Occurrence and Biosynthesis of β-Glucuronidic Linkages in Heparin

Abstract Heparin was degraded with nitrous acid to yield carbohydrate-serine compounds with uronic acid located at the terminal, nonreducing position. In addition, a tetrasaccharide with the proposed structure, uronosyl-N-acetylglucosaminyl-uronosyl-anhydromannose was isolated from heparin by a similar procedure. On digestion with liver β-glucuronidase, between one-third and all of the uronic acid located at the terminal position of these fragments was liberated as glucuronic acid. The results indicate that a large portion of the glucuronidic linkages of heparin have the β configuration. Furthermore, one of the β-glucuronidase-treated fragments served as acceptor for glucuronic acid when incubated with UDP-[14C]glucuronic acid and a particulate enzyme preparation derived from a heparin-producing mouse mastocytoma. Previous treatment of the heparin fragment with β-glucuronidase was mandatory for acceptor activity. On digestion of the radioactive product with β-glucuronidase, all of the radioactivity was released as [14C]glucuronic acid.


SUMMARY
Heparin was degraded with nitrous acid to yield carbohydrate-serine compounds with uranic acid located at the terminal, nonreducing position.
In addition, a tetrasaccharide with the proposed structure, uronosyl-N-acetylglucosaminyluronosyl-anhydromannose was isolated from heparln by a similar procedure.
On digestion with liver @-glucuronidase, between one-third and all of the uranic acid located at the terminal position of these fragments was liberated as glucuronic acid. The results indicate that a large portion of the glucuronidic linkages of heparin have the /3 configuration. Furthermore, one of the /3-glucuronidase-treated fragments served as acceptor for glucuronic acid when incubated with UDP-[14C]glucuronic acid and a particulate enzyme preparation derived from a heparin-producing mouse mastocytoma. Previous treatment of the heparin fragment with fi-glucuronidase was mandatory for acceptor activity. On digestion of the radioactive product with &glucuronidase, all of the radioactivity was released as [14C]glucuronic acid.
Heparin, a high molecular weight polymer, is thought to consist of alternating residues of uranic acid and glucosamine, joined by a-l -+ 4 glycosidic linkages (I). The regularity is interrupted at the reducing end of the molecule, which is linked to a polypeptide via a sequence of 3 neutral sugar molecules: 1 The abbreviations used are: UA, uranic acid; GlcUA, glucuronic acid; GlcNAc, %acetamido-2-deoxy-D-glucose, AM, anhydromannose; Fragments B1 and Bg, cf. the schematic structures, Fig. 1; Fraction Bl-p, Fragment B1 after treatment with liver p-glucuronidase and removal of the glucuronic acid monosaccharide formed (cj. Fig. 2~). protein linkage region have been shown to be N-acetylated (2). Also the more peripheral portions of the polysaccharide chain appear to contain a limited number of N-acetylglucosamine residues (3).
Previous work has established that the neutral sugar units as well as the glucuronosyl-galactose linkage are all of the p-anomerit configuration (4). However, the possibility of some glucuronidic linkages other than the GlcUA-Gal unit having the fi configuration has not been discounted. The optical molecular rotation of heparin and derivatives thereof (1) is lower than expected for an exclusive presence of ol-D-glycosidic linkages.
It has recently been suggested (5) that the low value may be due to the presence of L-iduronic acid, which constitutes a considerable portion of the uranic acid in heparin (5,6). It is uncertain, however, whether the iduronic acid content could account quantitatively for the optical rotation data.
In the present study, fragments obtained after degradation of heparin with nitrous acid (2, 7) have been investigated as substrates for liver fl-glucuronidase.
Evidence is presented demonstrating that the enzyme released glucuronic acid from several heparin fragments, in amounts varying between one-third and all of the uranic acid residues present at the terminal, nonreducing position of the molecule.
Furthermore, glucuronosyl transfer has been studied in a cellfree system derived from a heparin-producing mouse mastocytoma, and the formation of glucuronidic linkages susceptible to cleavage by ,&glucuronidase has been demonstrated.
