Selective Inhibition of Factor Xa in the Prothrombinase Complex by the Carboxyl-terminal Domain of Antistasin*

Studies of antistasin, a potent factor Xa inhibitor with anticoagulant properties, were performed wherein the properties of the full-length antistasin polypeptide (ATS-119) were compared with the properties of forms of antistasin truncated at residue 116 (ATS-116) and residue 112 (ATS-112). ATS-119 was 40-fold more potent than ATS-112 in prolonging the activated partial thromboplastin time (APTT), whereas ATS-119 inhibited factor Xa 2.2-fold less avidly and about 5-fold more slowly than did ATS-112. The decreased reactivity of ATS-119 suggests that the carboxyl-terminal domain of ATS-119 stabilizes an ATS conformation with a reduced reactivity toward factor Xa. The observation that calcium ion increases the reactivity of ATS-119 but not that of ATS-112 suggests that calcium ion may disrupt interactions involving the carboxyl terminus of ATS-119. Interestingly, ATS-119 inhibited factor Xa in the prothrombinase complex 2–6-fold more potently and 2–3-fold faster than ATS-112. These differences in affinity and reactivity might well account for the greater effectiveness of ATS-119 in prolonging the APTT and suggest that the carboxyl-terminal domain of ATS-119 disrupts interactions involving phospholipid, factor Va, and prothrombin in the prothrombinase complex. The peptide RPKRKLIPRLS, corresponding to the carboxyl domain of ATS-119 prolonged the APTT and inhibited prothrombinase-catalyzed processing of prothrombin, but it failed to inhibit the catalytic activity of isolated factor Xa. Thus, this novel inhibitor appears to exert its inhibitory effects at a site removed from the active site of factor Xa.

Factor Xa (fXa), 1 a serine protease that functions at the intersection of the intrinsic and extrinsic pathway for blood coagulation, activates prothrombin to thrombin. Thrombin in turn activates platelets and converts the soluble plasma protein fibrinogen to the insoluble fibrin matrix of blood clots. The central role of fXa in blood coagulation suggests that fXa inhibitors might have therapeutic utility as anticoagulants. It is important to note, however, that fXa functions in the blood coagulation cascade in a complex with calcium ion, factor Va (fVa) and an appropriate phospholipid membrane, and that the catalytic activity toward prothrombin of this prothrombinase complex is several orders of magnitude greater than that of isolated fXa (1,2). These observations suggest that certain inhibitors might exhibit different inhibitory potencies toward fXa in the prothrombinase complex and isolated fXa.
Many natural inhibitors of fXa are known. In plasma, antithrombin III and tissue factor pathway inhibitor regulate blood coagulation by inhibiting fXa and other enzymes in the blood coagulation cascade (3,4). Other fXa inhibitors, such as tick anticoagulant peptide and antistasin (ATS) have been isolated from blood-sucking animals (5,6). Antistasin, originally isolated from the salivary gland of the Mexican leech Haementeria officinalis (6), is a cysteine-rich polypeptide 119 amino acid residues in length containing a cationic carboxyl-terminal domain that is not part of the two internally homologous domains of antistasin (Fig. 1). Kinetic studies reveal that antistasin is a potent, slow, tight-binding inhibitor of fXa (7). The only other protease inhibited by antistasin appears to be trypsin (IC 50 , 5 nM) (7). Antistasin, a Kunitz-type protease inhibitor, bears a structural resemblance to the fXa substrate prothrombin, and as expected, the inhibitor undergoes fXa mediated cleavage (at Arg-34) to form a stable covalently bound enzyme-inhibitor complex (7).
Intact antistasin had been successfully expressed in the insect baculovirus host (8). The present study, wherein fulllength and two truncated antistasins (from yeast and an African green monkey kidney cell line) are characterized, suggests that the carboxyl-terminal domain of antistasin, which has little influence on inhibitory potency toward isolated fXa, is important for the anticoagulant activity of antistasin. Additionally, a small peptide encompassing the carboxyl-terminal domain of antistasin that selectively inhibits fXa in the prothrombinase complex is identified.
