Turnover of Heparan Sulfate Depends on 2-O-Sulfation of Uronic Acids*

To study how the pattern of sulfation along a heparan sulfate chain affects its turnover, we examined heparan sulfate catabolism in wild-type Chinese hamster ovary cells and mutant pgsF-17, defective in 2-O-sulfation of uronic acid residues (Bai, X., and Esko, J. D. (1996) J. Biol. Chem. 271, 17711–17717). Heparan sulfate from the mutant contains normal amounts of 6-O-sulfated glucosamine residues and iduronic acid and somewhat higher levels of N-sulfated glucosamine residues but lacks any 2-O-sulfated iduronic or glucuronic acid residues. Pulse-chase experiments showed that both mutant and wild-type cells transport newly synthesized heparan sulfate proteoglycans to the plasma membrane, where they shed into the medium or move into the cell through endocytosis. Internalization of the cell-associated molecules leads to sequential endoglycosidase (heparanase) fragmentation of the chains and eventual lysosomal degradation. In wild-type cells, the chains begin to degrade within 1 h, leading to the accumulation of intermediate (10–20-kDa) and small (4–7-kDa) oligosaccharides. Mutant cells did not generate these intermediates, although internalization and intracellular trafficking of the heparan sulfate chains appeared normal, and the chains degraded with normal kinetics. This difference was not due to defective heparanase activities in the mutant, since cytoplasmic extracts from mutant cells cleaved wild-type heparan sulfate chains in vitro. Instead, the heparan sulfate chains from the mutant were relatively resistant to degradation by cellular heparanases. These findings suggest that 2-O-sulfated iduronic acid residues in heparan sulfate are important for cleavage by endogenous heparanases but not for the overall catabolism of the chains.

Animal cells elaborate several membrane-associated proteoglycans bearing heparan sulfate, including betaglycan and one or more members of the syndecan and glypican families of proteoglycans (1)(2)(3)(4). After their appearance on the cell surface, the proteoglycans turn over either through a shedding process or by endocytosis (as reviewed in Ref. 5). The endocytic pathway involves several sequential steps of degradation including proteolysis of the core protein, heparanase cleavage of the heparan sulfate chains, and eventually complete degradation by proteases, exoglycosidases, and sulfatases. The heparanases cleave infrequently along the chain (6), giving rise to several fragments that accumulate in lysosomes or possibly prelysosomal compartments (7)(8)(9)(10)(11)(12)(13). Some proteoglycans (e.g. those containing a glycosylphosphatidylinositol anchor) may bypass the heparanases and degrade rapidly in lysosomes without the formation of intermediate sized fragments (13).
To study how the pattern of sulfation might affect degradation of a heparan sulfate chain in vivo, we have examined turnover in a mutant cell line defective in 2-O-sulfation of iduronic acid residues. Altered sulfation in the mutant does not affect the overall turnover of heparan sulfate, but the defect prevents the formation of intermediate breakdown products characteristic of heparanase action. The lack of intermediates results from reduced susceptibility of the chains to cellular heparanases, suggesting that the enzymes in CHO cells prefer substrates containing 2-O-sulfated uronic acids.

EXPERIMENTAL PROCEDURES
Cell Culture-Chinese hamster ovary cells (CHO-K1) were obtained from the American Type Culture Collection (CCL-61, Rockville, MD). The cells were grown under an atmosphere of 5% CO 2 , 95% 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 g/ml of streptomycin sulfate, and 100 units/ml of 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 (PBS) (26).
Pulse-Chase Experiments-Both wild-type CHO and mutant pgsF-17 cells were grown to near confluence and pulse-labeled for 1 h with 100 Ci/ml Na 35 SO 4 (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40) Ci/mg; NEN Life Science Products) in sulfate free F-12 medium. After removing the labeled medium, the cell layers were washed three times, and fresh F-12 medium supplemented with 1 mM Na 2 SO 4 was added. In some experiments, 100 g/ml chloroquine was included in the chase medium. At the times indicated in the figure legends, the medium was collected, and the cells were washed three * 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 should be addressed. E-mail: jesko@ ucsd.edu. times with PBS. Cell surface proteoglycans were released by treating the cells with 0.125% (w/v) trypsin for 5-10 min at 37°C. The cells were centrifuged at 800 ϫ g for 3 min, and the released proteoglycans were recovered in the supernatant. The pellets were designated as the intracellular pool.