EXPERIMENT.4L PROCEDURE Materials-Heparin was obtained from Wilson Laboratories, Chicago, Illinois, and purified as described (8). Hyaluronic acid was a gift from Professor T. C. Laurent of this institute.
p-Nitrophenyl-a-D-glucoside and p-nitrophenyl-O-D-glucuronide were purchased from Sigma.
3-O-B-D-Glucuronosyl-D-galactose was prepared from chondroitin g-sulfate as described (9). UDP-[14C]glucuronic acid (33 &i per pmole) was obtained from The Radiochemical Centre, Amersham, England. Liver P-glucuronidase (type R-l; 220 Fishman units per mg) and limpet fl-glucuronidase (type L-l ; 500 Fishman units per mg) were products of Sigma. Testicular hyaluronidase (14,000 units per mg) was obtained from Leo, Halsingborg, Sweden. A mast cell tumor originally described by Furth,Hagen,and Hirsch (10) was maintained as a solid tumor in (A/Sri X Leaden) FL mice by subcutaneous and intramuscular transplantation in the hind legs every 10 to 14 days.
Analytical Methods-Uranic acid was determined as described by Bitter and Muir (11). Hexosamine was estimated after hy- to that described (16). A sample (0.45 g) of p-nitrophenyl-ac-nglucoside was mixed with a suspension of freshly reduced plati-the structure indicated, are given in Table I. The yield of the num (10%)activated carbon catalyst (0.2 g) in 25 ml of water. pure compound was 0.5 mg per g of heparin. The reaction mixture was kept at 90", with an air flow of approxi-Isolation of Pentasacchariok from Hyaluronic Acid-Hyaluronic mately 180 liters per hour.
The pH was maintained at about 9 acid (0.3 g) was dissolved in 200 ml of 0.1 M acetate buffer, pH by intermittent addition of 5 ml of 0.5 M NaHC03. After 3 5.0, containing 0.15 M NaCl and digested at 37" with 2 mg of teshours, the reaction mixture was filtered, acidified to pH 3.0, and ticular hyaluronidase for a period of 24 hours. After being kept passed through a column (2 x 5 cm) of Dowex 50-X2 (Hf form, in a boiling water bath for 10 min, the digestion mixture was con-200 to 400 mesh). Upon concentration in the rotary evaporator, centrated, clarified by centrifugation and applied to a column the product crystallized.
The product reacted properly in the carbazole reac-effluent volume of 405 to 430 ml, was pooled, concentrated, and tion for uranic acids and migrated at an RGicUA-value of 0.74 on desalted on a column (4.5 x 25 cm) of Sephadex G-15. Further paper electrophoresis in Buffer C. purification was achieved by preparative paper chromatography Isolation of Heparin Fragments after Treatment with Nitrous for 96 hours in Solvent B on washed Whatman No. 3MM papers. Acid-Degradation of heparin with nitrous acid and the isolation The main spot had an RGicnA-oai-value of 0.30 in this solvent of Fractions B1 and B2 has been described (2). The tetrasaccha-and was homogeneous when analyzed by paper electrophoresis ride, UA-GlcNAc-UA-AM, was isolated by a similar procedure, (Buffer C; RGleU* = 0.85). As expected for a hexasaccharide, involving treatment of heparin with nitrous acid and gel chroma-one-third of the glucuronic acid was present at the nonreducing tography.
asaccharide with liver P-glucuronidase as described below. The effluent was concentrated with additions of methanol, and ly-Digestion with p-Glucuronidase--Unless indicated otherwise ophilized.
The dried material was suspended in methanolic HCl samples were dissolved in 0.05 M acetate buffer, pH 5.0, to give a and desulfated according to Kantor and Schubert (17). After 0.1 y. solution of oligosaccharide and digested with liver p-glucuremoval of methanol on a rotary evaporator, the solution was ad-ronidase (type B-l; 10 mg of enzyme per ml of digestion mixture). justed to pH 4.5 by addition of Dowex 3 (OH-form, 20 to 50 After 12 hours of incubation at 37" a second portion of the enzyme mesh) and lyophilized.
