Protease nexin-2/amyloid beta-protein precursor inhibits factor Xa in the prothrombinase complex.

Protease nexin-2/amyloid β-protein precursor (PN-2/AβPP) is a Kunitz-type protease inhibitor which has been shown to be a tight-binding inhibitor of coagulation factors XIa and IXa. Here we show that PN-2/AβPP and its KPI domain also inhibited isolated factor Xa with a Ki of 10−8M. On a solid phase binding assay, PN-2/AβPP formed a complex with factor Xa. Incubation of molar excess factor Xa to PN-2/AβPP produced a single cleavage within PN-2/AβPP's heparin binding domain liberating a 8.2-kDa amino-terminal peptide. PN-2/AβPP and its KPI domain equally inhibited factor Xa in the prothrombinase complex with a Ki of 1.9 × 10−8M and 1.3 × 10−8M, respectively. AβPP695 which does not contain the KPI domain was a substrate of factor Xa but did not inhibit it, indicating the PN-2/AβPP inhibition of factor Xa was not substrate inhibition. All of the factor Xa inhibition in the prothrombinase complex by PN-2/AβPP and its KPI domain on the chromogenic assay was accounted for by inhibition of release of prothrombin fragment F1+2 as determined on immunochemical assay. In the prothrombinase complex, PN-2/AβPP inhibited factor Xa with a kassoc = 1.8 ± 0.7 × 106M−1 min−1 similar to antithrombin III and heparin inhibition (kassoc of 3.0 ± 0.2 × 106M−1 min−1). These studies indicated that PN-2/AβPP in the assembled prothrombinase complex inhibited factor Xa comparable to antithrombin III in the presence of heparin. PN-2/AβPP's factor Xa inhibitory activity along with its known inhibition of factors XIa and IXa suggest that this protease inhibitor and related proteins could be regulators of hemostatic reactions on membranes of cells in the intravascular compartment.

Amyloid ␤-protein precursor (A␤PP), 1 a multidomain protein, is the parent protein of amyloid ␤ protein, a 39 -42 amino acid peptide that is deposited in senile plaques and in the walls of cerebral blood vessels of patients with Alzheimer's disease (Kang et al., 1987;Glenner and Wong, 1984). The single gene for A␤PP found on chromosome 21 encodes at least three dis-tinct mRNAs produced by alternative splicing that result in three different sized proteins (A␤PP 695 , A␤PP 751 , and A␤PP 770 ) (Ponte et al., 1988;Tanzi et al., 1988;Kitaguchi et al., 1988). Two of these mRNAs code for proteins (A␤PP 751 and A␤PP 770 ) which contain a domain homologous to Kunitz-type protease inhibitors (KPI) (Ponte et al., 1988;Tanzi et al., 1988;Kitaguchi et al., 1988). The secreted isoforms of A␤PP containing the KPI domain are identical to protease nexin-2 (PN-2) (Van Nostrand et al., 1989;Oltersdorf et al., 1989).
PN-2/A␤PP and its KPI domain have been recognized to be potent inhibitors of trypsin, chymotrypsin, epidermal growth factor binding protein, and the ␥ subunit of nerve growth factor (Van Nostrand et al., 1989, 1990bOltersdorf et al., 1989). PN-2/A␤PP which is present in high concentrations in platelets is a potent inhibitor of factor XIa (Van Nostrand et al., 1990a, 1990bSmith et al., 1990). PN-2/A␤PP also is an inhibitor of factor IXa (FIXa) in the assembly of the tenase complex on phospholipid vesicles (PSPC), platelets, and endothelial cells (Schmaier et al., 1993(Schmaier et al., , 1995. Similarly, a homologue of PN-2/ A␤PP, amyloid ␤-protein precursor-like protein-2, has been shown to have inhibitory activity against hemostatic enzymes factors XIa, IXa, and Xa (Petersen et al., 1994;Sprecher et al., 1993;Van Nostrand et al., 1994). These studies suggest that this family of proteins may have a regulatory role in hemostasis. While examining the ability of PN-2/A␤PP to inhibit factor IXa on non-biologic surfaces, we found in one assay that the degree of factor IXa inhibition could not be fully accounted for by its inactivation alone. The present investigation shows that PN-2/A␤PP also is an inhibitor of factor Xa alone and when assembled on PSPC in the prothrombinase complex. Recognition that PN-2/A␤PP's inhibits factor Xa enlarges its role as a regulator of hemostasis.

