Structure and affinity for antithrombin of heparan sulfate chains derived from basement membrane proteoglycans.

Metabolically 35S- or 3H-labeled heparan sulfate was isolated from murine Reichert's membrane, an extraembryonic basement membrane produced by parietal endoderm cells, and from the basement membrane-producing Engelbreth-Holm-Swarm mouse tumor. The polysaccharides were subjected to structural analysis involving identification of products formed on deamination of the polysaccharides with nitrous acid. The polysaccharide from Reichert's membrane contained N- and O-sulfate groups in approximately equal proportions. It bound almost quantitatively and with high affinity to antithrombin. A high proportion of antithrombin-binding sequence was also indicated by the finding that 3-O-sulfated glucosamine residues accounted for about 10% of the total O-sulfate groups. In contrast, at least 80% of the sulfate residues in the heparan sulfate isolated from the mouse tumor were N-substituents. Only a minor proportion of this polysaccharide bound with high affinity to antithrombin, and no 3-O-sulfated glucosamine residues were detected. These results are discussed in relation to the possible functional role of heparan sulfate in basement membranes.


1987)
. Whereas there is thus no apparent qualitative difference between the two types of polysaccharide, the sugar composition and sulfation pattern generally provide a quantitative distinction, heparin containing more sulfate and iduronic acid but less N-acetyl and glucuronic acid than heparan sulfate (Gallagher and Walker, 1985). The blood anticoagulant activity of heparin is due to the presence of a specific pentasaccharide sequence (Lindahl et al., 1984;Atha et al., 1985;see Fig. l), the distinguishing feature of which is a 3-0sulfated glucosamine residue . Polysaccharide chains possessing this sequence bind with high affinity to antithrombin and thus drastically increase the rate by which this proteinase inhibitor inactivates enzymes involved in the coagulation process (Bjork and Lindahl, 1982). Recent findings suggest that the same antithrombin-binding region may occur also in heparan sulfate (Marcum and Rosenberg, 1985;Lane et al., 1986;Marcum et al., 1986).
The proteoglycan forms of the two polysaccharides appear to be more clearly distinct. In the heparin proteoglycan, 10-15 polysaccharide chains are attached to a polypeptide core composed essentially of alternating serine and glycine residues (Robinson et al., 1978). At least 2 out of 3 serine residues carry a polysaccharide substituent, and the macromolecule is thus characteristically resistant to digestion by proteolytic enzymes such as Pronase. Heparan sulfates are generally assembled into more conventional proteoglycan structures, with polypeptide cores of complex composition and relatively sparsely distributed polysaccharide chains. The occurence of immunologically distinct core proteins has been reported ; see also Gallagher et al., 1986).
Basement membranes contain, in addition to other macromolecules, proteoglycans, in particular heparan sulfate proteoglycans. It has been proposed that these proteoglycans control the permeability properties of basement membranes by functioning as a charge barrier (Farquhar, 1981). Structural studies of heparan sulfate proteoglycans from the basement membrane-producing Engelbreth-Holm-Swarm tumor (EHS)' have demonstrated a high-and a low-density form of the proteoglycan (Fujiwara et al., 1984;Hassell et al., 1985).
The high-density form (Mr -130,000) contains a small protein core ( M , -10,000) to which four 30-nm-long heparan sulfate chains ( M , -29,000) are attached (Fujiwara et al., 1984). The low-density proteoglycan ( M , -600,000) carries three 90-nmlong heparan sulfate chains (Mr -43,000) connected to a large core protein ( M , -400,000) (Paulsson et al., 1986). High-and low-density heparan sulfate proteoglycans have also been isolated from Reichert's membrane, an extraembryonic basement membrane produced by parietal endoderm cells in the pregnant rodent uterus . Contrary to EHS material, Reichert's membrane is a product of normal cells, and since it can be cleanly isolated, it is a useful system for the study of basement membrane structure. In analogy with the findings for EHS proteoglycans, gel chromatography indicated a larger molecular size for the low-density proteoglycan from Reichert's membrane than for the high-density proteoglycan . Immunological characterization showed that high-and low-density EHS heparan sulfate proteoglycans shared common epitopes with each other and with the proteoglycans obtained from Reichert's membrane, but not with the heparan sulfate proteoglycan isolated from liver plasma membrane . It thus may be concluded that the heparan sulfate proteoglycans in EHS and in Reichert's membrane represent a unique class of proteoglycans with preferential localization to basement membranes.
