt concentration to ~0.1 M. The solution was applied to 0.5 ml column of DEAE-Sephacel prepared in a disposable polypropylene pipette tip plugged with glass wool. The column was washed with 20 mM sodium acetate buffer (pH 6.0) containing 0.25 M NaCl. Bound GAGs were eluted with 1 M NaCl in 20 mM sodium acetate (pH 6.0) and precipitated with 4 volumes of ethanol at 4 °C (2 h). The precipitate was dissolved in one ml of 0.5 M sodium acetate (pH 5.5) and reprecipitated with ethanol. The final material was dissolved in 20 mM Tris HCl (pH 7.4). [³⁵S]Chondroitin sulfate was removed by treating a sample overnight at 37 °C with 20 milliunits of chondroitinase ABC (Seikagaku) followed by DEAE-Sephacel chromatography.

GAG chains were treated at 4 °C for 24 h with 1 M NaBH₄ in 0.5 M NaOH to -eliminate the chains. The samples were diluted with water and purified by another round of DEAE chromatography. ³⁵S-GAGs were analyzed by anion-exchange HPLC using a 7.5 mm × 7.5-cm column of DEAE-3SW (TosoHaas, Montgomeryville, PA). The column was equilibrated in 10 mM KH₂PO₄ buffer (pH 6.0) containing 0.2% (w/v) Zwittergent 3-12 and 0.2 M NaCl. The GAGs were eluted with a linear gradient of NaCl (0.2-1 M) in the same buffer using a flow rate of 1 ml/min and by increasing the NaCl concentration by 10 mM/min. The effluent from the column was monitored for radioactivity with an in-line radioactivity detector (Radiomatic Flo one/beta, Packard Instruments) with sampling rates every 6 s and data averaged over 1 min.

N-Sulfate groups were removed from heparan sulfate by solvolysis (29). A solution of [³⁵S]heparan sulfate in water was passed through a 1-ml column of Dowex 50W-X8 prepared in disposable polypropylene pipette tips. The acidic fractions were pooled neutralized with pyridine and lyophilized. The sample was dissolved in 95% (v/v) Me₂SO₄ in water and heated at 50 °C for 2 h. The material was applied to a Sephadex G-50 column (29 × 1 cm) and eluted with 0.5 M pyridinium acetate (pH 5.0) at a flow rate of 6 ml/h to separate intact chains from inorganic sulfate. The eluate was collected in 0.5-ml fractions and counted by liquid scintillation chromatography.

2-O-Sulfate groups were preferentially removed by base-treatment as described by Liu and Perlin (30). [³⁵S]Heparan sulfate was resuspended in water, adjusted to pH 11.8 with 0.1 M NaOH, and lyophilized. The resulting yellow powdery residue was resuspended in 0.5 M pyridinium acetate (pH 5.0) and chromatographed on a Sephadex G-50 column as described above to separate inorganic sulfate from intact chains.

Depolymerization of the chains by cleavage at N-sulfated GlcN residues was achieved by treating [³H,³⁵S]heparan sulfate with nitrous acid at pH 1.5 (31, 32). The oligosaccharides were reduced with NaBH₄, dissolved in 0.5 ml of 0.5 M pyridinium acetate (pH 5.0), and chromatographed on a column of Bio-Gel P-10 (1.5 × 75 cm) in 0.5 M pyridinium acetate (pH 5.0) (15). The area under each peak was used to calculate the degree of GlcN N-sulfation (15).

