A Comparison of Three Heparin-binding Serine Proteinase Inhibitors*

The purpose of this study was to compare three hep- arin-binding plasma proteinase inhibitors in order to identify common and unique features of heparin bind- ing and heparin-enhanced proteinase inhibition. Experiments with antithrombin, heparin cofactor, and protein C inhibitor were performed under identical conditions in order to facilitate comparisons. Synthetic peptides corresponding to the putative heparin binding regions of antithrombin, heparin cofactor, and protein C inhibitor bound to heparin directly and interfered in heparin-enhanced proteinase inhibition assays. All three inhibitors obeyed a ternary complex mechanism for heparin-enhanced thrombin inhibition, and the op- timum heparin concentration was related to the apparent heparin affinity of the inhibitor. The maximum inhibition rate and rate enhancement due to heparin appeared to be unique properties of each inhibitor. In assays with heparin oligosaccharides of known size, only the antithrombin-thrombin reaction exhibited a sharp threshold for rate enhancement at 14-16 saccharide units. Acceleration of antithrombin inhibition of factor Xa, heparin cofactor inhibition of thrombin, and protein C inhibitor inhibition of thrombin, activated protein C, and factor Xa did not require a minimum saccharide size. The differences in heparin size dependence and rate enhancement of proteinase inhibition by these inhibitors might reflect differences in the importance of the ternary complex mechanism and other mechanisms, alterations in inhibitor reactivity, and orientation effects in heparin-enhanced proteinase inhibition. Heparin normalized by setting the inhibition rate in the absence of heparin to 1.0 and the inhibition rate obtained with the 20-mer to 2.0. Computer-aided Molecular Modeling-The three-dimensional structure of al-proteinase inhibitor (9) was obtained from the Protein Data Bank. The amino acid sequences of antithrombin, heparin cofactor, and protein C inhibitor were obtained from the literature (2) and modeled by mutating the corresponding residues in al-proteinase inhibitor using the molecular modeling software SYBYL (Tripos Associates, St. Louis, MO). The amino termini, insertions, and deletions that had no counterparts in al-proteinase inhibitor were not modeled. The energy of each mutated protein was minimized using the programs SYBYL or AMBER.'

inhibitors, in some cases up to several thousand-fold (3). The therapeutic action of unfractionated heparin is believed to depend primarily on heparin-accelerated inhibition of thrombin by antithrombin (4). The importance of antithrombin is documented by the incidence of thrombotic disease in patients with antithrombin deficiencies or abnormal antithrombin molecules (5). Similar clinical correlations are weak for heparin cofactor (6) and non-existent for protein C inhibitor; thus it is difficult to assign discrete physiological roles for these two inhibitors.
The fact that heparin enhances proteinase inhibition by all three heparin-binding serpins suggests that a common mechanism is involved. However, the heparin-antithrombin interaction has been extensively studied, and the heparin-heparin cofactor and the heparin-protein C inhibitor interactions less so, and it is clear that differences among these serpins do exist. Because of the difficulty in directly comparing results from a number of sources, the present study was undertaken. This report attempts to distinguish general rules for heparin stimulation of proteinase inhibition from the unique responses of each serpin. The experiments described below examine the heparin binding properties and proteinase inhibition reactions in the presence of heparin under identical conditions in order to facilitate comparisons. Some of the data appeared previously in abstract form (7). The preceding paper describes some heparin binding properties of protein C inhibitor (8).

EXPERIMENTAL PROCEDURES
Materials-All serpins and proteinases were purified from human plasma (8). All other reagents were identical to those in the preceding report (8). Heparin oligosaccharides (6-20 saccharide units) were the gift of Dr. Michel Canton, American Bioproducts.
Peptide Synthesis-Peptides were synthesized as described (8) according to the reported sequences of the inhibitors (2). Fidelity of synthetic peptides was confirmed either by amino acid sequence or composition analysis. Heparin affinity of peptides was measured by passing the peptide through a 1-ml column of heparin-agarose in 20 mM HEPES, 0.1% PEG 8000, pH 7.4, and eluting with a linear salt gradient.