EXPERIMENTAL PROCEDURES
Proteins-PN-2/A␤PP (A␤PP 751 ) was purified from fibroblast culture media using techniques of heparin affinity chromatography and immunoaffinity chromatography as described previously (Van Nostrand et al., 1990b). The KPI domain of PN-2/A␤PP, which was provided by Dr. Steven Wagner, Salk Institute Biotechnology/Industrial Associates, La Jolla, CA, was produced in a recombinant yeast expression system and purified as described previously (Wagner et al., 1992). The protease inhibitory activities of purified PN-2/A␤PP and KPI domain were determined by neutralization with active-site titrated trypsin (Van Nostrand et al., 1990b;Wagner et al., 1992). A␤PP 695 was obtained and purified like PN-2/A␤PP from culture media from human glioblastoma U-138 cells stably transfected to overexpress it (Davis-Salinas et al., 1994). Human factors IXa (FIXa), Xa (FXa), X, and II (FII) were purchased from Enzyme Research Laboratories, South Bend, IN. Human FIXa on nonreduced sodium dodecyl sulfate-13% polyacrylamide gel electrophoresis (SDS-PAGE) showed two bands at 52 and 33 kDa, and when reduced with 2% ␤-mercaptoethanol, four bands at 29, 25, 14, and 12 kDa. The 25-and 12-kDa bands seen on reduced SDS-PAGE represented only 5-10% of the total protein in all preparations. Human FXa was activated with Russel viper venom. On reduced 15% SDS-PAGE, FXa consisted of three majors bands at 36, 34, and 24 kDa. These bands * This work was supported by National Institutes of Health Grant HL49566 (to A. H. S. and W. E. V. N.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Phospholipid Vesicle Preparation-Phospholipid vesicles were prepared from a mixture of L-␣-phosphatidylserine (Sigma) and L-␣-phosphatidylcholine (Sigma) (25/75, mol/mol) that were dried in a glass test tube under a stream of nitrogen (Rawala-Sheikh et al., 1990). The dried material was resuspended in 0.05 M Tris-HCl, 0.175 M NaCl, pH 7.5, and sonicated for 30 s multiple times on ice over 60 min (Rawala-Sheikh et al., 1990). After sonication, some preparations were ultracentrifuged at 100,000 ϫ g to produce a homogenous suspension free of large particles and multilamellar liposomes (Barenholz et al., 1977). No difference in the cofactor activity of the phosphatidylserine/phosphatidylcholine vesicles (PSPC) was noted whether they were ultracentrifuged or not. The PSPC were stored at 4°C.
Measurement of Factor Xa Formation-The enzymatic activity of human FIXa was measured by its ability to activate human factor X using polylysine as an artifical surface (Schmaier et al., 1993(Schmaier et al., , 1995Lundblad and Roberts, 1982;Griffith et al., 1985;McCord et al., 1990). Factor IXa (4.45 nM) was incubated with factor X (400 nM) in 0.1 M triethanolamine, 0.1 M NaCl, pH 8.0, containing 0.1% polyethylene glycol (M r ϭ 8000), 0.2% bovine serum albumin, and 60 nM polylysine for 40 min at 20 -25°C. At the end of the incubation, an aliquot of the activated factor X solution was added to a solution of 0.4 mM tosyl-Gly-Pro-Arg-p-nitroanilide (Sigma). Hydrolysis proceeded for 60 min at 20 -25°C and the reaction was terminated by the addition of 50% acetic acid after which the optical density reading was obtained at 405 nm. FIXa alone had no amidolytic activity on the chromogenic substrate. The FIXa used in this assay activated ϳ0.5% of the added factor X. When inhibition studies were performed with PN-2/A␤PP and its KPI domain, the inhibitor was incubated with FIXa for 5 min at room temperature prior to the addition of factor X. All inhibition constants determined from the results of the chromogenic assay were calculated from the residual activity at end point.
Measurement of Factor Xa Activity-Factor Xa activity (1-2.5 nM) was measured in 0.1 M triethanolamine, 0.1 M NaCl, pH 8.0, containing 0.1% polyethylene glycol (M r ϭ 8000), 0.2% bovine serum albumin, and 60 nM polylysine or 0.02 M Hepes, 0.15 M NaCl, pH 7.4, containing 0.5 mg/ml bovine serum albumin, 2 mM Ca 2ϩ , and 0.1% polyethylene glycol using 0.4 mM tosyl-Gly-Pro-Arg-p-nitroanilide (Sigma) for 35 min at 20 -25°C. In certain experiments, the polylysine was removed from the buffer. In other experiments 25 M PSPC, 4.8 units/ml thrombin-activated factor VIII (FVIIIa), or 5 nM thrombin-activated bovine factor Va (FVa) were added to the reaction mixture in the presence of 2 mM Ca 2ϩ . The reaction was terminated by the addition of an equal volume of 50% acetic acid after which the optical density was obtained at 405 nm. Hydrolysis of the substrate was linear over the time of the reaction. When inhibition studies were performed with PN-2/A␤PP, its KPI domain, or antithrombin III, the inhibitor (2-10 nM) was incubated with FXa (1 nM) for 5 min at room temperature prior to the addition of the chromogenic substrate. All inhibition constants determined from the results of the chromogenic assay were calculated from the residual activity at end point. When investigations with antithrombin III were performed, 1 unit/ml heparin (Elkins-Sinn, Cherry Hill, NJ) was included in the reaction mixture.
