Regulation of Extrinsic Pathway Factor Xa Formation by Tissue Factor Pathway Inhibitor*

Transcription

1 THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 273, No. 8, Issue of February 20, pp , by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. Regulation of Extrinsic Pathway Factor Xa Formation by Tissue Factor Pathway Inhibitor* (Received for publication, November 6, 1997, and in revised form, December 9, 1997) Robert J. Baugh, George J. Broze, Jr., and Sriram Krishnaswamy From the Department of Medicine, Division of Hematology/Oncology, Emory University, Atlanta, Georgia and Division of Hematology, Jewish Hospital, Washington University, St. Louis, Missouri Tissue factor (TF) pathway inhibitor (TFPI) regulates factor X activation through the sequential inhibition of factor Xa and the VIIa TF complex. Factor Xa formation was studied in a purified, reconstituted system, at plasma concentrations of factor X and TFPI, saturating concentrations of factor VIIa, and increasing concentrations of TF reconstituted into phosphatidylcholine: phosphatidylserine membranes (TF/PCPS) or PC membranes (TF/PC). The initial rate of factor Xa formation was equivalent in the presence or absence of 2.4 nm TFPI. However, reaction extent was small (<20%) relative to that observed in the absence of TFPI, implying the rapid inhibition of VIIa TF during factor X activation. Initiation of factor Xa formation using increasing concentrations of TF/PCPS or TF/PC in the presence of TFPI yielded families of progress curves where both initial rate and reaction extent were linearly proportional to the concentration of VIIa TF. These observations were consistent with a kinetic model in which the rate-limiting step represents the initial inhibition of newly formed factor Xa. Numerical analyses of progress curves yielded a rate constant for inhibition of VIIa TF by Xa TFPI (>10 8 M 1 s 1 ) that was substantially greater than the value ( M 1 s 1 ) directly measured. Thus, VIIa TF is inhibited at near diffusion-limited rates by Xa TFPI formed during catalysis which cannot be explained by studies of the isolated reaction. We propose that the predominant inhibitory pathway during factor X activation may involve the initial inhibition of factor Xa either bound to or in the near vicinity of VIIa TF on the membrane surface. As a result, VIIa TF inhibition is unexpectedly rapid, and the concentration of active factor Xa that escapes regulation is linearly dependent on the availability of TF The activation of factor X by the extrinsic pathway of coagulation is considered the initiation step of the clotting cascade (1 4). The enzyme complex responsible for this reaction, referred to as the extrinsic Xase complex, is composed of the protease, factor VIIa, bound to tissue factor (TF) 1 in the presence of membranes and Ca 2 (5). This enzyme complex can activate either factor X or factor IX to their respective serine proteases by specific and limited proteolysis (5). Factor VIIa is the serine protease component of the extrinsic Xase which in itself exhibits very poor activity toward its protein substrates (5). Reversible interactions between factor VIIa and the integral membrane protein cofactor, tissue factor (TF), lead to a profound increase in the catalytic efficiency for factor X activation (5). TF does not require proteolytic activation before it can function as a cofactor in the extrinsic Xase complex. Thus, exposure of cells bearing TF to flowing blood following vascular damage is sufficient to recruit circulating factor VII or VIIa to these sites through the tight and reversible interaction between VII or VIIa and TF (6 11). These interactions lead to the proteolytic activation of the zymogen, factor VII (12), and yield the catalyst responsible for the initiation of coagulation by the extrinsic pathway. Once expressed, the activity of VIIa TF is down-regulated by plasma proteinase inhibitors including tissue factor pathway inhibitor (TFPI) (3). TFPI, a multivalent Kunitz-type inhibitor present at nanomolar concentrations in plasma, is a potent, slow, and tight-binding inhibitor of factor Xa and can also inhibit the VIIa TF complex (3). Of the three Kunitz-type domains in TFPI, the second domain functions in the inhibition of factor Xa, whereas the first is involved in the inhibition of VIIa TF (13). The significance and function of the third Kunitztype domain is unknown (13). The mechanism of action of TFPI has largely been inferred from kinetic studies of the individual reactions. Inhibition of factor Xa by TFPI is rapid and/or highly favored relative to the inhibition of VIIa or the VIIa TF complex (14 16). However, the reaction between TFPI and VIIa TF is greatly enhanced by the presence of factor Xa (13, 15). Thus, efficient inhibition of VIIa TF appears to require the prior reaction of TFPI with factor Xa, implying a stepwise inhibition of the individual enzymes (3, 13). The ultimate formation of a quaternary Xa TFPI VIIa TF complex is suggested by these data and the fact that free TF is not available following the action of TFPI on VIIa TF in the presence of factor Xa (17). These observations form the basis for the presumed mechanism of action of TFPI (Scheme I), indicating that inhibition of VIIa TF function by TFPI is achieved by the product activated feedback inhibition of the catalyst (3). The inferred kinetic mechanism of action of TFPI (Scheme I) implies an important physiological role in gating the initiation * This work was supported by Grants RO1 HL (to S. K.) and RO1 HL (to G. J. B.) from the National Institutes of Health and a grant from The Monsanto Corp. (to G. J. B.). 