The desulfation procedure was repeated was added and the digestion was continued for another 12 to 60 once. To check the completeness of the reaction, the product hours. Limpet P-glucuronidase was incubated together with was passed through a column (2 x 10 cm) of Dowex l-X2 (Cl-substrate in 0.05 M acetate buffer, pH 3.6, in a similar manner. form, 200 to 400 mesh). Approximately 80% of the carbazole-The digests were analyzed by gel chromatography on a column positive material appeared with the water wash, indicating that (1 x 95 cm) of Sephadex G-25, which was eluted with 1 br KCl. desulfation had occurred, with concomitant esterification of the Portions of the digests were also subjected to paper electrophorecarboxyl groups on the uranic acid residues. The material was sis (Buffer C), or paper chromatography (Solvent A  . After centrifugation at 10,000 x g for 10 min, the supernatant fluid was further centrifuged at 100,000 X g for 60 min. The resulting pellet, suspended in 15 ml of the buffer indicated above, was used as glucuronosyltransferase preparation. Enzyme assays were carried out by incubating UDP-['4C]glucuronic acid with appropriate acceptors, as described in the legend to Table III. The reaction mixtures were spotted on Whatman No. 3MM paper and subjected to electrophoresis in Buffer C (80 volts per cm, 75 min). The distribution of radioactivity was determined with a strip scanner. Preparation of Particulate Enzyme from Chick Embryo Cartilage -A particulate glucuronosyltransferase preparation was obtained from epiphyseal cartilage of la-day-old chick embryos essentially as described (18). The 100,000 x g pellet was suspended in Tris-acetate buffer with the composition given above and incubated with UDP-[14C]glucuronic acid and exogenous acceptors as outlined above.

RESULTS
Spec$city of Liver and Limpet Glucuronidases-In order to confirm the specificities of the enzyme preparations employed, p-nitrophenyl LY-and P-D-glucuronides were incubated with enzyme from bovine liver and from limpet. Table II shows that the liver /3-glucuronidase did not hydrolyze p-nitrophenyl-cY-n-glucuronide whereas the corresponding @-n-glucuronide was cleaved quantitatively under similar conditions.
With the limpet enzyme, complete hydrolysis of both glucuronides was obtained (Table  II).2 2 Similar results were obtained with methyl 01-and B-D-glUcuronides, respectively.
These were synthesized from the corresponding methyl glucosides by a procedure similar to that described under "Materials and Methods" for p-nitrophenyl-or-nglucuronide with the exception that the temperature was kept at 50°.
Although the methyl-a!-n-glucuronide was completely resistant to cleavage by liver &glucuronidase, only 18% of the methyl-/3-n-glucuronide was hydrolyzed under the same conditions, rendering these substrates less suitable for specificity studies. 2. Digestion of heparin fragments with liver P-glucuronidase. Samples containing 0.2 to 3.0 mg of material were incubated with b-glucuronidase in 0.05 M acetate buffer, pH 5.0. After 36 hours the tubes were heated in a boiling water bath and the precipitated protein was removed by centrifugation.
The supernatant fluid was applied to a column (1 X 95 cm) of Sephadex G-25 which was eluted with 1 M KC1 (3 ml per hour).
These results are in agreement with earlier reports that mammalian glucuronidases do not hydrolyze oc-glucuronidic linkages. In contrast, the limpet glucuronidase has been stated to cleave glucuronides of both anomeric configurations (for a review, see Reference 19).
Digestion of Fragments B1 and BB from Heparin with Liver P-Glucuronidase- Fig.  2a shows the pattern obtained on gel chromatography of Fraction B1 after treatment with ,&glucuronidase. Of the total uranic acid in the digest, 15'$$ emerged retarded on the Sephadex column and at the position expected for glucuronic acid. This value corresponds to 45yo of the uranic acid units in nonreducing terminal position of Fraction Bi (cf. the schematic structure, Fig. 1). The presence of glucuronic acid in the digest was confirmed by paper chromatography (Solvent A) and paper electrophoresis (Buffer C). On redigestion of Peak Bl-/3 ( Fig. 2a) with liver or limpet /I-glucuronidase, no further material was released. This result suggests that the remaining uranic acid units located at the terminal position of this fraction may be iduronic Issue of September 10, 1971 T. Helting and U. Lindahl acid since the latter has been shown previously to be an integral part of this fragment (2). Fragment Bz, which contains no iduronic acid ( Fig. 1) was subjected to a similar treatment and, as is seen from Fig. 2b, half of the uranic acid was released by the enzyme, corresponding to quantitative release of the glucuronic acid at the terminal position. The molar ratio, uranic acid to hexosamine, for Fraction Bs-/3 (Fig. 2b) was 1.5 as compared with a ratio of 2.8 for Compound Bz, again indicating that almost half of the glucuronic acid present in Compound Bz had been removed by /I-glucuronidase. 3 Treatment of Tetrasaccharide, UA-GlcNAc-UA-AM, with /3-Glucuronidase-To determine whether ,%glucuronidic linkages were a unique feature of the glucuronic acid residues in the immediate vicinity of the linkage region, the tetrasaccharide, UA-GlcNAc-UA-AM, was also treated with P-glucuronidase. This fraction represents a fragment from that portion of the polysaccharide chain which is distal to the first N-sulfated glucosamine moiety (7). Treatment of this material with liver P-glucuronidase released 14y0 of the total uranic acid (Fig. 2c) corresponding to 28% of the uranic acid located at the terminal, nonreducing position of the molecule.