Complete depolymerization of [³H]heparan sulfate to disaccharides was achieved by N-deacetylation and nitrous acid cleavage at low and high pH (31, 32). Briefly, the chains were deacetylated by treatment with 70% (w/v) aqueous hydrazine containing 1% (w/v) hydrazine sulfate at 96 °C for 4 h. Excess hydrazine was removed by repeated evaporation of the sample to dryness from toluene. After passing the deacetylated material through a column of Sephadex G-25, it was lyophilized and then treated with nitrous acid at pH 1.5 and at pH 3.9, followed by reduction with NaBH₄ (33). The disaccharides were purified by gel filtration chromatography (Bio-Gel P-2, ~75% yield) and analyzed by strong anion-exchange HPLC on a Whatman Partisil 5-SAX column (250 × 4.6 mm) using a step-wise gradient of KH₂PO₄: 12 mM for nonsulfated disaccharides, 26 mM for monosulfated disaccharides, and 150 mM for disulfated species (34). Individual peaks were identified by comparison with disaccharides from commercial heparin and with published data (16, 34).

Nonsulfated [³H]disaccharides were separated by reverse-phase ion-pairing chromatography (35) and pooled. The sample was analyzed by descending paper chromatography on Whatman No. 3MM paper in ethyl acetate/acetic acid/water (3:1:1, v/v) before and after treatment with -glucuronidase (type B-10, from bovine liver, Sigma). Some of the IdUA-[³H]aMan_R was cleaved to [³H]aMan_R due to contaminating -iduronidase in the -glucuronidase preparation as first described by Matalon et al. (36). Treatment of standard IdUA-[³H]aMan_R and analysis of the products on the Partisil 5-SAX column revealed that ~10% of the material was cleaved. Therefore, the content of IdUA-[³H]aMan_R was corrected for this loss.

Disaccharide standards were prepared in the following way. GlcUA-[³H]aMan_R was made from Escherichia coli K5 N-acetylheparosan by hydrazinolysis, high pH nitrous acid treatment, and borotritide reduction; ³H-aMan_R was obtained by treating a portion of GlcUA-[³H]aMan_R with -glucuronidase; IdUA-[³H]aMan_R was prepared by solvolyzing IdUA(2OSO₃)-[³H]aMan_R(6OSO₃) collected from a reversed-phase ion-pairing run of disaccharides derived from wild-type CHO cell [³H]heparan sulfate. Commercial heparin was sequentially depolymerized to disaccharides as described above and reduced with NaB³H₄.

The substrate for measuring sulfotransferase activity was prepared from porcine mucosal heparin. The material (50 mg) was first converted to the free acid by passing through a column of Dowex 50W-X8 (0.5 ml). The acidic fractions were pooled, neutralized with pyridine and solvolyzed with 90% (v/v) Me₂SO₄ in methanol at 100 °C for 7 h as described by Nagasawa et al. (37) and Lloyd et al. (38). Inorganic sulfate was removed by buffer exchange on a PD-10 column (Pharmacia). The sample was re-N-sulfated by incubation with trimethylamine-sulfur trioxide at 55 °C for 24 h (38), desalted, and lyophilized. N-Desulfoheparin was prepared by mild solvolysis (37).

The activities of PAPS:heparan sulfate GlcN N-sulfotransferase and PAPS:heparan sulfate hexuronic acid 2-O-sulfotransferase were measured in CHO cell homogenates as described in detail by Bame and Esko (15) and Bame et al. (16). CHO cells were grown to confluence, rinsed three times with cold phosphate-buffered saline (26), and detached with a rubber policeman in 50 µl of buffer containing 0.25 M sucrose, 50 mM Tris-HCl (pH 7.5), 1% (w/v) Triton X-100, 1 µg/ml leupeptin, 1 µg/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride. Aliquots of the cell extracts were stored at 20 °C. The sulfate donor, [³⁵S]PAPS was prepared using yeast homogenates as described by Renosto and Segel (39) and Robbins (40).