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. To whom correspondence should be addressed, after 2/98: Joseph Stokes, Jr. Research Institute, Children s Institute of Philadelphia, 310 Abramson, 324 South 34th St., Philadelphia, PA Tel.: ; Fax: The abbreviations used are: TF, recombinant tissue factor apoprotein; PCPS, vesicles composed of 75% (w/w) L- -phosphatidylcholine and 25% (w/w) L- -phosphatidylserine; PC, vesicles composed of L- phosphatidylcholine; SpXa, methoxycarbonyl cyclohexyl-gly-gly-argp-nitroanilide; TAP, recombinant wild type tick anticoagulant peptide; TF/PCPS, PCPS vesicles containing approximately 1 productively oriented TF/vesicle; TF/PC: PC vesicles containing approximately 1 productively oriented TF/vesicle; TFPI, recombinant tissue factor pathway inhibitor; Z-VVR-AMC, benzyloxycarbonyl-val-val-arg-aminomethylcoumarin; BSA, bovine serum albumin; PAGE, polyacrylamide gel electrophoresis. This paper is available on line at

2 Regulation of Factor X Activation by TFPI 4379 SCHEME I. Schematic illustration of the inhibitory mechanism of TFPI. VIIa TF (E) catalyzes the conversion of factor X (S) to factor Xa (P). For the purposes of simplicity, this reaction is illustrated as a classical two-step process governed by the kinetic constants K m and k cat. TFPI (I) inhibits this reaction through a two-step process. The first step (step 1) is considered to involve the inhibition of factor Xa through the reaction between I and P. The Xa TFPI binary complex (PI) formed in the initial step reacts efficiently with E (step 2) to yield an inhibited quaternary complex of VIIa TF, TFPI, and factor Xa. of coagulation through the extrinsic pathway (3). This concept is further supported by several of the following facts: the exposure of small amounts of TF to blood likely represents the controlling element of this pathway as the plasma concentration of factors VII/VIIa ( 10 nm) is saturating relative to the equilibrium dissociation constant for TF binding (6 11, 18); the concentration of the zymogen, factor X ( 170 nm), indicates a much higher achievable concentration of factor Xa relative to the maximum concentration of TFPI in plasma ( 2.4 nm) (19); the circulating concentration of TFPI is influenced by partitioning of the inhibitor to cell surfaces or other circulating proteins (20). These considerations imply that TFPI may function as a threshold regulator of the initiation of the coagulation cascade. Regulation might be expected to be most pronounced at low concentrations of TF which could be overcome by the exposure of TF at concentrations exceeding available TFPI. However, since inhibition of both factor Xa and VIIa TF by physiological (nanomolar) concentrations of TFPI is not instantaneous, the regulation of this pathway is likely to be heavily influenced by the kinetics determining the individual steps. Although the individual reactions of TFPI have been studied by a number of groups, there is uncertainty as to which inhibition reaction corresponds to the rate-limiting step (21, 22). In addition, the effects of TFPI on active factor Xa formation by the extrinsic pathway have not been experimentally documented. We have empirically assessed the ability of TFPI to regulate active factor Xa formation by VIIa TF and the adequacy of the surmised model for TFPI action to describe this process. Kinetics have been examined at reactant (factor X, TFPI, and VIIa TF) concentrations that may be relevant to the initiation of coagulation. These conditions are specifically chosen to assess the ability of the existing model to describe TFPI function and to provide an increased understanding of the reaction steps that are kinetically significant for the regulation of the extrinsic pathway. EXPERIMENTAL PROCEDURES Materials Hen egg L- -phosphatidylcholine (PC) and bovine brain L- -phosphatidylserine (PS) were from Sigma. The synthetic substrates methoxycarbonyl cyclohexyl-gly-gly-arg-p-nitroanilide (Spectrozyme Xa, SpXa) and benzyloxycarbonyl-val-val-arg-aminomethylcoumarin (Z-VVR-AMC) were obtained from American Diagnostica and Bachem, respectively. Stock solutions of SpXa were prepared in water, and the concentration was determined spectrophotometrically using E M M 1 (23). Working solutions of SpXa were prepared in 20 mm Hepes, 0.15 M NaCl, 10 mm EDTA, 0.1% (w/v) PEG-8000, 0.1% (w/v) BSA, ph 7.5. Stock solutions of Z-VVR-AMC ( 300 mm) were prepared in Me 2 SO and stored at 20 C. A working solution of Z-VVR-AMC was prepared daily in water, and the concentration was determined using E M ,200 M 1 (24). Patient plasma obtained by plasmapheresis was donated by the plasmapheresis laboratory of Emory University Hospital. Recombinant human factor VIIa was purchased from Novo-Nordisk (Gentofte, Denmark). Recombinant human tissue factor apoprotein (TF) was a kind gift from Genentech (South San Francisco, CA). Recombinant full-length TFPI was expressed and purified