Mechanisms Responsible for Catalysis of the Inhibition of Factor Xa or Thrombin by Antithrombin Using a Covalent Antithrombin-Heparin Complex*

Covalent antithrombin-heparin (ATH) complexes, formed spontaneously between antithrombin (AT) and unfractionated standard heparin (H), have a potent ability to catalyze the inhibition of factor Xa (or thrombin) by added AT. Although ≈30% of ATH molecules contain two AT-binding sites on their heparin chains, the secondary site does not solely account for the increased activity of ATH. We studied the possibility that all pentasaccharide AT-binding sequences in ATH may catalyze factor Xa inhibition. Chromatography of ATH on Sepharose-AT resulted in >80% binding of the load. Similar chromatographies of non-covalent AT + H mixtures lead to a lack of binding for AT and fractionation of H into unbound (separate from AT) or bound material. Gradient elution of ATH from Sepharose-AT gave 2 peaks, a peak containing higher affinity material that had greater anti-factor Xa catalytic activity (708 units/mg heparin) compared with the peak containing lower affinity material (112 units/mg). Sepharose-AT chromatography of the ATH component with short heparin chains (≤12 monosaccharides) resulted in active unbound (40%) and bound fractions (190 and 560 units/mg, respectively). Factor Xa-ATH or thrombin-ATH inhibitor complexes gave chromatograms on Sepharose-AT with more unbound material compared with that of free ATH. Also, ATH did not bind to Sepharose-heparin, and the intrinsic fluorescence due to activation of AT in ATH by its heparin chain was reversed at higher [NaCl] than that required to dissociate non-covalent AT·H complexes. Thus, exogenous AT can compete with the AT moiety of ATH for binding to the covalently linked heparin chain, leading to catalytic inhibition of factor Xa or thrombin. These data may suggest that access to pentasaccharide units in non-covalent AT·H complexes by free AT may be facile.

Unfractionated standard heparin (H) 1 is a glycosaminoglycan (GAG) that catalyzes inhibition of the coagulant enzymes factor Xa and thrombin by the serine protease inhibitor (serpin) antithrombin (AT) (1)(2)(3). Reaction occurs via the allosteric activation of AT, due to H binding, followed by attack of the enzyme on the reactive center of the inhibitor (4). In the case of thrombin, binding to the heparin chain by the enzyme must also occur for efficient reaction to take place (3). After formation of thrombin-AT or factor Xa-AT inhibitor complexes, affinity of the AT moiety for the heparin chain decreases, leading to release of the catalyst for further reactions with AT and enzyme (5,6). Although binding to thrombin is through nonselective interaction of negative charges on the GAG with the anion-binding exosite of the enzyme (7,8), AT binding to H occurs through high affinity to a specific pentasaccharide sequence on the heparin chain (2,9). Moreover, it has been shown that the rate-determining step for catalysis of thrombin (or factor Xa) inhibition involves the initial binding of AT and H (10).
Previously, we produced a covalent AT-heparin complex (ATH) to further study the mechanism of enzyme inhibition by H-activated AT (11,12). Surprisingly, although serpin and GAG cannot dissociate, it was observed that ATH could catalyze the inactivation of thrombin (or factor Xa) by added AT (11). In fact, the specific catalytic activity of ATH was ϳ4-fold greater than that measured for reaction of factor Xa/thrombin ϩ AT with starting H (11). Confirmation that inhibitory activity of ATH against either factor Xa or thrombin (in the presence of exogenous AT) was catalytic in nature became apparent from the fact that many-fold more molecules of enzyme were inactivated in the presence of excess added AT compared with that of ATH alone (11). Furthermore, plasma thrombin generation on fetal distal lung epithelial cells was inhibited more effectively in the presence of ATH than H due, in part, to the formation of thrombin-AT inhibitor complexes from plasma AT (13).
Investigations were carried out to determine components of the mechanisms involved in ATH catalysis of the AT ϩ factor Xa (or thrombin) reactions. AT in the ATH complex has been shown to exist in the activated form due to direct interaction with a high affinity (pentasaccharide) sequence on the covalently attached heparin chain (11). Thus, one possible mode for ATH catalysis of factor Xa/thrombin inhibition would be via the activation of added AT molecules by the pentasaccharide site proximal to the AT component of ATH. However, because of the rapid velocity of the direct thrombin (factor Xa) ϩ ATH reaction (second order rate constants for thrombin ϩ ATH and factor Xa ϩ ATH are 3.0 ϫ 10 9 and 2.4 ϫ 10 8 M Ϫ1 min Ϫ1 , respectively (14)), thrombin-ATH (or factor Xa-ATH), inhibitor complexes might be generated before several cycles of the catalytic reaction could take place. Once factor Xa or thrombin had formed a covalent bond with the AT portion of ATH, approach by free serpin and enzyme to the inhibited ATH may be restricted due to steric reasons. Another explanation for the potent catalytic activity of ATH is the possibility that during conjugate synthesis, AT may have selected for a small subpopulation of H molecules that have more than one pentasaccharide (15). There is strong evidence from a number of experiments that ATH preparations contain multipentasaccharide GAGs. First, titration of ATH by AT resulted in a 30 -40% increase in intrinsic fluorescence, which is characteristic of AT activation by pentasaccharide binding (11,14). Second, gel filtration of the non-covalent complexes formed between the heparin released from ATH by protease treatment and excess free AT showed that ϳ30% of the heparin molecules were capable of binding Ն2 AT molecules (14). However, assays of ATH preparations have consistently shown that both anti-factor Xa and anti-thrombin catalytic activities are ϳ2-fold greater than that of H with high affinity for AT (11). Thus, it would appear that ATH molecules that have a single pentasaccharide (ϳ70%) may also contribute to the observed catalytic potency.
We decided to study the possibility that all pentasaccharide units on ATH heparin chains may be available to activate added AT molecules for factor Xa (or thrombin) inhibition.

EXPERIMENTAL PROCEDURES
Chemicals-All reagents were of analytical grade. Standard heparin (H) was from Sigma (grade I-A, sodium salt, 15-kDa average molecular mass, from porcine intestinal mucosa (Mississauga, Ontario, Canada)). Human antithrombin (AT) was from Bayer (Mississauga, Ontario, Canada). Ninhydrin was from Fisher, and SnCl 2 ⅐2H 2 O and 2-methoxyethanol were from Sigma. Stachrom heparin kits (containing the CBS 31.39 substrate for factor Xa) were obtained from Diagnostica Stago (Asnières, France) and anti-IIa kits were from American Diagnostica Inc. (Greenwich, CT). Protamine sulfate was obtained as the solid from ICN (Cleveland, OH), and arginine was purchased from Sigma. Sephadex G-200 beads, CNBr-activated Sepharose 4B, and heparin-Sepharose CL6B were all from Amersham Biosciences. Molecular weight standards used to characterize Sephadex G-200 chromatograms are as follows: dextran 70,000 (Amersham Biosciences, catalogue number T-70), dextran 42,000 (Sigma, catalogue number D-4133), dextran 10 (Amersham Biosciences, catalogue number dextran 10, lot number To 5400), and dextran sulfate 8,000 (Sigma, sodium salt, catalogue number D-4911). Bio-Gel P-6 was from Bio-Rad. Purified heparin oligosaccharides, used as standards for characterizing heparin chains isolated from ATH fractions, were kindly provided by Dr. John Gallagher. Heparinase was from ICN. Human factor Xa and human thrombin were obtained from Enzyme Research Laboratories (South Bend, IN), and the S-2238 chromagenic substrate for thrombin titrations was from DiaPharma (West Chester, OH). ATH was prepared as described previously (11,16). Briefly, AT was incubated with H at 40°C in 0.15 M NaCl, 0.02 M phosphate buffer, pH 7.3, for 14 days, followed by purification of the covalent product by sequential chromatographies on butyl-agarose (Sigma) and DEAE-Sepharose Fast Flow (Amersham Biosciences). SDS-PAGE (reducing conditions) of the ATH produced showed that Ͻ5% of free unreacted AT or H were present in the final preparation. In order to ascertain the AT and H present in the conjugate, detailed chemical analyses were carried out (see below). Heparin chains from ATH were prepared according to a procedure similar to that described previously (14). In brief, ATH (equivalent to 8.66 mg of AT) ϩ 2 mg of protease P-5147 (Sigma) in 1 ml of 0.5 M Tris⅐HCl, pH 8.0, were heated at 37°C for 24 h. The incubate was centrifuged in a microcentrifuge and the supernatant dialyzed against 0.01 M Tris⅐HCl, pH 8.0. Dialyzed product was loaded onto DEAE-Sepharose Fast Flow (3 ml of packed beads) that was pre-equilibrated with 0.01 M Tris⅐HCl, pH 8.0. After washing the column with 0.25 M NaCl in 0.01 M Tris⅐HCl, pH 8.0, heparin chains were eluted with 2 M NaCl in 0.01 M Tris⅐HCl, pH 8.0. The eluted heparin (ϳ2 mg) was dialyzed exhaustively versus H 2 O and either stored at 4°C or freeze-dried. Heparin isolated from protease-treated ATH was designated as HЈ.