Paper chromatography (Solvent A) of a portion of the digest showed the presence of glucuronic acid as well as material migrating at the rate of the undigested tetrasaccharide.
In addition, a third, somewhat elongated spot with intermediate mobility was observed, presumably representing the trisaccharide product of the digestion. Insuflicient material precluded a more thorough, structural study of these fragments.

Transfer of Glucuronic Acid from UDP-glucuronic
Acid to Fraction &-&Incubation of the particulate enzyme from mouse mastocytoma with UDP-[%]glucuronic acid and Fraction B1-@ (Fig  2a) resulted in the formation of a product (1013 cpm; Table III) which migrated similarly to the undigested Fragment B1 on paper electrophoresis in Buffers C (Fig. 3) and D. This radioactive peak was absent in control experiments without added acceptor. Furthermore, on gel chromatography (Sephadex G-50 column (2 x 180 cm)) the radioactive product and Fragment B1 emerged with the same effluent volume (Fig. 4). Preliminary treatment of Fragment B1 with @-glucuronidase appeared to be mandatory for acceptor activity; the undigested fragment was essentially inactive as substrate (Table III).

SpeciJicity of Glucuronosyltransferase
Reaction-In order to investigate the specificity of the reaction, a pentasaccharide from hyaluronic acid, GlcNAc-GlcUA-GlcNAc-GlcUA-GlcNAc, was also incubated with the mastocytoma preparation and UDP-[14C]glucuronic acid (Table III).
However, no evidence for hexasaccharide synthesis was obtained.
It should be noted that whereas glucosamine is believed to occur in heparin linked via a-l -+ 4 bonds, the glucosaminidic linkage in hyaluronic acid is p-1 -+ 4 (1).
Furthermore, in view of recent evidence concerning limited synthesis of chondroitin sulfate in the mouse mastocytoma prep-aratiom4 it was important to establish whether transfer of glucuronic acid to Fraction Bl-/3 was due to the presence of a glucuronosyltransferase involved in the polymerization of chondroitin sulfate chains. For this reason, Fraction B,-/l was incubated with a particulate enzyme preparation previously utilized in studies on the biosynthesis of chondroitin sulfate (l&20).
However, no significant transfer of glucuronic acid from UDP-glucuronic acid to Fraction B& was observed under conditions where transfer of glucuronic acid to the endogenous acceptor present in the cartilage homogenate occurred (Table III).
These results suggest that the mastocytoma preparation contained a specific enzyme, presumably involved in the biosynthesis of heparin and capable of transferring [14Clglucuronic acid from UDP-[14C]glucuronic acid to Fraction B& to form Fraction B1. Digestion of [14C]Glucuronic Acid-labeled Fraction B1 with p-Glucuronidase-Treatment of the radioactive product obtained as &Glucuronidic Linkages in Heparin Vol. 246,No. 17 scribed above, with liver /3-glucuronidase, resulted in quantitative release of the labeled glucuronic acid, as evidenced by paper electrophoresis (Buffer C; Fig. 5) and paper chromatography (Solvent A).