as described (25). Recombinant tick anticoagulant peptide was expressed in Pichia pastoris and purified as described previously (26). All kinetic studies were performed in 20 mm Hepes, 0.15 M NaCl, 5 mm CaCl 2, 0.1% (w/v) PEG-8000, 0.1% (w/v) BSA, ph 7.5 (referred to as Assay Buffer), at ambient temperature (estimated at 23 2 C). Proteins Human factor X was isolated from plasmapheresis plasma, and bovine factor X was purified from bovine plasma as described previously (27). Trace amounts of factor Xa in human factor X preparations were removed by chromatography on soybean trypsin inhibitor Sepharose (Sigma), followed by gel filtration (Sephacryl S-300) to remove traces of inhibitor. These preparations were routinely assessed for residual Xa activity ( 0.005%) as well as unforeseen trace contamination by protease inhibitors. Factor Xa was prepared using factor X from either source by proteolytic activation using purified factor X activator (RVV x-cp ) from Russell s viper venom followed by purification as described (28). Kinetic titration of Xa using p-nitrophenol p -guanidinobenzoate (ICN Pharmaceuticals) (29) yielded 0.95 and 1.15 active sites/mol of protein for human and bovine Xa, respectively. The purity of all protein preparations was evaluated by SDS-PAGE (30). Protein concentrations were determined using the following molecular weights and extinction coefficients (E 0.1% 280 ): human X, 56,500 and 1.16 (31); bovine Xa, 45,300 and 1.24 (32); human Xa, 45,300 and 1.16 (31); human recombinant factor VIIa, 50,000 and 1.39 (33); TAP, 6,980 and 2.56 (34). The concentration of TFPI was determined using M r 0.1% 32,000 and E calculated from the primary structure by the method of Gill and von Hippel (35) and confirmed by quantitative amino acid analysis. Tissue factor was reconstituted into membranes containing 75% (w/w) PC and 25% (w/w) PS (PCPS) or 100% PC using n-octyl- -Dglucopyranoside (Calbiochem) and exhaustive dialysis at a controlled rate as described previously (36). The recovery of TF in the reconstituted mixture was assessed using known amounts of 125 I-labeled TF added as tracer to the initial reaction mixture. The final effective concentration of TF following dialysis was considered to be one-half the total concentration due to the two possible orientations of the protein (8). Phospholipid concentrations were determined following oxidation by a colorimetric assay for inorganic phosphate using Malachite Green (37). Quasi-elastic light scattering measurements (Nicomp 370, NICOMP Instruments) of the resultant TF/PCPS and TF/PC vesicles yielded gaussian distributions centered at d nm for TF/ PCPS and d nm for TF/PC. These dimensions, the total phospholipid concentration, and the published dimensions for head group surface area (38) were used to determine the vesicle concentration and the number of correctly oriented tissue factors per vesicle. Henceforth, TF/PCPS and TF/PC refer to tissue factor reconstituted into PCPS or PC membranes with approximately one (mean 1.02, range ) productively oriented TF molecule per vesicle. Stock solutions of TFPI ( 350 M, 11 mg/ml) stored in 2 M urea were diluted into distilled water to a final concentration not exceeding 1 M. Following brief centrifugation and absorbance measurements to determine concentration, the TFPI was further diluted in Assay Buffer to a useful working concentration of nm. The active concentration of TFPI was determined by titration with bovine factor Xa before every experiment (39). The active concentration of TFPI interpolated from these titrations ranged between 55 and 65% of the total protein present. The stated concentrations of TFPI correspond to the active concentration determined in these titration experiments. The inhibitory activity of TFPI was found to decrease with storage at 20 C with no obvious deterioration detected by SDS- PAGE. Preparations with reduced inhibitory activity yielded exactly the same rate constants when corrected for the active concentration of TFPI. Modification of Factor X with 125 I Factor X was labeled with 125 I using solid phase lactoperoxidase by modifications to procedures described previously (40, 41). Lactoperoxidase-Sepharose (2 IU/ml) (Worthington) was washed with 20 mm Hepes, 0.15 M NaCl, 0.1% (w/v) PEG, 5 mm CaCl 2, ph 7.5 (Labeling Buffer) by repeated centrifugation prior to use. The iodination mixture (50 l) contained 0.8 units/ml lactoperoxidase beads, 180 nm factor X, 8.6 M KI, and 1 mci of Na 125 I. The reaction was initiated by the addition of 28 M H 2 O 2, gently agitated at room temperature for 6 min, and quenched by the addition