The standard reaction mixture for measuring 2-O-sulfotransferase activity contained 50 mM HEPES (pH 7.4), 1% (w/v) Triton X-100, 10 mM MnCl₂, 10 mM MgCl₂, 50 µM [³⁵S]PAPS (~0.2 Ci/mmol), 25 µg of O-desulfoheparin, 25 µg of cell protein in a final volume of 25 µl. N-Sulfotransferase activity was measured by incubating 25 µg of cell protein with 50 mM HEPES (pH 7.4), 1% (w/v) Triton X-100, 10 mM MnCl₂, 1 mM MgCl₂, 50 µM [³⁵S]PAPS (~0.2 Ci/mmol), and 25 µg of N-desulfoheparin in a total volume of 25 µl. The mixtures were incubated for 60 min at 37 °C, and the reaction was stopped by adding 475 µl of 0.1 M EDTA (pH 7.4) containing 0.25 mg of heparin. The ³⁵S-labeled products were separated from unreacted [³⁵S]PAPS by anion exchange chromatography on 0.5-ml columns of DEAE-Sephacel packed in disposable pipette tips (15, 16). The products were depolymerized with nitrous acid at pH 1.5 and reduced with NaBH₄. The disaccharides were isolated by gel filtration chromatography and analyzed on a Partisil 5-SAX HPLC column as described above. Aliquots of the products were also subjected to alkaline 2-O-desulfation and the liberated ³⁵SO₄ was separated by gel filtration chromatography as described above.

RESULTS

In previous studies we had developed an autoradiographic technique for identifying proteoglycan-deficient mutants of CHO cells based on the incorporation of ³⁵SO₄ into colonies (10, 28). This method yielded mutants defective in xylosylation (11) and galactosylation (12), sulfate transport (20, 21), heparan sulfate polymerization (13, 14), and N-sulfation of heparan sulfate (15, 16, 17, 18). The defects generally corresponded to large changes in ³⁵SO₄ incorporation, which made it relatively easy to detect the mutants by colony autoradiography techniques. Mutants in other biosynthetic steps, for example 2-O-sulfation and uronic acid epimerization, were not found presumably because these alterations would not result in substantial changes in ³⁵SO₄ incorporation. To obtain these mutants, we developed a screening strategy based on binding of bFGF to heparan sulfate proteoglycans expressed on the surface of cells. The interaction of bFGF with heparin and heparan sulfate depends on an oligosaccharide sequence consisting of -HexUA-GlcNSO₃-HexUA-GlcNSO₃-IdUA(2OSO₃), in which the 2-O-sulfated IdUA residue is essential for binding (6). Thus, we reasoned that mutants defective in 2-O-sulfation or iduronic acid formation should fail to bind bFGF, but incorporate almost normal levels of ³⁵SO₄ into proteoglycans.

Using replica plating, we cloned mutagen-treated CHO cell colonies into two layers of polyester cloth (28, 41). One disk was used to measure ³⁵SO₄ incorporation into trichloroacetic acid-precipitable proteoglycans and the other was used to measure ¹²⁵I-bFGF binding. In the latter assay, the colonies were washed with 0.6 M NaCl to remove ¹²⁵I-bFGF bound with low affinity. Most of the colonies that failed to bind bFGF also did not incorporate ³⁵SO₄, but we occasionally found strains that incorporated normal levels of sulfate (incidence ~10⁵). Mutants that failed to bind bFGF, like the one indicated by the arrow in Fig. 1, stood out when the autoradiographic image of the colony was compared to the protein content of the colony measured by Coomassie Blue staining. One particular strain, mutant 17, was recloned and further characterized. Mutant 17 bears a unique recessive mutation based on complementation of previously identified mutations in proteoglycan synthesis (data not shown). Therefore, it was designated pgsF-17.

We first checked the synthesis of GAGs in the mutant by biosynthetic labeling studies with [6-³H]GlcN and ³⁵SO₄. The mutant produced a somewhat lower amount of [³H,³⁵S]heparan sulfate as wild-type cells (4.4 ± 1.2 × 10⁴ versus 6.7 ± 1.1 × 10^{4 35}S cpm/µg of cell protein and 2.6 ± 0.6 × 10⁴ versus 3.8 ± 0.7 × 10^{4 3}H cpm/µg of cell protein, respectively) but the ratio of ³⁵S/³H was almost the same (1.7 versus 1.8, respectively). Both strains secreted ~70% of the radiolabeled proteoglycans into the medium and compositional analysis showed that the proteoglycans contained ~70% heparan sulfate and ~30% chondroitin sulfate by ³⁵S counts.