Physicochemical Analyses of ATH-The AT and heparin content in ATH was measured using several methods in order to rigorously determine the AT:heparin mole ratio as a confirmation of earlier studies. ATH was hydrolyzed in 6 M HCl at 100°C for 20 h and, after evacuation under vacuum, analyzed for amino acid content (Beckman System 6300 High Performance Analyzer) against an amino acid reference that had been added as an internal standard. Given the known protein sequence and N-linked glycan content for human antithrombin (17,18), the molecular weight of AT was calculated to be 57,769. Analysis of results for amino acids that are stable to acid hydrolysis (i.e. alanine and arginine) was used to determine the number of moles of AT in the original sample (31 alanine residues and 22 arginine residues per AT molecule) and, thus, the number of milligrams of AT per unit volume of original ATH solution. Measurement of the absorbance at 280 nm of ATH stock solution and use of the milligram of AT/ml determined by amino acid analysis allowed for calculation of an extinction coefficient for ATH in terms of AT concentration. Three different methods were used to evaluate the heparin content in ATH. Heparin mass concentration in stock ATH solution was analyzed using the carbazole (19), Azure A (20), and Alcian blue (21) techniques. In each case, background measured in samples containing a similar concentration of purified AT was subtracted from values calculated for ATH. Standards were prepared from solid commercial Sigma heparin. Moles of heparin in ATH was derived by dividing the mass of heparin in ATH samples by the ATH heparin chain molecular weight. The number average molecular weight (Mn) of heparin released from ATH by exhaustive protease treatment was determined by end group analysis of the ␣-amino group on the remaining amino acid or short peptide linked to the heparin chains. In brief, ATH was incubated with protease P-5147 and the heparin chains (HЈ) isolated as described above. HЈ chains were dialyzed exhaustively against H 2 O and freeze-dried. Analysis of acid-hydrolyzed HЈ on the Beckman System 6300 High Performance Analyzer showed only a large free glucosamine peak and one unidentified peak (likely the lysyl-uronic acid linkage group reported previously (12)). To quantitate the HЈ amino acid end group, weighed out samples of HЈ (in 2 ml of H 2 O) were incubated with ninhydrin reagent (0.2 ml of a solution prepared from 0.2 g of ninhydrin ϩ 7.5 ml of 2-methoxyethanol ϩ 2.5 ml of 4 M sodium acetate, pH 5.5, ϩ 10 mg of SnCl 2 ⅐2H 2 O with constant stirring and bubbling of a stream of commercial grade N 2 ) at 100°C for 15 min. After cooling to 23°C, the absorbance at 570 nm was measured. Absorbance resulting from any primary amino groups present on intrachain glucosamine residues was determined by analysis of similar amounts of freeze-dried samples of disaccharides recovered outside dialysis bags containing reaction mixtures of ATH ϩ heparinase. Subtraction of the ninhydrin-glucosamine absorbance (a small proportion of the total) from the data gave results with absorbance exclusively due to reaction with the aldose-terminal amino acid. Results were compared with that of reactions with alanine standard to determine moles of terminal ␣-amino group. Division of the weight of sample by the moles of terminal amino groups gave the number average molecular weight per mol of HЈ. Heparin chains released from ATH by protease were also gel-filtered on a Sephadex G-200 column (Amersham Biosciences) to assess the polydispersity relative to commercial heparin. Finally, the number of moles of AT and heparin in ATH stock solutions was used to calculate the heparin:AT mole ratio in ATH.
Anti-Factor Xa and Anti-Factor IIa Assays-Anti-factor Xa activities were determined using the commercially available Stachrom heparin kit. Heparin standards (0, 0.4, and 0.8 anti-factor Xa IU/ml), controls, or samples (50 l), diluted in the presence (for catalytic activity) or the absence (for checking non-catalytic activity) of AT, were incubated with purified bovine factor Xa for 120 s at 37°C. After the incubation period, the chromogenic substrate CBS 31.39 is added, mixed, and incubated for 90 s before reading the absorbance at 405 nm. Control and unknown sample values were determined by interpolation from the linear heparin standard curve. The anti-factor IIa assay was determined chromogenically by commercially available kits called Actichrom heparin anti-IIa. Heparin standards (0.0 -0.6 USP units/ml), controls (1:16 dilution), or samples (1:16 dilution) were added to AT-containing reagent, followed by mixing and incubation at 37°C for 2 min. Thrombin reagent was then added, followed by mixing and incubation for a further 2 min at 37°C. Finally, Spectrozyme TH substrate was added, and the mixture was incubated for 1 min before reading the absorbance at 405 nm. Similar to the anti-Xa assay, control and unknown sample absorbance values were interpolated from the heparin standard curve. All assays were performed on an automated ACL 300ϩ machine (Instrumentation Laboratories, Milano, Italy). The heparin values were converted from units/ml to units/mg by dividing the heparin activity values by the heparin mass concentrations (mg/ml), as determined by protamine sulfate assay (see below).
Protamine Sulfate Assay-The protamine sulfate assay is an aggregation assay used to determine the heparin mass concentration in a sample (22)(23)(24). Briefly, 0.2 ml of 1.0 mg/ml protamine sulfate solution in H 2 O was added to 0.5 ml of H standards, ATH standards, or unknown samples, followed by immediate vortexing. After 10 min at room temperature, 1.0 ml of 0.1 M L-arginine was added to the mixture and vortexed, followed immediately by the addition of 2.3 ml of 0.1 M Tris⅐HCl, pH 8.0, and further vortexing. Absorbances of H standards, ATH standards, or unknown samples were read at 470 nm within 1 h, and unknown samples were read from the appropriate H or ATH standard curve.
Sepharose-AT Chromatography-Lyophilized AT powder was reconstituted with 10 ml of sterile H 2 O from the kit. It was then dialyzed against coupling buffer before conjugation to CNBr-activated Sepharose beads. Conjugation of AT to the beads was done according to the manufacturer's instructions and resulted in affinity matrix material containing 9.84 mg of AT/ml. Sepharose-AT columns (10 ml, pre-equilibrated with 0.15 M NaCl in 0.01 M phosphate, pH 7.3 buffer) were loaded with either ATH (equivalent to 2 mg of AT and 0.6 mg of H), H, AT ϩ H, HЈ, HMWATHF, LMWATHF, HMWH, or LMWH at similar loading levels (33.9 nmol of each species). After loading, the column was washed with 3 column volumes of 0.15 M NaCl buffer before elution of any bound material with a linear gradient (25 ml of 0.15 M NaCl buffer in the mixing chamber and 25 ml of 2 M NaCl buffer as limit solution). Finally the column was treated with 2 column volumes of 2 M NaCl buffer to ensure equilibration of the column with high salt. NaCl concentrations in the eluate were determined using a conductivity meter (E C Meter, Amber Science Inc., Eugene, OR). Fractions were analyzed for either protein or heparin by measuring absorbance (280 or 215 nm) or taking samples for assay with protamine (as described above), respectively. AT or heparin peaks were dialyzed against H 2 O, freeze-dried, and resuspended in 0.15 M NaCl for further assays for activity (antifactor Xa or anti-IIa, described above). Smaller columns of Sepharose-AT (1.5 ml) were constructed for chromatographies of factor Xa-ATH or thrombin-ATH complexes using the same washing and elution protocols as those for the larger columns.
