Aspartic acid residues 72 and 75 and tyrosine-sulfate 73 of heparin cofactor II promote intramolecular interactions during glycosaminoglycan binding and thrombin inhibition.

We used site-directed mutagenesis to investigate the role of Glu(69), Asp(70), Asp(71), Asp(72), Tyr-sulfate(73), and Asp(75) in the second acidic region (AR2) of the serpin heparin cofactor II (HCII) during formation of the thrombin.HCII complex with and without glycosaminoglycans. E69Q/D70N/D71N recombinant (r)HCII, D72N/Y73F/D75N rHCII, and E69Q/D70N/D71N/D72N/Y73F/D75N rHCII were prepared to localize acidic residues important for thrombin inhibition. Interestingly, D72N/Y73F/D75N rHCII had significantly enhanced thrombin inhibition without glycosaminoglycan (4-fold greater) and with heparin (6-fold greater), showing maximal activity at 2 microg/ml heparin compared with wild-type recombinant HCII (wt-rHCII) with maximal activity at 20 microg/ml heparin. The other rHCII mutants had lesser-enhanced activities, but they all eluted from heparin-Sepharose at significantly higher ionic strengths compared with wt-rHCII. Neutralizing and reversing the charge of Asp(72), Tyr-sulfate(73), and Asp(75) were done to characterize their individual contribution to HCII activity. Only Y73K rHCII and D75K rHCII have significantly increased heparin cofactor activity compared with wt-rHCII; however, all of the individual rHCII mutants required substantially less glycosaminoglycan at maximal inhibition than did wt-rHCII. Inhibition of either alpha-thrombin/hirugen or gamma(T)-thrombin (both with an altered anion-binding exosite-1) by the AR2 rHCII mutants was similar to wt-rHCII. D72N/Y73F/D75N rHCII and D75K rHCII were significantly more active than wt-rHCII in a plasma-based thrombin inhibition assay with glycosaminoglycans. These results indicate that improved thrombin inhibition in the AR2 HCII mutants is mediated by enhanced interactions between the acidic domain and anion-binding exosite-1 of thrombin and that AR2 may be a "molecular rheostat" to promote thrombin inhibition in the presence of glycosaminoglycans.

Heparin cofactor II has several features for thrombin inhibition and specificity that render it novel among heparin-binding serpins (24 -26). Heparin cofactor II has an atypical Leu 444 reactive center residue, whereas the more typical Arg is found in thrombin inhibitors like ATIII or protein C inhibitor (10,12,13). The inhibition of thrombin by HCII is enhanced by both heparin and dermatan sulfate, and the glycosaminoglycan binding site is primarily contained within the D-helix region of HCII (15,(27)(28)(29)(30)(31)(32)(33)(34). Although HCII is ϳ30% identical in sequence to antithrombin and other serpins, it has a unique N-terminal extension of ϳ80 residues that contains a tandem repeat of negatively charged acidic residues (9 Asp, 5 Glu, and 2 Tyrsulfate) (25,26). Interestingly, the acidic domain of HCII is homologous to the C terminus of the leech anticoagulant protein hirudin, which is a potent thrombin inhibitor. Within the acidic domain of HCII, the two regions are designated acidic region 1 (AR1) from residues 49 -62 ( 49 DWIPEGEEDD-DYLD 62 ) and acidic region 2 (AR2) from residues 63-75 ( 63 LE-KIFSEDDDYID 75 ). There is substantial evidence that the acidic domain of HCII is involved in facilitating thrombin inhibition, especially in the presence of glycosaminoglycans (27,29,30). Ragg et al. (29,30) initially examined the role of the acidic domain in HCII by either deleting or neutralizing various acidic amino acids and found that AR1 and AR2 have separate functions for thrombin inhibition. Van Deerlin and Tollefsen (27) demonstrated that, by deleting portions of the acidic domain (AR1 and AR2), HCII was a less effective inhibitor of thrombin; interestingly, these mutants also had substantially enhanced binding to heparin compared with wildtype (wt) recombinant (r)-HCII. These studies suggested that the acidic domain of HCII both is involved in and is imperative for effective thrombin inhibition. The two current models to describe the mechanism for glycosaminoglycan-accelerated inhibition of thrombin by HCII are the "double-bridge" and the "allosteric" mechanisms (31,35). In both models it is assumed that the acidic domain position is altered following glycosaminoglycan binding to the D-helix region of HCII, thereby allowing the N-terminal acidic domain to bind thrombin anion-binding exosite-1 (exosite-1) (27,28,31,33,34,35,37). This paper investigates the role of six AR2 acidic residues in HCII during thrombin inhibition in the absence and presence of glycosaminoglycans and with thrombin derivatives with and without a functional exosite-1. The results indicate that Asp 72 , Tyr-sulfate 73 , and Asp 75 promote intramolecular interactions of the acidic domain with HCII in the absence of glycosaminoglycans, and altering the charge of these acidic residues substantially changes glycosaminoglycan-dependent activities of HCII.