Measurement of Factor X Activation Peptide-Simultaneous samples of human FIXa activation of factor X in the presence of polylysine were prepared for both chromogenic and immunochemical determination of factor X activation. Immunochemical determination of activation of factor X by FIXa was measured as nanomoles of factor X activation peptide liberated as detected by radioimmunoassay using an antiserum directed to the factor X activation peptide (Bauer et al., 1989). These assays were generously performed by Dr. Kenneth A. Bauer, Beth Israel Hospital, Boston, MA. The percent liberation of factor X activation peptide in PN-2/A␤PPor its KPI domain-treated samples was calculated from the ratio of nanomole activation peptide released when the inhibitor was present versus the nanomole activation peptide re-leased when no inhibitor was present times 100. The percent liberated peptide in the inhibitor-treated sample was utilized as a measure of residual FIXa activity to calculate inhibition constants.
Measurement of Prothrombin Activation-Activation of human prothrombin (FII) (1000 nM) by FXa (1 nM) in the presence of 25 M PSPC, 5 nM bovine factor Va, and 0.7 mM HD-Phe-Pip-Arg-p-nitroanilide-2HCl (S2238) (Kabi-Pharmacia, Franklin OH) was performed at 37°C in 0.025 M Hepes, 0.175 M NaCl, pH 7.5, containing 5 mg/ml bovine serum albumin and 2 mM Ca 2ϩ (Rosing et al., 1980(Rosing et al., , 1993. After a 4-min hydrolysis, the reaction was stopped with an equal volume of 50% acetic acid and read in a microplate reader at 405 nm. Factor Xa itself did not hydrolyze this substrate. Preliminary studies revealed that at a 4-min activation time, there was sufficient substrate present such that hydrolysis was on an upward slope and there was no substrate exhaustion over the time of measurement. When investigations with inhibitors were performed, FXa (1 nM) and its inhibitor (PN-2/A␤PP, KPI domain, or antithrombin III) (2-10 nM) were preincubated or not with PSPC and FVa for 5 min followed by starting the reaction with the addition of FII and the chromogenic substrate.
Measurement of Prothrombin Fragment F 1ϩ2 -Simultaneous samples of FXa activation of FII in the presence of PSPC and FVa were prepared for both chromogenic and immunochemical determination of FII activation. Prothrombin fragment F 1ϩ2 (Lau et al., 1979) was measured using a commercial kit from Baxter Diagnostics, Inc., graciously provided by Viola Sotomayer of Baxter Diagnostics. The percent liberation of prothrombin fragment F 1ϩ2 in PN-2/A␤PP-or its KPI domaintreated samples was calculated from the ratio of nanomole of activation peptide released when the inhibitor was present versus the nanomole of activation peptide released when no inhibitor was present times 100. The percent liberated peptide in the inhibitor-treated sample was utilized as a measure of residual FXa activity to calculate inhibition constants.
Calculation of Kinetic Parameters and Constants-The K m and V max of human FXa activation of FII in the presence of PSPC (25 M) and FVa (5 nM) were determined by measuring the rate of FII activation (20 -1600 nM) by 1.0 nM FXa in four independent experiments. The mean Ϯ S.E. of each point performed in triplicate in four individual experiments were analyzed on a double reciprocal plot by linear regression. The K m and V max were determined from a substrate/velocity plot and the negative reciprocal of the x-and y-intercepts of the double reciprocal plots, respectively. The nanomolar ␣-thrombin formed was determined by comparing the level of hydrolysis seen in the present assay with the level of hydrolysis measured by known concentrations of human ␣-thrombin under identical assay conditions. The turnover numbers for factor IIa formation (k cat ) were determined by the ratio of the maximum concentration of factor IIa formed (V max ) divided by the concentration of the forming enzyme (FXa). The stoichiometry of FXa inhibition by PN-2/A␤PP and its KPI domain was determined by nonlinear regression as previously reported (Schmaier et al., 1993). Briefly, FXa at 1 nM was added to 2-100 nM inhibitor (PN-2/A␤PP or its KPI domain) and the residual FXa activity was determined. The plotted x-intercept of the inhibitor concentration versus the % inhibition of FXa activity indicated the concentration of added inhibitor to the known amount of added FXa.
The equilibrium inhibition constants (K i ) presented for PN-2/A␤PP and its KPI domain were calculated as previously reported (Van Nostrand et al., 1990b) by the procedure of Bieth (1984) for tight-binding inhibitors using the following equation: to yield an apparent K i where (I) is the inhibitor concentration, (E) is the factor Xa concentration, and a is the residual factor Xa activity after incubation with the inhibitor. The actual K i was calculated using the subsequent equation: is the concentration of the substrate, factor II, and K m is the Michaelis constant for the factor Xa-factor II (protease-substrate) reaction (Bieth, 1984). The second-order association rate constants (k assoc ) for each of the inhibitors were calculated using the integrated second-order rate equation: where E is the FXa concentration, I is the inhibitor concentration, EI is the concentration of the FXa-inhibitor complex, and t is the time of incubation in minutes (Gigli et al., 1970).