3 4380 Regulation of Factor X Activation by TFPI of 14 mm NaN 3. Factor X was separated from the reactants by the use of a centrifuge column (4 ml, Sephadex G-25) precoated with 10% (w/v) BSA and extensively washed with Labeling Buffer. Typical yields were dpm/ g of protein. Validation of the Use of 125 I-Labeled Factor X Reaction mixtures containing factor X and 125 I-labeled factor X added as a tracer in Assay Buffer were initiated with the preassembled extrinsic Xase complex (TF/PCPS; 1 nm TF, 130 M PCPS, 2 nm VIIa) or 5 nm RVV x-cp. Progress curves for factor Xa formation were determined in EDTA-quenched samples from discontinuous measurements of the initial velocity of SpXa hydrolysis as described previously (27). Samples were also analyzed by SDS-PAGE both with and without disulfide bond reduction using dithiothreitol. Dried gels were exposed to an imaging plate, and radioactivity was detected and quantitated using a PhosphorImager (Molecular Dynamics) following detection of luminescence by laser scanning. Quantitation was performed by integrating the volumes (density/area) of each band to yield the extent of factor X activation calculated as a ratio of the intensity of the factor X or Xa bands relative to the total integrated volume in each lane. Control experiments established a linear relationship between the radioactivity loaded on the gel and the integrated volume. Results obtained from discontinuous measurements of SpXa hydrolysis were indistinguishable from progress curves of factor X activation inferred from the radioactive bands. It was concluded that under steady state conditions, the kinetics of cleavage of 125 I- labeled factor X was equivalent to that of unlabeled factor X. Measurement of Xa Formation by Extrinsic Xase in the Presence of TFPI The final concentration of reactants was 170 nm factor X, 2.4 nm TFPI, 2 nm VIIa, and varying concentrations of TF/PCPS or TF/PC. Factor Xa formation was initiated in wells of a 96-well plate (Corning Assay Plate) by mixing equal volumes (50 l each) of preassembled VIIa TF with a mixture of factor X and TFPI in Assay Buffer. To minimize adsorption artifacts, the plates were pretreated with Assay Buffer lacking Ca 2 but containing 0.02% (v/v) Tween 20 for 1 h, centrifuged to remove solvent, and air-dried before use. Following incubation for varying periods ( min) at room temperature, further factor X activation was quenched by the addition of 100 l of0.4 mm SpXa in 20 mm Hepes, 0.15 M NaCl, 10 mm EDTA, 0.1% (w/v) PEG-8000, 0.1% (w/v) BSA, ph 7.5, and the initial velocity of SpXa hydrolysis was measured by continuously monitoring the change in absorbance at 405 nm in a kinetic plate reader (Molecular Devices). The concentration of factor Xa was inferred from the linear dependence of the rate on known concentrations of factor Xa assayed under the same conditions. Initial rates for factor X activation in the absence of TFPI were determined using reaction mixtures containing identical concentrations of enzyme components and substrate from serially quenched samples as described previously (27). In separate experiments, 125 I-labeled factor X was present as a tracer in the substrate solution to permit analysis by SDS-PAGE. Aliquots were quenched at various times following initiation by mixing with an equal volume of 62.5 mm Tris, 2% (w/v) SDS, 0.01% (w/v) bromphenol blue, 10% (v/v) glycerol, 16 mm EDTA, ph 6.8. Samples were then heated (5 min, 80 C) in the absence or presence of 100 mm dithiothreitol and subjected to electrophoresis. Approximately 30,000 dpm was applied per lane, and quantitation of factor Xa produced was performed as described above. Measurement of the Overall Dissociation Constant (K i *) for the Inhibition of Factor Xa Reaction mixtures (150 l) in wells of a 96-well plate contained 0.1 or 0.2 nm Xa and increasing concentrations of TFPI in Assay Buffer. Following incubation for 2, 4, 6, and 8 h, the concentration of free (uninhibited) factor Xa was determined from the initial steady state rate of SpXa hydrolysis initiated by the addition of 50 l of 0.4 mm SpXa prepared in the same buffer. The half-time for the dissociation of TFPI from Xa is established to be substantially greater than the 5 min required for the initial rate measurement (25). Therefore, corrections for the perturbation of the initial equilibrium by dilution or due to competitive effects of SpXa are not necessary. The final concentrations of Xa and TFPI used in the analysis were therefore those present in the initial 150- l incubation mixture. Measurement of the Rate Constant for the Inhibition of Factor Xa The rate constant for the inhibition of factor Xa was inferred from exponential decays in the rate of SpXa hydrolysis in the presence of TFPI. Reaction mixtures (150 l) containing SpXa and increasing concentrations of TFPI in Assay Buffer were initiated with 50 l of factor Xa in the same buffer. The final concentrations of reactants were 200 M SpXa, the indicated concentrations of TFPI, and 0.05 nm factor Xa. Following mixing by brief vibration, SpXa hydrolysis was monitored by continuously measuring absorbance at 405 nm. Data acquired over 2 h were corrected for the measured 30-s interval between the initiation and the onset of absorbance measurements. Measurement of the Overall Dissociation Constant (K i *) for the Inhibition of VIIa TF by TFPI Alone or by the Xa TFPI Binary Complex VIIa TF activity was measured by the hydrolysis of the fluorescent substrate Z-VVR-AMC. Fluorescence measurements were performed using an SLM 8000C fluorescence spectrophotometer with adapted hardware and software (OLIS, Bogart, GA) as described (41). Ratiometric fluorescence was measured in 1-cm 2 stirred quartz cuvettes maintained at 25 C using ex 380 nm and measuring broad band emission intensity ( em 400 nm) with a long pass filter (Schott KV 399) in the emission beam. The overall dissociation constant, K i *, for the inhibition of VIIa TF