EXPERIMENTAL PROCEDURES
Materials-Human plasma ␣-thrombin and HCII were prepared and purified in our laboratory as previously described (38,39). We prepared human ␥ T -thrombin by limited proteolysis of ␣-thrombin with trypsin as previously described (28). Monoclonal antibody 2-4-34 to purified human plasma HCII was made in our laboratory using standard techniques. Human antithrombin-deficient plasma (catalog no. 203) and normal hemostasis reference plasma (catalog no. 258N) were purchased from America Diagnostica (Greenwich, CT). Heparin was from Diosynth (Oss, The Netherlands). Dermatan sulfate was from Calbiochem (La Jolla, CA) and was nitrous acid-treated to remove contaminating heparin/heparan sulfate (40). Hi-Trap Heparin-Sepharose was from Amersham Biosciences, Inc. (Piscataway, NJ), and Q-Sepharose was from Sigma Chemical Co. Tosyl-Gly-Pro-Arg-4-nitroanilide acetate (tosGly-Pro-Arg-NA, Chromozym TH) was from Roche Molecular Biochemicals (Indianapolis, IN), and succinyl-Ala-Ala-Pro-Phe-p-nitroanilide (sucAla-Ala-Pro-Phe-NA) was from Sigma. Hirugen and the reverse peptide to HCII's acidic domain were synthesized and purified by the University of North Carolina at Chapel Hill Peptide Synthesis and Protein Core Facility.
Protein Expression and Purification-Each mutant rHCII and wt rHCII was cotransfected with linearized BaculoGold (BD PharMingen) Autographa californica nuclear polyhedrosis virus (AcNPV) into Spodoptera frugiperda (Sf9, Invitrogen) insect cells using established protocols in our laboratory (41). We expressed rHCII in HighFive insect cells (Invitrogen, Carlsbad, CA) maintained at 25°C in Excel 405 medium (JRH Biosciences, Lenexa, KS) as previously detailed. Three days post-baculovirus transfection, the media was collected and centrifuged to remove cell debris. The protein was purified essentially as described previously (41). Protein was then dialyzed against 20 mM HEPES, pH 7.4, 150 mM NaCl, 0.1% PEG8000 and 0.02% NaN 3 . Protein was aliquoted and stored at Ϫ80°C. The production of rHCII was verified by immunoblot analysis of whole cell lysates using a monoclonal antibody to HCII (mAb 2-4-34). The concentration of rHCII was determined by enzyme-linked immunosorbent assay using purified plasma HCII as the standard as described previously (32). We obtained ϳ70 g of recombinant protein of each mutant from 100 ml of HighFive cells in a 200-ml shaker flask infected with the various recombinant viral stocks. Immunoblot and SDS-PAGE confirmed that each mutant rHCII had the same electrophoretic mobility as wt rHCII (data not shown).