HPLC anion-exchange analysis of the GAG chains from wild-type cells resolved heparan sulfate (0.28-0.52 M NaCl) from chondroitin sulfate (0.52-0.62 M NaCl) (Fig. 2A). The GAG chains from the mutant also separated into two peaks of material, but the position of the earlier peak was shifted to higher salt (Fig. 2B). After treating the sample with chondroitinase ABC, the resistant material, tentatively identified as heparan sulfate, eluted at higher salt in the mutant than in the wild type (Fig. 2, filled circles). This finding suggested that the heparan sulfate from the mutant was more negatively charged than the material from the wild type. Chondroitin sulfate made by mutant and wild-type cells did not exhibit any striking differences in elution from the column, and the amount remained unchanged. Thus the decrease in bFGF binding correlates with a change in the structure of the heparan sulfate chains.

To determine how the mutation in pgsF-17 altered the structure of heparan sulfate, [³H,³⁵S]heparan sulfate was purified from cells and spent media (see ``Experimental Procedures''). Solvolysis of the material under conditions where only sulfate groups from GlcNSO₃ units were liberated (37) showed that heparan sulfate from wild-type cells contained ~45% N-sulfate groups and ~55% O-sulfate groups, which yielded a ratio of O-sulfation to N-sulfation of ~1.2. In contrast the mutant contained ~64% N-sulfate groups and ~36% O-sulfate groups, which yielded a ratio for O-sulfates and N-sulfates of ~0.6 (n = 2, S.D. 10%). Thus, the mutant appeared to make heparan sulfate with less O-sulfates compared to N-sulfates.

These data did not distinguish if the defect depressed O-sulfation, enhanced N-sulfation, or both. To study this problem, the absolute amount of N-sulfated GlcN residues was estimated by treating samples with nitrous acid at pH 1.5 (31). These conditions depolymerize heparan sulfate between GlcNSO₃ and HexUA residues, liberating oligosaccharides whose length reflects the spacing between N-sulfated GlcN units. Therefore, the proportion of ³H counts in each oligosaccharide corrected for the number of internal GlcNAc residues reflects the proportion of GlcNSO₃ units. Gel filtration profiles of the nitrous acid cleavage products are shown in Fig. 3. By this method, heparan sulfate from wild-type cells contained about 47% N-sulfated GlcN residues, whereas the material from the mutant contained about 64% GlcNSO₃. Thus, heparan sulfate from the mutant was more highly N-sulfated.

Heparan sulfate from CHO cells contains O-linked sulfate groups attached to the 6-OH group of GlcN residues and to the 2-OH position of IdUA and GlcUA residues (16). In order to examine the positional distribution of O-sulfate groups, samples of [³⁵S]heparan sulfate were treated with alkali under conditions that preferentially remove the sulfate groups from IdUA(2OSO₃) without disturbing N-sulfates or other O-sulfates (30). Treatment of wild-type [³⁵S]heparan sulfate in this way released 33 ± 3% of ³⁵S counts, but treatment of heparan sulfate from the mutant released only 10 ± 3% counts. Since O-sulfate groups represents ~55% of the ³⁵S counts in wild-type heparan sulfate and ~35% in material from the mutant, we calculated that 6-O-sulfate groups accounts for ~22% of ³⁵S counts in wild-type chains and ~26% in the mutant. This finding suggested that 6-O-sulfation was normal or slightly elevated in the mutant, and that 2-O-sulfation was decreased at least 3-fold.