Steady-state Labeling of Heparan Sulfate-Wild-type CHO and mutant pgsF-17 cells were labeled to constant radiospecific activity after 24 h with Na 35 SO 4 (50 Ci/ml) in sulfate-free F-12 medium. The lack of sulfate in the medium does not cause undersulfation of the chains, since CHO cells derive adequate sulfate from the catabolism of cysteine and methionine (27).
Purification of Heparan Sulfate-Radiolabeled glycosaminoglycan chains were isolated from various fractions in the following way. The medium, the cell surface pool, and the pellet were adjusted to 0.1 M NaOH and then neutralized with 10 M acetic acid. Chondroitin sulfate A (2 mg) and 1 ⁄6 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 were added. After overnight incubation, the solutions were diluted 5-fold with water to reduce the salt concentration to ϳ0.1 M and applied to a 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 glycosaminoglycans 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 0.5 M sodium acetate (1 ml, pH 5.5) and reprecipitated with ethanol. The final material was dissolved in 20 mM Tris-HCl (pH 7.4). [ 35 S]chondroitin sulfate was removed by treating a sample overnight at 37°C with 20 milliunits of chondroitinase ABC (Seikagaku). The remaining heparan sulfate was treated at 4°C for 24 h with 0.5 M NaOH containing 1 M NaBH 4 to ␤-eliminate the chains and reduce the terminal sugars to their corresponding alditols. Carrier chondroitin sulfate was added, the samples were diluted with water, and the glycosaminoglycans were purified by another round of DEAE chromatography and ethanol precipitation.
The [ 35 S]heparan sulfate chains were analyzed by gel filtration HPLC using a TSK G2000SW column (7.5 mm ϫ 30 cm column; Toso-Haas, Montgomeryville, PA). The column was equilibrated in 100 mM KH 2 PO 4 buffer (pH 6.0) containing 0.2% (w/v) Zwittergent 3-12 and 0.5 M NaCl and run at a flow rate of 0.5 ml/min (0.5-ml fractions). Blue dextran (Sigma) and [6-3 H]glucosamine-HCl (NEN Life Science Products) were used to determine the V o and V t of the column, respectively. The column was calibrated with standard chondroitin sulfate (ϳ50 kDa, K av ϳ0.13) and heparin fragments (13.5 kDa, K av ϳ0.4; 5-6 kDa, K av ϳ0.6; 3 kDa, K av ϳ0.8) (6,28). 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. Samples were counted using Ultima Gold XR scintillation fluid (Packard).
Digestion of Heparan Sulfate Chains in Microsomes-To prepare cell extracts, confluent monolayers of cells were washed three times with cold PBS. The cells were scraped with a rubber policeman into 50 mM Tris-HCl buffer (pH 7.4) containing 0.25 M sucrose, 1 g/ml leupeptin, 1 g/ml pepstatin A, 10 mM N-ethylmaleimide, and 1 mM phenylmethlsulfonyl fluoride. The cells were sonicated for 10 s (on 50% duty) and centrifuged at 1000 ϫ g for 10 min to remove nuclei and unbroken cells. The postnuclear supernatant was centrifuged at 14,000 ϫ g for 10 min, and the microsomal pellet was resuspended in a buffer of 50 mM citrate, 100 mM sodium phosphate (pH 5.5), and 0.15 M NaCl. Purified [ 35 S]heparan sulfate isolated from the cell surface pool of proteoglycans (ϳ5000 cpm) was mixed with the pellet prepared from ϳ5 ϫ 10 6 cells. After overnight incubation at 37°C, the sample was boiled and centrifuged for 10 min at 14,000 ϫ g to remove precipitated proteins. The supernatant was analyzed by gel filtration HPLC on a TSK G2000SW column.