In the presence of 10 g/ml glycosaminoglycan, either hirugen or a control peptide corresponding to the reverse sequence of the HCII acidic domain (residues 47-61) at 20 M and rHCII (10 nM) was incubated with 0.5 nM thrombin in the presence of 1 mg/ml Polybrene and 2 mg/ml BSA in HNPN, pH 7.4. In the presence of either heparin (5 or 20 g/ml) or dermatan sulfate (20 or 200 g/ml), 0.5 nM ␥ T -thrombin was incubated with 10 nM rHCII in the presence of 2 mg/ml BSA in HNPN, pH 7.4.
Association time courses ranged from 1 to 240 min for ␣-thrombin, ␥ T -thrombin, and chymotrypsin depending upon the assay conditions. Residual thrombin activity was measured with 150 M tosGly-Pro-Arg-NA and 1 mg/ml Polybrene in the absence of glycosaminoglycan or 2 mg/ml Polybrene in the presence of heparin or dermatan sulfate. Residual chymotrypsin activity was measured with 150 M sucAla-Ala-Pro-Phe-NA. Substrate cleavage was measured by color development at 405 nm on a V max Kinetic Microplate Reader (Molecular Devices). Assays were performed at least in triplicate on two or more recombinant protein preparations. Second order inhibition rate constants were calculated as described (32,42,43).
Heparin-Sepharose Affinity Chromatography-We determined the relative heparin affinities by using 1-3 g of rHCII protein diluted in 20 mM Hepes, 50 mM NaCl, 0.1% PEG, 0.05% NaN 3 , pH 7.4. We ran each protein on a 1-ml Hi-Trap heparin-Sepharose column equilibrated in 20 mM HEPES, 0.1% PEG, 50 mM NaCl, 0.05% NaN 3 , pH 7.4, using an fast-protein liquid chromatography system (Amersham Biosciences, Inc.). After the sample was loaded onto the column, protein was eluted with a 40-ml gradient of 20 mM Hepes, pH 7.4, from 50 to 1500 mM NaCl, and 0.5-ml fractions were collected. For each fraction, 100 l was aliquoted onto a 96-well microtiter plate and an enzyme-linked immunosorbent assay was performed. The peak elution ionic strength was determined by plotting 405-nm color development and NaCl concentration against the fraction number. All samples were run at least in triplicate using two or more protein preparations.
Human Plasma Assays-The assay used to evaluate rHCII in plasma was designed based on previously published methods (32,44,45). We performed this assay using normal hemostasis reference plasma (REF), human antithrombin-deficient plasma (DEF), or a 50:50 mixture of both types of plasma (REF/DEF) (American Diagnostica) at room temperature in 96-well microtiter plates previously coated with 2 mg/ml BSA (32). We incubated 10 nM wt rHCII or mutant rHCII with either 1 g/ml heparin or 5 g/ml dermatan sulfate, in the presence of each type of plasma (diluted 1:100). Thrombin inhibition was started when 1 nM thrombin was added to each well. After 15 s, 150 M tosGly-Pro-Arg-NA was added, and the residual thrombin activity was measured by color development at 405 nm on a kinetic microplate reader.
Statistical Analysis-The statistical significance of the data was evaluated using Tukey's Student t test; p values Յ 0.05 were considered significant.

Inhibition of ␣-Thrombin, ␥ T -Thrombin, and
Chymotrypsin by the AR2 rHCII Derivatives in the Absence of Glycosaminoglycans-Wild-type rHCII inhibits ␣-thrombin at a rate of ϳ1 ϫ 10 4 M Ϫ1 min Ϫ1 in the absence of glycosaminoglycans, whereas E69Q/D70N/D71N rHCII, D72N/Y73F/D75N rHCII, and QNNNFN rHCII inhibit ␣-thrombin 1.2-, 4.2-, and 3.2-fold faster than wt-rHCII, respectively (Table I). The results with these "shotgun" HCII mutants suggest that neutralizing Asp 72 , Asp 75 , and Tyr-sulfate 73 had the biggest influence on thrombin inhibition. Comparing the effect of these individual residues in HCII individually for thrombin inhibition (using neutral and reverse-charge mutations), we found only D75K had a significantly increased rate of ␣-thrombin inhibition (2.6-fold) when compared with wt-rHCII (Table I). ␥ T -Thrombin (␣-thrombin with an altered exosite-1) inhibition by wt-rHCII and all of the AR2 rHCII derivatives was ϳ2-3 ϫ 10 3 M Ϫ1 min Ϫ1 (Table I), which further demonstrates the importance of exosite-1 in thrombin during HCII inhibition. Chymotrypsin inhibition by HCII requires neither the acidic domain nor glycosaminoglycan binding, and we found no significant difference for any of the AR2 rHCII derivatives compared with wt-rHCII (Table I).