Determination of complex formation between Factor Xa and PN-2/ A␤PP-Complex formation between PN-2/A␤PP and FXa was demonstrated by solid phase binding assay. Microtiter plates were coated with PN-2/A␤PP (250 ng) in 0.1 M Na 2 CO 3 , pH 9.6, and then blocked with 1% radioimmunoassay grade bovine serum albumin (Sigma). After washing, the wells were incubated with FXa (50 ng) followed by a mouse anti-human FX/Xa antibody (1 g/ml) (American Diagnostica, Inc., Greenwich, CT) in 0.01 M sodium phosphate, 0.15 M NaCl, pH 7.4, containing 0.05% Tween 20. After further incubation and washing, a rabbit anti-mouse antibody conjugated with alkaline phosphatase (Sigma number 2429 at 1/1000) was added. The color reaction was initiated by the addition of p-nitrophenyl phosphate disodium (1 mg/ml) in 0.05 M Na 2 CO 3 , 1 mM MgCl, pH 9.8. An additional solid phase binding assay for complex determination was performed by linking FXa (50 ng) in 0.1 M Na 2 CO 3 , pH 9.6, to the microtiter plate. After blocking the cuvette wells with bovine serum albumin, they successively were incubated with PN-2/A␤PP (250 ng) and monoclonal antibody P2-1 to PN-2/A␤PP in ascites fluid (Van Nostrand et al., 1989) followed by detection with a rabbit anti-mouse antibody conjugated with alkaline phosphatase (Sigma number 2429 at 1/1000 dilution).
Protein and Peptide Sequencing-Factor Xa was incubated with PN-2/A␤PP (1 to 16 parts factor Xa to 1 part PN-2/A␤PP, mol/mol) in 0.05 M Tris-HCl, 0.15 M NaCl, pH 7.4, containing 2 mM Ca 2ϩ for 1 h at room temperature. The reaction was stopped with sample buffer for SDS-PAGE. The PN-2/A␤PP samples (1.0 g/lane) were electrophoresed on a 6% SDS-PAGE and then electroblotted onto nitrocellulose. After blocking with Blotto, the protein was detected using monoclonal antibody P2-1 followed by a second antibody conjugated with horseradish peroxidase (Van Nostrand et al., 1990b) using the chemiluminescence system of Amersham. Factor Xa cleaved PN-2/A␤PP in 4:1 (mol/mol) ratio of FXa to PN-2/A␤PP was electrophoresed on 15-22 and 6% SDS-PAGEs, respectively, followed by electroblotting onto a Problot membrane (Applied Bioscience) and detected by staining. The NH 2 -terminal amino acid sequences of an 8.2-kDa peptide seen on the 15-22% SDS-PAGE and of the cleaved PN-2/A␤PPs at 116, 97, and 90 kDa seen on the 6% SDS-PAGE were determined from the Problot membrane in the Protein and Carbohydrate Structure Facility at the University of Michigan, Ann Arbor, MI.
Statistics-Significance of difference in results between groups was measured by t test for groups of unpaired data.

PN-2/A␤PP Inhibits Factor
Xa-PN-2/A␤PP was found to produce a K i of FIXa of 1.5 ϫ 10 Ϫ10 M (a mean of two experiments) on chromogenic detection of a polylysine based-factor X activation assay versus a K i of FIXa of 6.9 ϫ 10 Ϫ9 M when the same samples were studied on a factor X activation peptide assay. This 46-fold difference in inhibitory ability of the same samples on two different factor IXa enzymatic assays suggested that on the chromogenic assay, whose results are determined indirectly through factor X activation, additional inhibition must have occurred than that of FIXa alone. This result prompted the present investigation to determine if PN-2/A␤PP and its KPI domain were direct inhibitors of FXa. In independent experiments, the K m of FXa for the chromogenic substrate N-tosyl-Gly-Pro-Arg-p-nitroanilide was found to be 0.26 Ϯ 0.1 mM. Using conditions identical to the polylysine-based FX activation assay by FIXa, PN-2/A␤PP inhibited FXa with a K i of 1.6 Ϯ 0.4 ϫ 10 Ϫ10 M (Table I). If polylysine were excluded from the buffer of the assay, PN-2/A␤PP inhibited FXa with a K i of 4.5 Ϯ 2.3 ϫ 10 Ϫ8 M (Table I). These data indicated that poly-lysine itself potentiated the inhibition of FXa by PN-2/A␤PP as measured on a chromogenic assay. A␤PP 695 , an isoform of A␤PP which does not contain the KPI domain, did not inhibit FXa when used in up to 10-fold molar excess to enzyme. The discordance in inhibition of FIXa by PN-2/A␤PP of the same samples between the chromogenic and FXa activation peptide assays also can be explained by the inhibitor blocking generated FXa. Additional investigations were performed to determine the degree of FXa inhibition by PN-2/A␤PP and its KPI domain under various conditions. Similar to the results seen with isolated FXa and PN-2/A␤PP, the presence of PSPC, FVIIIa, and/or FVa did not influence the degree of inhibition of FXa by PN-2/A␤PP and its KPI domain (Table I). These data indicated PN-2/A␤PP and its KPI domain were equipotent inhibitors of FXa. Furthermore, the stoichiometry of FXa inhibition by PN-2/A␤PP was 1:1. However, at 4 orders of magnitude molar excess KPI domain to FXa, FXa activity was not reduced to zero.
