Location of the Antithrombin-binding Sequence in the ... · Location of the Antithrombin-binding Sequence in the ... Casu et al., 1981; Choay et al., 1981 ... 298 The Antithrombin-binding

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 264, No. 1, Issue of January 5, pp. 296304,1989 Printed in U.S.A.

Location of the Antithrombin-binding Sequence in the Heparin Chain* (Received for publication, August 9, 1988)

Lars-Goran Oscarsson, Gunnar Pejler, and Ulf Lindahl From the Department of Veterinary Medical Chemistry, The Swedish University of Agricultural Sciences, The Biomedical Center, S-751 23 Uppsala, Sweden

The antithrombin-binding region of heparin is a pentasaccharide sequence with the predominant structure -GlcNAc(6-OSOs)-GlcA-GlcNSO~(3,6-di-OS03)- IdoA(2-OSOs)-GlcNSOs(6-OS03)-. By using the 3-0- sulfated glucosamine residue as a marker for the antithrombin-binding sequence, the location of this sequence within the heparin chain was investigated. Heparin with high affinity for antithrombin (HA-heparin) contains few N-acetyl groups located outside the antithrombin-binding region, and cleavage at such groups was therefore expected to be essentially restricted to this region. HA-heparin was cleaved at N- acetylated glucosamine units by partial deacetylation followed by treatment with nitrous acid at pH 3.9, and the resulting fragments with low affinity for antithrombin (LA-fragments) were recovered after affinity chromatography on immobilized antithrombin. The LA-fragments were further divided into subfractions of different molecular size by gel chromatography and were then analyzed with regard to the occurrence of the nonreducing terminal GlcA-GlcNSOs(3,6-di-OS- Os)- sequence. Such units were present in small, inter- mediate-sized as well as large fragments, suggesting that the antithrombin-binding regions were randomly distributed along the heparin chains. In another set of experiments, HA-heparin was subjected to limited, random depolymerization by nitrous acid (pH 1.5), and the resulting reducing terminal anhydromannose residues were labeled by treatment with NaBSH4. The molecular weight distributions of such labeled LA- fragments, determined by gel chromatography, again conformed to a random distribution of the antithrombin-binding sequence within the heparin chains. These results are in apparent disagreement with previous reports (Radoff, S., and Danishefsky, I. (1984) J. Biol. Chem. 259, 166-172; Rosenfeld, L., and Danishefsky, I. (1988) J. Biol. Chem. 263,262-266) which suggest that the antithrombin-binding region is preferentially located at the nonreducing terminus of the heparin molecule.

Heparin is a sulfated glycosaminoglycan that is synthesized by connective tissue-type mast cells and is well known for its ability to prevent the coagulation of blood (Bjork and Lindahl, 1982). Interaction between this polysaccharide and anti-

* This work was supported by Grant 13X-2309 from the Swedish Medical Research Council; the Faculty of Veterinary Medicine, Swed- ish University of Agricultural Sciences; the Swedish National Board for Technical Development; KabiVitrum AB, Stockholm; and Car- meda AB, Stockholm, Sweden. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

thrombin (AT)’ results in a dramatic acceleration of the rate by which this protease inhibitor binds to and thereby inacti- vates the enzymes of the coagulation process. The AT-binding region in heparin has been identified as a pentasaccharide sequence with the predominant structure, -GlcNAc(G-OS03)- GlcA-GlcNS0~(3,6-di-OS03)-IdoA(2-OS0~)-GlcNSO~(6- oso3)- (see Fig. 2, upper sequence, units 2-6), the 3-0-sulfate group on the internal glucosamine unit constituting a marker component that is either absent or very rare elsewhere in the molecule (Lindahl et al., 1980, 1984; Casu et al., 1981; Choay et al., 1981; Thunberg et al., 1982a). Only about one-third of the molecules in commercial heparin preparations contain this sequence and have high affinity for AT (HA-heparin); the majority of the chains lack the 3-0-sulfate group and thus show low affinity for the inhibitor (LA-heparin) and only weak anticoagulant activity (see Bjork and Lindahl, 1982).

While structure/function relationships have been defined in some detail for the actual AT-binding pentasaccharide sequence (Riesenfeld et al., 1981; Thunberg et al., 1982a; Lindahl et al., 1983; Atha et al., 1984, 1985, 1987; Petitou et al., 1988), less attention has been given to the relation between this sequence and other portions of the HA-heparin molecule. However, Radoff and Danishefsky (1984) and Rosenfeld and Danishefsky (1988) have proposed that the AT-binding sequence is preferentially located toward the nonreducing end of the molecule or, more explicitly, “within 20% of the heparin chain length from the nonreducing terminus.” Their conclusions were based on studies of saccharide fragments generated from heparin chains that were covalently linked at their reducing termini to derivatized Sepharose beads. The matrix- linked heparin was subjected to limited degradation with nitrous acid and, after the released fragments had been removed, the sections contiguous with the original reducing ends were liberated from the gel by a specific cleavage process. After fractionation of these products by gel chromatography, anticoagulant activity was found to be preferentially associ- ated with the larger fragments.

In the present study we have addressed the same problem in a different manner. HA-heparin was subjected to selective or random cleavage and the resulting fragments were fractionated on AT-Sepharose. The LA-fractions containing either the reducing or nonreducing termini of the parent HA- chains were characterized with regard to molecular size. The results, in disagreement with the conclusions of Danishefsky

The abbreviations used are: AT, antithrombin; HA- and LA- saccharides, heparin-related molecules with high and low affinity for AT, respectively; HexA, unspecified hexuronic acid; GlcA, D-glUCU- ronic acid; IdoA, L-iduronic acid; GlcNAc, 2-deoxy-2-acetamido-~- glucose (N-acetyl-D-glucosamine); aManR, 2,5-anhydro-D-mannitol formed by reduction of terminal 2,5-anhydromannose residues with NaBH4; -NS03, N-sulfate group; -OSOS, 0-sulfate, ester sulfate group (the locations of 0-sulfate groups are indicated in parentheses); HPLC, high performance liquid chromatography.

296

The Antithrombin-binding Sequence in the Heparin Chain 297

and co-workers (1984, 1988) indicated that the AT-binding sequence is randomly located in the heparin chain.