Human factor Xa (fXa) was prepared from human fX (Hematologic Technologies, Inc., Essex Junction, VT) by the method of Bock et al. (11) or purchased from Hematologic Technologies Inc. fXa concentration was determined using an ⑀ 280 nm ϭ 1.16 (mg/ml) Ϫ1 and a molecular weight of 46,000 (12,13). The activated fXa was homogeneous (Ͼ90% ␣-fXa), as judged by sodium dodecyl sulfate electrophoresis in the Laemmli buffer system (14); the active site concentration was determined by titration with fluorescein mono-p-guanidinobenzoate (11) and by titration using the tick anticoagulant peptide double mutant, Y1W/ D10R (K i ϭ 10 pM) (15). Human fVa was purchased from Hematologic Technologies, Inc. fVa concentration was determined using an ⑀ 280 nm ϭ 1.74 (mg/ml) Ϫ1 and a molecular weight of 168,000 (16). Human prothrombin was isolated from citrated human plasma (17), and the concentration was determined using ⑀ 280 nm ϭ 1.38 (mg/ml) Ϫ1 and a molecular weight of 72,000 (18). Concentrations of recombinant truncated forms of ATS were determined by quantitative amino acid analysis and verified by titration with fXa. Small peptides were prepared by solidphase synthesis (19) and purified by preparative high pressure liquid chromatography on Vydac reversed phase C18 silica columns using an aqueous 0.1% trifluoroacetic acid/acetonitrile gradient. Identity and homogeneity were confirmed by quantitative amino acid analysis, high pressure liquid chromatography, and fast atom bombardment mass spectral analysis.
The Activated Partial Thromboplastin Time (APTT) Assay-This assay was performed using a Medical Laboratory Automation coagulation timer (Electra 900). A 10 l aliquot of inhibitor or buffer was added to 100 l of freshly prepared platelet-poor human plasma in a disposable cuvette (n ϭ 2). The resulting sample was mixed with 100 l of Actin reagent (Baxter Diagnostics Inc., McGaw Park, IL) and 100 l of 20 mM CaCl 2 at 37°C according to the standard protocol provided by the reagent manufacturer. Concentrations of inhibitors that doubled clotting times (2ϫ APTT) are reported as the final concentration in the assay mixture.
Preparation of Phospholipid (PL) Vesicles-Phospholipid was prepared by a modification of the procedure of Barenholz et al. (20). Synthetic 1,2-dioleoyl-phosphocholine (PC) and 1,2-dioleoyl phosphoserine (PS) (purchased from Avanti Polar Lipids, Inc., Alabaster, AL) were mixed (molar ratio, PC:PS ϭ 75:25) and evaporated to dryness under a nitrogen stream. The resulting residue was suspended in HBS buffer containing 50 mM HEPES, 150 mM NaCl, pH 7.4, at a phospholipid concentration of 10 mg/ml. The solution was sonicated (Braun-Sonic 2000 or Cole Palmer Ultrasonic Homogenizer) in an ice bath for 6 -8 cycles of 1.5 min of sonication spaced by 1-min intervals under a continuous nitrogen stream. The sonicated solution was centrifuged (Beckman XL-90 Ultracentrifuge) in Beckman Ultra-Clear tubes for 3-4 h at 4°C in an SW 50.1 swinging-bucket rotor, at 190,000 ϫ g (40,000 rpm). The upper 50% of the final solution was collected, and stored under nitrogen at 4°C. A phosphate assay (21) was used to determine the phospholipid concentration of the vesicle preparation.
Determination of k cat and K m for the Action of fXa on Prothrombin-Prothrombin was reacted with fXa (0.1 pM-2 nM), fVa (5 nM), and/or phospholipid vesicle (5 M) with calcium chloride (2 mM) in HBS buffer in a PEG-20,000 precoated 96-well microtiter plate for 0.5-2 h at room temperature, after which time, processing was quenched by addition of a solution containing 25 mM EDTA and 100 nM tick anticoagulant peptide. The concentration of the active thrombin produced was then determined from the rate of hydrolysis of S2238 (final concentration, 50 M). Values of k cat /K m were derived from the dependence of the rate of thrombin activation on concentrations of the prothrombin and fXa.
Inhibition of fXa with Recombinant Antistasin Using a Small Substrate as Probe-Assay solutions contained recombinant ATS (final concentration, 1 nM), calcium chloride (2 mM) and IEGR-AMC (40 M, K m ϭ 750 M) in HBSP buffer containing 50 mM HEPES, pH 7.4, 150 mM NaCl, and 0.1% PEG-8000 in a PEG-20,000 precoated 96-well fluorogenic microtiter plate (Dynatech MicroFluor) at room temperature. A solution of fXa (final 0.2 nM) containing 0.02 nM recombinant hirudin was mixed with the assay solution, and the hydrolysis rate was measured directly from the rate of increase of fluorescence intensity using a fluorescence plate reader (FluroStar, SLT Labinstruments, Salzburg, Austria). Excitation and emission wavelengths were set at 390 and 460 nm, respectively. The pseudo-first order rate constant for the inhibition of fXa by ATS was determined from the first order approach of the velocity to its final inhibited velocity, using Equation 1 as described previously (22,23). Because ATS is a competitive inhibitor of fXa (7) , the inhibition constant (K i ) and the second order association rate constant (k on ) could be obtained using Equations 2 and 3, respectively, where F o , F t , V i , and V s represent the initial fluorescence, the fluorescence at time t, and the initial and final rate of change of fluorescence, respectively. V o is the rate of change of fluorescence in the absence of inhibitor.