Peptide Competition Assays-Peptides (0.1-100 p~) in 20 mM HEPES, 150 mM NaCl, 0.1% PEG, pH 7.4, were added to assay mixtures containing 2 mg/ml bovine serum albumin, 0.05-0.15 pg/ml heparin (for antithrombin and heparin cofactor assays) or 1 pg/ml heparin (for protein C inhibitor assays), and 50 nM inhibitor in the same buffer at 25 "C. Reactions were started with the addition of 5 nM thrombin or activated protein C. After incubation, remaining proteinase activity was measured using 0.15 mM chromogenic substrate with 2 mg/ml Polybrene. Substrate hydrolysis was linearly related to proteinase concentration. Exceptions to the above procedure were: 500 nM antithrombin was used with 50 nM factor Xa; 100 nM protein C inhibitor was used with 10 nM activated protein C, and 1 pg/ml dermatan sulfate was used in some heparin cofactor assays. Control experiments performed in the absence of heparin or in the absence of inhibitor verified that peptides had no direct effect on proteinase or inhibitor activity.
Effect of Heparin on Thrombin Inhibition-Assays contained 10 8795 This is an Open Access article under the CC BY license. This is an Open Access article under the CC BY license.
nM antithrombin, protein C inhibitor, or heparin cofactor, 2 mg/ml bovine serum albumin, and 0.01-1000 pg/ml heparin in 20 mM HEPES, 150 mM NaCl, 0.1% PEG, pH 7.4 at 25 "C. Reactions were started with the addition of 1 nM thrombin. Following incubation, thrombin activity was determined by hydrolysis of chromogenic substrate containing 2 mg/ml Polybrene. Second order inhibition rate constants were calculated as -In a/t [A where a is the fractional proteinase activity remaining relative to the uninhibited control, t is the time of incubation, and [A is the inhibitor concentration. For assays comparing a-thrombin and 7,-thrombin, the conditions were 5 nM antithrombin, 5 nM protein C inhibitor, 20 nM heparin cofactor, and 0.5 nM thrombin. Heparin affinity chromatography of antithrombin, heparin cofactor, and protein C inhibitor was performed using a Pharmacia LKB Biotechnology Inc. fast protein liquid chromatography system and a 5-ml column of heparin-Sepharose in 20 mM HEPES, 10 mM NaC1, 0.1% PEG, pH 7.4. Samples were eluted with a 1 ml/min linear salt gradient from 10 mM to 1.2 M NaCl. Proteins were detected by absorbance at 280 nM. Salt concentration corresponding to peak elution was the average of six to nine runs.
Effect of Saccharide Chain Length-Assays contained 62.5 nM antithrombin, protein C inhibitor, or heparin cofactor, 2 mg/ml bovine serum albumin, and 500 mM heparin oligosaccharide (0.9 pg/ ml for the hexamer and 3.0 pg/ml for the 20-mer). Reactions were started with the addition of 5 nM thrombin, activated protein C, or factor Xa. After incubation, proteinase activity was determined by hydrolysis of the appropriate chromogenic substrate. Inhibition rate constants were calculated as described above. The results were normalized by setting the inhibition rate in the absence of heparin to 1.0 and the inhibition rate obtained with the 20-mer to 2.0.
Computer-aided Molecular Modeling-The three-dimensional structure of al-proteinase inhibitor (9) was obtained from the Protein Data Bank. The amino acid sequences of antithrombin, heparin cofactor, and protein C inhibitor were obtained from the literature (2) and modeled by mutating the corresponding residues in alproteinase inhibitor using the molecular modeling software SYBYL (Tripos Associates, St. Louis, MO). The amino termini, insertions, and deletions that had no counterparts in al-proteinase inhibitor were not modeled. The energy of each mutated protein was minimized using the programs SYBYL or AMBER.'

Models of Heparin-binding Sites-Peptides corresponding
to the putative heparin-binding sites of antithrombin, heparin cofactor, and protein C inhibitor were synthesized, and their identity as heparin-binding peptides was confirmed by their ability to bind heparin directly and indirectly in proteinase inhibition assays. Table I lists the apparent affinity of the three serpins and synthetic peptides for immobilized heparin. The peptides corresponding to the putative heparin binding region of antithrombin (AT 124-140) bound with lower affinity than native antithrombin, possibly because the peptide did not assume the same conformation as the corresponding structure in the native molecule, or because additional structural elements were required. The peptides corresponding to the putative heparin binding regions of heparin cofactor (HC 183-200) and protein C inhibitor (PC1 264-283) bound with greater affinity than the native molecules, indicating that the heparin-binding sites in the native molecules might be masked. In two cases, peptides containing the same residues as the putative heparin binding region but in random sequence, bound with lower affinity (AT 124-140 random and PC1 264-283 random), suggesting that the exact sequence of the peptide, and thus its structure, was more important than charge alone for binding to heparin.