To confirm these findings, we analyzed the disaccharide composition of heparan sulfate from mutant and wild-type cells. [³H]Heparan sulfate was N-deacetylated and completely depolymerized by treatment with nitrous acid at pH 1.5 and 4, and the liberated disaccharides were reduced with NaBH₄. This type of analysis yields nonsulfated, monosulfated, and disulfated disaccharides with characteristic elution positions from a Partisil 5-SAX HPLC column (Fig. 4). Both disulfated disaccharides (peaks 8 and 9, GlcUA(2OSO₃)-aMan_R(6OSO₃) and IdUA(2OSO₃)-aMan_R(6OSO₃)) and the monosulfated disaccharide (peak 7, IdUA(2OSO₃)-aMan_R) were missing entirely in the mutant. In contrast, peaks 5 and 6 (GlcUA-aMan_R(6OSO₃) and IdUA-aMan_R(6OSO₃)) accumulated in the mutant. The nonsulfated material (peaks 1, 2, and 3) was not adequately resolved under these conditions, but subsequent analysis by reversed-phase ion-pairing chromatography and paper chromatography (Fig. 5) showed that their relative amounts (GlcUA-aMan_R, IdUA-aMan_R, and aMan_R) did not differ significantly. These results are summarized in Table II. When this information was combined with the results of solvolysis and low pH nitrous acid depolymerization, we obtained values for the various modifications that heparan sulfate undergoes (Table III). As shown, the proportion of GlcUA and IdUA did not differ significantly. However, the total amount of 2-O-sulfation was dramatically depressed in the mutant, whereas the extent of glucosaminyl N-sulfation and 6-O-sulfation was enhanced.

Table II. Disaccharide composition of heparan sulfate from mutant and wild-type cells [3H]GlcN-labeled heparan sulfate from mutant 17 and wild-type was cleaved by nitrous acid-catalyzed deamination (see ``Experimental Procedures''). The different monosaccharides and disaccharides were identified by anion-exchange and paper chromatography and quantitated by scintillation counting. Each value was expressed as a fraction of total material recovered. The data represent a single exemplary analysis. In duplicate experiments, the values varied by 20% of the those shown. Disaccharides Wild type pgsF-17 % of total aManR 7.5 9.8 GlcUA-aManR 41.5 48.5 IdUA-aManR 18.1 17.7 GlcUA-aManR (6OSO3) 3.6 9.1 IdUA-aManR (6OSO3 4.3 14.5 GlcUA (2OSO3)-aManR 0.5 <0.1 IdUA (2OSO3)-aManR 10.8 0.1 GlcUA (2OSO3)-aManR (6OSO3) 2.0 <0.1 IdUA (2OSO3)-aManR (6OSO3) 11.7 0.2

Table III. Compositional analysis of heparan sulfate [3H, 35S]Heparan sulfate was analyzed for N-sulfated GlcN residues, 2-O-sulfates, 6-O-sulfates, GlcUA, and IdUA acid by nitrous acid catalyzed depolymerization of the chains as described in the text. Note that the sum of GlcNAc and GlcNSO3 equals 100% of the total glucosaminyl residues, the sum of N-sulfates, 2-O-sulfate, 6-O-sulfates equals 100% of the total sulfate groups and the sum of GlcUA and IdUA equals 100% of the total HexUA acids. The values represent the average of two or more experiments, which varied by 10%. Residue Wild-type pgsF-17 per 100 disaccharides GlcNAc 55 35 GlcNSO3 45 65 2-O-Sulfate 30 <1 6-O-Sulfate 25 34 GlcUA 53 58 IdUA 47 42