PN-2/A␤PP and Factor Xa Interactions-Investigations next were performed to determine if FXa and PN-2/A␤PP formed a complex as determine by solid phase binding assay (Fig. 1). When PN-2/A␤PP was coupled to microtiter plate wells, FXa specifically bound to the PN-2/A␤PP as detected by an antibody to FXa followed by a second antibody conjugated with alkaline phosphatase (Fig. 1, top). Likewise, when FXa was linked to microtiter plate wells, PN-2/A␤PP specifically bound to the FXa as detected by an antibody to PN-2/A␤PP followed by a secondary antibody conjugated with alkaline phosphatase (Fig. 1, bottom). These studies indicated that FXa and PN-2/A␤PP formed a complex characteristic of Kunitz-type inhibitors.
Additional investigations showed that PN-2/A␤PP was a substrate of FXa (Fig. 2). When PN-2/A␤PP was incubated with increasing concentrations of FXa (1-16-fold molar excess), there was a decrease in the large, dark 124-kDa band of the starting material on an immunoblot of a nonreduced 6% SDS-PAGE and the appearance of 3 new bands at 116, 97, and 90 kDa, respectively, that migrated further into the gel ( Fig. 2A). At a presumed 1:1 molar ratio of FXa to PN-2/A␤PP some cleavage in PN-2/A␤PP occurred ( Fig. 2A). As the concentration of FXa to PN-2/A␤PP increased from 4:1 to 16:1, all of the 124-kDa starting material was converted into lower molecular mass bands at 116, 97, and 90 kDa. Four-fold molar excess FXa to PN-2/A␤PP liberated a single 8.2-kDa peptide which was detected on a reduced 18% SDS-PAGE (data not shown). The amino terminus sequence of this peptide was LEVPTDG-NAG . . . which is the known amino-terminal sequence of PN-2/A␤PP after cleavage of its signal peptide at alanine 17 (Fig.  3). The liberated peptide was a single peptide because on multiple gel electrophoreses using different percentage acrylamide gels (15-22%), only this single amino-terminal sequence was obtained. These data indicated that FXa liberated an aminoterminal peptide from PN-2/A␤PP. Further investigations sought the FXa cleavage site on the amino-terminal side of PN-2/A␤PP. When PN-2/A␤PP (a major band at 124 kDa and two minor bands at 105 and 98 kDa) was cleaved by 4-fold molar excess FXa, three new corresponding lower molar mass bands were detected (a major one at 115 kDa and two minor bands at 97 and 90 kDa) when the sample was reduced and electrophoresed on a 6% SDS-PAGE, as detected by a Coomassie Blue staining (Fig. 2B). The amino-terminals of each of these three bands were sequenced and a single amino acid sequence (KQCKTHPHFV . . . ) was found for all three of these bands (Fig. 3). These data indicated that molar excess FXa to PN-2/A␤PP cleaved PN-2/A␤PP at a single site after arginine 102. Additional investigations showed that A␤PP 695 , a a FXa (1 nM) was incubated in 0.1 M triethanolamine, 0.1 M NaCl, pH 8.0, containing 0.1% polyethylene glycol and 0.2% bovine serum albumin in the absence or presence of 60 nM polylysine as indicated. After a 5-min incubation with PN-2/A␤PP (2-10 nM), the reaction was started by the addition of substrate (see "Experimental Procedures").
b FXa (1 nM) was incubated in 0.02 M Hepes, 0.15 M NaCl, pH 7.4, containing 0.1% polyethylene glycol and 0.5 mg/ml bovine serum albumin in the presence of 2 mM Ca 2ϩ and PSPC. After a 5-min incubation with PN-2/A␤PP or its KPI domain (2-10 nM), the reaction was started by the addition of substrate (see "Experimental Procedures"). c PSPC were used at 25 M; FVIIIa was used at 4.8 units/ml; and FVa was used at 5 nM.
form of A␤PP which does not contain the KPI domain and does not inhibit FXa, also was a substrate of the enzyme (Fig. 2C). These data indicated that inhibition of FXa was independent of the inhibitor being a FXa substrate. Further isolated KPI do-main was not cleaved by FXa (data not shown).