Fractionation of ATH or H into High and Low Molecular Weight Fractions-Fractionation of ATH by molecular weight was performed on a Sephadex G-200 column. The Sephadex G-200 column was prepared swollen and poured according to the manufacturer's instructions. The column (300 ml) was equilibrated with 2 M NaCl, and 20 mg of ATH was loaded, with elution at 2 M NaCl. Fractions were collected (Ϸ3.9 ml/fraction) and analyzed for protein concentration (A 280 ). The column fractions of high (the first to be eluted from the column) and low molecular weight species were designated high molecular weight ATH fraction HMWATHF and low molecular weight ATH fraction LM-WATHF, respectively. The first 2-9% (HMWATHF) and the last 2-9% (LMWATHF) of the eluted fractions were separately pooled and then dialyzed against 0.15 M NaCl in 0.01 M phosphate buffer, pH 7.3, before further fractionation on Sepharose-AT. To confirm the molecular weight of LMWATHF, the chain length of heparin released from LM-WATHF was assessed. LMWATHF was incubated for up to 24 h at 37°C with catalytic amounts of protease (Ͻ10% by mass) in 0.05 M Tris⅐HCl, pH 8.0. The reaction mixture was chromatographed on a Bio-Gel P-6 column (1 ϫ 49 cm), with elution at 0.15 M NaCl (1-ml fractions) and detection by a refractive index detector (Jasco RI-1531, Tokyo, Japan). The peak containing LMWATHF heparin chains was pooled and freeze-dried. Freeze-dried material was resuspended in a small volume of H 2 O, followed by precipitation of LMWATH heparin chains with 10 volumes of absolute ethanol and recovery of the precipitate pellet by centrifugation. The pellet was resuspended in H 2 O and reprecipitated in ethanol two more times to complete the removal of any remaining NaCl. The isolated amino acid capped LMWATHF heparin chains (along with commercial heparin and heparin oligosaccharide standards) were subjected to fluorophore-assisted carbohydrate electrophoresis (25), with staining by Safranin O. Comparison of the lane containing LMWATHF heparin chains to that of oligosaccharide standards allowed for definition of chain length in terms of number of saccharide units. The HMWATHF (equivalent to 1.4 mg of AT) and LM-WATHF (equivalent to 1.6 mg of AT) samples were subjected to Sepharose-AT chromatography (10 ml column) that was pre-equilibrated with 0.15 M NaCl buffer at pH 7.3. After loading, the column was washed with 2 column volumes of 0.15 M NaCl buffer before elution of any bound material, using a linear NaCl gradient and a 2 column volume of 2 M NaCl wash as described above. Fractions were analyzed for protein by measuring absorbances at 280 and 215 nm. Sepharose-AT chromatographic peaks were pooled and then dialyzed against H 2 O, freeze-dried, and resuspended in 0.15 M NaCl for further activity assays (anti-factor Xa or anti-IIa, described above). HMWH and LMWH were prepared in the same way as described for HMWATHF and LM-WATHF, except that 5 mg of heparin was gel-filtered, and heparin analyses were by the protamine sulfate assay.
Sepharose-AT Chromatography of Serine Protease-ATH Inhibitor Complexes-Factor Xa-ATH or thrombin-ATH inhibitor complexes were prepared as follows. Each serine protease was used to titrate ATH to Ϸ100% equivalence, as determined by the presence of a small amount of activity against its chromogenic substrate (CBS 31.39 substrate from the Stachrom heparin kit for factor Xa; S-2238 substrate for thrombin). Factor Xa or thrombin was added to 0.25 mg of ATH (in terms of AT) and, after adjusting the total volume to 1 ml, immediately loaded onto a 1.5-ml Sepharose-AT column. Identical washing and elution protocols were used as those for the larger (10 ml) columns. The various eluted peaks, as well as preformed factor Xa-ATH and thrombin-ATH complexes, were assayed for anti-factor Xa activity as described above. In order to investigate the possibility that excess free AT may actually protect the AT of ATH from attack by protease, competition experiments were carried out. Either factor Xa or thrombin was reacted for various periods with ATH Ϯ100-fold molar excess of added AT at 37°C. Reaction was stopped by heating at 100°C for 5 min, and the ATHcontaining material was purified from free thrombin, factor Xa, or AT by chromatography on DEAE-Sepharose using the same method as that for ATH purification given above (11). ATH ϩ enzyme-ATH inhibitor complexes were treated with heparinase for 2 h at 37°C, followed by SDS-PAGE and staining of the gels with Coomassie Blue R-250 for protein as described previously (11). Laser densitometry of bands on dried gels was carried out to determine the proportion of free ATH (as AT) compared with enzyme complexed ATH (as either factor Xa-AT or thrombin-AT). Comparison of results from lanes of experiments that had no added AT to those with added AT revealed the effect of added AT on direct reaction of ATH with either factor Xa or thrombin.
Determination of Binding Affinities Using Intrinsic Fluorescence-All fluorometric determinations were made with stirring in a 1 ϫ 1-cm quartz fluorescence cuvette placed in a cuvette holder heated at 25°C and using a PerkinElmer Life Sciences LS50B luminescence spectrometer. Experiments were carried out by adding either 1 ml of 100 nM AT, 100 nM AT ϩ 234 nM H, 100 nM ATH, or 100 nM LMWATHF in 0.02 M Tris⅐HCl, pH 7.4, to the cuvette, followed by titration with 5 M NaCl, 0.02 M Tris⅐HCl, pH 7.4, containing either 100 nM AT, 100 nM AT ϩ 234 nM H, 100 nM ATH, or 100 nM LMWATHF, respectively. Thus, the NaCl concentration of AT, AT ϩ H, ATH, or LMWATHF was increased from 0 to 2.25 M. Protein intrinsic fluorescence was measured (after each high salt solution addition (added in 10 -100-l increments)) with an excitation at 280 nm and emission detected at 340 nm (with a 290 nm cut-off filter). Excitation and emission slit widths were 5 and 7 nm, respectively. The effect of NaCl on intrinsic fluorescence was determined as the fluorescence intensity after each high salt addition minus the value at equilibrium end point (2.25 M NaCl) and calculated as a percentage of the difference in fluorescence intensity at 0 and 2.25 M NaCl. Values for the percent difference in fluorescence for AT ϩ H, ATH, and LMWATHF were corrected for change in fluorescence of AT alone at similar NaCl concentrations.
Sepharose-Heparin Chromatography-The heparin Sepharose CL-6B column was prepared according to the manufacturer's instructions. The column (10 ml) was equilibrated with binding buffer (0.15 M NaCl in 0.01 M phosphate buffer, pH 7.3). ATH (equivalent to 2 mg of AT), AT (2 or 4 mg of protein), AT (2 mg) ϩ H (0.5 mg), or AT (2 mg) ϩ HЈ (0.6 mg) were loaded onto the column, followed by washing with 3 column volumes of binding buffer before elution of bound material with a linear NaCl gradient (0.15-2.0 M) and final washing with 2 column volumes of 2.0 M NaCl. Absorbance at 280 and 215 nm of the collected fractions was used to determine protein concentration, and protamine sulfate assay was used to measure H concentration. Fractions in the different peaks were pooled, dialyzed against H 2 O, freeze-dried, and resuspended in 0.15 M NaCl for activity testing (anti-factor Xa or anti-IIa).