These compositional studies suggested that mutant 17 had a deficiency in a sulfotransferase responsible for the generation of 2-O-sulfated hexuronic acid residues. To test this hypothesis, an assay was designed and optimized using N,O-desulfated, re-N-sulfated heparin as a substrate, and crude cell homogenates from wild-type and mutant CHO cells as a source of enzyme. In the wild-type, enzyme activity was proportional to time for ~1 h and to the amount of cell protein up to at least 100 µg. Analysis of the reaction products by nitrous acid depolymerization and Partisil 5-SAX HPLC chromatography showed only one disaccharide (IdUA(2OSO₃)-aMan_R, 83% yield) and ³⁵SO₄ (17% yield), which was released presumably from GlcN³⁵SO₃ residues generated by simultaneous GlcNAc N-deacetylation/N-sulfation in the crude extract (Fig. 6A). The residual product produced by the mutant migrated almost entirely as ³⁵SO₄ (92% of ³⁵S counts). The few remaining counts migrated in the position of IdUA-aMan_R(6OSO₃). Thus, the mutant lacks > 98% 2-O-sulfotransferase activity. The decrease in enzyme activity was not caused by a soluble inhibitor in the mutant or from a lack of soluble activator present in wild-type cells since enzyme activity was additive when extracts of wild-type (110 ± 11 pmol/min/mg) and mutant cells (17 ± 4 pmol/min/mg) were mixed and assayed (53 ± 4 pmol/min/mg) (data uncorrected for GlcN sulfation). The defect was specific for the 2-O-sulfotransferase since GlcN N-sulfotransferase activity was the same in mutant (108 ± 8 pmol/min/mg) and wild-type cells (116 ± 8 pmol/min/mg) (n = 2).

DISCUSSION

In this report we describe a new mutant altered in 2-O-sulfation of heparan sulfate. Previous CHO cell mutants found by replica plating included strains with defects in the formation of the core protein linkage tetrasaccharide (11, 12), heparan sulfate polymerization (13, 14), GlcN N-sulfation (15, 16, 17, 18), and sulfate transport (20, 21). All of these strains were identified through colony autoradiography, which measured sulfate incorporation into proteoglycans in colonies replica plated to polyester cloth. More specific assays were needed to detect mutants altered in O-sulfation since these modifications account for only a small proportion of sulfate incorporated into heparan sulfate chains. De Agostini et al. (22) approached this problem by blotting colonies with ¹²⁵I-antithrombin, which yielded mutants altered in the expression of heparan sulfate with high affinity for antithrombin, but normal levels of sulfate incorporation (23). In this report we used ¹²⁵I-bFGF blotting to detect mutants altered in binding to heparan sulfate. The identification of mutant pgsF-17 and the demonstration that it is specifically altered in hexuronic acid 2-O-sulfation expands the repertoire of biochemically defined CHO mutants altered in glycosaminoglycan assembly and suggests that the blotting technique may yield other mutants altered in uronic acid epimerization and 6-O-sulfation.

The lack of 2-O-sulfation of IdUA residues in the heparan sulfate chains and the depression of enzyme activity in vitro suggests that pgsF-17 most likely bears a mutation in the structural gene encoding the enzyme. However, we cannot exclude the possibility that the mutant contains a lesion in a regulatory component or a noncatalytic subunit required for activity. Mixing experiments suggest that soluble inhibitors of enzyme activity are most likely absent. Nevertheless, cloning of the wild-type allele, correction of the mutant phenotype by transfection, and subsequent expression studies of the enzyme will be needed to establish this point.

The interaction of bFGF with heparan sulfate has attracted considerable attention because of its importance in growth factor activation and mitogenesis, and as an avenue for finding antagonists that might have useful clinical applications (6, 42, 43, 44). A number of studies have pointed to the importance of 2-O-sulfated iduronic acid in binding of bFGF to heparin (45) and heparan sulfate (5, 46). More recently, Maccarana et al. (47) found that the pentasaccharide, -HexUA-GlcNSO₃-HexUA-GlcNSO₃-IdUA(2OSO₃)-, was the minimal sequence for binding. Studies of the mutant confirm this observation since cell surface heparan sulfate on mutant cells bound bFGF less avidly than polysaccharide on wild-type cells. In other studies, we have found that heparan sulfate from the mutant elutes from a bFGF affinity column with less salt compared to chains from wild-type cells, consistent with the idea that the 2-O-sulfate groups in heparan sulfate confer high affinity binding characteristics to the heparan sulfate chains.² Current studies are focused on whether the lower affinity binding can facilitate interaction of bFGF with high affinity signal transducing receptors introduced into the mutant by transfection. Despite the decrease in 2-O-sulfation and diminished binding to bFGF, heparan sulfate from the mutant bound to anion-exchange resin more tightly than chains from wild-type cells. Presumably, the tighter binding was due to enhanced N-sulfation (and possibly 6-O-sulfation) in the mutant (Table III).