CPC Degradation Assay of CHO Cell Heparanase-Heparanase activity was partially purified (ϳ100-fold) from CHO-K1 cells as described (28). Purified [ 35 S]heparan sulfate (ϳ5000 cpm) was incubated for 24 h at 37°C in a buffer of 50 mM sodium citrate and 100 mM sodium phosphate (pH 5.5, 75 l) with enzyme (1-1.5 g of protein). Carrier heparin (100 g) was added, and the reaction mixture was treated with 1% (w/v) cetylpyridinium chloride in 0.32 M NaCl, 40 mM sodium acetate (pH 5.5) at 37°C for 2 h. Under these conditions, uncleaved heparan sulfate chains precipitate, whereas the cleavage fragments do not (28). The samples were centrifuged at 1000 ϫ g for 10 min, and the 35 S counts in the supernatant were assayed by liquid scintillation counting. The size of the heparanase-cleaved heparan sulfate also was examined by gel filtration HPLC.
Assay for Recycling of Biotinylated Cell Surface Proteoglycans-Cell surface proteoglycans were biotinylated as described (29). The cells were first chilled on ice and then washed four times at 4°C with a solution containing 10 mM KH 2 PO 4 (pH 7.0), 0.137 M NaCl, 10 mM MgCl 2 , and 10 mM EDTA. The cells were gently stirred at 4°C for 30 min with 1 mg/ml of Sulfo-NHS-Biotin (Pierce) in buffer. The cells were brought to room temperature, washed once with warm buffer (37°C), twice with sulfate-free F-12 medium, and then labeled with 35 SO 4 (100 Ci/ml) in sulfate-free F12 medium for 24 h. After biotinylation, the medium was removed, and the cell layers were washed three times with PBS and incubated at 4°C for 10 min with 1 ml of buffer containing 2% Triton X-100, 10 mM K 2 HPO 4 (pH 7.4), 150 mM NaCl, 10 mM EDTA, 10 mM N-ethylmaleimide, 1 g/ml of pepstatin A, and 1 mM phenylmethlsulfonyl fluoride. The extract was centrifuged at 1000 ϫ g for 5 min, and 1 mg of chondroitin sulfate A and 50 g/ml bovine serum albumin were added to the supernatant. The sample was chromatographed on a small DEAE-Sephacel column to separate the proteoglycans from other proteins. The column was washed with 0.3 M NaCl in the same buffer and eluted with 2.5 ml of 1 M NaCl. A sample (0.5 ml) was diluted to 0.5 M NaCl with water, and 30 l of strepavidin-agarose beads (Sigma) was added. The mixtures were incubated at 4°C overnight with end-overend mixing. The agarose beads were then washed three times by centrifugation with a buffer composed of 0.1% Triton X-100, 50 mM Tris-HCl (pH 7.4), and 0.5 M NaCl. An aliquot was taken for counting, and another sample was digested with chondroitinase ABC. The digestion products were removed by centrifugation, and an aliquot of the beads was counted as a measure of heparan sulfate proteoglycans containing biotin. To calculate the extent of recycling, the counts recovered on the beads were expressed relative to the counts recovered from cells that were first labeled with 35  Transfectants were selected in growth medium containing 400 g/ml G418 (effective concentration). Resistant colonies were replica-plated to polyester cloth (31) and incubated with 125 I-labeled basic fibroblast growth factor (bFGF) (32). Colonies that bound bFGF were detected by autoradiography, picked with glass cloning cylinders, and expanded in culture. Cell extracts from the transfectants were assayed for 2-O-sulfotransferase activity as described (32). Cells were also labeled with 50 Ci/ml of 35 SO 4 for 8 h, and the [ 35 S]heparan sulfate was isolated and analyzed by gel filtration HPLC as described above.
Cell Surface Radioiodination-Confluent monolayers of wild-type and mutant pgsF-17 cells were washed three times with PBS. Sodium [ 125 I]iodide (0.5 mCi) was added in PBS with 20 units of lactoperoxidase (Sigma), and three 20-l aliquots of 0.1% H 2 O 2 were added with gentle agitation in 5-min intervals (33). The reaction was stopped by sedimenting the cells and washing them three times with PBS.