The peptides were also tested for their ability to compete with serpins for heparin binding in proteinase inhibition assays. Increasing concentrations of peptides were added to Software, version 3.1, obtained from U. C. Sing, P. K. Weiner, J. W. Caldwell, and P. A. Kollman, P.A., Dept. of Pharmaceutical Chemistry, University of California, San Francisco. mixtures of serpin, proteinase, and heparin such that competition between peptide and inhibitor for heparin would decrease the efficiency of proteinase inhibition. Table I1 presents the results of these experiments, expressed as the concentration of peptide required to block half the proteinase inhibition activity of the serpin. Peptides corresponding to the putative heparin-binding sites of the three serpins competed with the parent serpin in proteinase inhibition assays. Furthermore, heparin cofactor-derived peptides and protein C inhibitor-derived peptides competed with antithrombin, and antithrombin-derived peptides competed with protein C inhibitor. The peptide corresponding to the heparin-binding site in antithrombin competed with antithrombin in assays for thrombin or factor Xa inhibition, and peptides from protein C inhibitor competed with protein C inhibitor in assays for thrombin or activated protein C inhibition. The ability of the various peptides to compete for heparin binding regardless of the particular serpin or proteinase involved indicates a general ability to bind heparin, and rules out specific effects of the peptides on these inhibitors or proteinases.
Dermatan sulfate, in addition to heparin, accelerates thrombin inhibition by heparin cofactor, but not antithrombin or protein C inhibitor (8, 10). Two peptides from the putative glycosaminoglycan binding region of heparin cofactor (HC 173-190 and HC 183-200) competed for heparin binding, with HC 183-200 being more effective. However, both peptides were equally effective in dermatan sulfate-containing thrombin inhibition assays. These results suggest that the structural determinants for heparin or dermatan sulfate binding to heparin cofactor differ somewhat. As a positive control, the known heparin-binding site of platelet factor 4 (PF4 74-85, Ref. 11) was synthesized and tested for its ability to compete for heparin binding in proteinase inhibition assays. The platelet factor 4-derived peptide competed in all cases except when dermatan sulfate was used, which might reflect the ability of platelet factor 4 to bind heparin but not dermatan sulfate. In most cases, the peptides with random sequences were less effective at competing for heparin binding than their native sequence counterparts (antithrombin-derived peptides in antithrombin assays and protein C inhibitorderived peptides in protein C inhibitor assays). In only one case (antithrombin-derived peptides in the protein C inhibitor-thrombin assay) was the random peptide more effective than the native peptide. On the whole, the results suggest that the specific sequence of the peptide, not merely its charge, is important for heparin binding. The peptide with the greatest apparent affinity for immobilized heparin, PC1 264-283, was also one of the most effective at competing for heparin binding in proteinase inhibition assays. The ability of the serpin-derived peptides to bind heparin, as measured by two different methods, provides proof that these are models of the heparin-binding sites in the parent inhibitors. Effect of Heparin on Thrombin Inhibition-The accelerating effect of heparin on thrombin inhibition by antithrombin, heparin cofactor, and protein C inhibitor is well-documented. In this study, reactions with each serpin were compared under identical conditions of buffer, temperature, heparin preparation, and proteinase, so that the only variable was the inhibitor. Constant protein concentrations were used since they can affect the results (12). Fig. 1 shows the effect of heparin concentration on the rate of thrombin inhibition by each of the serpins. The shapes of these curves are consistent with the ternary complex model of heparin action (13), in which lower concentrations of heparin bind both proteinase and inhibitor, thereby bringing them into closer proximity for a a Heparin affinity is given as the salt concentration required for elution from a 1-ml column of heparin-agarose. This value is an indication of heparin affinity, but is not identical to an affinity constant, which requires more rigorous measurement.