These results indicate that altering the charge of AR2 in HCII creates a better ␣-thrombin inhibitor in the absence of glycosaminoglycans, and that Asp 72 , Tyr-sulfate 73 , and Asp 75 contribute the most to this effect.
Glycosaminoglycan-accelerated ␣-Thrombin Inhibition and Heparin-Sepharose Binding by the AR2 rHCII Derivatives-With heparin, each of the shotgun rHCII derivatives inhibit ␣-thrombin 3-to 6-fold faster than wt-rHCII (Fig. 1, top panel). 2 Interestingly, less heparin was required for each rHCII mutant to reach maximal thrombin inhibition (1-2 g/ml) compared with wt-rHCII (ϳ20 g/ml) (Table II). In the presence of dermatan sulfate, wt-rHCII inhibits thrombin at a rate of 3.3 ϫ 10 8 M Ϫ1 min Ϫ1 (Fig. 1, bottom panel). D72N/Y73F/D75N rHCII inhibits thrombin at a rate 3.3 times faster than that of wt-rHCII whereas E69Q/D70N/D71N rHCII and QNNNFN rHCII inhibit thrombin similar to wt-rHCII ( Fig. 1, lower panel and Table II). However, the amount of dermatan sulfate needed for maximal thrombin inhibitory activity is also significantly decreased for the shotgun AR2 rHCIIs (5-20 g/ml) as compared with wt-rHCII (100 -200 g/ml) ( Table II). The individual AR2 rHCII derivatives of Asp 72 , Tyr-sulfate 73 , and Asp 75 all required less heparin and dermatan sulfate to reach maximal thrombin inhibition compared with wt-rHCII (Fig. 2); however, only Y73K rHCII and D75K rHCII have significantly increased activity in the presence of heparin whereas none have significantly increased activity in the presence of dermatan sulfate (Table II and Fig. 2). Because the individual Asp 72 , Tyr-sulfate 73 , and Asp 75 rHCII mutants do not have increased glycosaminoglycan-accelerated thrombin inhibitory activity to the same extent, as does D72N/Y73F/D75N rHCII, this implies an additive influence of these AR2 residues to accelerate the HCIIglycosaminoglycan thrombin inhibition reaction.
Because the AR2 rHCII derivatives required substantially less glycosaminoglycan to reach maximal thrombin inhibition, we next addressed whether this implied a more accessible glycosaminoglycan-binding site. We used NaCl gradient elution from heparin-Sepharose to determine whether the AR2 rHCII mutations have altered apparent affinity for immobilized heparin (Table II). We found that the AR2 rHCII deriva- 2 We determined the stoichiometry of inhibition values (SI; the number of serpin molecules consumed before an inactivated serpin⅐protease complex is formed) for the shotgun rHCII mutants compared to wt-rHCII in the presence of heparin and dermatan sulfate. SI values for wt-rHCII, E69Q/D70N/D71N rHCII, D72N/Y73F/D75N rHCII, and QNNNFN rHCII ranged from 1.9 to 2.1 (similar to values reported previously (42)). This suggests that the kinetics of inhibition for these rHCII mutants were very similar to wt-rHCII.  tives (except D75N) required more NaCl to elute from heparin-Sepharose compared with wt-rHCII, ranging from 525-550 mM NaCl for wt-rHCII and D75N rHCII to 650 -925 mM NaCl for the remainder of the AR2 rHCII derivatives. Collectively, the activity data and the heparin-Sepharose elution results suggest that altering the charge of AR2 allows heparin to bind with higher affinity to the D-helix region of HCII.