Density Gradient Fractionation-Wild-type and pgsF-17 cells labeled with 35 SO 4 for 24 h or by radioiodination were washed twice at 4°C in 20 mM HEPES buffer (pH 7.4) containing 150 mM NaCl and 2 mM CaCl 2 and scraped into 1 ml of buffer. The cells were centrifuged at 800 ϫ g for 5 min at 4°C. The cell pellets were resuspended in 1 ml of buffer containing 1 g/ml pepstatin A, 1 g/ml leupeptin, and 1 mM phenylmethlsulfonyl fluoride and forced through a 28-gauge needle 10 times. Postnuclear supernatants were obtained by centrifuging the samples at 800 ϫ g for 10 min at 4°C. Percoll (20%, w/v) (Pharmacia Biotech Inc.) was prepared in 10 mM Tris-HCl (pH 8.0) containing 0.15 M NaCl. A sample of postnuclear supernatant (1 ml) was loaded on top of 11 ml of Percoll solution and centrifuged for 50 min at 4°C at 20,000 ϫ g using a 75Ti fixed-angle rotor (Beckman). Fractions (1 ml) were collected from the top of the gradient and counted. The distribution of plasma membranes was determined be the recovery of 125 I counts from surfacelabeled proteins. The lysosomes were found in the higher density fractions, as assessed by the distribution of ␤-hexosaminidase activity.
[ 35 S]Heparan sulfate chains were isolated from pooled fractions and analyzed by gel filtration HPLC as described above.

Mutant pgsF-17 Accumulates Large Heparan Sulfate
Chains-Cell surface proteoglycans in CHO cells are either shed into the growth medium or taken up by endocytosis. The internalized proteoglycans undergo degradation with accumulation of intracellular oligosaccharides of intermediate (10 -20-kDa) and small (4 -7-kDa) size prior to complete degradation in lysosomes (6,7,28). One or more cellular heparanases catalyze the formation of these intermediate and small size fragments (6). These enzymes apparently require the presence of N-sulfated glucosamine residues in the chains, but O-sulfate groups appear to be dispensable (14,18,23,24). However, much of the evidence derives from in vitro studies of partially purified enzymes with defined substrates. To test this hypothesis directly in cells, we examined the turnover of heparan sulfate chains in a CHO cell mutant (pgsF-17) defective in 2-O-sulfation of uronic acid residues. This mutant produces heparan sulfate chains with normal levels of 6-O-sulfated glucosamine residues and iduronic acid and somewhat higher levels of N-sulfated glucosamine residues (60% GlcNSO 3 in the mutant versus 45% in the wild-type). In contrast, the mutant fails to make any 2-O-sulfated iduronic acid residues due to a deficiency in a 2-O-sulfotransferase (32).
Analysis of steady-state labeled [ 35 S]heparan sulfate chains by gel filtration HPLC showed that wild-type cells contain large, intermediate, and short chains, in agreement with previous studies (6,7). In contrast, the mutant contained only large chains that eluted near the void volume of the TSK 2000 column (Fig. 1). Their elution position on a TSK 4000 column (K av ϳ0.4) indicated an M r of ϳ8 ϫ 10 4 (28). The lack of intermediates in the mutant indicated that their formation somehow depended on 2-O-sulfation of the uronic acid residues. To confirm this hypothesis, mutant cells were transfected with a cDNA clone encoding the CHO 2-O-sulfotransferase (30). Stable transfectants were selected by replica plating using 125 I-bFGF blotting to detect colonies that produced cell surface heparan sulfate (32). Several colonies that bound bFGF strongly were purified, and one was characterized in greater detail. By enzymatic assay, the transfectant contained a normal level of 2-O-sulfotransferase activity (118 Ϯ 10 pmol/ min/mg of cell protein versus 98 Ϯ 10 in the wild-type and 15 Ϯ 5 pmol/min/mg of cell protein in pgsF-17 cells). 2 (Fig. 2). These findings confirmed that the altered processing of the chains in the mutant was due to reduction in 2-O-sulfation of the uronic acid residues. Further studies were therefore performed to determine the cause for the degradation defect (e.g. altered secretion, endocytosis, heparanase cleavage, or lysosomal degradation).