In the present study, heparan sulfate proteoglycans isolated from EHS and from Reichert's membrane have been characterized with regard to polysaccharide structure. The heparan sulfate from Reichert's membrane showed a higher O/Nsulfate ratio than did the corresponding tumor material. Furthermore, it bound with high affinity to antithrombin and contained large amounts of the 3-0-sulfated glucosamine residue previously implicated as a unique component of the antithrombin-binding region of heparin.
Sephadex G-25 (superfine grade), Sephadex G-15, and Sepharose CL-4B were purchased from Pharmacia P-L Bio chemicals; bovine liver (3-D-glucuronidase (type BlO), chondroitin ABC lyase, and Pronase were from Sigma; DEAE-cellulose (DE52) was from Whatman; and NaB3H4 (5-15 Ci/mmol) was from New England Nuclear. Antithrombin covalently bound to Sepharose 4B was prepared as described (Hook et al., 1976) Metabolically [35S]sulfate-labeled heparan sulfate proteoglycans were isolated from cultures of mouse Reichert's membrane as described  and from mouse EHS tumor by a highly similar procedure (Timpl et al., 1987). Tumor-bearing mice were injected with 0.8 mCi of carrier-free [35S]sulfate (H~%304; 0.5 mCi/ml, 43 Ci/mg; New England Nuclear) into the tumor tissue and were killed after 24 h. Dissected tumor tissue (about 5 g/mouse) was extracted twice overnight with 6 M guanidine HCI, 0.05 M Tris-HC1 (pH 7.4) containing 10 mM EDTA, 2 mM N-ethylmaleimide, and 2 mM phenylmethanesulfonyl fluoride as protease inhibitors. Purification was done by consecutive DEAE-cellulose chromatography in 7 M urea, gel filtration on Sephacryl S-400 in 6 M guanidine HCI, and finally CsCl density gradient centrifugation in 6 M guanidine HCI using a starting density of 1.50 g/ml. The low-density preparations from both Reichert's membrane and EHS tumor were pure heparan sulfate proteoglycans. The high-density proteoglycan preparations from both sources contained contaminating galactosaminoglycan, which was generally (see "Results") removed by digestion with chondroitinase ABC, followed by reisolation of heparan sulfate proteoglycan by gel filtration on Sephacryl S-400 . On occasion, polysaccharide chains were first released by alkaline elimination, followed by chondroitinase digestion , and heparan sulfate chains were isolated as a void volume peak after chromatography on Sephadex G-75. Purification of [3H]glucosaminelabeled proteoglycans from EHS tumor followed the same procedure using as starting material tumors harvested 24 h after injection of the mice with 0.5 mCi of D-[6-3H]glucosamine (10-30 Ci/mmol; New England Nuclear). Nonradioactive EHS high-density heparan sulfate proteoglycan was purified as described by Fujiwara et al., (1984). 'H-Labeled heparan sulfate proteoglycan, isolated from rat hepatocytes incubated in vitro with D-[3H]glucosamine (Oldberg et al., 1979), was kindly given by E. Unger (Department of Veterinary Medical Chemistry, Swedish University of Agricultural Sciences, The Biomedical Center).
Methods-Digestion of disaccharides with ~-D-g~ucuronidase (Jacobsson et al., 1979a) was performed as described. Polysaccharide chains were released from proteoglycans by treatment with 0.5 M NaOH at 4 "C. After 15 h, the samples were neutralized with 4 M HCl and were then desalted by dialysis. Saccharides were N-deacet-ylated by treatment with hydrazine/hydrazine sulfate at 100 "C for 2 h (Thunberg et at., 1982). The products were desalted by passage through a column (1 X 64 cm) of Sephadex G-15, equilibrated with 10% ethanol.