by the Xa TFPI binary complex was assessed at 0.25 nm VIIa, 1 nm TF/ PCPS, 0.6 nm TFPI, and increasing concentrations of Xa after incubation for 2hatroom temperature. Uninhibited VIIa TF activity was assessed by the addition of 1 M TAP to rapidly inhibit traces of free factor Xa followed by 60 M Z-VVR-AMC. The initial velocity for peptide hydrolysis was determined by monitoring the linear, steady state increase in fluorescence over 5 min. Since the concentration of TFPI was in excess and well above the K i * for the inhibition of factor Xa, the concentration of the Xa TFPI binary complex was considered approximately equal to the limiting concentration of factor Xa. To accommodate the uncertainty with this assumption, the overall dissociation constant was also determined using increasing concentrations of TFPI at a single saturating concentration of factor Xa (1 nm) or by simultaneously varying both Xa and TFPI maintained at a constant ratio (1:0.9) to each other. Measurement of the Rate Constant for VIIa TF Inhibition by the Xa TFPI Complex Reaction mixtures (2 ml) contained the indicated concentrations of the preformed Xa TFPI complex (a fixed, saturating concentration of TFPI and increasing concentrations of factor Xa preincubated for 2 5 h) and 1 nm TF/PCPS in Assay Buffer. Following the initial incubation, 60 M Z-VVR-AMC was added, and substrate hydrolysis was initiated with varying concentrations of factor VIIa (0.1, 0.2, and 0.3 nm). The inhibition of VIIa TF was inferred from exponential decays in the rate of Z-VVR-AMC hydrolysis in the presence of Xa TFPI complex over a 20-min period. In these experiments, (Xa TFPI) 3 (VIIa TF) to maintain pseudo first-order conditions. Data Analysis Data were analyzed according to the indicated functions by nonlinear least squares regression analysis using the Marquardt algorithm (42). The uncertainty of the fitted terms is represented by 95% confidence limits. Linear regression analysis of replicated measurements was performed by weighting the data using the reciprocal of the standard deviation (42). Simulations according to the stated ordinary differential equations was performed using the program DynaFit (43), generously provided as a gift by Petr Kuzmic, BioKin, Madison, WI. This program permits the combination of numerical solutions with nonlinear least squares regression analysis to yield fitted values of selected rate constants (43). Determination of the Overall Equilibrium Dissociation Constant (K i *) Initial velocity measurements of peptidyl substrate hydrolysis following prolonged incubation of enzyme (E) with increasing concentrations of inhibitor (I) were analyzed according to Equations 1 and 2. E i n I E K i* n I E K i * 2 4 n I E 2 (Eq. 1) v obs v E i v o E E i (Eq. 2) where E and I refer to total concentrations of E and I; E i is the inhibited concentration of E; n reflects the moles of I that combine per mol of E at saturation to yield the inhibited species, and K i * is the overall equilibrium dissociation constant for the interaction. The observed initial velocity (v obs ) is related to the concentration terms through the specific activity of free E (v o ) and of E saturated with I (v ). Initial velocities measured at increasing concentrations of I at two or more fixed concentrations of E were analyzed by Equations 1 and 2 to yield fitted values of n, K i *, v o, and v. For limited data sets, the analysis was conducted assuming n 1. Determination of the Rate Constant for Inhibition Progress curves for the exponential decay in peptidyl substrate hydrolysis by enzyme E in the presence of inhibitor I were analyzed according to Equations 4 6 of Williams et al. (44) describing the (pseudo) first-order decay from an instantaneous initial velocity at zero time to a steady state rate and including corrections for inhibitor depletion and competition by the indicator substrate. The significance of the derived rate constant is heavily dependent on the presence of multiple steps in the inhibition

4 pathway, the activity contributions of the different species, and the suitability of the rapid equilibrium assumption. However, despite the possible presence of multiple steps in the inhibition scheme, when I is substantially less than the equilibrium dissociation constant for the initial binding of E to I, the derived k obs is related to I (44) by Equations 3 and 4. Regulation of Factor X Activation by TFPI 4381 I effective K i* 1 S/K m E I 2 4 E I 1 S/K m (Eq. 3) k obs k k INH I effective (Eq. 4) where S and K m refer to the concentration of the indicator peptidyl substrate and its Michaelis constant; K i * is the overall equilibrium constant, and E and I refer to total concentrations of enzyme and inhibitor. The observed rate constant is predicted to vary linearly with the effective concentration of inhibitor. The derived rate constants k INH and k represent the overall forward association rate constant for inhibition (k INH ) and the first-order rate constant for the regain of activity from the inhibited species (k ) which adequately describe the inhibition reaction without ascribing any specific physical significance to the rate constants. For inhibition studies with factor Xa measured with SpXa, I effective was calculated using determined values of K i * 25.4 