Secretion and Turnover of Heparan Sulfate Proteoglycans Occurs Normally in the Mutant-The turnover of [ 35 S]heparan sulfate was measured by pulse-chase experiments in which cells were labeled with 35 SO 4 for 1 h and chased for different periods of time (Fig. 3). [ 35 S]Heparan sulfate proteoglycans were isolated from the growth medium (secreted or shed material in solution), the cell surface (trypsin-releasable), and intracellular pools that remained with the cell pellet after trypsin treatment of the cells and centrifugation (see "Experimental Procedures"). The heparan sulfate chains found on cell surface proteoglycans were large in size (M r ϳ8 ϫ 10 4 ) (28) and decreased in amount very rapidly during the chase in both wild-type and mutant cells (Fig. 3, A and B). During the chase, chains were recovered in increasing amounts from the growth medium, but the chains remained large in size in both cells.
The intracellular pool behaved very differently in mutant and wild-type cells. At the start of the experiment, most of the label was in large chains, presumably associated with newly made proteoglycans in the Golgi and proteoglycans recently internalized from the cell surface. In addition, wild-type cells contained a small amount of intermediate and small sized chains (Fig. 3E, filled circles). These fragments accumulated during the chase as the large chains decreased. The large chains in the mutant also declined with time, and a small shift in elution position occurred, suggesting slow, incomplete cleavage (Fig. 3F). Some intermediate and small sized chains accumulated as well, but the extent of cleavage was greatly reduced compared with the wild-type. Thus, the mutant produced few degradation intermediates by pulse-chase, in agreement with the steady-state labeling experiments shown in Figs. 1 and 2.
The overall extent of degradation was determined by measuring the disappearance of 35 S-pulse-labeled heparan sulfate chains from the cells and the medium. Degradation occurred with similar kinetics in mutant and wild-type cells (Fig. 4), taking place in two phases. One was rapid and accounted for about one-third of the chains labeled during the pulse, and a 2 The products generated in crude extracts of wild-type cells consisted of ϳ85% 2-O-sulfated iduronic acid residues and 15% GlcNSO 3 or 6-O-sulfated glucosamine residues. The sulfated products made in the mutant were entirely GlcNSO 3 and a small amount of 6-O-sulfate esters (32). second occurred more slowly. The initial rapid phase may reflect turnover of material en route to lysosomes, whereas the slow phase represents degradation of material that slowly enters the endocytic pathway from the cell surface or from the growth medium. Chloroquine inhibited degradation by ϳ70%, consistent with the idea that degradation occurred in a low pH compartment, most likely in lysosomes (results not shown). These findings indicated that overall degradation occurred independently of 2-O-sulfation, whereas formation of intermediate and small size fragments did not.
Heparan Sulfate Proteoglycans and Chains Are Distributed Normally in Subcellular Organelles-The failure to generate intermediate fragments in the mutant could reflect differences in trafficking of the chains through subcellular compartments where the heparanases reside. To test this idea, we examined the internalization of proteoglycans in mutant and wild-type cells. In one method, cell surface proteins were biotinylated and then the cells were metabolically labeled with 35 SO 4 for 24 h to detect both newly made proteoglycans and those that were sulfated after internalization from the cell surface and recycling through the Golgi (29,34). As shown in Table I, both wild-type and mutant cells sulfated a small amount of biotinylated heparan sulfate proteoglycans, suggesting that some transfer had occurred to the Golgi compartment where sulfation takes place. About twice as much material was recovered from the mutant, but the fraction of cell surface heparan sulfate proteoglycans that recycled was quite low in both cell lines, representing ϳ0.2% of cell surface heparan sulfate proteoglycans. 3 Trypsin treatment of the cells prior to biotinylation did not yield any 35 S-labeled biotinylated proteoglycans, indicating that the radiolabeled material did not result from biotinylation of proteoglycans inside the cells. Although recycling is a quantitatively minor process, its occurrence in both mutant and wild-type cells indicated that this part of the endocytic pathway continued in the absence of 2-O-sulfation and in fact may be somewhat accentuated when 2-O-sulfation is depressed.

FIG. 4. Degradation of heparan sulfate.
Wild-type and mutant cells were pulse-labeled for 1 h with 35 SO 4 (100 Ci/ml) and chased for the indicated times. The total amount of [ 35 S]heparan sulfate recovered from cells plus medium was measured (see "Experimental Procedures"). q, wild-type; E, pgsF-17.