Structural studies of heparan sulfate from the mutant and enzymatic assays showed that pgsF-17 lacks a hexuronic acid 2-O-sulfotransferase. This enzyme occupies a central role in both heparin and heparan sulfate assembly. The polymer modification reactions initiate by the removal of acetyl groups from a subset of GlcNAc residues with rapid, nearly stoichiometric addition of a sulfate group to form GlcNSO₃ units³ (48, 49). The adjacent GlcUA residue can then undergo epimerization to L-IdUA and subsequent 2-O-sulfation. However, unlike N-deacetylation/N-sulfation, tight coupling between IdUA formation and 2-O-sulfation does not seem to occur, since the mutant continues to produce IdUA almost normally (Table III). Similarly, the addition of 6-O-sulfate groups to glucosamine residues proceeds independently of 2-O-sulfation. Studies of heparin biosynthesis in mastocytoma suggested that the 2-O-sulfotransferase and 6-O-sulfotransferase activities may share a common subunit or reside in a single protein (50). However, purification of the 6-O-sulfotransferase from the culture medium of CHO cells lacks 2-O-sulfotransferase activity (51). The somewhat elevated level of 6-O-sulfation in the CHO cell mutant argues that 2-O- and 6-O-sulfotransferases are most likely independent enzymes.

A rather intriguing finding concerns the apparent increase in the extent of GlcNAc N-deacetylation and N-sulfation from 40-45% of the residues in the wild-type to 60-65% in the mutant (Table III). These reactions initiate the polymer modification sequence, creating the preferred substrate for the epimerase that converts adjacent D-GlcUA to L-IdUA residues (52). The 2-O-sulfotransferase then adds sulfate groups to the 2-OH position of IdUA units. The increase in GlcNAc N-deacetylation/N-sulfation in the mutant suggests an apparent coupling of these earlier reactions with 2-O-sulfation of IdUA. Since N-sulfotransferase activity in mutant and wild-type cells was comparable, it seems likely that the presence of 2-O-sulfated residues somehow decreases the rate or extent of the N-deacetylase/N-sulfotransferase reaction. Alternatively, the decrease in 2-O-sulfation may increase the availability of PAPS, the active sulfate donor, which enhances the extent of GlcNAc N-deacetylation/N-sulfation and 6-O-sulfation. Changing the availability of UDP-Gal apparently alters the composition and sulfation of glycosaminoglycans in mouse 3T3 cells (53), suggesting that compensatory changes may occur when biosynthesis is altered.

CHO cells produce small amounts of oligosaccharide sequences containing the disaccharide GlcUA(2OSO₃)-GlcN(Ac/SO₃)(6OSO₃) and lesser amounts of GlcUA(2OSO₃)-GlcN(Ac/SO₃) (Fig. 6 and Table II). Sulfation of GlcUA residues at the 2-OH position is a relatively minor modification in both heparan sulfate and heparin biosynthesis (34, 54). Microsomal incorporation studies using endogenous substrates have shown that the formation of GlcUA(2OSO₃) occurs directly by sulfation of GlcUA, and not by epimerization of IdUA(2OSO₃) residues (55). A 2-O-sulfotransferase that acts on IdUA residues was purified recently from CHO cells, but its action on GlcUA moieties was not measured (56). The final resolution of this problem awaits development of an in vitro assay for GlcUA 2-O-sulfation and cloning and expression of the various O-sulfotransferases. Interestingly, the defect in pgsF-17 cells diminishes 2-O-sulfation of GlcUA as well as IdUA. This finding indicates that the same enzyme is most likely responsible for adding sulfate groups to both GlcUA and IdUA residues.