Degradation of saccharides with nitrous acid at pH 1.5 (deamination of N-sulfated glucosamine residues; 10 min at room temperature) or at pH 3.9 (deamination of N-unsubstituted glucosamine residues; 10 min at room temperature) was carried out as described (Thunberg et al., 1982), with some modification. In these reactions, the glucosamine target residues are converted into 2,5-anhydro-~-mannose units, with cleavage of the corresponding glucosaminidic linkages (see Shively and Conrad, 1976). Reduction of these units to terminal anhydromannitol residues was achieved by adding, per each 200-p1 portion of pH 1.5 reaction mixture, either 75 pl of 1 M Na2C03 containing 13 mg of unlabeled NaBH4/ml or 75 pl of 1 M Na2C03, followed by 5 mCi of NaB3H4 (specific activity, 2.5-7.5 Ci/mmol) in 200 g1 of 0.01 M NaOH. The products obtained after deamination at pH 3.9 (600-pl reaction mixture) were neutralized with 75 pl of 2 M Na2C03 containing 13 mg of unlabeled NaBH4/ml. After reduction of deamination products at room temperature for -15 h, the pH was lowered to -4 with glacial acetic acid (3H-containing samples in a fume hood) and was then adjusted to -pH 7 with 4 M NaOH.
Labeled di-and tetrasaccharides were separated by high-performance ion-exchange chromatography on a Partisil-10 SAX column (Whatman), eluted at a rate of 1 ml/min using step gradients with increasing concentrations of aqueous KH2P04 (Bienkowski and Conrad, 1985). The column was connected to a model Flo-One HS radioactive-flow detector (Radiomatic Instruments & Chemical Co. Inc., Tampa, FL), equipped with a 2.5-ml cell, using Flo-Scint 111 (Radiomatic) as scintillation medium (5 ml/min). The radioactivity was integrated continuously and recorded on a strip chart. The retention times of labeled components were related to those of diand tetrasaccharide standards. The identity of 35S-labeled sample components was ascertained by use of 3H-labeled internal standards, by rerunning samples after digestion with p-glucuronidase, or, in the case of tetrasaccharides, by identifying disaccharides formed by Ndeacetylation, followed by deaminative cleavage (pH 3.9 reaction) of the tetrasaccharide.
Hexuronic acid was detected by the carbazole reaction (Bitter and Muir, 19621, and radioactivity was detectedeither by the flow detector described above or by liquid scintillation spectrometry. Labeled samples separated by paper chromatography or paper electrophoresis were quantified by cutting the paper strips in I-cm pieces which were then soaked with 1 ml of water in the counting vial before the addition of scintillation medium (4 ml of Packard Emulsifier Scint 299).

Characterization of intact Polysaccharide
Macromolecular Properties-Gel chromatography on Sepharose 4B of high-density, 35S-labeled heparan sulfate proteoglycan from Reichert's membrane showed a major peak (KaV 0.34) and, in addition, a shoulder corresponding to smaller components (Fig. 2 A ) . Alkali treatment of this material resulted in a pronounced shift of the elution profile to a position (Kav 0.65), indicating single polysaccharide chains. In contrast, the same material was only marginally affected by digestion with Pronase ( Fig. 2 A ) . Similar digestion of heparan  (0). For Pronase digestion, the samples were mixed with 1 mg of enzyme (Sigma) in 250 p1 of 0.1 M Tris-HCI, 0.002 M CaClz (pH 8.0) and incubated at 50 "C for 18 h. Effluent fractions of 1 ml were collected and analyzed for radioactivity. For additional experimental details, see "Methods." The sample from Reichert's membrane had been digested with chondroitinase ABC prior to analysis; the hepatocyte sample was claimed to be free from chondroitin sulfate. sulfate proteoglycan derived from cultured rat hepatocytes yielded single polysaccharide chains only, as indicated by the nearly identical elution patterns obtained with alkali-treated and with Pronase-digested material (Fig. 2B). These results suggest that the heparan sulfate proteoglycan from Reichert's membrane is relatively resistant to proteolytic degradation. The product obtained on digesting the corresponding EHS proteoglycan with Pronase showed a peak elution position intermediate to those of the intact proteoglycan and the single heparan sulfate chains (not shown).