Recycling of biotinylated cell surface heparan sulfate proteoglycan
Wild-type and pgsF-17 cells were biotinylated and subsequently labeled for 24 h with 35 SO 4 (100 Ci/ml) (see "Experimental Procedures"). The proteoglycans were isolated from the cell layer by guanidine extraction, and a portion was purified on strepavidin beads (see "Experimental Procedures"). The counts recovered in heparan sulfate proteoglycans bound to the beads are given for biotinylated and control samples. We also measured the distribution of heparan sulfate proteoglycans and chains in different subcellular organelles isolated by density gradient centrifugation in Percoll. Plasma membranes were recovered in fractions 1-3 of the gradients, based on the recovery of 125 I counts from cells that were surface-radioiodinated (Fig. 5B). Lysosomes sediment in the heavier fractions (10 -12) based on the recovery of acid Nacetylhexosaminidase (data not shown). As shown in Fig. 5A, the distribution of [ 35 S]heparan sulfate proteoglycans and chains was identical in wild-type and mutant membrane fractions, with peaks of material in the plasma membrane and lysosomal pools. These results suggested that the uptake and subcellular distribution of heparan sulfate chains in the mutant was not altered significantly.
[ 35 S]Heparan sulfate from each subfraction (Fig. 5A, solid bars) were analyzed by gel filtration HPLC (Fig. 6). The chains from wild-type plasma membrane fractions were mostly large, but some smaller fragments were present possibly due to crosscontamination by lysosomal membranes. 4 The lysosomal pool contained mostly intermediate and small chains, consistent with the idea that these accumulate in lysosomes or in a late endosomal compartment. In contrast to these findings, the subcellular fractions prepared from the mutant did not contain short oligosaccharides, except for a small amount of material in the lysosomal fraction (Fig. 6B). Instead, most of the material migrated as large chains. The peak was somewhat broader in the lysosomal fractions, suggesting that limited cleavage may have occurred.
Mutant pgsF-17 Has Normal Heparanase Activity-Another explanation for the lack of intermediate and small oligosaccharide fragments in the mutant was a secondary mutation affect-ing heparanase activity. To check this possibility, microsomal membranes of wild-type and mutant cells were mixed with large [ 35 S]heparan sulfate chains from wild-type cells (Fig. 7A). After incubation at 37°C for 24 h, the reaction products were analyzed by gel filtration HPLC. As shown in Fig. 7B, the extract prepared from the mutant degraded the chains to oligosaccharide fragments to the same extent as the extract prepared from wild-type cells. Furthermore, partially purified heparanase from the mutant cleaved wild-type chains to the same extent as enzyme prepared from wild-type cells. 5 Thus, the heparanase activity appeared normal in the mutant.
Heparan Sulfate Chains from the Mutant Resist Heparanase Degradation in Vitro-Another possibility was that the heparan sulfate in the mutant was simply a poor substrate for the intracellular heparanases. To test this possibility, [ 35 S]heparan sulfate chains from cell surface proteoglycans of mutant and wild-type cells were compared as substrates for partially purified heparanase from wild-type CHO cells. As shown in Fig. 8, chains from the mutant degraded more slowly than chains from the wild-type, and the extent of degradation was not as great. Furthermore, when the products of the reaction were analyzed by gel filtration HPLC, intermediate and small sized fragments were less prevalent in the mutant (Fig. 9). There was greater heterogeneity in the products, and some chains did not appear to be cleaved at all. These differences suggested that the deficiency in 2-O-sulfation rendered the chains less susceptible to FIG. 5. Separation of intracellular organelles by density gradient centrifugation. Postnuclear supernatants were prepared from wild-type and pgsF-17 cells that had been labeled for 1 h with 35 SO 4 (100 Ci/ml) or by surface radioiodination (see "Experimental Procedures"). Samples were fractionated by Percoll density gradient centrifugation, and the amount of [ 35 S]heparan sulfate or 125 I-labeled surface proteins recovered in each fraction was determined (see "Experimental Procedures"). Fractions representing plasma membranes and lysosomes were pooled as indicated by the bars. A, recovery of 35 S counts; B, recovery of 125 I counts. q, wild-type; E, pgsF-17.