Statistical Analysis-Data were compared for significant differences using either the Student's t test (in the case of two groups) or by analysis of variance (ANOVA, for more than two groups). Upon finding a significant difference within several groups by ANOVA, testing between two groups within that set was carried out by t test. A p value of Ͻ0.05 was considered significant, and results were expressed as mean Ϯ S.E.

RESULTS
Physicochemical Analysis of ATH-Stringent analyses of the protein and heparin content of ATH preparations were carried out to verify the heparin:AT mole ratio present in the conjugate. Aliquots of ATH stock solution were treated with HCl, and the hydrolysate was analyzed to determine amino acid content. Given the known sequence for human AT, the number of moles of acid-stable amino acids (alanine and arginine) recovered were used to calculate the molar concentration (in terms of AT) of the original ATH solution. Typical ATH stock solutions were 1.4 ϫ 10 Ϫ4 M in AT. Given a molecular weight for AT of 57,769 (calculated from the amino acid sequence and known carbohydrate content (17,18)), absorbance readings at 280 nm for dilutions of ATH stock solutions gave an extinction coefficient of 0.641 for ATH concentrations of 1 mg of AT/ml. A value of 0.630 obtained for purified human AT was in agreement with that found previously (26). Three separate methods were used to determine the mass concentration of heparin in ATH solution. Background contribution due to AT in the ATH sample was assessed using AT solutions of similar concentration. Although AT control values were low in azure A and Alcian blue heparin assays, a significant value was obtained when the carbazole assay was applied (Table I). The relatively high signal given by AT controls in the carbazole assay was not surprising given that neutral sugars in the N-linked glycans of AT give H 2 SO 4 dehydration products that condense with carbazole in the assay procedure (19,27). Nevertheless, after correction for AT control values, heparin:AT mass concentration ratios were similar for all three assay methods (although precision was reduced for the carbazole procedure). Heparin:AT mole ratios for ATH were calculated from the heparin mass assays, given the number of moles of AT in the stock solution (as determined above) and a number average molecular weight for heparin chains in ATH of 16,900 Ϯ 200 (average Ϯ S.E.; amino acid end group analysis of chains released by protease). Results from the three heparin mass analysis procedures indicated that the heparin:AT mole ratio for ATH was close to 1:1 (Table I). Heparin mass concentration analyses of HMWATHF and LMWATHF using the Azure A method gave results that were proportional to the relative molecular weights of the conjugate heparin chains.
Sepharose-AT Fractionation of ATH and H-Chromatography of ATH on immobilized AT resulted in binding of Ͼ80% of the load (Ͼ70% as a high affinity fraction, Fig. 1A). In order to ensure that the Sepharose-AT column was not overloaded, chromatographies with different loading amounts were run. Similar chromatographies of H gave sizable peaks of unbound material (ϳ40% of the total recovery (Fig. 1B)). Sepharose-AT chromatographies of either ATH or H resulted in three distinct peaks as follows: peak 1 appeared in the wash fraction as unbound material, and peaks 2 and 3, which represent low and high affinity products, respectively, were eluted from the column by linear NaCl gradient (0.15 to 2 M). This three peak  followed by 2 column volumes of 2 M NaCl in buffer. Eluted material was detected by A 280 (protein) or protamine sulfate assay (heparin (A 470 )) and appeared as either unbound, low affinity, or high affinity peaks (peaks 1-3, respectively). In the case of AT ϩ H chromatographies, unbound (peak 1) AT (q) and H (ƒ) materials chromatographed separately. pattern (no affinity (unbound peak 1), low affinity (peak 2), and high affinity (peak 3)) was typical of separations of heparincontaining species on immobilized AT. Chromatography of noncovalent AT⅐H complexes on the AT column led to dissociation of protein and GAG, with free AT running slightly ahead of unbound H, followed by the usual low (peak 2) and high (peak 3) affinity gradient eluted peaks (Fig. 1C). The relative proportions and elution positions of peaks produced by Sepharose-AT fractionation of AT ϩ H were essentially the same as those of H alone (compare Fig. 1, B and C) and in agreement with the 45-55% high affinity AT binding observed previously for this commercial H (14). Fractions in the peaks obtained from Sepharose-AT chromatographies were pooled and concentrated. Initial testing of the Sepharose-AT peaks showed that all ATH peaks had significant, direct non-catalytic activity against factor Xa, whereas peaks containing only H were completely inactive. Further assays were done to determine the ability to catalyze inhibition of factor Xa by added AT (catalytic anti-factor Xa assay). All assays for catalytic activity showed high sensitivity and reproducibility. As expected, although unbound and low affinity material from either H or AT ϩ H chromatographies had very low catalytic anti-factor Xa activities (Ͻ10 units/mg heparin (Table II)), H with high affinity for AT had significant catalytic activity (463 or 447 units/mg for H or AT⅐H, respectively (Table II)). Alternatively, assays of unbound and low affinity material from Sepharose-AT chromatograms of ATH had moderate anti-factor Xa catalytic activities (231 and 112 units/mg heparin for peaks 1 and 2, respectively (Table II)), and high affinity ATH contained 1.53 times the activity of high affinity H (t test, p Ͻ 0.01). Thus, the vast majority of ATH with potent activity to catalyze the factor Xa ϩ AT reaction was capable of strong AT binding, prior to complexation with a serine protease (i.e. factor Xa or thrombin). In order to verify that the heparin chains of ATH were responsible for the significant binding affinity to immobilized AT, ATH was treated with protease, and the heparin released (HЈ) was purified on DEAE-Sepharose. The anti-factor Xa activity of HЈ was measured to be 644 units/mg. Gel filtration of HЈ showed that the ATH heparin chains had a much higher proportion of molecules with longer chain length than standard H ( Fig. 2A). Chromatography of HЈ on Sepharose-AT resulted in 83% of the material having high affinity binding and potent (660 units/ mg) anti-factor Xa activity (Fig. 2B and Table II). Therefore, the vast majority of heparin chains in ATH contained catalytically active, high affinity AT-binding sites.