EXPERIMENTAL PROCEDURES

Materials-Pig mucosal heparin was obtained from Inolex Phar- maceutical Division, Park Forest South, IL, and purified by repeated precipitation with cetylpyridinium chloride from 1.2 M NaCl as described (Lindahl et al., 1965). This procedure removed heparin chains of low Mr. The product was fractionated into HA- and LA-fractions on AT-Sepharose as described (Thunberg et al., 1982a). The HA- heparin was N-acetylated (Hook et al., 1982) to block any unsubstituted amino groups present. Gel chromatography of this material on Sephadex G-100 showed a fairly symmetrical peak centered around K., 0.25 (Fig. 4A); this value corresponds to an approximate M. of 15,000 (see Table 1 in Laurent et al., 1978). The number of N- acetylated glucosamine residues per HA-heparin chain, based on this M, value, was -1.5 as indicated by compositional analysis (disaccharide/tetrasaccharide ratio after HNOZ (pH 1.5) treatment) and 'H NMR spectroscopy.' AT covalently attached to Sepharose 4B was prepared as described earlier (Hook et al., 1976). Sephadex G-15, G- 25, G-50, G-100, and Sepharose 4B were purchased from Pharmacia, bovine liver @-glucuronidase (type B-10) from Sigma, and NaB3H4 (5-15 Ci/mmol) from Du Pont-New England Nuclear (Dreieich, West Germany). 3H-Labeled HexA-aManR standard disaccharides with 0- sulfate groups at various positions were prepared from heparin as described (Thunberg et al., 1982a). Unlabeled, even-numbered heparin oligosaccharides (degrees of polymerization 2-24) were isolated by partial deaminative cleavage of the polysaccharide with nitrous acid, followed by gel chromatography on Sephadex G-50 as described (Pejler et al., 1988).

Methods-Digestion of saccharides with @-glucuronidase (Jacobs- son et al., 1979), degradation of saccharides by nitrous acid at pH 1.5 or at pH 3.9, with or without subsequent reduction of the products with NaB3H4 (Pejler et al., 1987a; see also legends to Figs. 1, 3, and 6) and N-deacetylation of N-acetylglucosamine residues by hydrazinolysis (Thunberg et al., 1982a; see legend to Fig. 3) were all performed according to the published procedures.

Gel chromatography was performed on a column of Sephadex G- 100 (1 X 145 cm), eluted with 1 M NaCl at -2 ml/h. Affinity chromatography on AT-Sepharose was done as described previously (Thunberg et al., 1982a; see also legends to Figs. 3 and 6). Immu- noaffinity chromatography of heparin fragments was performed as described (Pejler et al., 1988). Briefly, radiolabeled HA- and LA- fragments were applied to a column (1 X 3 cm) of monoclonal antibodies (539/4H6) covalently attached to Affi-Gel 10. Unbound material was removed by washing with Tris-HC1 (pH 7.5) containing 0.14 M NaC1, and the column was then eluted with the same Tris buffer containing 3 M NaCl. Hexuronic acid was determined by the carbazole reaction (Bitter and Muir, 1962).

Labeled di- and monosaccharides were separated by anion exchange HPLC on a Partisil-10 SAX column (Whatman), eluted with aqueous KHZPO4 of stepwise increasing concentrations at a rate of 1 ml/min (Bienkowski and Conrad, 1985). The column was connected to a radioactive-flow detector (Flo-One HS; Radiomatic Instruments) equipped with a 2.5-ml cell, using Flo-Scint I11 (Radiomatic) as scintillation medium. 3H-Labeled compounds were identified by com- parison of their retention times with those of 3H-labeled reference standards (Pejler et al., 1987a).

RESULTS

The strategy adopted to define the location of the AT- binding site in the HA-heparin chain is outlined schematically in Fig. 1. Two different approaches were employed. One procedure (Fig. lA) was initiated by limited deacetylation of N-acetylglucosamine residues and followed by cleavage of the polysaccharide chain at the resulting N-unsubstituted unit by treatment with nitrous acid at pH 3.9 (Shively and Conrad, 1976). The process was designed to induce no more than one cleavage site per polysaccharide chain, thus involving only a fraction of the total initial HA-molecules. Products formed by cleavage of the AT-binding site were recovered as LA-

' M. Kusche, U. Lindahl, G. Torri, and B. Casu, unpublished observations.

I HydrazinolyslslHNO, (pH 3.9)

LSO' I I 3.0S0, WAC)

Fragment NRT I I Fragmenl RT

(6)

HAC 1 -osq

NAc 3-0S0,

I HNO, (pH1.5)lNaB H, 3

I I- Fragment HA'

- Fragmmt LA -

Fragmenl LA' I N A c 1 .0sq

1 Fragment HA

FIG. 1. Scheme of strategies employed to define the location of the AT-binding sequence in HA-heparin chains. A, cleavage of HA-heparin at the AT-binding sequence; B, random cleavage of HA-heparin. The symbols used are: 0, glucosamine unit (N-sulfated unless specifically indicated to be N-acetylated); 0, hexuronic acid unit; c , a , original reducing terminus before and after reduction with unlabeled NaBH4, respectively; M, .*, 2,5-anhydromannose unit before and after reduction with NaB3H4, respectively. Cleavage sites in the polysaccharide chains are indicated by solid arrowheads; second- ary cleavage sites that may influence the interpretation of the results are indicated by open arrowheads. The boxed pentasaccharide sequence represents the AT-binding region. For additional information see the text.

material after affinity chromatography on AT-Sepharose and were separated further by gel chromatography. LA-fragments containing the original reducing terminus of the parent HA- heparin chains (denoted Fragment RT in Fig. 1) were identified by the occurrence of a nonreducing terminal GlcA- GlcNS0~(3,6-di-OS03)-disaccharide sequence, generated by cleavage of the -GlcNAc(6-0SO3)-GlcA- linkage in the AT- binding sequence (Fig. 2). The size of these LA-fragments would define the distance between the AT-binding site and the reducing terminus of the original chain.