DISCUSSION
Several important conclusions emerge from the study of heparan sulfate turnover in mutant pgsF-17. First, the depression in 2-O-sulfation of uronic acid residues in heparan sulfate decreases the susceptibility of the chains to heparanase cleavage in vitro (Figs. 8 and 9). This finding explains in part the failure of the cells to produce intermediate and small sized fragments in vivo (Figs. 1 and 2). These fragments are generally considered catabolic intermediates in the degradation of heparan sulfate chains, but the data indicate that the overall degradation of heparan sulfate to inorganic sulfate occurs independently of their formation (Fig. 3). Thus, 2-O-sulfation may play a critical role in processing of the chains, but it has little if any effect on overall degradation. This raises several interesting issues.
Altered Susceptibility to Heparanase Cleavage-The reduced cleavage of the chains suggests that 2-O-sulfation of the uronic acid residues plays an important role in substrate recognition by CHO heparanases. Similar enzymes have been detected from several sources, including cultured human skin fibro-blasts (35), human placenta (36) and platelets (14,15), murine mastocytoma (17,18) and melanoma (20), and rat hepatocytes (9) and liver (21). All of the enzymes act as endo-␤-glucuronidases, cleaving the chains on average at 1-3 sites and leaving a GlcUA residue on the reducing end of the fragments. Using heparin biosynthetic intermediates with different patterns of sulfation, Oldberg et al. (14) found that platelet heparanase does not require O-sulfate groups for cleavage. Studies of melanoma and rat hepatoma heparanases using chemically modified forms of heparin as inhibitors indicated that O-sulfate groups were dispensable as long as N-sulfated glucosamine residues were present (23,24). In contrast, Thunberg et al. (18) showed that the heparanase from mastocytoma (which produces heparin) required both N-and O-sulfate groups for cleavage of heparin. Thus, the CHO cell enzyme behaves more like the activity found in mastocytoma, since 2-O-sulfated uronic acid residues enhance the rate and extent of cleavage of the chains.
The cleavage site for a heparanase from any source has not been elucidated in full detail. In general, the "H1 fragment" (i.e. the section of the chain containing the GlcUA residue on the reducing end) contains a high proportion of GlcNAc versus GlcNSO 3 immediately adjacent to the GlcUA residue (6,9,17,18). The preponderance of GlcNAc residues in this position is consistent with biosynthetic studies that indicate that N-deacetylation/N-sulfation of GlcNAc is prerequisite for the conversion of the D-GlcUA toward the reducing side (i.e. at the cleavage site) to L-iduronic acid (37,38). Additional studies using nitrous acid and bacterial heparinase to cleave the chains at N-sulfated glucosamine residues (6,9,14) suggest a preferred structure for the H1 fragment, age site and the first modified residue, suggesting that multiple heparanases may exist with different specificities.
Less information is available about the H2 fragment (i.e. the fragment to the right of the cleavage site). Heparanases from multiple sources generally require N-sulfated glucosamine residues for activity, but the location of this unit relative to the cleavage site is not known with certainty. The requirement for glucosamine N-sulfation was confirmed in vivo by Bame (7), who showed that the heparan sulfate chains from a CHO cell mutant defective in GlcNAc N-deacetylase/N-sulfotransferase were relatively resistant to cleavage. The general requirement for N-sulfated glucosamine residues suggests that heparanase may prefer GlcNSO 3 as the nonreducing terminal residue on the H2 fragment. Supporting data for this idea derives from studies of the platelet heparanase, which will cleave heparin octasaccharides within the disaccharide unit, -GlcUA-GlcNSO 3 (3SO 3 )- (18). Our studies of pgsF-17 suggest that the next residue should be IdoA-2SO 3 since the reduced level of 2-O-sulfation in the mutant diminishes cleavage of the chains. Thus, we propose that the H2 fragment has a nonreducing terminal sequence, Defective 2-O-Sulfation Alters Turnover of Heparan Sulfate in Vivo-The altered susceptibility of heparan sulfate from the mutant to heparanase explains in part the diminution of cleavage fragments in the mutant. However, the lack of intermediate or small sized oligosaccharides is somewhat surprising given that in vitro assays showed that reduced 2-O-sulfation of the chains only led to a decrease in the rate of cleavage and not to complete resistance (Figs. 8 and 9). One way to explain these findings is to consider how proteoglycans traffic through intracellular compartments. Proteoglycans are thought to be endocytosed through endosomes and exposed to proteases, heparanases, and eventually lysosomal glycosidases and sulfatases (5). If the heparanases were located in a prelysosomal compartment, the exposure of the chains to the enzymes might be temporally limited as the chains migrate from this compartment to a lysosome. Thus, if the rate of cleavage is slow (as in the mutant) and the transit time is relatively rapid, little or no cleavage of the chains would occur.