Sepharose-AT Chromatography of Different Molecular Weight Fractions of ATH and H-In order to further distinguish the sites within ATH heparin chains which were binding to immobilized AT, ATH was fractionated according to molecular weight by gel filtration on Sephadex G-200 under high ionic strength conditions. As a rule, ATH was gel-filtered under high salt conditions (2 M NaCl) to prevent possible binding of the AT moiety in one ATH molecule with the pentasaccharide on the heparin chain in another ATH molecule. However, no differences in size exclusion profiles (which would be indicative of complexes forming due to intermolecular ATH-ATH interactions) were observed under low ionic strength (0.15 M NaCl) conditions. SDS-PAGE analysis of the fractions from ATH material gel-filtered on Sephadex G-200 indicated that subpopulations of ATH molecules with discrete molecular weight ranges could be obtained across the peak (Fig. 3). ATH fractions with high molecular weight (first 2-9% of eluted material) or low molecular weight (last 2-9% of eluted material) were concentrated and designated as HMWATHF or LMWATHF, respectively (Fig. 4A). Because polydispersity of ATH results from variation in length of the heparin chains, HMWATHF and LMWATHF contained covalently linked heparin with greater and smaller numbers of saccharide units, respectively. Previous gel filtration analyses of the heparin chains from HM-WATHF and LMWATHF (isolated after protease treatment of the conjugate) indicated that the heparin moieties had Ͼ83 and Ͻ10 saccharide units, respectively. To further confirm the molecular weight range of heparin in the LMWATHF preparation, protease-treated LMWATHF was gel-filtered on a calibrated Bio-Gel P-6 column (Fig. 5). The peak containing heparin chains was pooled and, after concentration, subjected to fluorophore-assisted carbohydrate electrophoresis (FACE) analysis for determination of the chain length. The molecular weight for heparin prepared from the LMWATHF was observed to range from 12 to 4 saccharide units (Fig. 6). Sepharose-AT chromatography of HMWATHF gave trace amounts (3% of recovery) of unbound material, followed by low and high affinity bound peaks (Fig. 4B) that had significant anti-factor Xa activities (210 and 762 units/mg heparin, respectively (Table III)). Thus, the ATH fraction with longer heparin chains tended to have slightly improved binding to exogenous AT with somewhat increased catalytic activities compared with the parent unfractionated preparation. The LMWATHF was further fractionated on Sepharose-AT into a significant amount of unbound material (ϳ40% of recovery), a low affinity peak, and (relative to ATH and HMWATHF) a reduced amount of high affinity material (ϳ30% of recovery (Fig. 4C)). Interestingly, although nearly half of the LMWATHF did not have affinity for exogenous AT, this unbound material possessed significant ability to catalyze reaction of factor Xa and AT (190 units/mg heparin (Table III)). In contrast to the HMWATHF, catalytic activity of the high affinity peak of the LMWATHF was considerably reduced, tending toward the level of that for H (ANOVA for specific activity of LMWATHF versus H, p Ͼ 0.05). As a control, H with similar chain lengths to that within HMWATHF and LMWATHF were prepared by gel filtration (Fig. 7A). Chromatography of high molecular weight H (HMWH) on the AT column resulted in Ͻ10% in the unbound fraction and an ϳ1:3 ratio of low:high affinity peaks (Fig. 7B). Anti-factor Xa catalytic activities of unbound, low affinity, and high affinity HMWH peaks (0.5, 1.7, and 436 units/mg heparin, respectively (Table III)) were similar to the values for the corresponding H peaks (ANOVA, p Ͼ 0.05). In comparison, Sepharose-AT fractions of the HMWATHF had increased catalytic activities. The  Elution profiles (determined either by protein absorbance or protamine sulfate heparin mass assay) appeared as three peaks: unbound (peak 1), low affinity (peak 2), and high affinity (peak 3). Fractions comprising each peak were pooled, concentrated, and analyzed for heparin mass and activity (anti-factor Xa kit). Catalytic activities were determined as the ability to accelerate reaction of excess added AT with factor Xa (versus commercially available standard H). The units of activity were divided by the heparin mass to give specific activities (units/mg). Results are given as mean Ϯ S.E. (n Ն 2). See "Experimental Procedures" for details.

Column load
Anti-factor Xa catalytic activity in units/mg (proportion of each peak as a percent of total eluate given in parentheses) majority of LMWH molecules (ϳ75%) was unable to bind to AT and had no activity (Fig. 7C), whereas the small proportion of LMWH molecules with high AT affinity had significant activity (277 units/mg heparin (Table III)). Analysis of Anti-thrombin (Anti-IIa) Catalytic Activity-All peaks obtained from chromatographies on Sepharose-AT were analyzed for the ability to catalyze reaction of added AT with thrombin (anti-IIa). The relative anti-IIa catalytic activities of unbound, low affinity, and high affinity peaks eluted from Sepharose-AT were directly proportional to AT-binding strength (Table IV). Furthermore, relative specific activities of peaks from ATH, H, and AT ϩ H chromatographies were similar to those measured by the anti-factor Xa assays (Table II). Anti-IIa catalytic activities (Table IV) of the high affinity ma-terial (peak 3) of LMWATHF and particularly LMWH were greatly reduced compared with high affinity peaks of other species (in each case p Ͻ 0.05 (ANOVA)). This result would be expected given that thrombin requires longer chain heparin molecules in order to bridge both AT and the enzyme (3). The fact that the LMWATHF had any significant anti-IIa activity was interesting, given that the heparin chain length was not likely to bridge both AT and thrombin. Subtraction of the activity due to direct reaction of LMWATHF with thrombin (non-catalytic activity) gave a value of 106 units/mg for the high affinity material (peak 3).
Sepharose-AT Chromatography of Factor Xa-ATH and Thrombin-ATH Inhibitor Complexes-Covalent inhibitor complexes were formed by titration of ATH with either factor Xa or thrombin to ϳ100% equivalence, as shown by the detection of a small amount of remaining activity against chromogenic substrates. Treatment of the Xa-or thrombin-titrated ATH with heparinase, followed by SDS-PAGE, showed that Ͼ95% of the AT (as ATH) had been converted to either factor Xa-AT or thrombin-AT bands. Factor Xa-ATH and thrombin-ATH complexes were used to test the effect of linkage to a serine protease on the affinity of ATH for the immobilized AT. Sepharose-AT chromatographic profiles for factor Xa-ATH and thrombin-ATH were compared with those for ATH fractionated on the same column (Fig. 8). An increase in the unbound fraction and a corresponding small decrease in bound material was noted for the inhibitor complexes relative to that for ATH

FIG. 2. Analysis of chain length and Sepharose-AT binding affinity of heparin chains from covalent ATH.
Heparin chains released from ATH using protease (HЈ) were gel-filtered on Sephadex G-200 (2.6 (inner diameter) ϫ 49 cm long) using 2 M NaCl as irrigant (A). Gel filtration of standard heparin on the same column is shown for comparison. HЈ was chromatographed on Sepharose-AT (column and conditions the same as in Fig. 1) to give unbound, low, and high affinity peaks 1-3, respectively (B).

FIG. 3. Analysis of fractions from size exclusion chromatography of covalent ATH. ATH was chromatographed on a Sephadex
G-200 column (2.6 (inner diameter) ϫ 43 cm (long)), followed by SDS-PAGE of alternating fractions (3.9 ml). The gel was stained for heparin using Alcian blue, followed by silver for increased sensitivity. Alternating fraction number of the ATH peak increases (and molecular mass decreases) for lane numbers going from left to right and molecular weight on the gel decreases going from top to bottom. (39,28, and 12% as unbound material for factor Xa-ATH, thrombin-ATH, and free ATH chromatographies, respectively). There was, however, no significant change in the position of elution for either low affinity or high affinity peaks due to reaction with factor Xa or thrombin. Analysis of peaks for anti-factor Xa activity showed that for both factor Xa-ATH and thrombin-ATH, unbound (peak 1) and low affinity (peak 2) peaks had significantly decreased activities (Յ120 units/mg), whereas high affinity material (peak 3) had high activity (700 -1000 units/mg). Anti-factor Xa assay of preformed factor Xa-ATH or thrombin-ATH complexes showed that activities were decreased relative to that for free ATH (ϳ20% for factor Xa-ATH and ϳ5% for thrombin-ATH). Furthermore, reaction of factor Xa or thrombin with ATH in the presence of added AT resulted in a 20 -30% decreased formation of factor Xa-ATH or thrombin-ATH complexes.
Fluorescence Titrations of ATH and AT⅐H with NaCl-Intrinsic fluorescence of the protein in ATH or AT ϩ saturating H (2.34-fold molar excess to AT) was measured at increasing NaCl concentrations, and the values were corrected for any changes in the intrinsic fluorescence of control AT that was titrated with NaCl under the same conditions (Fig. 9). For both ATH and AT ϩ H, [NaCl] was inversely proportional to the AT intrinsic fluorescence induced by heparin pentasaccharide binding. However, significantly greater NaCl concentrations were required to reduce the intrinsic fluorescence intensity of ATH compared with that for AT ϩ H (Fig. 9). Material in pooled fractions containing either HMWATHF (first 9% of peak) or LMWATHF (last 9% of peak) was separately fractionated on Sepharose-AT (B and C, respectively). Chromatography (as in Fig. 1) gave unbound, low, and high affinity peaks 1-3, respectively.