Similarly, the product was unstable in the mastocytoma preparation utilized for its synthesis, indicating the presence of a /3-glucuronidase in the particulate enzyme. This finding was substantiated by control experiments with p-nitrophenyl-p-D,glucuronide which was hydrolyzed by a 10,000 x g supernatant solution derived from mouse mastocytoma.
Under the conditions tested, no hydrolysis of the corresponding a-glucuronide was observed.  Fig. 3 with liver @-glucuronidase.
The radioactive product (330 cpm) was dissolved in 0.2 ml of buffer, pH 5.0, and digested with 1 mg of enzyme for 24 hours.
The reaction was stopped and the digest was analyzed by paper electrophoresis in Buffer C. The guide strip inserted below the tracing shows: (I) glucuronic acid; (II) Fraction B1. DISCUSSION Assignments of configuration to the glycosidic linkages in heparin have rested heavily on molecular rotation data (1). Such data obtained on disaccharides isolated after acid hydrolysis of carboxyl-reduced heparin, seem to establish the presence of a-1 --f 4 glucosaminidic and glucuronidic bonds (21). However, it is clear that the fragments obtained by acid hydrolysis of heparin are not representative for the entire parent molecule. For instance, the hydrolytic conditions initially used by Wolfram, Vercelotti, and Horton (21) were apparently too drastic to permit the isolation of derivatives of L-iduronic acid, which constitutes a substantial portion of the uranic acid present in heparin (5,6,22).
Studies on the structure of heparin have been greatly facilitated by the introduction of novel degradation methods, particularly the deaminative cleavage with nitrous acid. Treatment of heparin with nitrous acid affords conversion of N-sulfated but not of N-acetylated, hexosamine residues to anhydromannose units (7). The simultaneous cleavage of the corresponding glucosaminidic bond results in the formation, in high yield, of fragments containiug uranic acid at the terminal, nonreducing positions.
In the present study such fragments were investigated as substrates for liver P-glucuronidase.
On treatment with this enzyme Fraction B1 and the tetrasaccharide, UA-GlcNAc-UA-AM, were partially degraded and released glucuronic acid (Fig. 2, a and c). The structure of those fragments which were resistant to digestion with ,&glucuronidase is unknown, and the presence of cr-glucuronidic linkages in these fractions cannot be excluded.
It seems probable, however, that L-iduronic acid units were located at the nonreducing terminal position of these fragments, since no further glucuronic acid was released by treatment of Fraction Bl-P with limpet glucuronidase, which was shown to contain cr-glucuronidase activity (Table II).
Furthermore, digestion with liver P-glucuronidase of Fragment &, which contains no iduronic acid (2), resulted in essentially complete hydrolysis of the glucuronidic linkages at the nonreducing end of the molecule. Further proof for the occurrence of P-glucuronidic linkages in heparin was obtltined in studies 011 the biosynthesis of this polysaccharide in a cell-free system from mouse mastocytoma. ,4 particulate fraction from this tumor, which produces considerable amounts of heparin (23) was used to catalyze the transfer of glucuronic acid to well-defined heparin fragments, with the aim of investigating the configuration of the glucuronidic linkage formed.
Apparently all of the glucuronic acid which was transferred from UDP-glucuronic acid to Fraction B1-/3 had the p-anomerit configuration, and the synthesis of an cr-glucuronidic linkage in heparin thus remains to be demonstrated.
The linkage of glucuronic acid to the nonreducing terminal galactose moiety of the neutral trisaccharide in the heparin-protein linkage region (4)  Issue of September 10, 1971 T. He&g and U. Lindahl 5447 ases involved in the formation of the more peripheral portions of the polysaccharide chain (cf. Reference 20).
The results of the present investigation are pertinent to the finding of Perlin et al. (24) that L-iduronic acid residues of heparin have the Lu-anomeric configuration. Although the biosynthesis of the iduronic acid moieties in glycosaminoglycans such as heparin has not been studied with cell-free systems, it is generally believed that UDP-cr-D-glucuronic acid is the precursor of the D-glucuronic acid as well as the L-iduronic acid residues in these polysaccharides (25). Assuming that the mechanisms of incorporation into heparin of glucuronic acid and iduronic acid differ merely by an epimerization process at C-5, the presence in this polymer of a-L-iduronic acid residues would require the Dglucuronic acid units to have the p-anomeric configuration (see Fig. 6).