Numbers refer to the amino acid sequence of the parent protein, antithrombin. Numbers refer to the amino acid sequence of the parent protein, protein C inhibitor. Numbers refer to the amino acid sequence of the parent protein, heparin cofactor. e Numbers refer to the amino acid sequence of the parent protein, platelet factor 4. >loo a Assays contained the indicated inhibitor and proteinase as well as heparin, as described under "Experimental Procedures." The peptide concentration that blocked 50% of inhibition activity was measured using inhibition activity in the absence of peptide as the starting (0%) point and proteinase activity in the absence of inhibitor as the end (100%) point. Plots of proteinase activity uersus peptide concentration were typically sigmoidal.
Results were taken from Ref. 8. Assay contained dermatan sulfate rather than heparin.
faster rate of reaction. Higher concentrations of heparin act to sequester the reactants, restoring the reaction rate to normal. The optimum heparin concentration, which is the concentration corresponding to maximum inhibition rate, is expected to be a function of the affinity of the inhibitor for heparin, since all other variables were constant. This was in fact the case, as shown in Table 111: the optimum heparin concentration was inversely related to apparent heparin affinity for each ~e r p i n .~ Unfractionated heparin and a fraction of heparin with high affinity for antithrombin produced the same degree of inhibition rate  Table 111.
If the ternary complex mechanism were the only one involved in heparin-enhanced thrombin inhibition, then the degree of rate enhancement would be expected to be similar for all serpins. Table I11 shows that this is not the case, as there was no obvious relationship between heparin affinity and maximum thrombin inhibition rate or degree of rate enhancement. For example, heparin cofactor, which had the lowest heparin affinity, exhibited the greatest increase in thrombin inhibition rate, approximately 9000-fold.
Other studies (14, 15) have proposed that heparin, in addition to serving as a template for the inhibitor-proteinase reaction, causes the exposure of a secondary binding site for thrombin in heparin cofactor. yT-Thrombin, which lacks the structure necessary for binding to this site, is unable to react with heparin cofactor as rapidly as a-thrombin in the presence of heparin (16). Table IV lists the rate constants for inhibition enhancement and the same heparin optimum in an antithrombinthrombin assay system, although the high affinity heparin showed greater rate enhancement at lower concentrations, relative to unfractionated heparin. In an antithrombin-factor Xa assay system, heparin with high affinity for antithrombin produced maximal rate enhancement at a lower concentration than the unfractionated heparin, but the maximum inhibition rate was the same (results not shown). Thus, fractionation of heparin by affinity for antithrombin does not influence the parameters examined here, namely optimum heparin concentration and maximum thrombin inhibition rate. 4.6 9000 Heparin affinity is given as the salt concentration required for peak elution from heparin-Sepharose using a fast protein liquid chromatography system. The apparent heparin affinities given here are higher than those in Table I. The differences are most likely due to the higher heparin content of the heparin-Sepharose compared to heparin-agarose produced in the laboratory.
The optimum heparin is the concentration at which the maximum inhibition rate occurs, as shown in Fig. 1.