An alternative approach (Fig. 1B) involved random cleavage of the polysaccharide chain, by treatment with nitrous

298 The Antithrombin-binding Sequence in the Heparin Chain 1 2 3 4 5 6

Fwm*n( NRT I F r 8 g m e n t RT

FIG. 2. Scheme of analysis involv- I HMO, (pH 1.5)iN#BaY, ing cleavage of HA-heparin in the AT-binding sequence. For additional information see the text and the legend to Fig. 3.

bn

acid at pH 1.5, which selectively attacks N-sulfated glucosamine residues (Shively and Conrad, 1976). Again the process was restricted to involve only a fraction of the total HA- heparin chains. The starting material was reduced with unlabeled NaBH4 before the cleavage reaction, and treatment of the degradation products with NaB3H4 thus afforded labeling exclusively of the newly formed reducing end groups. The labeled fragments were fractionated on AT-Sepharose, and the resulting HA- and LA-fractions were analyzed by gel chromatography. The location of the AT-binding region was assessed primarily on the basis of the size distribution of labeled LA-fragments.

Cleavage of HA-Heparin at the AT-binding Sequence-Pre- vious studies have indicated that the glucosamine unit 2 of the AT-binding sequence (see Fig. 2) in pig mucosal heparin is largely (-70%) N-acetylated, the remainder being N-sulfated (Lindahl et al., 1984). Since HA-heparin contains on the average 1.5 N-acetyl groups/chain (see “Materials”), approximately 50% of the total N-acetyl groups would be located outside the AT-binding region.” The location of these addi-

same HA-heparin preparation, based on HNO, (pH 1.5)/NaB3Hl This estimate is in agreement with compositional analysis of the

treatment (M. Kusche and U. Lindahl, unpublished data), and does not account for the N-acetylglucosamine unit(s) present in the polysaccharide-protein linkage region (see Lindahl, 1966). Of the total H~xA-G~cNAc-G~cA-[~H]~M~~R tetrasaccharide formed, approximately 50% contained a terminal 3-0-sulfated or 3,6-di-O-sulfated, labeled anhydromannose unit. In the remaining tetrasaccharides this unit was nonsulfated or 6-0-sulfated to about the same extent. N- Deacetylation/deamination of the latter type of parent sequence thus would generate fragments containing a nonreducing terminal GlcA- GlcNS03(6-OSOa)- structure. It is noted that the glucuronic acid unit of GlcNAc(1-4)GlcA- sequences is not recognized as a substrate by the hexuronosyl C5-epimerase that converts D-glucuronic acid into L-iduronic acid units during heparin biosynthesis (Jacobsson et al., 1984), hence the D-ghco configuration for the internal hexuronic acid unit of the tetradaccharide structure above.

HNO,(pH 1.5YNaBaY,

tional N-acetyl groups is not known, and it can not be ex- cluded that some may occur between the AT-binding region and the reducing terminus of the HA-heparin chain. Since simultaneous N-deacetylation at both the AT-binding region (filled arrowhead in Fig. LA) and a second site closer to the reducing terminus (open arrowhead) would invalidate the strategy outlined in Fig. 1, it was decided to restrict the extent of deacetylation so that only a fraction of the chains were affected. The time of hydrazinolysis was therefore limited to 30 min, after which the polysaccharide was recovered and specifically cleaved at the sites of the N-deacetylated glucosamine residues by treatment with HNOz at pH 3.9. Chro- matography of the products on AT-Sepharose showed that -40% of the material had lost the affinity for AT, whereas -60% remained high affinity (Fig. 3).

Gel chromatography on Sephadex G-100 of the HA-fraction isolated after the deacetylation-deamination procedure gave an elution profile indistinguishable from that of the starting material (Fig. 4A). This finding underlines the specificity of the deamination reaction employed, which apparently did not involve N-sulfated glucosamine units to any significant extent. Furthermore, these residual HA-components had apparently escaped significant N-deacetylation, not only in the AT- binding region but altogether, since deacetylation outside the binding region would have resulted in the generation of trun- cated HA-fragments during the subsequent deamination. In contrast, the LA-fraction was appreciably depolymerized, the peak elution position approximately corresponding to that of a heparin eicosasaccharide (Fig. 4B). It was anticipated that this elution profile would represent a composite of the fragments formed on cleavage of the HA-heparin at the AT- binding sequence, along with any fragment sequestered from the molecule by cleavage outside this sequence. However, judging from the apparently unchanged chain length of the


8 u)

a

0,08 I 13

c LA

0,02

0,oo 0

I h I

I

5"

k

Y

0 2

0.1

0 e9 v)

Fraction number FIG. 3. Chromatography on AT-Sepharose of products ob-

tained after limited N-deacetylation/deamination of HA-heparin. A sample (-30 mg) of HA-heparin was reacted (100 "C, 30 min) with 15 mg of hydrazine sulfate in 1.5 ml of anhydrous hydrazine. The partially deacetylated HA-heparin was desalted by passage through a column of Sephadex G-15 equilibrated with 0.2 M NH,HC03 followed by lyophilization. The product was specifically cleaved at N-deacetylated glucosamine units by treatment with HN02 at pH 3.9 as described (Pejler et al., 1987a). A sample (-15 mg) of the product was applied to a column (5 X 25 cm) of AT-Sepharose which was eluted with a linear salt gradient. Effluent fractions (-50 ml) were collected and analyzed for hexuronic acid (0). LA and HA material was recovered as indicated by horizontal bars and was further analyzed. - - -, NaC concentration.

residual HA-heparin as compared to that of the starting material (Fig. 4A), the latter type of cleavage would seem to be relatively infrequent. It is conceivable that the 6-0-sulfated N-acetylglucosamine residue of the AT-binding region (unit 2 in Fig. 2) was more susceptible to hydrazinolysis than were the N-acetylated glucosamine units in other regions of the heparin chain.