FIG. 5. Chromatography of protease-treated low molecular weight fraction of covalent antithrombin-heparin complex on
Bio-Gel P-6. LMWATHF was incubated with catalytic amounts of a general protease (P-5147 from Sigma) at 37°C and chromatographed on a Bio-Gel P-6 column (1 (inner diameter) ϫ 49 cm long). Elution was with 0.15 M NaCl, and 1-ml fractions were collected (void volume ϭ fraction 11). Heparin chains (fractions [12][13][14][15][16][17][18][19][20] were well separated from complex-type Asn-linked glycans (fractions 26 -31) or free amino acids (fractions [35][36][37][38][39][40][41][42][43][44][45], as confirmed by control digests of uncomplexed antithrombin. Eluted material was detected by a refractive index detector and measurements given as refractive index units (RIU) relative to 0.15 M NaCl in the reference cell. whereas the latter half of the curve was more coincident with that of ATH (Fig. 9). The proportion of species containing heparin with either low or high non-covalent affinity for AT (as indicated by the fluorescence data for the LMWATHF) was consistent with the Sepharose-AT chromatographic data in which ϳ40% of the LMWATHF was unbound (low affinity for exogenous AT) and had no catalytic activity ( Fig. 4C and Table III).
Chromatography of ATH and H on Sepharose-Heparin-ATH, AT, and AT ϩ H were chromatographed on columns of Sepharose-heparin to determine whether heparin pentasaccharide-binding sites could interact with the AT moiety in covalent or non-covalent complexes of AT and heparin (Fig. 10). In contrast to the results from chromatography on Sepharose-AT, Ͼ95% of ATH was unable to bind to the immobilized heparin (Fig. 10A). Alternatively, application of AT to the heparin column gave 87% binding of the load (Fig. 10B). The small amount of AT that did not bind to the immobilized heparin (10 -12%) was not a result of column capacity because application of different loading amounts gave similar results. Fractionation of non-covalent mixtures of AT ϩ H on Sepharose-heparin caused dissociation of the AT⅐H complexes (Ͼ84% of AT molecules bound and ϳ90% of H molecules unbound (Fig. 10C)) reminiscent of chromatography on Sepharose-AT. Thus, AT in non-covalent complexes with H could interact with immobilized heparin, whereas AT in covalent ATH complexes could not. Evidence that the lack of affinity of AT in ATH for Sepharoseheparin was due to the covalent linkage to heparin was obtained by chromatography of a mixture of free AT with HЈ (heparin released from ATH by protease). Unlike ATH, AT in AT⅐HЈ complexes bound to Sepharose-heparin whereas the HЈ passed through unretarded (similar to that for AT⅐H (Fig. 10C)). DISCUSSION Inhibition of factor Xa or thrombin by AT is potentiated by H due, in part, to binding of the serpin to a pentasaccharide sequence on the GAG which, in turn, allosterically activates the inhibitor (4). After reaction of factor Xa/thrombin with AT⅐H, the enzyme-serpin complex dissociates from H leaving the GAG available for catalysis of another factor Xa/thrombin ϩ AT reaction (5). We have studied a highly active covalent complex of AT and heparin (ATH) to investigate the reaction steps involved in the turnover of heparin during reaction of factor Xa and thrombin with permanently stabilized AT⅐H.
ATH has been shown to exhibit potent catalytic activity in the reaction of AT with factor Xa and thrombin (11). This finding was surprising given that covalently linked AT and heparin were unable to completely dissociate after formation of inhibitor complexes by direct reaction with factor Xa or thrombin (11,12). One possible mechanism that might explain the catalytic activity of ATH was the presence of a second ATbinding pentasaccharide sequence on the covalently linked heparin chain that was separate from the one that activates the conjugate's own AT moiety. Although studies showed that ϳ30 -40% of ATH complexes contained 2 pentasaccharide units per molecule (14), specific catalytic activities of the conjugate were ϳ1.8 -2-fold greater than that of the H fraction with high affinity for AT (11). Thus, in order to investigate further the basis for the catalytic properties of ATH, experiments were performed to probe the accessibility of the pentasaccharide site that interacts with the covalently linked AT.
Deductions from results of interaction studies with ATH TABLE III Anti-factor Xa activities of high and low molecular weight fractions of covalent antithrombin-heparin complex and standard heparin chromatographed on Sepharose-AT Covalent antithrombin-heparin complex (ATH) or standard heparin (H) were gel filtered to give high molecular weight fractions of ATH (HMWATHF) and H (HMWH) or low molecular weight fractions of ATH (LMWATHF) and H (LMWH). The separate fractions were chromatographed on Sepharose-AT. Elution profiles (determined either by protein absorbance or protamine sulfate heparin mass assay) appeared as three peaks as follows: unbound (peak 1), or low (peak 2), and high affinity (peak 3) gradient eluted material. Each peak was concentrated and analyzed for heparin mass and activity (anti-factor Xa kit). Activities were determined as either non-catalytic (direct reaction alone with factor Xa) or catalytic (ability to accelerate reaction of excess added AT with factor Xa (versus commercial standard H)). Activity was divided by heparin mass to give specific activity (units/mg). Results are given as mean Ϯ SE (n Ն 2).

Column load
Anti-factor Xa catalytic activity in units/mg (proportion of each peak as a percent of total eluate given in parentheses)  (18) FIG. 7. Preparation of HMWH and LMWH, followed by chromatography on Sepharose-AT. H (5 mg) was gel-filtered on a 2.6 (inner diameter) ϫ 43-cm (long) column of Sephadex G-200 with 2 M NaCl as irrigant (A). Material in pooled fractions containing either HMWH (first 9% of peak) or LMWH (last 9% of peak) was separately chromatographed on Sepharose-AT (B and C, respectively). Chromatography (as in Fig. 1) gave unbound, low, and high affinity peaks 1-3, respectively. relied on precise determination of the structural components of the conjugate. Previously, we have analyzed the content of heparin in ATH by Alcian blue staining of SDS-PAGE gels of protease-treated ATH and compared the stain density with that of known amounts of standard H using laser densitometry (11). This methodology using Alcian blue staining for heparin quantitation has been validated previously by a number of investigators. We have analyzed a large molecular weight range (1000 -30,000) of heparin isolated from heparin starting material (Sigma) or other commercial LMWHs and HMWHs, and we found no significant difference in stain bound per mg of heparin loaded. In fact, we have found previously that the intensity of Alcian blue stain bound per mg of GAG is the same for heparin, heparan sulfate, dermatan sulfate, chondroitin 4-sulfate, and chondroitin 6-sulfate, in agreement with the work of Bartold and Page (28). Many other reports support the validity of cationic staining (Alcian blue and toluidine blue) for quantitation of GAGs with varying molecular weights (29 -32). Small fraction samples were analyzed for heparin using a protamine sulfate turbidimetric assay. This protamine test for heparin was employed because of its very high sensitivity (Ͻ1 g/ml could be detected) compared with other known methods. Also, data showing that similar protamine turbidimetric responses are given for a wide molecular weight range (300 -25,000) of Sigma heparin and other heparins have been reported previously with this method (23).
Rigorous analyses of ATH for protein and heparin content gave further verification that the conjugate contained, on the average, one heparin chain per AT molecule. The mole concentration of ATH solutions in terms of AT (calculated from amino acid analyses of acid hydrolysates and the known amino acid sequence) was divided into the number of moles of ATH heparin (number average molecular weight ϭ 16,900) determined from three different mass assay procedures. Analysis of all data resulted in the conclusion that the heparin:AT mole ratio was statistically consistent with that of a 1:1 complex. This outcome verified analyses carried out previously by different methods (Chan et al. (11) found the heparin:AT molar ratio in ATH to be 1.1). Given that ATH molecules contained one heparin chain per AT, the proportions of ATH molecules with different binding affinities could be easily compared by measuring the amount of AT (by absorbance) in each peak. Heparin mass analyses (although less sensitive) of pooled fractions with different affinities for Sepharose-AT confirmed this assumption.