An alternative explanation for the failure to produce few cleavage fragments is that the lack of 2-O-sulfation may change the way that heparan sulfate chains move through intracellular compartments. Bame (7) has suggested that CHO heparanases will act on native proteoglycans, liberating chains from the core protein. If cleavage of the chains from the core protein were reduced, then the proteoglycans might remain intact, which in turn might raise the possibility that the chains recycle to the cell surface as part of a proteoglycan. Recycling of uncleaved proteoglycans could raise the likelihood of their being shed, which would predict that a greater portion of material should be recovered from the growth medium in the mutant. Indeed, we have observed that a larger proportion of heparan sulfate chains are secreted from mutant cells than in the wild type, and the elevated amount of secreted material in the mutant can account for the missing intermediate and small fragments inside the cell. 6 Degradation Proceeds at a Normal Rate-Although interme-diate oligosaccharides are not present in the mutant, cell-associated heparan sulfate degrades at a normal rate and to the same extent as heparan sulfate in wild-type cells (Fig. 3). Lysosomal degradation of heparan sulfate occurs by the concerted action of three exoglycosidases, an acetyl transferase, and several sulfatases working at the nonreducing end of the chain (39). Thus, one would predict that heparanase cleavage should increase the number of nonreducing termini and therefore enhance the overall rate of degradation (9,39). However, this argument presumes that the exolytic cleavage of the chains is slow, which in turn predicts that partial degradation products should occur as the chains degrade. The absence of these catabolic intermediates in the mutant suggests that exolytic degradation proceeds rapidly. One resolution to this apparent paradox is to consider the possibility that lysosomal enzymes associate into complexes, which degrade the chains rapidly in a processive manner. If the number of functional complexes is limiting, then increasing the number of chains by heparanase action would not have any effect on the overall rate of degradation. Obviously, additional studies are needed to test if the lysosomal enzymes exist in complexes. We also need to establish the physical location of the heparanases. What Is the Physiological Role for Intracellular Heparanase?-Our finding that degradation of heparan sulfate proceeds at a normal rate in the absence of heparanase cleavage of the chains calls into question the significance of this enzyme in catabolism. Other roles for the cleavage reaction have been considered based on the secretion of heparanases by tumor cells, where they may participate in remodeling the extracellular matrix (40). However, inside the cell heparanases may perform other roles, such as the production of biologically active oligosaccharide fragments. Intracellular heparanases might liberate ligands that were bound and internalized by way of heparan sulfate proteoglycans, giving rise to complexes of the ligand and a heparan sulfate oligosaccharide. For example, cells internalize bFGF while bound to cell surface heparan sulfate proteoglycans (41)(42)(43), and the formation of complexes could protect bFGF from protease degradation, increasing its availability (44,45). Site-directed mutagenesis of the heparinbinding region of bFGF increases its degradation after endocytosis (46), suggesting that inside cells heparan sulfate and bFGF may transiently exist as a stable complex. Finally, it is interesting to note that the smaller oligosaccharide fragments (4 -7 kDa) consist of 8 -14 disaccharide units, which are more than adequate to interact with most heparin-binding proteins described to date (47,48). Tumova and Bame (28) have recently shown that bFGF can block heparanase cleavage sites and that the small fragments generated in vivo still have binding sites for bFGF. Determination of whether these oligosaccharides have particular oligosaccharide sequences that facilitate selective binding to other protein ligands awaits further study.