In order to assess the distribution of Fragment RT (see Fig. 1) within the elution profile of the LA degradation products (Fig. 4B), this material was separated into four pools as indicated, and each pool was subjected to compositional analysis by the HN02 (pH 1.5)/NaB3H4 procedure (see "Methods"), before and after digestion with bovine liver P-glucuronidase. The rationale behind this approach is apparent from the schemes in Figs. 1 and 2. Fragment RT, generated by N- deacetylation of unit 2 of the AT-binding region followed by deaminative cleavage of the corresponding glucosaminidic linkage, is distinguished by a nonreducing terminal GlcA- GlcNS03(3-OS03 or 3,6-di-oso3)- disaccharide unit, that is converted into the corresponding GlcA-[ l-3H]aMan~ disaccharide after HN02/NaB3H4 treatment. The location of this marker sequence at the nonreducing terminus was ascertained by the P-glucuronidase digestion, which would remove the terminal glucuronic acid residue from the polysaccharide fragment before the HN02/NaB3H4 step and thus preclude the formation of 3-0-sulfated labeled disaccharide (Fig. 2). These predictions are verified by the chromatograms in Fig. 5, which illustrate the separation of labeled disaccharides derived from Pool IV. The marker disaccharide, Gl~A-[~H]aMan~(3,6-di- osos), emerges somewhat after the major disaccharide component, I~OA(~-OSO~)-[~H]~M~~R(G-OSO~), the 3- but not 6- 0-sulfated analogue occurring in much smaller amounts (Fig. 5A) . Digestion of the labeled disaccharides with P-glucuronidase before the HPLC analysis abolished the peaks of GlcA- [3H]aMan~(6-OS03), Gl~A-[~H]aMan~(3-0S0~) and GlcA- [3H]aMan~(3,6-di-OS03), yielding instead labeled anhydro- mannitol-sulfate monosaccharides (not shown in the figure)

a

20 40 60 80

I

IV L 100 120

Effluent volume (ml) 100000 t

t

100 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Kav FIG. 4. Gel chromatography on Sephadex G-100 of HA- and

LA-fractions obtained after limited N-deacetylationldeami- nation of HA-heparin. A, superimposed chromatograms of the starting material (0) and of the HA-fraction recovered from the reaction products (m); B, the LA-fraction recovered from the reaction products. The LA-fraction was separated into four pools as indicated by the roman numerals, and each pool was desalted and analyzed further as described in the text. C, a calibration curve was constructed by plotting log M, uersus K., for a series of heparin oligosaccharide standards (from 2 to 24 monosaccharide units) and extending the relation to include the HA-heparin starting material (K, 0.25; M, - 15,000).

and [3H]aManR(3,6-di-OS03) (Fig. 5B). Analysis of polysaccharide fragments digested with @-glucuronidase before HN02/NaB3H4 treatment showed selective loss of the labeled GlcA-aManR(3,6-di-OSO3) disaccharide (the lack of [3H] aMan~(3,6-di-OSO~) is probably due to loss of monosaccharide components during the workup of reduction products) (Fig. 5 C ) , in accord with the predicted location of the 3-0- sulfated parent sequence at the nonreducing terminus of Fragment RT.

The compositional analysis of Pools I-IV and of the HA-

300 The Antithrombin-binding Sequence in the Heparin Chain

n

E p. 0

I W

cc)

8000

6000

4000

2000

2000

1000

15000

10000

5000

n

- (C)

6 +

1 2 3 4 5 7 - f f f f f f

20 40 60 80

Time (min) 100

FIG. 5. Anion-exchange HPLC of 'H-labeled disaccharides derived from Pool IV. A, disaccharides obtained by HN02 (pH 1.5)/NaB3H4 treatment of Pool IV; B, the same disaccharides, digested with @-glucuronidase before HPLC analysis; C, disaccharides obtained by HN02 (pH 1.5)/NaB3H4 treatment of @-glucuronidase- digested Pool IV. Samples of isolated disaccharides (-300,000 cpm 3H) were analyzed on a Partisil-10 SAX column as described under "Methods". The elution positions of standard mono- and disaccharides are indicated by numbered arrows: 1, GlcA(2-OSO3)-aMan~ (tentatively identified; see Bienkowski and Conrad, 1985); 2, GlcA- aMan~(6-OSOa); 3, IdoA-aMan~(G-OS03); 4, IdoA(2-OS03)-aMan~; 5, GlcA-aMan~(3-OSO3); 6, IdoA(2-OSO3)-aMan~(6-0SO3); 7, GlcA- aMan~(3,6-di-OS03); 8, aMan~(3,6-di-OS03). - - -, KHzPO, concentration.

heparin starting material (Table I) may be summarized as follows. (i) The contents of the marker disaccharide, GlcA- aMan~(3,6-di-OSO~), increased strikingly from Pool 1 to Pool IV, in agreement with the postulated terminal location of the corresponding intact sequence. Pretreatment of the polysaccharide fractions with 8-glucuronidase invariably precluded the subsequent formation of this particular disaccharide but did not significantly affect the generation of any of the other disaccharide components. (ii) The disaccharide GlcA- aMan~(6-OSOs) did not increase in amount with decreasing M, of the parent LA-saccharide fragments, nor did 8-glucuronidase digestion of these fragments prior to HN02/NaB3H4 treatment result in decreased yields of this disaccharide. This disaccharide thus would not seem to represent any nonreducing terminal structure, in agreement with the notion that N- deacetylation had been largely restricted to the AT-binding sequence? (iii) The relative amounts of the disaccharide IdoA- aMan~(6-OSOa) showed a 3-fold increase from Pool 1 to Pool IV and consistently exceeded those present in the HA-heparin

starting material. This disaccharide presumably represented units 1 and 2 of the original AT-binding sequence, the latter unit being the target of hydrazinolysis and the subsequent deamination reaction at pH 3.9, and thus the reducing terminus of Fragment NRT (see Figs. 1 and 2). (iv) The major disaccharide unit, recovered as IdoA(2-OS03)-aMan~(6- OSOS), as well as a low frequency unit converted into GlcA(2- OsO3)-aMan~ (tentative identification, see Bienkowski and Conrad, 1985), showed no apparent variation in abundance between the various LA-fragments. The relatively lower yield of the disaccharide IdoA(2-OS03)-aMan~ from the smallest LA-fragments (Pool IV) is intriguing, and may conceivably indicate a preferential location of the parent disaccharide unit in the middle portion of the intact HA-heparin molecule.

Taken together, the results suggest that all four LA-fractions contain both Fragment NRT and Fragment RT, apparently in approximately equimolar proportion^.^ If the AT- binding region had been located at or near the nonreducing terminus, cleavage of the HA-heparin would yield predomi- nantly small-sized Fragment NRT along with a large-sized Fragment RT, and vice versa (assuming an essentially mon- odisperse starting material). The results seem incompatible with the notion that the AT-binding region be preferentially located close to the nonreducing terminus of the heparin chain but instead strongly suggest a random distribution.