Ion-exchange chromatography of polysaccharide chains, released from proteoglycans by alkali treatment, yielded essentially symmetrical peaks that emerged somewhat retarded compared to standard chondroitin sulfate, yet clearly before heparin (Fig. 3). The elution position of heparan sulfate isolated from Reichert's membrane (Fig. 3A) did not differ significantly from that of EHS polysaccharide (Fig. 3B). The polysaccharides derived from the low-density proteoglycans (not shown) appeared slightly before those of the high-density proteoglycans (in Fig. 3).
Affinity for Antithrornbin-"35S-Labeled heparan sulfate proteoglycans from Reichert's membrane and from EHS tissue were treated with alkali to release heparan sulfate chains. The labeled chains were mixed with unlabeled heparin and were then subjected to affinity chromatography on a column of antithrombin-Sepharose that was eluted with a linear salt gradient. The heparin preparation was separated into approximately one-third of high-affinity and two-thirds of low-affinity components (Fig. 4), as is commonly seen with commercially available heparin preparations (see Bjork and Lindahl, 1982). In contrast, about 80% of the polysaccharide chains from the high-density heparan sulfate proteoglycan of Reichert's membrane (Fig. 4A) and -70% of the chains from the corresponding low-density proteoglycan (not shown) emerged with 1 mg of standard pig mucosal heparin and fractionated on a column (3 ml) of antithrombin-Sepharose as described (Thunberg et al., 1982). Effluent fractions of 2 ml were collected and analyzed for radioactivity (0) and for hexuronic acid (carbozole reaction; -). The heparin standard is separated into fractions of low and high affinity for antithrombin, as indicated. ---, NaCl concentration.
as retarded as or even later than the high-affinity component of the heparin standard. The heparan sulfate from EHS tissue, on the other hand, showed largely low affinity for the immobilized antithrombin, although -30% of the material appeared at an elution position intermediate to the high-affinity and low-affinity heparin components ( Fig. 4B; chromatogram of polysaccharide from high-density proteoglycan, the elution profile of polysaccharide from the corresponding low-density proteoglycan being essentially similar (not shown). The unretarded peak in Fig. 4B probably represents at least partly galactosaminoglycan since this sample had not been digested with chondroitinase.

Compositional Analysis
Heparan Sulfate from Reichert's Membrane-Biosynthetically 35S-labeled heparan sulfate proteoglycan, digested with chondroitinase ABC, was depolymerized by treatment with nitrous acid (pH 1.5 reaction), and the products were reduced with NaBH,. Chromatography of the deamination products obtained from the high-density proteoglycan on Sephadex G-25 (Fig. 5A) showed a major fraction (-60% of the total label) corresponding to disaccharides and inorganic sulfate, but also appreciable amounts of tetrasaccharides (-30%) and larger oligosaccharides (-10%). The corresponding pattern relating to the low-density proteoglycan (not shown) was essentially similar, with fractions of disaccharides (including inorganic sulfate), tetrasaccharides, and larger oligosaccharides constituting -50, -30, and -20%, respectively, of the total 35S label.  A and B; -500,000 cpm; preparative runs) or [3H]glucosamine-labeled ( C ; -15,000 cpm; analytical run) chondroitinase ABC-digested high-density proteoglycans were treated with nitrous acid at pH 1.5, and the products were reduced with NaBH4 and subjected to gel chromatography (see "Methods"). Effluent fractions of -2 ml were collected, analyzed for radioactivity, and combined into pools corresponding to disaccharides (also including inorganic sulfate) and tetrasaccharides as indicated by the horizontal bars. Of the two peaks in the disaccharide/inorganic sulfate fraction in A , the most retarded one corresponds largely to inorganic sulfate and mono-0-sulfated disaccharides, whereas the less retarded peak is due to di-0-sulfated disaccharides. The inorganic [35S]sulfate accounted for 74% of the total label in this fraction as determined by high-performance ion-exchange chromatography. Elution patterns highly similar to those in A-C were obtained with polysaccharides isolated from the corresponding low-density proteoglycans (not shown). Disaccharides corresponding to the most retarded peak in C were isolated by preparative gel chromatography on Sephadex G-15 (not shown); the pooled fractions were lyophilized before further analysis.