[ATS] o is the inhibitor concentration.
Inhibition of fXa Using Prothrombin as Substrate-Truncated forms of ATS (final concentration, 0.5-10 nM) and prothrombin (20 nM-1 M) were added to an assay solution of fXa (0.025-2 pM), fVa (0.5-5 nM), and phospholipid vesicle (1-5 M) with calcium chloride (2 mM) in HBSP buffer in a PEG-20,000 precoated 96-well microtiter plate at room temperature. For reactions in the absence of inhibitors, a diluted solution of fXa (final concentration, 0.02-0.05 pM) was used. The rate of thrombin activation appeared to be constant using 0.02-2 pM fXa. At various times (up to 4 h), a 20-l aliquot of the reaction mixture was removed and quenched with 105 l of 10 mM EDTA in Tris-buffered saline with PEG-8000. The amount of thrombin generated was determined by an activity assay using 50 l of S2238 (250 M in the quenching buffer). A fit of the data to Equation 4 yielded the pseudo-first order rate constant for reaction of ATS with fXa, where V i , V s , and P t represent the initial and final rate of thrombin generation and amount of thrombin generated at time t, respectively.
The inhibition constant (K i ) and the second order association rate constant (k on )were determined using Equations 2 and 3, respectively. fXa Chromogenic Substrate Assay for Determination of the Inhibitory Potency of Small ATS Peptides-fXa, 0.5 nM, was preincubated with a small ATS peptide (final concentration, 0 -400 M) at room temperature in a reaction buffer (Tris-buffered saline with PEG-8000) containing 50 mM Tris, pH 7.4, 150 mM NaCl, and 0.1% PEG-8000 in a PEG-20,000 precoated microtiter plate at room temperature for 30 min. Spectrozyme Xa (final concentration, 100 M) was then added, and the velocity of substrate hydrolysis was determined from the time-dependent increase in absorbance at 405 nm.
fXa Clotting Assay-This assay was performed using a fibrinometer (BBL FibroSystem) at 37°C. A solution of 80 l of human plasma deficient in fX and fVII was mixed with 80 l of rabbit brain cephalin (Sigma), 40 l of fXa (final concentration, 10 nM), and a small ATS peptide (0 -500 M). The mixture was equilibrated at 37°C, and 100 l of 100 mM CaCl 2 was added to initiate measurement of the clotting time. Tris-buffered saline containing 50 mM Tris, pH 7.4, 150 mM NaCl was used to dilute peptides and to obtain clotting time of the control without peptides.
The Human Prothrombinase Assay for Determination of the Inhibitory Potency of Small ATS Peptides-Antistasin peptides (final concentration, 0 -400 M) were preincubated with 1 pM fXa, 0.1-5 nM fVa, 2 mM CaCl 2 , 0.5-5 M PL in HBS buffer at 37°C for 30 min, and prothrombin (final concentration, 20 nM) was then added. At various times, aliquots of the reaction mixture were quenched with EDTA (final concentration, 25 mM). The concentration of thrombin in the quenched reaction mixture was determined using the chromogenic thrombin substrate S2238 (9).

RESULTS
Characterization of Truncated Forms of Antistasin-Fulllength antistasin (ATS-119), an antistasin derivative truncated at residue 116 (ATS-116), and an antistasin derivative truncated at residue 112 (ATS-112) were purified from secreted cell-free products of insect, yeast, and African green monkey kidney cells, respectively, in which recombinant antistasin was expressed (8,10). The molecular weights of the ATS polypeptides and/or the amino acid sequences of carboxyl-terminal tryptic peptides established the identity of each form of ATS (Table I). Data shown in Table I indicate the substantial effect of carboxyl-terminal truncation on the potency of the truncated forms of ATS in prolonging the APTT coagulation time. Surprisingly, the difference in potency of the three truncated forms of ATS was not reflected in their inhibitory activity toward isolated fXa. In fact, the most truncated antistasin (ATS-112), which, relative to the full-length antistasin (ATS-119), exhibited a 40-fold reduction in potency in prolonging the APTT, bound fXa 2.2-fold more tightly and reacted with fXa 5-fold faster than did ATS-119 ( Fig. 2 and Table II). Interestingly, calcium ion increased the affinity of all three forms of ATS for fXa by about 5-fold ( Fig. 2 and Table II). Additionally, calcium ion increased the value of the rate constant for formation of the fXa-ATS inhibitory complex by about 3-fold for full-length ATS-119, whereas little or no change in the rate constant was observed with ATS-112.