␣-Thrombin/Hirugen and ␥ T -Thrombin Inhibition by the AR2 rHCII Derivatives-We further studied the role of thrombin exosite-1 during AR2 rHCII inhibition (with heparin or dermatan sulfate). Consistent with past studies, hirugen drastically reduced the rate of ␣-thrombin inhibition with heparin or dermatan sulfate by both wt-rHCII and all of the AR2 rHCII derivatives (Fig. 3, shown for dermatan sulfate in the top panel). Although the inhibition rates for ␥ T -thrombin are much reduced compared with ␣-thrombin, E69Q/D70N/D71N rHCII, D72N/Y73F/D75N rHCII, and D75K rHCII have slightly increased rates of ␥ T -thrombin inhibition as compared with wt-rHCII with heparin or dermatan sulfate (Fig. 3, shown for dermatan sulfate in the bottom panel). Overall, these results imply that the process of thrombin recognition by all of the rHCII's is similar whether using hirugen to sterically hinder access to thrombin exosite-1 or ␥ T -thrombin where the ␤-loop of exosite-1 is absent.
Physiological-based Thrombin Inhibition Studies by the ARII rHCII Derivatives-Our results presented so far show that some AR2 rHCII mutants are more active than wt-rHCII. We next assessed the potential of three AR2 rHCII derivatives a Values in parentheses are the ratio calculated as mutant rHCII/wt-rHCII. b p Յ 0.05 compared with wt-rHCII.

FIG. 2. Inhibition of ␣-thrombin by wild-type and individual AR2 rHCIIs in the presence of heparin and dermatan sulfate. ␣-Thrombin was incubated with rHCII in the presence of increasing amounts of heparin (top panels) or dermatan sulfate (bottom panels).
For all graphs, wild-type rHCII is indicated by ‚, in top and bottom left panels, the neutral AR2 residues are D72N (), Y73F (q), and D75N (f); in the top and bottom right panels, the basic AR2 residues are D72K (), Y73K (q), and D75K (f). The data represent an average of at least three separate protein preparations performed in triplicate.
(E69Q/D70N/D71N rHCII, D72N/Y73F/D75N rHCII, and D75K rHCII) to inhibit ␣-thrombin with a more physiologically relevant assay setting (Table III). With heparin or dermatan sulfate, E69Q/D70N/D71N rHCII and D72N/Y73F/D75N rHCII both exhibited a 2-to 3-fold greater thrombin inhibitory rate than wt-rHCII in the presence of all three types of plasma (normal reference, antithrombin-deficient, and a 50:50 mix to mimic a heterozygous antithrombin deficiency), whereas the inhibitory activity of D75K rHCII was slightly less than the other rHCII mutants (Table III). These results show that the AR2 rHCII mutants are potent thrombin inhibitors in the presence of glycosaminoglycans and plasma and imply that they may have potential as therapeutic anticoagulants. DISCUSSION Although HCII is ϳ30% identical to ATIII and other serpins, it contains an N-terminal extension of ϳ80 residues that is found in no other serpin (10,14,25,26). The N terminus contains a tandem repeat of two acidic stretches that have the following sequences: Asp 49 -Trp 50 -Ile-Pro-Glu-Gly-Glu 55 -Glu-Asp-Asp-Asp-Tyr 60 -Leu-Asp 62 and Leu 63 -Glu-Lys 65 -Ile-Phe-Ser-Glu-Asp 70 -Asp-Asp-Tyr-Ile-Asp 75 . Metabolic labeling of HepG2 cells demonstrated that the Tyr residues in each acidic repeat are sulfated (46). These acidic repeat elements are homologous to a thrombin-binding region of hirudin, the extremely potent thrombin inhibitor from leech salivary glands. HCII deletion mutants lacking either the first (⌬67-rHCII) or both (⌬74-rHCII) acidic domains were found to have greatly increased affinity for heparin as assessed by salt elution from heparin-Sepharose (27). This observation has led to the hypothesis that the D-helix of HCII is part of an "internal binding site" for the acidic domain. Because the nature of heparin-HCII interaction is mostly anionic, the increased heparin affinity could also be due to the removal of negative charges concentrated in the acidic domain. The goal of this study was to study the structure-activity relationships of the second acidic region (AR2) in HCII using site-directed mutagenesis. Our working hypothesis was that, by altering the charge of AR2 acidic residues in HCII, we would increase both progressive antithrombin activity (inhibition in the absence of added glycosaminoglycans) and heparin-and dermatan sulfate-cofactor activities (thrombin inhibition in the presence of added glycosaminoglycans).