PN-2/A␤PP Inhibits Factor Xa in the Prothrombinase Complex-The possible importance of PN-2/A␤PP and its KPI domain to inhibit FXa is dependent upon whether these proteins produce inhibition in biologic assays. Studies were performed to determine if PN-2/A␤PP and its KPI domain could inhibit FXa in the prothrombinase complex. Initial experiments determined the K m and k cat of prothrombin activation on phospholipid vesicles in the presence of FVa. By double reciprocal plot, FXa activation of FII in the presence of PSPC and FVa was shown to have a mean K m of 0.62 M (range 0.23-1.6 M) with a V max of 7 nM ␣-thrombin formed min Ϫ1 (range 3.6 -19.1 nM min Ϫ1 ) (Fig. 4). These values calculated to a k cat of 7 min Ϫ1 and k cat /K m of 10.2 M Ϫ1 min Ϫ1 . Both PN-2/A␤PP and its KPI domain blocked FXa activity in the prothrombinase complex (Table II). When FXa and PN-2/A␤PP or its KPI domain were FIG. 1. Solid phase binding assay between PN-2/A␤PP and FXa. Top, purified PN-2/A␤PP (250 ng) was linked to microtiter plate cuvette wells (see "Experimental Procedures"). After blocking with bovine serine albumin, purified FXa (50 ng), antibody to Factor X, and a second antibody to detect the antibody to Factor X was added in sequential order. In various wells, one of the components of the complex assay was excluded. No APP represents wells where no PN-2/A␤PP were linked to the cuvette wells; No FXa indicates wells where no FXa was added; No AntiFXa represents wells were no primary antibody to Factor X was added; No 2nd Ab indicates wells where no alkaline phosphatase-conjugated second antibody were added; and ALL represents cuvette wells were all components were added. The data presented represents the mean Ϯ S.E. of nine experiments. Bottom, purified FXa (50 ng) was linked to microtiter plate cuvette wells (see "Experimental Procedures"). After blocking with bovine serine albumin, purified PN-2/A␤PP (250 ng), antibody to PN-2/A␤PP, and a second antibody to detect the antibody to PN-2/A␤PP was added in sequential order. In various wells, one of the components of the complex assay was excluded. No FXa represents wells where no Factor Xa was linked to the cuvette wells; No APP indicates wells where no PN-2/A␤PP was added; and No AntiAPP represents wells were no primary antibody to PN-2/ A␤PP was added. The data presented represents the mean Ϯ S.E. of nine experiments.  (Ponte et al., 1988). The arrow after arginine 102 represents the FXa cleavage site in PN-2/A␤PP. preincubated for 5 min with PSPC and FVa prior to the addition of FII and chromogenic substrate, the K i were 1.9 Ϯ 1.5 ϫ 10 Ϫ8 M and 1.3 Ϯ 0.8 ϫ 10 Ϫ8 M, respectively (Table II). These results only were 2.8-and 4.9-fold more inhibition for PN-2/ A␤PP and its KPI domain, respectively, than the K i determined (5.3 Ϯ 2.5 ϫ 10 Ϫ8 M and 6.4 Ϯ 2.3 ϫ 10 Ϫ8 M, respectively) when the incubation mixture included the inhibitor, PSPC, FVa, and FII and the reaction was started by the addition of FXa (Table  II). Under these latter conditions, 100-fold molar excess of substrate (i.e. 1 M FII) had little influence on PN-2/A␤PP (2-10 nM) blocking 1 nM FXa. Regardless of the order of addition of reactants, both PN-2/A␤PP and its KPI domain inhibited FXa to the same degree, indicating that the inhibitory domain of PN-2/A␤PP for FXa was confined to the KPI region and not due to being a competing substrate. In the prothrombinase complex, 4 orders of magnitude molar excess KPI domain did not bring the level of FXa activity to zero. When samples were prepared for measurement of FXa enzymatic activity by a chromogenic assay and measurement of prothrombin fragment F 1ϩ2 , the degree of inhibition seen was the same (Table III). These data indicated that PN-2/A␤PP and its KPI domain did not inhibit ␣-thrombin, consistent with other reports (Van Nostrand et al., 1990a;Smith et al., 1990;Schmaier et al., 1993).
Investigations were next performed to determine the comparative inhibitory abilities of PN-2/A␤PP versus antithrombin III and heparin on isolated FXa and FXa in the prothrombinase complex (Table IV). When PN-2/A␤PP was incubated for 5 min with isolated FXa in the presence of polylysine, its second-order association rate constant (k assoc ) (1.4 Ϯ 0.4 ϫ 10 7 M Ϫ1 min Ϫ1 ) for FXa inhibition was 5-fold higher than that of antithrombin III and heparin (k assoc ϭ 3.0 Ϯ 3.3 ϫ 10 6 M Ϫ1 min Ϫ1 ). Interestingly if heparin were present in the reaction mixture with PN-2/A␤PP, no inhibition of FXa was detected (data not shown). Alternatively, if polylysine was absent from the reaction mixture, the inhibitory abilities of PN-2/A␤PP and antithrombin III were reversed. PN-2/A␤PP inhibited FXa with a k assoc ϭ 3.0 Ϯ 2.0 ϫ 10 5 M Ϫ1 min Ϫ1 ; antithrombin III and heparin blocked FXa with a k assoc ϭ 1.3 Ϯ 0.3 ϫ 10 7 M Ϫ1 min Ϫ1 . The data indicated that polylysine had opposite effects on both PN-2/A␤PP and antithrombin III. In the prothrombinase complex, the biologically relevant assay, the k assoc ϭ 1.8 Ϯ 0.7 ϫ 10 6 M Ϫ1 min Ϫ1 for PN-2/A␤PP was essentially the same as the k assoc ϭ 3.0 Ϯ 0.2 ϫ 10 6 M Ϫ1 min Ϫ1 seen with antithrombin III and heparin. These data indicated that in the prothrombinase complex, PN-2/A␤PP in the absence of heparin and antithrombin III and heparin were equipotent inhibitors of FXa.