Random Cleavage of HA-Heparin-A different approach to defining the location of the AT-binding region, outlined in Fig. lB, was based on limited random degradation of HA- heparin, using the HN02 (pH 1.5) procedure. In this reaction N-sulfated glucosamine residues are converted into anhydromannose units, with cleavage (arrowheads in Fig. 1B) of the corresponding glucosaminidic linkages (Shively and Conrad, 1976). Since the starting material had been reduced with unlabeled NaBH4 before degradation, reduction of the cleavage products with NaB3H4 was expected to yield selective labeling (anhydro-[ l-3H]mannitol units indicated by aster- isks) of fragments containing the original nonreducing termini of the polysaccharide chains. Such fragments would be of HA or LA type, depending on the location of the cleavage site in relation to that of the AT-binding sequence. Labeled LA- fragments would also be generated whenever two cleavage sites (second site indicated by open arrowhead) would occur between the AT-binding region and the original reducing terminus of a chain. In order to facilitate the interpretation of the results it was therefore decided to minimize the extent of cleavage and thus the proportion of chains cleaved at more than one site.

Reduced HA-heparin was treated with nitrous acid as described in detail in the legend to Fig. 6. In a series of pilot experiments, various amounts of the deamination reagent were tested with regard to effects on the molecular size of the polysaccharide. The procedure adopted resulted in significant yet limited depolymerization of the HA-heparin, as evidenced by gel chromatography (Fig. 6A), whereas treatment with a

4Estimate based on quantitative relation between putative end groups and molecular size. For instance, in the major LA-fraction (Pool 111, average degrees of polymerization -18) approximately 1 out of 10 disaccharide units should occupy a nonreducing terminal position. Table I shows that GlcA-aMan(3,6-di-OSO~), the marker for Fragment RT, accounts for about 1 out of 20 (5.1%) disaccharide units and thus constitutes about half of the nonreducing termini. Conversely, Fragment NRT, as represented by the disaccharide IdoA- aManR(6-OSO3), would account for about half of the reducing termini (the yield, 7.5 mol %, of this disaccharide should be adjusted for some contribution by internal -IdoA-GlcNS03(6-OS03)- sequences, as expressed in the analysis of the HA-heparin starting material). Similar calculations apply to the other three LA-pools.


TABLE I Disaccharides formed on deaminationlreduction of HA-heparin and LA-fragments

Deamination products”

HA-heparin 50 0.7‘ -d 9.9‘ -d 2.1‘ -d 6.4‘ -d <0.5‘ - d 78‘ -d 3.1‘“ -d

LA-fragments (starting material)

Pool I 40-50 0.9 1.3 9.8 14 4.2 5.2 9.0 7.8 <0.5 <0.5 74 71 1.9 <0.5 Pool I1 28 0.8 0.8 11 11 5.8 6.8 8.1 9.4 <0.5 <0.5 71 72 2.8 <0.5 Pool 111 18 0.7 1.2 9.4 9.2 7.5 10 6.3 9.2 0.7 1.7 70 68 5.1 <0.5 Pool I V 8 0.7 0.7 7.7 8.1 11 15 3.6 5.6 1.3 0.5 63 69 13 C0.5

Determined by anion-exchange HPLC as described under “Methods.” The results are expressed as mol % of total sulfated disaccharides formed on deamination. The amount of labeled tetrasaccharides formed (not shown) corresponded to -7% of the label recovered in the disaccharide fractions from the LA-fragments (corrected for “anomalous ring contraction” as described by Pejler et al., 1987b).

d.p., degree of polymerization, denotes the approximate average number of monosaccharide units per polysaccharide or fragment molecule. The value for the HA-heparin starting material was calculated assuming a M, of 15,000 for the polysaccharide chains (see “Materials”), and M, 600 for the average disaccharide unit. The values for the LA-fractions are approximate and refer to the midpoint elution volumes for each pool (see Fig. 4B); they are based on calibration data obtained using well characterized oligosaccharide reference standards (extrapolated for Pools 1 and 11).

e Data in this column obtained after HN02/NaB3H4 treatment of undigested saccharide fractions. Data in this column obtained after &glucuronidase digestion of parent saccharide fractions (-, not performed

for the HA-heparin starting material). e This value can not be directly related to the amounts of the same disaccharide derived from the LA-fragments.

The latter disaccharide was generated through N-deacetylation of the major form of the AT-binding region, whereas that obtained on direct deamination of HA-heparin represented a minor, exclusively N-sulfated variant of the binding region (Lindahl et al., 1984). This structure would presumably not be affected by hydrazinolysis, hence would escape cleavage in the subsequent deamination step, and would be recovered in the fraction of residual HA-heparin after affinity chromatography.

10-fold diluted deamination reagent failed to induce any de- tectable change of the elution pattern. The peak elution volume of the depolymerized material, K,, -0.37, would cor- respond to an M , of about 11,000 (see Fig. 4C), i.e. indicating fragments more than half the size of the average starting material HA-heparin molecule. Chromatography on AT- Sepharose (not shown) of the labeled products (indicated Fragment HA* and Fragment L A * in Fig. 1B) obtained after reduction with NaB3H4 showed 60% LA- and 40% HA-fragments. Gel chromatography showed, as expected, a larger average molecular size for Fragment HA* (Fig. 6B) than for Fragment LA* (Fig. 6C). While a small-sized Fragment HA* would presumably be derived from a chain with the AT- binding site located close to the nonreducing terminus, a large- sized Fragment HA* would be less informative since the deaminative cleavage could have occurred at any locus between the binding site and the original reducing terminus (see Fig. 1B). Instead, each Fragment LA* would serve to indicate the minimal distance, in the corresponding parent chain, between the original nonreducing terminus and the AT-binding region. Indeed, Fragment LA* was markedly polydisperse, its size extending from that of a pentasaccharide (see below) toward that of the HA-heparin starting material. As much as 22% of the 3H label resided in fragments of M, > 7,500, suggesting that a significant proportion of the AT-binding regions had been located in the reducing terminal halves of the original chains. Again, these findings speak strongly against a preferential location of the binding region close to the nonreducing terminus but rather point to a random distribution.

The strategy adopted in this approach relied heavily on the notion that the label incorporated represented new reducing termini formed due to the deaminative cleavage. The validity of this assumption was tested using a recently developed monoclonal antibody which specifically recognizes a tetrasac-

charide sequence of the structure, -IdoA(2-OS03)-GlcNSO3(6- OS03)-IdoA(2-OS0~)-aMan~(6-OS03) (Pejler et al., 1988). This sequence, which represents the most abundant tetrasaccharide structure in heparin, is required in its entirety, in- cluding the anhydromannose end group, for interaction with the antibody. Approximately 50% of Fragment LA* was ad- sorbed to a column of the immobilized antibodies and was recovered by salt elution (see “Methods”). Gel chromatography of this material gave an elution profile virtually indistinguishable from that of the unfractionated fragment (Fig. 6C), hence in accord with prediction.