␣-Thrombin inhibition rates of two of the three shotgun AR2 rHCII mutants in the absence of glycosaminoglycans were significantly increased suggesting that the largest change in activity was influenced by the final three acidic residues (Asp 72 , Tyr-sulfate 73 , and Asp 75 ). There was an additive effect of these acidic residues, as only D75K (when residues were individually mutated) had a substantially increased rate compared with wt-rHCII yet this inhibition rate was still lower when compared with D72N/Y73F/D75N rHCII. The increased inhibition activity appears to be caused by an enhanced interaction between the HCII acidic domain and exosite-1 of thrombin, because inhibition rates of both ␥ T -thrombin and ␣-thrombin in the presence of hirugen were all reduced to rates similar to wt-rHCII. Finally, the comparable inhibition rates of wt-rHCII and the AR2 rHCII mutants with chymotrypsin (which requires neither the HCII acidic domain nor glycosaminoglycans for inhibition) indicate that the reactive site loop has not been altered to an "activated" conformation. The results further implicate a role for the acidic domain due to altered HCII intramolecular interactions that increased activity to ␣-thrombin.
We initially hypothesized that the AR2 rHCII mutants would have significant changes related both to glycosaminoglycan binding by HCII and to glycosaminoglycan-mediated inhibitory activity of HCII with thrombin. This hypothesis is due to the unique role postulated for AR2 to be an intramolecular mimic of a glycosaminoglycan (27,29,30). The majority of the AR2 rHCII mutations did affect both heparin-and dermatan sulfate-accelerated thrombin inhibition. Interestingly, the glycosaminoglycan-dependent inhibition curves for most of the AR2 rHCII mutants were "left-shifted" when compared with wt-rHCII. The optimal glycosaminoglycan concentration, which is the concentration corresponding to the maximum inhibition rate, is expected to be a function of the affinity of HCII for glycosaminoglycan because ␣-thrombin was used for all experiments and all other variables were constant. We found that AR2 mutants, like D72N/Y73F/D75N rHCII and Y73K rHCII, had maximum heparin and dermatan sulfate optima clearly reduced from wt-rHCII, in some cases as much as 10-to 20-fold less glycosaminoglycan was required for rates that were increased 2-to 6-fold compared with that for wt-rHCII. Additionally, the inhibition values were still enhanced for the AR2 rHCII mutants compared with wt-rHCII when tested with a mixture of plasma and buffer with added glycosaminoglycans. Even with ␥ T -thrombin or ␣-thrombin/hirugen in the presence of glycosaminoglycans, some AR2 mutants were more active than wt-rHCII (although all of these rates were drastically reduced compared with ␣-thrombin), most likely because the acidic domain binding site on exosite-1 is neither totally removed in ␥ T -thrombin nor completely blocked by hirugen in ␣-thrombin (for instance, with hirugen (20 M) and heparin (10 g/ml) the inhibition rates (k 2 , M Ϫ1 min Ϫ1 ) for D72N/Y73F/ D75N rHCII and D75K rHCII were reduced to 2.0 Ϯ 0.5 ϫ 10 8 and 1.0 Ϯ 0.3 ϫ 10 8 , respectively, as compared with 0.4 Ϯ 0.1 ϫ

FIG. 3. Inhibition of ␣-thrombin/hirugen and of ␥ T -thrombin by wild-type and AR2 rHCIIs in the presence dermatan sulfate.