b Human factor Xa (FXa) was used at 1.0 nM in all studies; phospholipid vesicles (PSPC) were used at 25 M; bovine Factor Va (FVa) was used at 5 nM; human Factor II (FII) was used at 1 M; and both the KPI domain and PN-2/A␤PP were used at 2-10 nM.
c Preincubation means that the FXa and inhibitor (KPI domain or PN-2/A␤PP) were preincubated with PSPC, FVa, and chromogenic substrate for ␣-thrombin for 5 min prior to the addition of FII. No preincubation means that inhibitor (KPI domain or PN-2/A␤PP), PSPC, FVa, chromogenic substrate, and FII were mixed together and the reaction was started by the addition of FXa.
d The values in parentheses represent the number of independent determinations performed for each of the assay conditions. Each value is the mean Ϯ S.D. of all the independent determinations. e p Ͻ 0.0000007 between the two values. f p Ͻ 0.00008 between the two values.  (5 nM) in the presence of inhibitors (10 nM) (see "Experimental Procedures"). In these experiments, FXa and the inhibitor were preincubated for 5 min after which, FII were added to start the reaction. Multiple aliquots from three different experiments were frozen for future assay.
b The K i were calculated from residual activity determined at the end point of the reaction. The values are the mean Ϯ S.D. of three experiments performed at different times.
c The K i were calculated from the % liberated prothrombin fragment F 1ϩ2 activation peptide at the end point of the reaction which was determined by the ratio of nanomoles of peptide liberated in an inhibited sample over nmoles peptide liberated in an uninhibited sample times 100. The values are the mean Ϯ S.D. of three experiments performed at different times.
ity to tissue factor-factor VIIa Lazarus, 1994a, 1994b). Initial reports suggested that PN-2/A␤PP was not an inhibitor of FXa to any great extent Van Nostrand et al., 1990b); however, other reports, consistent with coagulant assays, suggest otherwise (Kitaguchi et al., 1990;Petersen et al., 1994;Schmaier et al., 1993). Our investigations indicate that PN-2/A␤PP is a direct inhibitor of FXa both as an isolated protein and in the prothrombinase complex. The inhibitory activity of PN-2/A␤PP resides completely in its KPI domain because both the parent protein and its isolated KPI domain inhibit FXa to the same degree. Although the stoichiometry of inhibition appears to be 1:1, inhibition by the KPI domain does not appear to be active site-directed. Four orders of magnitude for the KPI domain to FXa does not reduce the FXa activity to zero, both in assays of isolated FXa and FXa in the prothrombinase complex. These data are different from those found with FIXa. Infinite concentrations of KPI domain abolished FIXa activity in the tenase complex (Schmaier et al., 1995). Nonactive site-directed FXa inhibitors also recently have been described for the hookworm-derived inhibitor of human FXa (Cappello et al., 1995).
We found that when using a polylysine-based FX activation assay, some of the measured inhibition of FIXa by PN-2/A␤PP can be accounted for by PN-2/A␤PP inhibiting generating FXa. Since the degree of FXa generated in this assay is small (1-2 nM), the concentration of PN-2/A␤PP or KPI domain present in the assays would have been sufficient to inhibit both FIXa and the generated FXa (Schmaier et al., 1993(Schmaier et al., , 1995. It also was of interest to learn that polylysine itself potentiated the degree of inhibition of FXa by PN-2/A␤PP and reduced that of antithrombin III/heparin. The mechanism for this independent activity of polylysine is not known. Since polylysine itself can be an independent variable contributing to PN-2/A␤PP's inhibitory ability, it should probably be avoided in assays of FIXa. In addition to inhibition of enzymatic activity, we were able to demonstrate a physical interaction between PN-2/A␤PP and FXa. On a solid phase binding assay, specific complex formation was detected between PN-2/A␤PP and FXa. This information suggests that PN-2/A␤PP is an inhibitor of FXa of the slow, tight class characteristic of Kunitz type inhibitors. PN-2/A␤PP was also a substrate of FXa when the enzyme was in molar excess to inhibitor. It appears that FXa proteolyzes the major band of PN-2/A␤PP at 124 kDa and the two minor bands (105 and 98 kDa) into corresponding lower molecular mass species (116,97,and 90 kDa,respectively), each with the same new amino terminus as seen on immunoblot and Coomassie-stained gels. It is possible that FXa also cleaves PN-2/A␤PP at a single point on the carboxyl-terminal side of the protein liberating an approximate 30 -34-kDa protein. This result would explain the intensification of the post-cleaved 90-kDa band of PN-2/A␤PP seen in Fig. 2, A and B. However, we never have found any evidence of such a band on our Coomassie-stained gels since it would be migrating with one of the subunits of FXa and thus be hidden. The fact that PN-2/A␤PP is a substrate to molar excess FXa does not indicate that its mechanism of inhibition of the enzyme is substrate inhibition. First, A␤PP 695 is a substrate of FXa but it is not a FXa inhibitor. Second, the isolated KPI domain of PN-2/A␤PP which does not contain the FXa cleavage site inhibits FXa to the same degree as its parent protein.