To explore the source of the 40-fold enhanced potency of ATS-119 in prolonging the APTT, we studied the effect of truncated forms of ATS on fXa-catalyzed prothrombin processing in the presence of the other components of the prothrombinase complex. Table III illustrates the well documented increase in catalytic efficiency of fXa upon assembly of the prothrombinase complex. Thus, fVa/Ca and Ca/PL vesicles increased k cat /K m 150-and 2900-fold, respectively. PL appears to selectively effect K m (K m ϭ 0.042 M at [PL] ϭ 5 M) for the interaction of fXa with prothrombin, because the value of k cat (0.01 s Ϫ1 ) in the presence of excess PL is similar to the value reported for fXa acting alone (2). The values of 0.19 Ϯ 0.02 M for K m and 170 Ϯ 20 s Ϫ1 for k cat for the fully assembled human prothrombinase complex indicate a 10 7 -fold increase in catalytic efficiency (k cat /K m ) relative to that of isolated fXa. These enhancements are similar to those reported previously (2, 24) for bovine fXa.
As shown in Fig. 3, full-length ATS-119 exhibits greater inhibitory potency and faster association with fXa in the prothrombinase complex than does either truncated ATS-112 or ATS-116. Comparison of the inhibition constants determined for inhibition of the activity of isolated fXa plus calcium ion (Table II) with those for inhibition of fXa in the prothrombinase complex (Table IV) indicates that (depending on the fVa and PL concentration) full-length ATS-119 is 9 -18-fold more potent in inhibiting prothrombinase than in inhibiting fXa/Ca. The truncated forms ATS-116 and ATS-112 show reduced selectivity toward prothrombinase versus fXa of 6 -9-fold and 1.3-1.4-fold, respectively (Tables II and IV). ATS-119 inhibits prothrombinase 2-6-fold more potently and binds to prothrombinase 2-3fold faster than does truncated ATS-112 (Table IV).
It is interesting to note that PL selectively increases the rate of inhibition of fXa/Ca by ATS-119 so as to alter the relative rate of inhibition of fXa/Ca by ATS-119 and ATS-112 from 0.76 in the absence of PL to 2.5 in the presence of 1 M PL (Fig. 4). Addition of fVa to fXa/Ca/PL leads to a further enhancement in the rate of inhibition by both ATS-119 and ATS-112 (Table IV). Interactions involving the carboxyl-terminal domain of ATS-119 and components of the prothrombinase complex may ac-  b Calculated value is based on primary structure of recombinant ATS 1-119 residues, with all cysteine residues oxidized to disulfide bonds and the amino terminus as pyroglutamate residue.
c Calculated value is based on primary structure of recombinant ATS 1-116 residues, with all cysteine residues oxidized to disulfide bonds, methionine residues oxidized to methionine sulfoxide, and the N terminus as pyroglutamate residue (9). d Calculated value is based on primary structure of ATS 1-112 residues, with all cysteine residues oxidized to disulfide bonds and the amino terminus as pyroglutamate residue. count for the enhanced reactivity of ATS-119 with prothrombinase and the enhanced anticoagulant activity of ATS-119 in the APTT assay. As shown by the data in Table IV, the enhanced interaction of ATS-119 with prothrombinase appears to depend upon the concentrations of fVa and PL used to form the prothrombinase complex. Increasing the fVa concentration from 0.5 nM to 5 nM increases the rate constant for inhibition by 3-5-fold, whereas increasing the PL concentration from 1 M to 5 M decreases the rate constant by 1.6 -2.9-fold.