The results of the random cleavage approach described above, in particular the formation of large-size Fragment LA* species, support the concept of a randomly distributed AT- binding region. However, such fragments could also arise through cleavage at two sites located between the AT-binding region and the original reducing terminus (effect of cleavage at open arrowhead in Fig. lB) , and would then be compatible with a nonreducing terminal location of the binding region. In order to discriminate between these possibilities, hypo- thetical models were devised for each alternative and the experimental data were compared with those predicted on the basis of these models. Both models were based on 100 polysaccharide chains, M, 15,000, each containing 25 disaccharide units. The glucosaminidic linkages of these chains, numbered in a consecutive fashion from 1 to 2500, were cleaved by random selection, and the resulting fragments were sorted with regard to molecular size and affinity for AT. In one of the models (“nonreducing terminal model”), the AT-binding sequence was restricted to the nonreducing terminal region of each chain, unit 2 of the binding sequence (N-acetylated and hence resistant to deaminative cleavage) being set as the third glucosamine unit from the nonreducing terminus. In the other model (“random model”) the AT-binding region was shifted by one disaccharide unit from one chain to the next. In each

302

0 c3

am

n

E

I' p. 0

m

The Antithrombin-binding Sequence in the Heparin C h i n

0.15 I

0.10

0.05

300

200

100

10000

5000

0 30 50 70 90 110 130

Effluent volume (ml) FIG. 6. Gel chromatography on Sephadex G- 100 of products

obtained after random deaminative cleavage of HA-heparin. To a sample (5 mg) of HA-heparin in 3 ml of ice-cold HzO, acidified to pH 1.5 with dilute H2S04, was added 12.5 pg of NaNOz in 50 pl of HzO. The reaction mixture was kept in an ice bath for 3 h and was then neutralized with 1 M NazC03. After addition of 5 mCi of NaB3H4 in 200 p1 of 0.01 M NaOH, the sample was left a t room temperature for -15 h, was then acidified to pH -4 with glacial acetic acid, and finally was neutralized with 4 M NaOH. The products were desalted by passage through a column of Sephadex G-15 equilibrated with 0.2 M NH~HCOI and were lyophilized. Radiolabeled HA- and LA-fragments (corresponding to Fragment HA* and LA*, respectively, in Fig. 1B) were recovered after affinity chromatography on a column (1 X 3 cm) of AT-Sepharose eluted with a linear salt gradient. A, superimposed chromatograms of the HA-heparin starting material (.) and of the total products obtained after limited cleavage with HNOZ (pH 1.5) (0). The chromatography was performed as described under "Methods", and effluent fractions (-1 ml) were analyzed for hexuronic acid by the carbazole reaction. B, Fragment HA*; C, Fragment LA*. Fragment LA* was analyzed either directly (O), or after adsorp- tion to the immobilized monoclonal antibody 539/4H6 (e see "Meth- ods"). Effluent fractions indicated by the horizontal bur were combined and analyzed further as described in Fig. 8. For additional information see the text and Fig. 1.

model the cleavage process was pursued until a ratio of Fragment LA* to Fragment HA* of 60 to 40 was achieved, i.e. the value obtained under the experimental conditions employed. This point was reached when -0.5 linkage/polysaccharide chain had been cleaved in the random model, and

-1.75 linkages/chain in the nonreducing terminal model. By this criterion alone, the modest overall extent of degradation recorded by gel chromatography (Fig. 6A; average size of degradation products more than half that of the starting material) would tend to favor the random model. The Frag- ments LA* generated at this stage were displayed in a cumulative M, distribution, obtained by summating the numbers of the various fragments from the low", extreme upward. As seen from Fig. 7, the curve representing the nonreducing terminal model was shifted toward smaller sized fragments, compared to the random model curve. The data from the actual experiment were adopted to the same treatment, by subdividing the elution curve in Fig. 6C (material not subjected to immunoabsorption) into segments corresponding to odd-numbered (see below) oligosaccharides of defined M, (using the calibration curve described in the legend to Fig. 4C). The population obtained was then transformed into a similar cumulative M, distribution, using the area under each segment as a measure of the number of the corresponding oligosaccharide molecules. The curve obtained clearly con-

" 0 2000 4000 6000 8000 10000'

Molecular weight

FIG. 7. The molecular weight distribution of Fragment LA* as determined by gel chromatography. The amount of eluted material has been summated from the tail end of the chromatogram shown in Fig. 6C and plotted us molecular weight (A), using the calibration curve shown in Fig. 4C. Also shown are two analogous curves representing simulated model degradations of HA-heparin chains, in which the AT-binding region is either located close to the nonreducing terminus (El) or randomly distributed (.). For additional information see the text.

3000

n

E 2ooo t Q 0 = ift

c3 L

-

1000 -

0 90 100 110 120 130 140 150 160

Effluent volume (ml) FIG. 8. Gel chromatography on Sephadex G-SO of low mo-

lecular weight components of Fragment LA*. Fractions retarded on gel chromatography on Sephadex G-100 (indicated by the horizontal bur in Fig. 6C) were combined, concentrated, and applied to a column (1 X 190 cm) of Sephadex G-50, eluted with 1 M NaCl at -3 ml/h. Effluent fractions (-1 ml) were collected and analyzed for radioactivity. The numbered urrows indicate the peak elution positions of heparin disaccharides (2), tetrasaccharides (41, and hexasac- charides (6).

The Antithrombin-binding Sequence in the Heparin Chain

formed to that of the random rather than the nonreducing terminal model (Fig. 7).