Top panel, inhibition of ␣-thrombin by AR2 rHCII mutants in the presence of 10 g/ml dermatan sulfate with hirugen (black bars), buffer (gray bars), or a control acidic peptide (white bars). The data represent an average of at least two separate protein preparations performed in triplicate. Bottom panel, ␥ T -thrombin inhibition by the AR2 rHCII mutants in the presence of 20 g/ml dermatan sulfate (black bars) or 200 g/ml dermatan sulfate (white bars). The data represent an average of at least two separate protein preparations performed in triplicate. 10 8 for wt-rHCII). We have shown previously that dermatan sulfate accelerates ␥ T -thrombin inhibition by HCII by 30-fold, indicating that the acidic domain still interacts with the remaining portions of exosite-1 in a productive manner. These results imply that for wild-type HCII, Asp 72 , Tyr-sulfate 73 , and Asp 75 help maintain critical intramolecular interactions that attenuate thrombin inhibition activity until glycosaminoglycans bind, which expels the acidic domain to then contact exosite-1 of thrombin.
To complement the above results, the optimum glycosaminoglycan concentration was inversely related to apparent heparin affinity for each AR2 rHCII mutant as found using heparin-Sepharose elution profiles. Because the intermolecular interactions between glycosaminoglycans and HCII are thought to be primarily ionic, the concentration of NaCl required to elute the proteins from heparin-Sepharose is a measure of their relative affinity. We found that some of the AR2 rHCII mutants required from 1.4-to 1.7-fold higher NaCl concentrations to elute from heparin-Sepharose when compared with wt-rHCII. What is remarkable about these AR2 rHCII mutants is that the majority of the negatively charged acidic domain is present while still requiring increased NaCl concentrations to elute from heparin-Sepharose. These elution differences are somewhat analogous to N-terminal deletions when the negative charges were removed from the intact HCII molecule, where ⌬1-67 rHCII and ⌬1-74 rHCII eluted from heparin-Sepharose at 1.4-and 2.1-fold higher NaCl concentrations, respectively, compared with wt-rHCII (27,32). We found previously that a synthetic peptide corresponding to the glycosaminoglycan binding site in HCII (from residues 183 to 200) bound with 2.6-fold greater affinity to heparin-Sepharose when compared with native HCII, indicating that the glycosaminoglycan-binding site was either masked or in an altered conformation compared with the peptide (47). It is intriguing to speculate that both Y73K and D75K (which had the highest increase in NaCl elution from heparin-Sepharose) might be directly contributing to glycosaminoglycan binding and be juxtaposed to the D-helix glycosaminoglycan region of HCII. Compared with wt-rHCII, the AR2 rHCII mutants are generally more permissive to glycosaminoglycan binding. These results indicate that for wild-type HCII, Asp 72 , Tyr-sulfate 73 , and Asp 75 support crucial intramolecular interactions that hinder glycosaminoglycan binding.