Third, the isolated KPI domain is not cleaved by FXa. It is of interest that FXa cleaves PN-2/A␤PP through its heparin binding domain. Since heparin neutralizes PN-2/A␤PP's inhibitory activity on FXa, cleavage through this domain may preserve the inhibitory function of PN-2/A␤PP for FXa.
PN-2/A␤PP is a potent anticoagulant of FXa in the prothrombinase complex. In our laboratory the K m and k cat /K m ratio of prothrombin activation by FXa is 0.62 M and 10.2 M Ϫ1 min Ϫ1 , respectively, results which are comparable to the findings of other investigators (Krishnaswamy et al., 1987) using a fluorescent marker instead of a chromogenic substrate to monitor prothrombin activation. The degree of inhibition of FXa by PN-2/A␤PP and its KPI domain is to the same order of magnitude in the prothrombinase complex as with the isolated pure enzyme. Regardless of the order of addition of reactants, PN-2/A␤PP and its KPI domain inhibit FXa on PSPC. PSPC in the presence of FVa must have oriented FXa such that it was susceptible to inhibition by PN-2/A␤PP or its KPI domain even though the inhibitors were competing with 5 orders of magnitude more substrate. The finding that PN-2/A␤PP and its KPI domain inhibit FXa in the prothrombinase complex make this class of inhibitors more important than what would be appreciated by just examining isolated FXa inhibition. Since PN-2/ A␤PP is not a plasma protein but rather a cell surface-associated protease inhibitor, influencing FXa activity in the prothrombinase complex suggests that this class of Kunitz-type protease inhibitors may be important regulators of various hemostatic enzymes. In fact, PN-2/A␤PP and its homologue, amyloid ␤-protein precursor-like protein-2, may constitute a new class of serine protease inhibitors modulating hemostasis (Sprecher et al., 1993).
Although our investigations show that artificial agents can influence the degree of FXa inhibition by PN-2/A␤PP or antithrombin III and heparin, in the prothrombinase complex assembly, PN-2/A␤PP and antithrombin and heparin were comparable inhibitors. In the absence of added heparin, the degree of antithrombin III inhibition of FXa was orders of magnitude less potent than that seen with PN-2/A␤PP. In plasma, antithrombin III would be the predominant inhibitor because its plasma concentration is 4 M versus the 30 nM level of PN-2/ A␤PP which may be achievable in plasma when platelets are activated (Van Nostrand, et al., 1991). However, it is not known which may be the predominant inhibitor on cell membranes. The second-order rate constants for antithrombin III and heparin inhibition of FXa that we obtained were 1-2 orders of magnitude higher than that reported by other investigators (Olson et al., 1992;Ellis et al., 1982). Differences in heparin preparations and concentrations and ionic strengths in the buffers may account for these variations. Alternatively, in the presence of heparin, PN-2/A␤PP did not inhibit FXa. PN-2/ A␤PP is known to have a heparin binding domain which allows 3.0 Ϯ 3.3 ϫ 10 6 (9) b,c 1.4 Ϯ 0.4 ϫ 10 7 (10) c Factor Xa NO polylysine a 1.3 Ϯ 0.3 ϫ 10 7 (7) 3.0 Ϯ 2.0 ϫ 10 5 (6) Prothrombinase d complex 3.0 Ϯ 0.2 ϫ 10 6 (4) 1.8 Ϯ 0.7 ϫ 10 6 (4) a FXa was added at 1 nM; PN-2/A␤PP and antithrombin III were added at 2-10 nM. When using antithrombin III, the reaction mixture was made 1 unit/ml with heparin. The FXa and PN-2/A␤PP or antithrombin III were incubated 5 min prior to adding the chromogenic substrate for FXa. When polylysine was present, it was added at 60 nM.
b Each value is the mean Ϯ S.D. of all determinations. The values in parentheses represent the number of determinations performed for the individual assay conditions. c p Ͻ 0.000013 between the two values. d FXa (1 nM) was incubated with 10 nM PN-2/A␤PP or 10 nM antithrombin III and heparin (1 unit/ml) for 5 min in the presence of PSPC (25 M) and FVa (5 nM) at 37°C. The enzymatic reaction was initiated by the addition of FII (1 M) and chromogenic substrate for ␣-thrombin.
heparin to potentiate its inhibition of factor XIa but not factor IXa Van Nostrand et al., 1990b;Schmaier et al., 1993). It is of interest that the FXa cleavage site on PN-2/ A␤PP is at arginine 102 which is in the heparin binding region of the protein (Small et al., 1994). Excess FXa could be preventing PN-2/A␤PP from associating with heparin on cell membranes. These investigations show that PN-2/A␤PP in the absence of heparin and antithrombin III in the presence of heparin are the naturally occurring inhibitors of FXa. Isolated KPI domain of PN-2/A␤PP is a fragment from a naturally occurring human protein which may have potential use as an anticoagulant since its inhibitor activity is equal to the tick anticoagulant peptide (Waxman et al., 1990).