Basement Membra
The label recovered in di-and oligosaccharides represented the total incorporated 0 -[ 3 5 S ]~~l f a t e groups, whereas the released inorganic [35S]~ulfate derived from N-[35S]~~lfate groups in the intact polysaccharide (Shively and Conrad, 1976). Further separation of the most low-molecular-weight fraction by ion-exchange chromatography (see below) allowed an assessment of the amount of inorganic [35S]sulfate relative to total [35S]sulfate incorporated and thus of the N/O-sulfate ratio of the intact polysaccharide. Such analyses showed that N-sulfate constituted -50 and 40% of the total labeled sulfate groups of the high-and low-density heparan sulfate proteoglycans, respectively.
The di-and tetrasaccharide fractions were analyzed further by high-performance ion-exchange chromatography. The labeled HexA-aManR disaccharides thus identified would correspond to various 0-[35S]sulfated (-GlcNSOJ-HexA-Glc-NSO, sequences in the parent polysaccharide; whereas tetrasaccharides of the general structure, HexA-GlcNAc-GlcA-aManR, would represent (-GlcNS03)-HexA-GlcNAc-GlcA-GlcNS0; sequences (Shively and Conrad, 1976;Thunberg et al., 1982;Bienkowski and Conrad, 1985). The larger oligosaccharides should contain 2 or more consecutive N-acetylated disaccharide units, the internal units presumably lacking 0-["S]lsulfate groups (Jacobsson and Lindahl, 1980;Gallagher et al., 1986). It should be noted that tetrasaccharides and larger oligosaccharides may be formed in the deamination reaction not only due to the presence of resistant, N-acetylated glucosamine residues, but also due to an aberrant ring contraction reaction which converts initially N-sulfated glucosamine residues into 3-aldehydo-pentose units, without cleavage of the corresponding glycosidic linkage (Shively and Conrad, 1976;Thunberg et al., 1982;Bienkowski and Conrad, 1985). Finally, it should be emphasized that the analytical procedure employed relates exclusively to incorporated [35S] sulfate groups and thus will not provide any information regarding nonsulfated saccharide sequences.