pm and K m 59.2 M. For measurements with VIIa TF and Z-VVR-AMC, the determined K i * was 8.1 pm, and the concentration of the peptidyl substrate was established to be well below the K m thereby permitting the simplification (1 S/K m ) 1. RESULTS AND DISCUSSION Experimental Design TFPI-dependent regulation of active factor Xa formation by the extrinsic Xase complex was studied at approximate plasma concentrations of factor X (170 nm) and TFPI (2.4 nm) using saturating concentrations of factor VIIa (2 nm) and increasing concentrations of TF (0 1 nm) reconstituted into PCPS membranes. Since the TF/PCPS concentration is limiting and the factor VIIa concentration was high relative to the K d for the VIIa TF/PCPS interaction (6 11), the concentration of the extrinsic Xase complex is determined by the concentration of TF/PCPS. This experimental design is based on the prevailing idea that exposure of TF triggers the initiation of coagulation by the extrinsic pathway. Although the precise concentration range of TF relevant to the initiation of coagulation in vivo is unknown, the conditions chosen in the present study maintain limiting catalyst (VIIa TF/PCPS) concentrations relative to the concentration of TFPI, thereby restricting observations to the domain where the kinetics of active factor Xa formation is expected to be strongly influenced by both inhibitory reactions of TFPI. Therefore, these conditions are suitable for testing the ability of quantitative information developed from the studies of the isolated component reactions and Scheme I to describe the TFPI-dependent regulation of factor Xa production by the extrinsic Xase complex. Effect of TFPI on Active Factor Xa Formation A comparison of progress curves for active factor Xa production by the VIIa TF/PCPS complex under these conditions in the presence and absence of 2.4 nm TFPI is presented in Fig. 1. In contrast to the rapid and near complete conversion of factor X to factor Xa within approximately 5 min in the absence of the inhibitor, a blunted progress curve was observed in the presence of TFPI indicating the rapid and potent inhibition of the activation reaction. Although initial rate was largely independent of TFPI, the active factor Xa concentration rapidly reached a constant steady state value within approximately 30 s. The maximum concentration of active factor Xa formed represented a small fraction ( 20%) of the total activable factor X present. Analysis of factor X cleavage by SDS-PAGE (Fig. 1) confirmed these interpretations. The agreement between factor X cleavage determined by direct physical measurements and the active concentration of factor Xa formed suggests that the significantly reduced amplitude of the progress curve in the presence of TFPI arises from rapid inhibition of VIIa TF/PCPS during FIG. 1. Factor X activation by the extrinsic pathway in the presence of TFPI. Factor Xa formation was determined in reaction mixtures containing 170 nm factor X, 2 nm factor VIIa, and 1 nm TF/PCPS (1 nm VIIa TF/PCPS) either in the presence ( ) or absence (Œ) of 2.4 nm TFPI. Active factor Xa produced at the indicated times following initiation was determined as described under Experimental Procedures. Factor X cleavage in identical reaction mixtures either in the presence (E) or absence ( ) of TFPI was also measured separately using 125 I-labeled factor X and SDS-PAGE analysis. factor X activation. Regulation of Factor Xa Formation by TFPI Systematic studies of the fate of factor Xa formed in the presence of TFPI following initiation of factor X activation at different concentrations of the extrinsic Xase complex (VIIa TF/PCPS) are illustrated in Fig. 2A. The progress curves were characterized by a rapid increase in the active Xa concentration and reached an essentially constant concentration within approximately 30 s. The amplitude of each progress curve represented a small fraction of the total activable factor X present and increased with increasing concentrations of catalyst (VIIa TF/PCPS). Equivalent results were obtained when the factor X concentration was doubled or at higher fixed concentrations of factor VIIa (not shown). Progress curve amplitude was inversely dependent on the fixed concentration of TFPI. However, a similar family of progress curves as illustrated in Fig. 2A was obtained at different fixed TFPI concentrations. These observations indicate that the qualitative behavior of this system does not result from the fortuitous selection of a single set of reactant concentrations. Comparable effects on factor Xa formation by the extrinsic Xase complex were observed when the enzyme was assembled using TF reconstituted into PC membranes (Fig. 2B). However, the maximal limiting amplitude of active factor Xa produced was established more slowly and reached a lower amplitude in comparison to the equivalent concentrations of catalyst assembled on PCPS membranes (Fig. 2A). In agreement with previous observations (45), the initial rate of factor Xa formation in the absence of TFPI was approximately 30-fold lower under these conditions when compared with initial rates observed with PCPS membranes (not shown). Thus, provided the concentration of extrinsic Xase does not exceed the inhibitor concentration, regulation by TFPI leads to a dependence of the amplitude of product formation on the rate of factor X activation rather than the concentration of available substrate, factor X. This conclusion is supported by the linear dependence of the amplitude of product formed as a function of increasing concentrations of initiating catalyst on both types of membranes (Fig. 2C). These observations establish the regulatory effect of TFPI on factor Xa formation by this pathway. In contrast to the unregulated reaction where initial rate is dependent on the