The rate increase is calculated as the ratio of the maximum rate to the rate in the absence of heparin. of a-thrombin and 7,-thrombin by the three heparin-binding serpins in the presence of optimal concentrations of heparin. For a given serpin, the optimum heparin concentration was the same for a-thrombin and yT-thrombin. Only heparin cofactor exhibited a dramatic difference in heparin-catalyzed inhibition rate for the two forms of thrombin (1760-fold difference), while antithrombin and protein C inhibitor exhibited modest differences (2.5-fold for antithrombin and 7.8fold for protein C inhibitor). Thus, heparin cofactor appears t o be unique among the three serpins by virtue of its secondary binding mechanism. Effect of Heparin Saccharide Chain Length-The influence of heparin saccharide chain length on the rate of proteinase inhibition by the three heparin-binding serpins was examined in order to determine the importance of the ternary complex mechanism for proteinase inhibition, which involves simultaneous binding of the inhibitor and proteinase to heparin. If the ternary complex mechanism were the only one involved, then small heparin oligosaccharides would not accelerate thrombin inhibition until a minimum heparin saccharide chain length, capable of binding both serpin and proteinase, were reached. If other mechanisms were involved, such as changes in reactivity of the serpin or secondary binding sites as described above for heparin cofactor, then the saccharide size dependence of the inhibition reaction might not exhibit a threshold phenomenon. Some previous studies have reported a variety of minimum saccharide chain lengths required to accelerate antithrombin (17) and heparin cofactor (18, 19) reactions. The present study examined the three heparin-binding serpins under identical conditions'using heparin oligosaccharides of known chain length, from 6 to 20 saccharide units. (Saccharides of this length would not be capable of binding more than two protein molecules; Ref. 20.) Because the same concentration of heparin accelerates throm-bin inhibition to various degrees depending on the serpin (as shown in Fig. l), the results were normalized by setting the inhibition rate in the absence of heparin to 1.0 and the inhibition rate obtained with the 20-mer to 2.0. The results are shown in Fig. 2. Only the antithrombin-thrombin reaction showed a sharp threshold with increasing size of the heparin oligosaccharide: inhibition activity increased at 16 saccharide units. The antithrombin-factor Xa reaction exhibited an almost linear relationship between inhibition rate and saccharide chain length. The effect of heparin chain length on the heparin cofactor-thrombin and protein C inhibitor-thrombin reactions was intermediate to that of the two antithrombin reactions. The heparin saccharide chain length dependence for the protein C inhibitor-activated protein C and protein C inhibitor-factor Xa reactions was almost identical to that of the protein C inhibitor-thrombin reaction. One interpretation of these results is that the template mechanism is clearly important for the antithrombin-thrombin reaction, but less so for the heparin cofactor and protein C inhibitor reactions, and possibly not at all for the antithrombin-factor Xa reaction.

DISCUSSION
In order to distinguish general features of heparin-binding serpins from features unique to each inhibitor, the three proteins were studied in parallel. The putative heparin-binding sites in antithrombin, heparin cofactor, and protein C inhibitor have been identified on the basis of the following: 1) high charge density; 2) results of chemical modification of residues that affect heparin interactions in antithrombin (21,22) and heparin cofactor (23); 3) natural mutations that affect heparin binding in antithrombin (21,24) and heparin cofactor (25); 4) site-directed mutagenesis in heparin cofactor (26), and 5) heparin binding activity of synthetic peptides (8). The heparin binding regions in antithrombin and heparin cofactor are located primarily in the D helix, and in fact, these proteins exhibit sequence homology in this region (residues 107-136 2.0 .  and protein C inhibitor-factor Xa (0). The results were normalized to give a range of rates from 1.0 to 2.0. The actual rate constants span the following values (from 6 to 20 saccharide units): antithrombin-thrombin: 0.04-7.74 X 10' M" min"; heparin cofactor-thrombin: 0.04-5.62 X lo6 M" rnin"; protein C inhibitor-thrombin: 1.01-1.80 X lo6 M" rnin"; antithormbin-factor Xa: 0.25-4.40 X lo7 M" rnin"; protein C inhibitor-activated protein c: 0.92-3.94 X lo6 M" min"; protein C inhibitor-factor Xa: 1.78-2.64 X lo6 M" rnin". in antithrombin and 165-195 in heparin cofactor). Protein C inhibitor is not homologous to either antithrombin or heparin cofactor in this region, and the putative heparin-binding site in protein C inhibitor is assigned to the H helix (residues 264-273, Refs. 8, 27).The homology among serpins (2) and the similarity of crystal structures of other serpins (28,29) allow the heparin-binding serpins to be modeled with confidence. The peptide backbone structures of antithrombin, heparin cofactor, and protein C inhibitor are shown in Fig. 3.
Arginine and lysine residues in the primary heparin-binding sites in the three serpins are shown with van der Waals radii highlighted, Not shown are additional residues that are involved in heparin binding, such as the amino termini of antithrombin and protein C inhibitor, which have been proposed to participate in heparin binding on the basis of other molecular modeling studies (2,27). In any case, the primary heparin-binding sites in the three serpins are assigned to surface residues that form a-helices in the proteins, and the location of the helix is the same in antithrombin and heparin cofactor, but not protein C inhibitor.
The ability of synthetic peptides from antithrombin, heparin cofactor, and protein C inhibitor to bind heparin (and dermatan sulfate, in the case of heparin cofactor-derived peptides) provides additional proof that these sequences correspond to the heparin-binding sites in the three serpins. This is the first demonstration that heparin-binding peptides derived from different serpins are functionally similar.