Previous studies have identified the 3-0-sulfated glucosamine unit as a unique component of the antithrombin-binding region of heparin, present only in heparin molecules with high affinity for the proteinase inhibitor (Bjork and Lindahl, 1982). One of the glucosamine residues (unit 2 in Fig. 1) of the pentasaccharide-binding region may be either N-acetylated or N-sulfated, whereas the two remaining glucosamine residues (units 4 and 6) are invariably N-sulfated (Lindahl et al., 1984). Deaminative cleavage of this sequence converts the unique 3-0-sulfated (or 3,6-di-0-sulfated) glucosamine unit 4 into a terminal 2,5-anhydromannose residue, which is recovered in a disaccharide when unit 2 is N-sulfated but in a tetrasaccharide when unit 2 is N-acetylated. Clearly, the tetrasaccharide variety predominated among the deamination products of heparan sulfate from Reichert's membrane (Table  I); only a minor proportion (-10%) of the antithrombinbinding regions in the high-density proteoglycan yielded the disaccharide and thus would appear to have been exclusively N-sulfated. The total amounts of 3-0-sulfate groups in the heparan sulfates were estimated from the overall yields of 3mono-0-sulfated and 3,6-di-O-sulfated anhydromannose residues, as calculated from the gel chromatographic distribution of the deamination products and the compositional analysis of the resulting fractions. Assuming that no 3-0-sulfate groups occurred in the mono-0-sulfated tet.rasaccharide fraction or in oligosaccharides larger than tetrasaccharides, the 3-0-[3sS] sulfate groups would account for -5% of the total incorporated sulfate groups or -10% of the 0-sulfate groups. These values would essentially apply to both the high-and lowdensity heparan sulfate proteoglycans. Due to the lack of information regarding nonsulfated regions, it is not possible to calculate the number of antithrombin-binding regions per polysaccharide chain. However, assuming a minimum average of approximately one sulfate group/disaccharide unit, which seems reasonable in view of the ion-exchange chromatography properties of the intact polysaccharide chains (Fig. 3A), we estimate that the heparan sulfate from Reichert's membrane contains a 3-0-sulfate group, and hence an antithrombinbinding site, for each 10-20 disaccharide units. The antithrombin-binding capacity of this polysaccharide thus matches or exceeds that of conventional heparin with high affinity for antithrombin.
Heparan Sulfate from EHS-The sulfation pattern of the tumor heparan sulfate differed drastically from that of the Reichert's membrane polysaccharide. Biosynthetically 35Slabeled tumor heparan sulfate, freed from chondroitin sulfate by digestion with chondroitinase ABC, was treated with nitrous acid at pH 1.5; and the products were separated by gel chromatography on Sephadex G-25. Contrary to the heparan sulfate from Reichert's membrane which yielded 0-[3sS]su1fated oligosaccharides and inorganic [35S]~ulfate (derived from N-sulfate groups) in about equal amounts (Fig. 5A), the chromatograms relating to EHS heparan sulfate were dominated by peaks of inorganic [35S]sulfate, with only minimal amounts of labeled dior oligosaccharides (Fig. 5B). Calculations of N / 0 -[ 3 5 S ]~~l f a t e ratios, following more accurate determination of inorganic [35S]~ulfate by high-performance ion-exchange chromatography (not shown) of the retarded gel chromatography fraction, indicated that at least 80% of the total sulfate groups in the high-and low-density tumor heparan sulfate proteoglycans were N-substituents. These Nsulfate groups accounted for -50% of the total N-substituents, as calculated (Jacobsson et al., 1979b) from the size distribution of oligosaccharides obtained by deaminative cleavage of heparan sulfate biosynthetically labeled with [3H]glucosamine (Fig. 5C). Di-and tetrasaccharides each accounted for -25% of the total labeled products, the remainder being hexasaccharides or larger fragments containing 2 or more consecutive N-acetylated disaccharide units, The disaccharides isolated after deamination of the [3H] glucosamine-labeled heparan sulfate were analyzed further by high-voltage paper electrophoresis at pH 5.3 (Fig. 7). About 70% of the disaccharides obtained from either highor lowdensity proteoglycan were nonsulfated, 20-25% were mono-0-sulfated, and only -5% were di-0-sulfated. The amounts of radioactivity incorporated into EHS heparan sulfate by the biosynthetic labeling procedures employed were too low to permit a more detailed characterization of the deamination products. Therefore, an alternative route was exploited, involving deamination of unlabeled polysaccharide at pH 1.5, followed by reduction of the products with NaB3H, (see "Methods"). The labeled disaccharides obtained showed the same separation into