5 4382 Regulation of Factor X Activation by TFPI FIG. 2.Regulation of active factor Xa formation by TFPI. Panel A, progress curves for factor Xa formation were determined in reaction mixtures containing 170 nm factor X, 2 nm factor VIIa, 2.4 nm TFPI, and increasing concentrations of TF reconstituted into PCPS membranes. Progress curves (from top to bottom) were obtained using TF/PCPS concentrations of 1024, 512, 384, 256, 192, 128, 64, and 32 pm. Panel B, factor Xa formation was measured using reaction mixtures containing 170 nm factor X, 2 nm factor VIIa, 2.4 nm TFPI, and increasing concentrations of TF reconstituted into PC membranes. Progress curves (from top to bottom) were obtained using TF/PC concentrations of 1024, 512, 256 and 128 pm. The lines in panels A and B were drawn by numerical integration using the constants listed in Table I. Panel C, dependence of concentration of catalyst and the extent of product formation is largely determined by the plasma concentration of factor X, the action of TFPI as a slow, tight-binding, product-activated feedback inhibitor of the catalyst yields a response that is linearly proportional to the concentration of VIIa TF. Therefore, the active concentration of factor Xa that escapes to participate as a catalyst in the subsequent step of coagulation is linearly dependent on the magnitude of TF exposure or the initiating stimulus. Agreement with Existing Quantitative Information Quantitative modeling of the regulatory behavior of TFPI requires an adequate kinetic description of the process of factor Xa formation by VIIa TF. Previous studies have suggested the need to include steps relevant to membrane binding by the substrate and the product as well as possible conformational changes in the enzyme in the kinetic pathway for factor X activation by VIIa TF (45, 46). Despite this complexity, factor Xa formation by VIIa TF can adequately be described by the Michaelis- Menten equation under a wide range of conditions even though the derived K m and k cat terms may not be interpretable in the classical way (45, 47). Since regulation by TFPI results in the rapid inhibition of VIIa TF following the accumulation of small amounts of product, kinetic accounting for factor X activation requires the prediction of product formation at short times following initiation without significant substrate depletion. We have therefore represented the process of factor X activation as a classical two-step process, approximated by the empirical terms K m and k cat, which can adequately predict the initial velocity of factor Xa formation by VIIa TF. Numerical simulations performed according to Scheme I and literature values for the rate constants for the inhibition of factor Xa and for the inhibition of VIIa TF by the preformed Xa TFPI complex (21) are presented in Fig. 3A. There was a large discrepancy between the observed data and progress curves simulated according to Scheme I. Although a comparison between simulated and observed results is provided at one concentration of the extrinsic Xase complex for the purposes of illustration (Fig. 3A), this discrepancy was sustained over the entire range of TF concentrations used (not shown). These findings indicate that Scheme I does not adequately describe the behavior of TFPI. An alternative possibility is that the existing inhibition rate constants determined in studies of the isolated component reactions cannot be extrapolated to describe the individual reactions in the reconstituted system. Previous studies have indicated that the second inhibition step (inhibition of VIIa TF by Xa TFPI, Scheme I) is obviously rate-limiting in the inhibition pathway (21). Numerical simulations (not shown) suggested that the empirical findings were more consistent with the alternative case where the inhibition of factor Xa by TFPI represents the rate-limiting step. This possibility was tested by examining the effects of the preformed Xa TFPI complex on progress curves of factor Xa generation (Fig. 3B). Preincubation of increasing concentrations of factor Xa (0 1 nm) with TFPI prior to initiation of the reactions systematically decreased the extent of factor Xa formation (Fig. 3B). The addition of 1 nm factor Xa without preincubation with TFPI had no obvious effect on the profile of factor Xa formation (not shown). Therefore, prior incubation of factor Xa with TFPI overcomes a rate-limiting step in the inhibition process. These observations suggest that it is the initial reaction between progress curve amplitude on the concentration of extrinsic Xase. Mean concentrations of factor Xa S.D. were determined by averaging data points between 120 and 900 s using TF/PCPS ( ) or 200 and 900 s for TF/PC (E). The lines were drawn by weighted linear regression analysis.