The mechanism of glycosaminoglycan-accelerated thrombin inhibition by HCII is unique among thrombin-inhibiting serpins (15,16,47). HCII appears to employ an allosteric mechanism whereby binding of glycosaminoglycans to the D-helix region is thought to alter the acidic domain, which then serves as a "tethered-ligand" for exosite-1 of thrombin (27,28,(31)(32)(33)(34)(35)(36)(37). The acidic domain/exosite-1 interaction is a critical component of the inhibition mechanism in the presence of glycosaminoglycans, whereas ternary complex formation bridging HCII, glycosaminoglycan, and thrombin appears to play a more minor role in the inhibition mechanism. The proposed interaction between the acidic domain and exosite-1 is well supported by a variety of data. A synthetic peptide corresponding to residues 54 -75 of HCII inhibited thrombin cleavage of fibrinogen but not thrombin chromogenic substrate activity (48). Because proteolysis of fibrinogen requires binding between exosite-1 and fibrinogen, this suggested that the HCII acidic peptide interacts with exosite-1. Deletion of HCII residues 1-52 by recombinant DNA technology did not affect thrombin inhibition in either the absence or presence of glycosaminoglycans (27). By contrast, deletion of residues 1-67 (first acidic region) greatly decreased HCII inhibition of thrombin in the presence of glycosaminoglycans but only slightly reduced it in the absence of glycosaminoglycan, suggesting that the first acidic domain interacted with thrombin (27). Interestingly, deletion of the entire acidic domain (residues 1-74) caused no further decreases in activity, suggesting that only the first acidic domain interacts directly with thrombin. Our laboratory (28,32,34) and others (27,33) showed that glycosaminoglycan-accelerated HCII inhibition of ␥ T -thrombin (in which exosite-1 has been removed by limited proteolysis) was reduced by greater than 1000-fold as compared with ␣-thrombin, whereas inhibition by antithrombin was relatively unaffected by the absence of exosite-1. Site-directed mutants of the basic residues in exosite-1 showed specificity similar to where hirugen binds exosite-1, and large decreases in activity with HCII-glycosaminoglycans were seen for some of the mutants (31,35,37).
The report by Baglin et al. (49) of the determination of the structures of native HCII and S195A thrombin-complexed HCII revealed some of the expected features based on the above discussion; however, the location of the acidic domain in HCII was unequivocal. The resolved native structure of HCII was a dimer formed by anti-parallel ␤-sheet interaction between strands 1C of each monomer with the acidic domain and the N terminus undefined. Recently, Liaw et al. (33) employed the exact opposite strategy to our study, where they replaced the basic residues in the D-helix region of HCII with neutral residues. They found an increase in progressive antithrombin activity of more than 100-fold compared with wt-rHCII (33). As expected, thrombin exosite-1 was essential for inhibition by these rHCII mutants (33). More recently, Brinkmeyer et al. (36) characterized an rHCII mutant with a re-formable disulfide bond between P52C and F195C. The oxidized form of this mutant was unable to form stable HCII⅐thrombin complexes; however, after reduction with ␤-mercaptoethanol the reduced HCII mutant was an active thrombin inhibitor in the presence of glycosaminoglycans. Collectively, these data allow us to provide an updated proposal of the mechanism of thrombin inhibition by HCII (Fig. 4). The historical depiction of HCII is given Fig. 4A, and this shows the acidic domain docked to the D-helix region (27,29,30,32,36). The mutants and data described by Liaw et al. (33) suggest that the acidic domain is no longer bound to the Dhelix, and its location is altered and may resemble that shown in Fig. 4B. Following glycosaminoglycan binding, the acidic domain is fully extended and interaction with exosite-1 in thrombin is favored (Fig. 4C). The structural data from Baglin et al. (49) show that the HCII⅐S195A thrombin complex resembles that predicted with the acidic domain and exosite-1 in direct contact with the expected interaction between the reactive center of HCII and active site of thrombin (Fig. 4D). The AR2 rHCII mutants described here would appear to be in equilibrium between the HCII structures depicted in Fig. 4, A and B (comparing differences in activity found here to those reported by Liaw et al. (33)). The enhanced sensitivity of the AR2 rHCII mutants (especially Asp 72 , Tyr-sulfate 73 , and Asp 75 ) to glycosaminoglycan interactions is connected to increased thrombin inhibition activity, which is consistent with a more permissive glycosaminoglycan binding site due to altered intramolecular interactions. Thus, AR2 may act as a "molecular rheostat" to help maintain HCII in a relatively inactive form in the absence of glycosaminoglycans, but after glycosaminoglycan binding it allows the acidic domain to tether and recruit exosite-1 of thrombin that leads to favorable thrombin inhibition.