nonsulfated, monosulfated, and disulfated species on paper electrophoresis (not shown) as was observed for the [3H]glucosamine-labeled sample. The nonsulfated components were isolated by preparative paper electrophoresis and were then separated further by paper chromatography along with standards (Lindahl et al., 1984) of ['4C]G1~A-aManR and 1d0A-[~H]aMan~ (not shown). Quantification of the separated glucuronic acid-and iduronic acidcontaining disaccharides gave molar ratios of 1.4/1 and 1.6/1 for samples derived from high-and low-density heparan sulfate proteoglycans, respectively. These ratios thus are representative for disaccharide units that lack 0-sulfate substituents but are located within N-sulfated block regions of the polysaccharide chains. The 0-sulfated disaccharides were quantified by high-performance ion-exchange chromtography, which showed GlcA  Fig. 5B). These observations indicate a ratio of glucuronic acid to iduronic acid of -0.3/1 for the 0-sulfated disaccharide units, again with no significant difference between high-and low-density proteoglycans. Finally, similar chromatography of "-labeled tetrasaccharides showed these components to be largely nonsulfated, with trace amounts of mono-, di-, and tri-0-sulfated species (data not shown). Migration distance ( c m l FIG. 7. High-voltage paper electrophoresis of disaccharides isolated from EHS heparan sulfate. High-density [3H]glucosamine-labeled heparan sulfate proteoglycan was degraded with nitrous acid at pH 1.5, followed by reduction of the products with NaBH, and isolation of disaccharides by gel chromatography on Sephadex G-15 (see Fig. 5C). Electrophoresis at pH 5.3 was performed as described under "Methods." The arrows indicate the migration distances of nonsulfated ( I ) , mono-0-sulfated (Z), and di-0-sulfated (3) HexA-[3H]aManR disaccharide standards derived from heparin.

DISCUSSION
The immunochemical distinction between various heparan sulfate proteoglycans reflects the structural diversity of the core proteins. The basement membrane proteoglycans isolated from EHS tissue and from Reichert's membrane share antigenic determinants that are lacking in similar species produced by other types of cells, e.g. hepatocytes . A difference in core protein structure is further indicated by the finding that the basement membrane heparan sulfate proteoglycans partially resisted digestion by Pronase, whereas the hepatocyte proteoglycan was completely degraded to single polysaccharide chains. Resistance toward proteolysis appears to be a characteristic of proteoglycans in which the protein core is densely substituted by polysaccharide chains (Robinson et al., 1978;Bourdon et al., 1985). In fact, the high-density heparan sulfate proteoglycan from Reichert's membrane appeared almost unaffected by Pronase (Fig. 2 A ) and thus mimicked the heparin proteoglycan from mast cells (Robinson et al., 1978).
The similarity between the heparan sulfate proteoglycan from Reichert's membrane and the heparin proteoglycan includes also the presence of the antithrombin-binding sequence in the polysaccharide chains. The occurrence of this highly specific structure in heparan sulfate from certain tissues was recently reported (Marcum and Rosenberg, 1985;Lane et al., 1986;Marcum et al., 1986) and was found to be associated with anticoagulant activity. However, in all of these preparations, the fraction with high affinity for antithrombin amounted to at most a few percent of the total heparan sulfate molecules available. For instance, <1% of the heparan sulfate produced by cloned bovine aortic endothelial cells binds to antithrombin, yet accounts for most of the anticoagulant activity expressed by the unfractionated starting material . Even in commercially available heparin preparations, only about one-third of the molecules bind with high affinity to antithrombin (Bjork and Lindahl, 1982). The heparan sulfate produced by Reichert's membrane is thus exceptional in that practically all of the material has high affinity for antithrombin. Moreover, the 3-0-sulfated glucosamine residue, which is a marker for the antithrombinbinding region, is at least as abundant in this unfractionated heparan sulfate as in the fraction of heparin molecules with high affinity for antithrombin. These findings applied to both the high-and low-density forms of the proteoglycan; in fact, compositional analysis of the released polysaccharide chains merely suggested marginal differences in the N-substituent patterns of the antithrombin-binding regions (indicated by the occurrence of the disaccharide GlcA-aMan~(3,6-di-OSO~) only in the deamination products of the high-density proteoglycan).
The functional role of antithrombin binding is unclear. The The results apply to polysaccharides from both high-and lowdensity proteoglycans.