The harmonic MI distribution of the Fragment LA* oligosaccharides is interrupted at the low”, extreme by the ac- cumulation of a 3H-labeled component, identified in separate gel chromatography (Fig. 8) as a pentasaccharide. This finding is in agreement with the notion that Fragment LA* contains the original nonreducing terminus of the chain, which would be largely constituted by a glucosamine residue (see “Discus- sion”). About half of the pentasaccharide molecules bound to the monoclonal antibody referred to above (data not shown), indicating that the corresponding intact HA-chains had contained the following nonreducing terminal sequence, Glc-

NS03(6-OS03)-. In this sequence, the nonreducing terminal glucosamine unit has to be N-sulfated, due to the substrate specificity of the hexuronosyl C5-epimerase required in the formation of the adjacent iduronic acid unit (Jacobsson et al., 1984); however, it is not known whether this glucosamine residue is 6-0-sulfated or not. Also, the preferential release of a terminal pentasaccharide fragment is intriguing, as there is no obvious reason why the third glucosamine unit from the terminus should be particularly susceptible to deamination. However, on the basis of these findings it seemed reasonable to assume that the majority of the Fragment LA* molecules contained the same, or an analogous, pentasaccharide sequence, and hence that they would be composed of an odd number of monosaccharide units.

NS~~-I~OA(~-OSO~)-G~CNS~~(~-OS~~)-I~OA(~-OSO~)-G~C-

DISCUSSION

The results of the present study indicate that the AT- binding region is randomly distributed along the HA-heparin chain, and thus disagree with the notion (Radoff and Danish- efsky, 1984; Rosenfeld and Danishefsky, 1988) that this region is preferentially or exclusively located at the nonreducing terminus. The reason for this discrepancy is unclear, but may be due to the fact that Danishefsky and co-workers (1984, 1988) restricted their study to a small, high molecular weight fraction of heparin molecules, whereas the present investigation included the bulk of the HA-heparin chains. Further- more, their conclusions were based exclusively on determi- nations of anticoagulant activity, whereas we have relied on structural data. The validity of anticoagulant assays in this context appears questionable, considering the inherent de- pendence of anticoagulant activity (especially with thrombin as target enzyme; Lane et al., 1984) on the length of the polysaccharide chain.

It should be emphasized that the random distribution of the AT-binding region applies to the polysaccharide fragments recovered in commercially available heparin preparations but not necessarily to the native proteoglycan. Heparin is synthesized as a proteoglycan, in which the constituent polysaccharide chains, MI 60,000-100,000, are much more extended than are those of commercial heparin (Robinson et al., 1978). Studies on heparin proteoglycans isolated from rat skin indicated that the AT-binding regions were accumulated in a small fraction of the total native polysaccharide chains, each HA-chain containing multiple binding sites (Jacobsson et al., 1986). In murine mastocytoma cells, the polysaccharide chains of newly synthesized proteoglycans are degraded by an endo-8-D-glucuronidase to fragments similar in size to the commercially available polysaccharide (Jacobsson and Lin- dahl, 1987). These relationships are indicated in schematic form in Fig. 9. It seems likely that heparin obtained from more conventional sources, such as porcine intestinal mucosa or bovine lung, has been processed in a similar manner, since

-GLy-

*

.

303

I x 1 0 ‘ H, 6- 25

FIG. 9. Schematic representation of part of a heparin proteoglycan. The broader segments on some of the polysaccharide chains represent the AT-binding pentasaccharide regions. These regions appear to be accumulated in a fraction of the chains only (Jacobsson et al., 1986). The cleavage sites for the endoglucuronidase are indicated by arrowheads. The results of the present study does not point to any defined location of these sites in relation to the AT- binding regions. For additional information see the text.

even in preparations isolated by mild procedures only a fraction of the heparin molecules contain the covalently bound neutral sugars and amino acids indicative of a polysaccharide- protein linkage region (Lindahl and Roden, 1965; Lindahl et al., 1965). The location of the AT-binding regions in the fragments (heparin) released by the endoglucuronidase thus would be determined by the distribution of these regions in the polysaccharide chains of the native proteoglycan, relative to that of the cleavage sites for the enzyme. The substrate specificity of the endoglucuronidase appears to be quite ex- acting, since substrate recognition by the enzyme requires the presence of both N- and 0-sulfate groups, presumably in specific although as yet undefined configuration (Thunberg et al., 1982b). The partial identification of the nonreducing terminal pentasaccharide sequence in a fraction of Fragment LA* would seem to provide novel information regarding the substrate recognition site for the enzyme. However, the results of the present study do not point to any predetermined topographic relation between the AT-binding regions in the heparin proteoglycan and the cleavage sites for the endoglucuronidase. On the other hand, our study has not focused on any particular type of HA-heparin, and it is therefore possible, as claimed by Rosenfeld and Danishefsky (1988), that the first AT-binding region of a native heparin chain regularly occurs at some distance from the linkage to the proteoglycan core protein.

The present investigation has some bearing on the properties of the so-called low molecular weight heparins that cur- rently attract considerable interest as potentially improved antithrombotic agents (Holmer et al., 1986). Such prepara-

304 The Antithrombin-binding Sequence in the Heparin Chain

(A) - NAC 3-050,

' ' ' , , I

FIG. 10. Scheme showing the possible number of HA-octa- decasaccharides that can be generated by random cleavage (e.g. by deamination) of a heparin chain in which an AT- binding region is located either close to the nonreducing terminus (A) or at a more internal site (B, representing random distribution of the binding region). The symbols used are the same as in Fig. 1. For additional information see the text.

tions, with an average M, of 5000-6000 (8-10 disaccharide units), are generally obtained by random, limited cleavage of commercial heparin. Due to their small size, these fragments would be just barely capable of accelerating the inhibition of thrombin by antithrombin, while retaining essentially full activity against Factor X, (Lane et al., 1984). This discrepancy is ascribed to the mechanisms of interaction between heparin, antithrombin, and the two proteases. While heparin is able to accelerate the formation of antithrombin-Factor X. complex by interacting with antithrombin alone, an effect on thrombin inhibition would appear to require simultaneous binding of both enzyme and inhibitor to the polysaccharide chain. In order to accommodate both proteins in such a ternary complex the polysaccharide chain must contain at least 18 monosaccharide units, a major portion of this saccharide sequence apparently being required for thrombin binding (Danielsson et al., 1986). The thrombin-binding sequence has not been defined, nor is it known whether it is contiguous to the reducing or the nonreducing terminus of the AT-binding region. However, it must be assumed that only those 18 saccharides are capable of promoting thrombin inhibition, in which the AT-binding region is maximally shifted toward one (or the other) end of the molecule. If the AT-binding region is initially located close to the nonreducing terminus of the HA-heparin chain, the thrombin-binding sequence would by necessity extend toward the reducing terminus (Fig. 1OA). Moreover, a relatively large proportion of the HA-18-saccharides generated by random cleavage of such chains will pro- mote the inhibition of thrombin by antithrombin, since the AT-binding region will be preferentially restricted to the nonreducing terminal region of these fragments. In contrast, random cleavage of HA-heparin molecules containing randomly distributed AT-binding regions (as favored by the present study) should produce a large number of HA-18- saccharides in which the AT-binding regions are differently located (Fig. 10B). Presumably, only a small proportion of these fragments will be able to bind thrombin along with the antithrombin molecule. These considerations suggest that the

functional properties of low molecular weight heparin preparations may vary, in spite of uniform M, distributions, due to differences in the cleavage patterns of the reactions used to depolymerize the starting material.