The effect of heparin on the inhibition of thrombin by antithrombin, heparin cofactor, and protein C inhibitor also shows that the three inhibitors are similar in obeying the ternary complex mechanism. According to this theory (13) both inhibitor and proteinase bind to heparin so that colocalization of reactants increases the reaction rate. The rate reaches a maximum and then decreases as the heparin concentration increases. The optimum heparin concentration appears to be related to the apparent heparin affinity of the serpin. In these experiments, thrombin was used, but the preceding paper reports that the affinity of the proteinase also influences the optimum heparin concentration (8). Other mechanisms must also govern the effect of heparin on serpin reactivity, since the maximum inhibition rate and rate enhancement varied among the serpins. The size of the heparin oligosaccharide capable of accelerating thrombin inhibition by the three serpins provides information about possible mechanisms of heparin action. The sharp threshold in saccharide size dependence in the antithrombin-thrombin reaction is consistent with a mechanism in which both inhibitor and proteinase bind heparin; ternary complex formation appears to be less important for protein C inhibitor or heparin cofactor inhibition reactions, at least with the small saccharides tested in this study. No conclusions can be drawn regarding the size of the heparin-binding sites in protein C inhibitor or heparin cofactor, since these serpins did not exhibit a threshold phenomenon. Because thrombin binds approximately 6 saccharide residues (30), antithrombin must accommodate 8-10 saccharide units in its heparinbinding site (14-16 minus 6).
We propose that while antithrombin, heparin cofactor, and protein C inhibitor share some features of heparin-enhanced proteinase inhibition, the molecular events that determine maximum inhibition rate, rate enhancement, and heparin size dependence are unique to each inhibitor. Possible mechanisms for heparin action are discussed below for each serpin, considering the findings of this study and other work. Three factors are taken into account: ternary complex or template effects, changes in reactivity of the serpin, and orientation of the serpin relative to the proteinase.
Antithrombin differs dramatically in inhibition of thrombin and factor Xa. The heparin size dependence data suggest that a template mechanism is important in thrombin inhibition. These data are also consistent with the possibility that binding of 14-16 saccharide units is required to elicit a change in antithrombin reactivity. This mechanism seems unlikely since the heparin dependence of thrombin inhibition shows the bell-shaped curve characteristic of a ternary complex (13,31). Furthermore, the antithrombin-factor Xa reaction mechanism, in which the ternary complex mechanism is not important (32), showed no threshold at 14-16 residues. Other studies have concluded that a heparin-induced conformational change in antithrombin is not important in thrombin inhibition (33)(34)(35). We propose that during thrombin inhibition by antithrombin, heparin acts as a template to increase the rate of reaction of inhibitor and proteinase. In addition, the orientation of the reactants when they are bound to heparin may be optimized so that a 2500-fold increase in reaction rate is possible, and this probably masks the more limited effect of a heparin-induced increase in antithrombin reactivity. The optimal orientation of inhibitor and proteinase might be such that the reactive site of antithrombin would be directly apposed to the active site of thrombin when the molecules, still bound to heparin, collide. This scenario is consistent with a finely evolved system for limiting thrombin activity in vivo, which is believed to be the primary role of antithrombin (3).
Antithrombin inhibition of factor Xa, in contrast, depends almost entirely on the heparin-induced increase in reactivity of antithrombin (36). This is consistent with the lack of a heparin oligosaccharide size threshold the larger the saccharide, the greater the probability that it contains the pentasaccharide sequence that binds with high affinity to antithrombin (37). The binding of this sequence elicits a conformational change in antithrombin, detected by fluorescence changes, that is correlated with its increased reactivity (36). The role of heparin as a template in the antithrombin-factor Xa reaction is minimal (32). Although the affinity of factor Xa for immobilized heparin is significant (eluting at 445 mM NaC1, Ref. 8) and a ternary complex might be expected to form, the orientation of the enzyme and inhibitor may not favor interaction. This might also be responsible for the limited heparininduced rate enhancement for factor Xa compared to thrombin inhibition (31). The relative contributions of ternary complex formation and increased antithrombin reactivity appear to differ among antithrombin-proteinase pairs. For example, the increase in kallikrein inhibition is modest (about &fold) and does not follow a template mechanism (results not shown), while inhibition of factor IXa is accelerated significantly by heparin and a ternary complex mechanism seems to be involved (31). The differences might be determined by the orientation of the particular proteinase when it is bound to heparin. Other reports suggest that the orientation effect might include neutralization of positive charges on the inhibitor by heparin, which could contribute to more favorable interactions with cationic proteinases (38).