Fractionation of ATH on immobilized AT resulted in Ͼ74% high affinity binding (Fig. 1A). Thus, because the vast majority of ATH molecules could form ATH⅐AT complexes but only a relative minority of the heparin chains in ATH contain two pentasaccharides (14), most of the ATH that possesses only one pentasaccharide was able to bind tightly to exogenous AT. Although the heparin component in ATH remains covalently attached to AT, added AT molecules are able to compete for binding to the pentasaccharide sequence that causes the AT moiety in ATH to be in an active conformation. Similar to covalent ATH, immobilized AT was able to compete for binding to H in AT⅐H complexes, resulting in displacement of the AT (Fig. 1C). Heparin chains from ATH (HЈ) were isolated after protease treatment of the conjugate. Chromatography of HЈ on Sepharose-AT showed that the vast majority of ATH heparin contained high affinity binding sites (Fig. 2), verifying that the GAG component of ATH has pentasaccharide sites that would be capable of binding to exogenous AT. Further analysis of peaks from the Sepharose-AT chromatographies revealed that binding affinity was directly proportional to the specific catalytic activity. For non-covalently linked heparin, only the material with high affinity binding (peak 3) likely contained heparin molecules with high specificity AT-binding sites (only high affinity peak 3 had significant anti-factor Xa activity (Table  II)). In the case of ATH, because the pentasaccharide that interacts with the AT of ATH might be sterically hindered (due to the covalently linked AT), lower affinity material (peak 2) might retain significant catalytic activity that would be exhibited in the anti-factor Xa assays once factor Xa-ATH is formed. In fact, significant catalytic activities were observed in ATH fractions with decreased AT affinity as evidence of this (Tables  II and III). ATH fractions with high AT affinity were ϳ1.53 times greater in anti-factor Xa activity than that of high affinity H material (Table II). Rosenberg et al. (15) have shown previously that the subfraction in commercial H that has two AT-binding sites per molecule has a greater specific activity than that for H with only 1 interaction site for AT (738 USP units/mg compared with 363 USP units/mg) due to greater pentasaccharide density along the chain. Comparison of our present results with those of Rosenberg et al. (15) indicates that the specific activities of high affinity ATH and H are in the range of that for 2 pentasaccharides and 1 pentasaccharide containing H chains, respectively. Closer inspection reveals, however, that the ratio of catalytic activity for 2 pentasaccharide heparins to that for 1 pentasaccharide heparin is significantly greater than the ratio of high affinity ATH activity to high affinity H activity (2.0 compared with 1.53). This would be expected, given that the AT-binding fraction of ATH contains significant amounts of 1 pentasaccharide heparin chains and the AT-binding fraction of H must contain some 2 pentasaccharide molecules.
An alternative hypothesis for the Sepharose-AT binding results was that covalently linked AT may be capable of intermolecular binding to the second (free) pentasaccharide in 2 pentasaccharide ATH molecules. Thus, in some cases, the immobilized AT might be simply dissociating ATH dimers. In an attempt to address this possibility, as well as to confirm the direct interaction of exogenous AT with the intramolecular pentasaccharide-binding site for the AT of ATH, the ATH fraction containing heparin chains that were Յ12 monosaccharides in length (Fig. 6) were isolated. Because ATH of this size (representing Յ5% of ATH preparations) cannot contain 2 pentasaccharides, no excess (free) AT-binding sites are available. Sepharose-AT chromatograms of LMWATHF showed that ϳ50% bound to AT (Fig. 4C), which gave strong evidence for the direct competition of exogenous AT for the intramolecular pentasaccharide-binding site of the ATH. Furthermore, the specific catalytic activity of heparin chains in LMWATHF complexes that bound to AT (560 units/mg (peak 3, Table III)) was closer to that for high affinity H (463 units/mg (Table II)) than that for  (ATH) and heparin (H) fractions chromatographed on Sepharose-AT ATH or H were gel filtered to give high (HMWATHF and HMWH) or low (LMWATHF and LMWH) molecular weight fractions. Sepharose-AT chromatography gave unbound (peak 1), low (peak 2), and high (peak 3) affinity material. Peaks were tested for heparin mass and activity (catalysis of thrombin inhibition by AT (anti-IIa kit)). Activity/ mass gave specific activity (units/mg; mean Ϯ SE (n Ն 2)).

Column load
Anti-IIa catalytic activity in units/mg (proportion of each peak as a percent of total eluate given in parentheses) the corresponding AT-binding peak of ATH (708 units/mg (Table II)), which is in agreement with the fact that most high affinity H molecules have only ϳ1 pentasaccharide. However, a significant proportion of the LMWATHF was unable to bind to Sepharose-AT. It is possible that the covalent linkage of heparin to AT in ATH may sometimes occur at more internal lysyl residues or the aldose linkage residue may be located at the start of the actual pentasaccharide sequence. Steric hindrance in the conjugates arising from either of these linkage situations might be too difficult for the Sepharose-AT to overcome. Chromatography of HMWATHF leads to almost complete binding of the load (Fig. 4B). Previously, it has been found that multipentasaccharide H tends to occur on long chain molecules (15,34). Analyses of the specific activity of the HMWATHF high affinity peak gave results (762 units/mg (Table III)) suggesting that a high proportion of the conjugates contain 2 pentasaccharide chains. Thus, a significant amount of the interactions between the HMWATHF and immobilized-AT may have occurred through a second pentasaccharide on the covalently linked heparin chain. Control experiments using low and high molecular weight fractions of H gave relatively similar results on Sepharose-AT to those for LMWATHF and HMWATHF. Whereas Ͻ30% of LMWH bound to immobilized AT (Fig. 7C), Ͼ70% of HMWH bound to the column (Fig. 7B). Interestingly, the specific activity data given in terms of units/mg illustrate a fascinating property of HMWATHF (or HMWH) chains that have two pentasaccharides compared with LMWATHF (or LMWH) chains with one pentasaccharide. Logically, a heparin molecule that contains one pentasaccharide should have the same activity in units/mg as another heparin molecule that has two pentasaccharides but is twice the chain length. However, Rosenberg et al. (15) showed that although 20,000 molecular weight heparin with two pentasaccharide AT-binding sites had a specific activity of 738 units/mg, 7000 molecular weight heparin with one pentasaccharide had a specific activity of 363 units/mg. Thus, in terms of activity/heparin molecule, Rosenberg and co-worker's (15) 2 pentasaccharide heparin was 14.8 units/nmol and the 1 pentasaccharide heparin was 2.54 units/nmol! Our results were consistent with this finding, in that high affinity fractions of the HMWATHF and HMWH both had higher units/mg than their low molecular weight counterparts (Table III). Rosenberg et al. (15) recognized this problem and proposed a cooperative mechanism for two pentasaccharide heparin chains, whereby the two AT-binding sites operated synergistically by giving a reduced off-rate. Previously, we have proposed the same rationale to explain why AT selects for enrichment of two pentasaccharide heparin chains during ATH formation (mean free distance of intramolecular diffusion between pentasaccharides in 2-pentasaccharide heparin molecules is less than that for intermolecular diffusion).
As expected, catalytic anti-IIa activity of high affinity fractions of ATH chromatographed on Sepharose-AT was severalfold higher than that for H or AT ϩ H (Table IV). Previously, it has been shown that heparin chains of Ͼ18 saccharides in length are required to bridge both AT and thrombin during catalysis of thrombin inhibition (35). Thus, it was surprising that LMWATHF (with heparin chains Յ12 saccharides in length (Fig. 6)) possessed significant (albeit reduced) anti-IIa catalytic activity. We do not have a definitive explanation for this result. One possibility is that the heparin chains in LM-WATHF, although short in length, have a higher negative charge density that may assist in greater electrostatic attraction to thrombin.