The prothrombin concentration dependence of the inhibitory potency of antistasin was characterized in an attempt to elucidate the nature of the interaction between antistasin and human prothrombin. The physiological concentration of prothrombin (1.4 M) is approximately 7-fold greater than the K m determined for prothrombin processing by prothrombinase (Table III). For a case of simple competitive antagonism, the ap-parent dissociation constants for inhibitors should be reduced by a factor of 1 ϩ [S]/K m , where [S] represents the substrate concentration. Thus, at 0.2 and 1 M prothrombin, we should observe 2-fold and 6-fold decreases, respectively, in the apparent value of K i for inhibition of prothrombinase by antistasin. In the presence of 5 M PL and 5 nM fVa, 8-and 29-fold reductions of inhibitory potency were actually observed for ATS-119 (Table V). Similar effects were observed for ATS-116 (8.5-and 42-fold, respectively) and ATS-112 (19-and 52-fold, respectively). With a lower concentration of PL (1 M), prothrombin concentration-dependent decreases of inhibitory potency are somewhat less pronounced (Table V).     at the highest peptide concentration studied (400 M). ATS(109 -119) did, however, inhibit the activation of prothrombin by the prothrombinase complex (Table VI). As indicated by the data in Table VI

DISCUSSION
The ability of 6 nM ATS-119 to prolong the APTT 2-fold is impressive when one considers that 19 nM hirudin is required to double the APTT with human plasma 2 and that the K i for inhibition of prothrombinase by ATS-119 is approximately 6 pM, whereas the K i for inhibition of thrombin by hirudin is 22 fM (25). The greater potency of ATS-119 in prolonging the APTT despite the weaker interactions with its target may reflect the fact that ATS-119 acts earlier in the blood coagulation cascade than does hirudin.
The observation that truncated ATS-112 reacts with isolated fXa faster than does full-length ATS-119 suggests the possibility that the cationic carboxyl-terminal domain of ATS-119 may be stabilizing an antistasin conformation that is unfavorable for binding to fXa. The observation that calcium ion increases the rate of reaction of full-length ATS-119 with isolated fXa but not that of truncated ATS-112 is consistent with the possibility that the cationic carboxyl-terminal domain of ATS-119 may be interacting with certain internal anionic carboxylate groups. Complexation of such carboxylate groups with calcium ion could destabilize their putative interaction with the cationic carboxyl-terminal domain of antistasin and thereby favor an antistasin conformation that reacts more rapidly with isolated fXa. The 5-fold increase in the inhibitory potency toward isolated fXa exhibited by both the full-length and truncated forms of antistasin in the presence of calcium ion may well reflect an effect of calcium ion on the conformation of fXa.
The observation that ATS-119 inhibits prothrombinase more potently and more rapidly than does ATS-112 in a phospholipid and fVa-dependent fashion is consistent with the view that the carboxyl-terminal domain of antistasin interacts with these components of the prothrombinase complex. Thus, in addition to interacting with the active site of fXa, ATS-119 may be disrupting interactions involving fXa, fVa, PL, and prothrombin in the prothrombinase complex. Interestingly, addition of PL and fVa to fXa/Ca increases both affinity for full-length ATS and prothrombin processing activity, albeit fVa and PL enhance prothrombin processing activity to a much greater extent than affinity for ATS. It is tempting to speculate that interactions of ATS with prothrombinase may resemble to some extent those of prothrombin. With regard to this possibility, it is interesting to note that a sequence homology between the carboxyl-terminal domain of ATS-119 and residues 52-62 of prothrombin (Fig. 5) may provide a structural basis for some of the prothrombin-like interactions of ATS.
The observation that the antagonist effect of prothrombin on antistasin inhibition was greater than that expected for a competitive antagonist suggests the possibility that at high concentrations, more than one molecule of prothrombin can interact with each prothrombinase complex. Because each prothrombinase complex contains several molecules of PL, the excess prothrombin may bind to and thereby reduce the fraction of PL that is free to interact with ATS in the prothrombinase complex. Further studies of the complex equilibria involving ATS, prothrombin, and the components of the prothrombin complex will be required to properly characterize the antagonism of ATS by prothrombin.
The observation that the peptide ATS(109 -119), RP-KRKLIPRLS, inhibits prothrombinase but not isolated fXa supports the contention that the carboxyl-terminal domain of antistasin does not interact with the active site of fXa, but 2 S.-S. Mao, unpublished observations.   rather antagonizes interactions between fXa and its cofactors that give rise to the 10 7 -fold increase in catalytic efficiency of fXa in prothrombinase. The observation that the peptide ATS(119 -109) with the inverted sequence of ATS(109 -119) does not inhibit prothrombinase or prolong clotting times makes it difficult to ascribe the inhibition of prothrombinase observed with ATS(109 -119) to nonspecific charge-charge interactions. Thus, PRKRKLIPRLS exemplifies a new class of synthetic anticoagulants that selectively inhibits the catalytic activity of prothrombinase but not that of fXa.