Ackrwwledgments-We are grateful to Prof. Torvard C. Laurent for helpful discussions.

REFERENCES Atha, D. H., Stephens, A. W., and Rosenberg, R. D. (1984) Proc.

Atha, D. H., Lormeau, J.-C., Petitou, M., Rosenberg, R. D., and

Atha, D. H., Lormeau, J.-C., Petitou, M., Rosenberg, R. D., and

Bienkowski, M. J., and Conrad, H. E. (1985) J. Biol. Chem. 2 6 0 ,

Bitter, T., and Muir, H. M. (1962) Anal. Biochem. 4 , 330-334 Bjork, I., and Lindahl, U. (1982) Mol. Cell. Biochem. 48 , 161-182 Casu, B., Oreste, P., Torri, G., Zoppetti, G., Choay, J., Lormeau, J.-

C., Petitou, M., and Sinay, P. (1981) Biochem. J. 197 , 599-609 Choay, J., Lormeau, J.-C., Petitou, M., Sinay, P., and Fareed, J.

(1981) Ann. N. Y. Acad. Sci. 370,644-649 Danielsson, A., Raub, E., Lindahl, U., and Bjork, I. (1986) J. Biol.

Chem. 261 , 15467-15473 Holmer. E.. Soderbere. K.. Bereavist. D.. and Lindahl. U. (1986)

Natl. Acad. Sci. U. S. A. 8 1 , 1030-1034

Choay, J. (1985) Biochemistry 2 4 , 6723-6729

Choay, J. (1987) Biochemistry 26,6454-6461

356-365

"

Haemostkis 16, Suppl. 2, 1-7 , . .

Hook. M.. Biork. I.. HoDwood. J.. and Lindahl. U. (1976) FEBS Lett. , I - , ,

66,' 90193"

236-245

E., and von Figura, K. (1979) Biochem. J. 179 , 77-87

Feingold, D. S. (1984) J. Biol. Chem. 259, 1056-1063

, . .

Hook, M., Riesenfeld, J., and Lindahl, U. (1982) Anal. Biochem. 119 ,

Jacobsson, I., Hook, M., Pettersson, I., Lindahl, U., Larm, O., Wirin,

Jacobsson, I., Lindahl, U., Jensen, J. W., Rodin, L., Prihar, H., and

Jacobsson. K.-G.. Lindahl. U.. and Horner. A. A. (1986) Biochem. J. , I

240,625-632 '

. .

Jacobsson. K.-G.. and Lindahl. U. (1987) Biochem. J. 246.409-415 Lane, D. A., Denton, J., Flynn; A. M., Thunberg, L., and Lindahl, U.

Laurent, T. C., Tengblad, A., Thunberg, L., H&k, M., and Lindahl,

Lindahl, U. (1966) Biochim. Biophys. Acta 130 , 368-382 Lindahl, U., and Rodin, L. (1965) J. Biol. Chem. 2 4 0 , 2821-2826 Lindahl, U., Cifonelli, J. A., Lindahl, B., and Rodin, L. (1965) J. Biol.

Lindahl, U., Backstrom, G., Thunberg, L., and Leder, I. G. (1980)

Lindahl, U., Backstrom, G., and Thunberg, L. (1983) J. Biol. Chem.

Lindahl, U., Thunberg, L., Backstrom, G., Riesenfeld, J., Nordling, K., and Bjork, I. (1984) J. Biol. Chem. 2 6 9 , 12368-12376

Pejler, G., Backstrom, G., Lindahl, U., Paulsson, M., Dziadek, M., Fujiwara, S., and Tim 1, R. (1987a) J. Biol. Chem. 262,5036-5043

Pejler, G., Danielsson, x., Bjork, I., Lindahl, U., Nader, H. B., and Dietrich, C. P. (1987b) J. Biol. Chem. 2 6 2 , 11413-11421

Pejler, G., Lindahl, U., Larm, O., Scholander, E., Sandgren, E., and Lundblad, A. (1988) J. Biol. Chem. 263,5197-5201

Petitou, M., Duchaussoy, P., Lederman, I., Choay, J., and Sinay, P. (1988) Carbohydr. Res. 179 , 163-172

Radoff, S., and Danishefsky, I. (1984) J. Biol. Chem. 2 6 9 , 166-172 Riesenfeld, J., Thunberg, L., Hook, M., and Lindahl, U. (1981) J.

Robinson, C. H., Horner, A. A., Hook, M., Ogren, S., and Lindahl, U.

Rosenfeld, L., and Danishefsky, I. (1988) J. Biol. Chem. 2 6 3 , 262-

Shively, J. E., and Conrad, H. E. (1976) Biochemistry 15,3932-3942 Thunberg, L., Backstrom, G., and Lindahl, U. (1982a) Carbohydr.

Thunberg, L., Backstrom, G., Wasteson, A., Robinson, H. C., Ogren,

(1984) Biochem. J. 218,725-732

U. (1978) Biochem. J. 175 , 691-701

Chem. 240,2817-2820

Proc. Natl. Acad. Sci. U. S. A. 7 7 , 6551-6555

258,9826-9830

Biol. Chem. 266,2389-2394

(1978) J. Biol. Chem. 253, 6687-6693

266

Res. 100,393-410

S., and Lindahl, U. (1982b) J. Biol. Chem. 267,10278-10282

Documents

Location of the Antithrombin-binding Sequence in the ... · Location of the Antithrombin-binding Sequence in the ... Casu et al., 1981; Choay et al., 1981 ... 298 The Antithrombin-binding