Thrombin inhibition by heparin cofactor follows the template mechanism as shown in the present study and other reports (39). The shape of the heparin size dependence curve initially suggests that a heparin-induced change in heparin cofactor reactivity also occurs. However, the evidence is consistent with a secondary thrombin-binding site in heparin cofactor, consisting of acidic residues near the amino terminus that are made accessible when a glycosaminoglycan binds to heparin cofactor (14, 15). This secondary binding mechanism is illustrated by the failure of heparin to accelerate inhibition of yT-thrombin, which lacks the structure required to bind this acidic region of heparin cofactor (16). Since no minimum saccharide size was required for heparin acceleration of thrombin inhibition by heparin cofactor, even small saccharides must be able to expose the acidic region (by displacing it from the heparin binding region of heparin cofactor, Refs. 14, 15). This is consistent with a report that small dermatan sulfate saccharides are capable of increasing heparin cofactor reactivity without supporting a template mechanism (40). Studies that have found a minimum size requirement of 18-26 heparin saccharide units in heparin cofactor reactions (18, 19) might have used different heparin fractionation techniques or different approaches to assigning threshold inhibition rate values. The combination of the secondary binding site mechanism, which might also orient inhibitor and proteinase, and the ternary complex mechanism is able to accelerate thrombin inhibition 9000-fold, even though the reactive site sequence of heparin cofactor is Leu-Ser, which is not expected to be a good substrate for thrombin (41). It is also possible that the geometry of the ternary complex formed with heparin differs from that formed with dermatan sulfate. Although dermatan sulfate has lower affinity than heparin for heparin cofactor, it enhances thrombin inhibition to a greater extent (42).
The results of the present study and those of the preceding report (8) indicate that protein C inhibitor obeys the template mechanism for heparin-enhanced proteinase inhibition. However, the absence of a sharp threshold for heparin oligosaccharide size dependence suggests that small saccharides might also directly alter protein C inhibitor reactivity, as is the case for the antithrombin-factor Xa reaction. The poor proteinase inhibition rate enhancement elicited by heparin (relative to antithrombin or heparin cofactor inhibition rates) might reflect less favorable orientation of inhibitor and proteinase when they are bound to heparin. Thus, when heparin-bound inhibitor and proteinase collide, additional steps, such as dissociation from the heparin template, might occur before a productive reaction could occur. This is particularly intriguing since the putative heparin-binding site of protein C inhibitor (the H helix) is in a location different from the heparinbinding sites in antithrombin or heparin cofactor (the D helix, Fig. 3). It is possible that the expected template effect of heparin is dampened by geometrical factors that hinder the approach of inhibitor and proteinase. This is one explanation for the low maximum thrombin inhibition rate by protein C inhibitor, whose reactive site sequence, as in antithrombin, is Arg-Ser, a likely target for thrombin (43). Steric hindrance or other geometrical factors also might explain why inhibition of different proteinases is accelerated to various degrees by a given concentration of heparin (8).
Full confirmation of the proposed mechanisms for heparin acceleration of proteinase inhibition by the three heparinbinding serpins will require additional study, including an assessment of the contribution of geometric effects. In conclusion, the ternary complex mechanism for heparin-enhanced thrombin inhibition and the characteristics of the heparin-binding sites of antithrombin, heparin cofactor, and protein C inhibitor are common features of these serpins.
However, the location of the heparin-binding site and the importance of the ternary complex mechanism and other mechanisms for heparin-enhanced proteinase inhibition are unique features of these serpins. It remains to be seen how closely the lesser-studied heparin-binding proteinase inhibitors, such as C1 inhibitor (44) and protease nexin I (45), follow the rules described above. A recent report on protease nexin I suggests that its heparin-binding site resembles that in antithrombin, and heparin-enhanced proteinase inhibition obeys a ternary complex mechanism (46).