Further characteristics of the ATH catalytic mechanism FIG. 8. Chromatography of inhibitor complexes of covalent ATH on Sepharose-AT. Covalent inhibitor complexes of ATH with either factor Xa (factor Xa-ATH) or thrombin (thrombin-ATH) were prepared by titration of ATH to equivalence with the appropriate enzyme. Resultant inhibitor complexes (0.25 mg in terms of AT) were loaded onto 1.5-ml columns of Sepharose-AT (pre-equilibrated with 0.15 M NaCl in 0.01 M phosphate buffer, pH 7.3). Unbound material (peak 1) was washed off with 40 ml of 0.15 M NaCl, followed by elution of low affinity (peak 2) and high affinity (peak 3) material with a linear gradient (25 ml of buffered 0.15 M NaCl in the mixing chamber and 25 ml of buffered 2 M NaCl as limit solution) and 30 ml of 2 M NaCl. Fractions (1.3 ml) were collected, and the percent of total eluate recovered from each chromatography was calculated. were delineated. The effect of ATH reaction with factor Xa or thrombin on binding to Sepharose-AT was investigated in order to understand the capability of ATH-inhibitor complexes to catalyze further inhibition. Although Ͼ60% of either enzyme-ATH or ATH alone was bound by immobilized AT (Fig. 8), significantly more unbound material was recovered in the case of thrombin-ATH (28%) and, particularly, factor Xa-ATH (39%) inhibitor complexes compared with ATH (18%). Thus, complexation with factor Xa or thrombin may cause added steric hindrance of the ATH pentasaccharide toward exogenous AT. This hypothesis was confirmed by the fact that enzyme-ATH complexes (particularly factor Xa-ATH) had reduced anti-factor Xa activities compared with that of free ATH. Conversely, the presence of a vast excess of added AT inhibited the formation of enzyme-ATH inhibitor complexes, presumably by binding of the exogenous AT to the pentasaccharide site occupied by the AT in ATH. Increased physical obstruction by bound factor Xa or thrombin may be more critical in the case of molecules in LMWATHF that have smaller chain lengths for the initial electrostatic attraction to the immobilized AT (compare with Fig. 4, B and C). These latter results for ATH of varying molecular weights or in the form of inhibitor complexes led to studies of the relative affinity between the pentasaccharide and AT moieties. Direct determination of the binding of AT and heparin in ATH was accomplished by measuring the loss of intrinsic AT fluorescence when activating heparin is displaced (14,36). Greater [NaCl] was required for 50% reversal of the heparin-induced intrinsic fluorescence in ATH (Fig. 9), which is compliant with the fact that because AT and heparin are covalently linked, complete dissociation is prohibited (regardless of ionic strength, AT would always be in close proximity to the heparin-binding site). Fluorescence titrations of the LM-WATHF with NaCl gave a complicated profile. The fluorescence of the LMWATHF decreased rapidly with small additions of NaCl, followed by a more gradual reduction in fluorescence (similar to ATH) at higher ionic strengths (Fig. 9). Thus, a portion of the molecules in the LMWATHF contain weakly interacting protein and GAG, whereas the remainder have AT and heparin which have strong intramolecular interactions. These data fit with the heterogeneity in affinity of different subfractions of LMWATHF for immobilized AT in that some heparin chains of the LMWATHF are more easily bound by exogenous AT, possibly due to decreased intramolecular association of AT and heparin. Binding of exogenous AT to molecules of LMWATHF that have strong intramolecular AT-pentasaccharide interactions may require longer GAG chains for initial intermolecular-electrostatic attractions. The molecules of the LMWATHF that did not bind to AT represented Ͻ3% of all ATH molecules. Finally, probing with exogenous heparin showed that although the majority of AT bound to immobilized heparin, almost all ATH passed freely through the Sepharoseheparin column (Fig. 10). Lack of ATH binding to immobilized heparin was probably due to strong negative charge repulsion between ATH GAG chains and heparin on the column. Because AT and heparin in ATH cannot dissociate, the heparin in ATH would likely be in too close a proximity (on a charge basis) for the immobilized heparin to bind to the AT moiety of ATH. The likelihood that heparin was unable to access the AT in ATH because of electrostatic effects was further evidenced by the binding of Sepharose-heparin to AT in AT⅐H complexes via dissociation of the H (Fig. 10C). Furthermore, Sepharoseheparin chromatography of a mixture of AT ϩ the heparin chains released from ATH by protease (HЈ) resulted in binding of the AT and a lack of affinity for the HЈ. Finally, absence of ATH affinity for Sepharose-heparin was a further confirmation that intermolecular binding of ATH to other ATH molecules does not occur.
Originally, it was expected that the pentasaccharide bound by the AT in ATH might be hindered from other molecules because the AT in ATH is also covalently linked and cannot completely diffuse away from the ATH heparin chain (which can occur with AT that is only bound to a heparin molecule via the pentasaccharide (not covalently linked)). Although fluorescence data showed a resistance of the AT in ATH to be displaced from its pentasaccharide-binding site (presumably because the covalent linkage still keeps the AT tethered to the heparin), displacement from these non-covalent interactions by NaCl does occur (Fig. 9). Also, experiments with the heparin produced from protease-treated ATH (HЈ) verified that ATH heparin chains, on their own, interact with immobilized AT and immobilized H in a similar way to that of standard H (pentasaccharide units of ATH heparin are the same as that in standard H).
Results from the study of ATH catalytic mechanisms have several implications. Because exogenous AT can bind to co- ϩ H (0.5 mg) (C) were chromatographed on 1 (inner diameter) ϫ 12-cm (long) columns of Sepharose-heparin using elution conditions similar to those for Sepharose-AT given in Fig. 1. Eluted material was detected by A 280 (protein) or protamine sulfate assay (heparin (A 470 )) and appeared as either unbound, low affinity or high affinity peaks 1-3, respectively. In the case of AT ϩ H chromatographies, unbound (peak 1) AT (q) and H (ƒ) materials chromatographed separately. valently linked AT-heparin, interchange of AT in AT⅐H complexes with free AT may occur through a displacement model. Alternatively, a mechanism can be envisaged in which free AT electrostatically attracts the GAG in ATH, from the side opposite to that of the covalently linked AT, and causes a rotation of the heparin about its helical axis so that the pentasaccharide is now in the correct orientation for ionic/hydrogen bonding to the exogenous AT. Also, because the off-rate of AT bound to the pentasaccharide is relatively rapid (37), the AT in ATH may frequently dissociate from its non-covalent interaction with the pentasaccharide on the covalently linked heparin chain so that exogenous AT may bind and be activated. Notwithstanding which model describes transition states involved in AT complexation of ATH, it is clear that long distance dissociation of AT⅐H (which is impossible for ATH) is unnecessary in order for exchange with free AT to occur. Free AT may be able to access (or share) the pentasaccharide in AT⅐H complexes. Previous work (38 -44) has mapped out some of the AT residues involved in binding to the heparin pentasaccharide. Recent experiments using an AT mutant have shown that Arg 129 binds the heparin pentasaccharide cooperatively with other residues leading to an induced fit to the heparin molecule that locks the AT into its activated state (45). Binding of AT to the pentasaccharide occurs by interaction with the first three saccharide residues from the non-aldose terminus (46), which causes charge neutralization and helix D elongation leading to the induced fit, which is stabilized by interaction of the remaining two saccharide residues to Arg 46 and Arg 47 (47). Thus, initial stages of exogenous AT binding to ATH may involve reversal of the charge interactions between the aldose disaccharide unit of the pentasaccharide and Arg residues in the ATH molecule. Further charge attraction of the approaching AT to the ATH heparin moiety would then allow for the proper alignment of groups in the GAG for tight binding to the added AT. Studies with model compounds have shown that addition of saccharides (2 units) at the pentasaccharide aldose terminus gives a possible shift in positioning on the AT to an extended heparinbinding site at Ն0.2 ionic strength and a 2-fold increase in affinity (33). These findings may, in part, explain the strong ATH intramolecular pentasaccharide binding observed by fluorescence titration if the pentasaccharide in ATH is a few saccharide units away from the covalent linkage point on the AT. Sequence determinations are in progress to ascertain the specific AT lysyl group(s) in ATH that is conjugated to the heparin chain. Once the specific covalent heparin-linkage position(s) in the ATH protein sequence is established, solution phase models of heparin-activated AT can be devised.