6 Regulation of Factor X Activation by TFPI 4383 FIG. 3.Test of the ability of existing kinetic constants to describe the observations. Panel A, a progress curve for the activation of factor X (170 nm) by VIIa TF/PCPS (2 nm VIIa, 256 pm TF/PCPS) in the presence of 2.4 nm TFPI ( ) is compared with the expected results obtained by numerical simulation according to Scheme I. The constants used for the simulation were k INH, Xa 16 M 1 s 1, k,xa s 1, k INH, VIIa TF 10.4 M 1 s 1, and k,viia TF s 1 (21). To provide an appropriate comparison, the values for k cat and K m used were those determined in the present experimental system (Table I). Panel B, progress curves for Xa formation by 128 pm VIIa TF/PCPS (2 nm VIIa, 128 pm TF/PCPS) in the presence of 170 nm factor X and 2.4 nm TFPI were determined following prolonged incubation of TFPI with 0 ( ), 0.25 nm (E), 0.5 nm (Œ), and 1.0 nm ( ) factor Xa. The lines are drawn following numerical analysis using the values listed in Table I. TFPI and factor Xa rather than the subsequent reaction with VIIa TF that kinetically controls the inhibition of the extrinsic pathway. Characterization of the Reaction between TFPI and Factor Xa Initial velocity measurements using SpXa were used to infer the free (uninhibited) concentration of factor Xa remaining following prolonged incubation of two fixed concentrations of factor Xa and increasing concentrations of TFPI. The achievement of equilibrium in these experiments was suggested by the identity of the titration curves measured following incubation for 4, 6, or 8 h. Analysis of the data presented in Fig. 4 according to a bimolecular equilibrium yielded the overall equilibrium dissociation constant K i * pm and a stoichiometry of mol of TFPI/mol of Xa at saturation. The stoichiometry was also independently verified by titrations using a high fixed concentration of Xa relative to the K i * (34). These data are in reasonable agreement with those previously published for this isolated reaction (25). FIG. 4.Inhibition of factor Xa by TFPI. Residual factor Xa activity was determined using SpXa following prolonged incubation of mixtures containing 0.1 nm ( ) or 0.2 nm (E) factor Xa with the indicated concentrations of TFPI. The lines are drawn following analysis according to Equations 1 and 2 with the fitted constants K i * pm, n mol of TFPI/mol of Xa, v o ( ) 10 3 A 405 / min/nm Xa, and v fixed at near zero ( ). Mean velocities 1 S.D. were derived from initial velocities determined following incubation periods of 4, 6, and 8 h. The residuals to the fitted lines are illustrated in the upper panel. Inset, the pseudo first-order rate constant for the inhibition of factor Xa by TFPI was determined at increasing concentrations of TFPI as described under Experimental Procedures and analyzed according to Equations 3 and 4. The line was drawn by weighted linear regression analysis to yield k INH,Xa M 1 s 1 and k,xa ( ) 10 4 s 1. The rate constant for the inhibition of factor Xa was inferred from exponential decays in progress curves for SpXa hydrolysis following the addition of factor Xa to reaction mixtures containing substrate and increasing concentrations of TFPI. Following the required corrections for inhibitor depletion during the reaction and competitive effects of SpXa (Equations 3 and 4), k obs was found to increase linearly with increasing concentrations of TFPI (Fig. 4, inset). Although the reaction between TFPI and factor Xa has been established to proceed through a two-step mechanism, the equilibrium dissociation constant for the first step is significantly higher than the concentration range of TFPI used (25). Therefore, a linear dependence of k obs on increasing concentrations of TFPI is expected. However, the physical significance of the slope of this plot is related in a complex, mechanism-specific way to the intrinsic rate constants for the individual steps (48). We have therefore chosen to utilize the resulting slope (k INH,Xa ) as an index of the overall second-order rate constant for the inhibition of factor Xa by TFPI, without necessarily attaching a specific physical significance to this term. The determined value, k INH,Xa M 1 s 1, adequately describes the rate constant for the inhibition of factor Xa over the complete range of TFPI concentrations relevant to the present experimental design. The y axis intercept yields the first-order rate constant for the regaining of activity from the preformed Xa TFPI binary complex. As the physical significance of this term is also mechanism-dependent (48), we have chosen to utilize the intercept (k,xa ) as an adequate descriptor of the unimolecular processes that lead to TFPI dissociation from factor Xa. Both determined rate constants are consistent with some previous studies (25, 49). However, the value for k INH,Xa is 15-fold lower than that determined in another study using an equivalent approach (21). Assuming a simple bimolecular reversible equilibrium between TFPI and factor Xa, the overall equilibrium constant can be estimated by trivial division of k,xa by k INH,Xa. The

7 4384 Regulation of Factor X Activation by TFPI FIG. 5.Equilibrium constant for the inhibition of VIIa TF by Xa TFPI. Residual VIIa TF activity was determined using Z-VVR-AMC following prolonged incubation of 0.25 nm extrinsic Xase (1.0 nm TF/ PCPS, 0.25 nm VIIa) with 0.6 nm TFPI and increasing concentrations of factor Xa. Because the TFPI concentration is well above the overall equilibrium dissociation constant for the interaction with factor Xa, the concentration of Xa TFPI was considered equal to the limiting concentration of factor Xa. The line is drawn following analysis according to Equations 1 and 2 with the fitted constants K i * pm, n mol of TFPI/mol of Xa, v o F/s/nM VIIa TF, and v F/s/nM VIIa TF. Inset, inhibition of VIIa TF by increasing concentrations of TFPI in the absence of factor Xa. calculated value ( 0.4 nm) is substantially greater than the measured K i *of 25 pm for the inhibition of factor Xa by TFPI. This discrepancy clearly supports previous conclusions of a multistep reaction pathway describing the inhibition of factor Xa by TFPI (25). Inhibition of the Extrinsic Xase Complex Previous studies have relied on measurements of factor X activation to infer the kinetics of inhibition of VIIa TF by the preformed Xa TFPI binary complex (21, 22). Because these measurements involve the generation of factor Xa in situ and usually require the use of low concentrations of Xa TFPI relative to the overall equilibrium dissociation constant, it is difficult to distinguish the contribution of the kinetics of inhibition of VIIa TF by preformed Xa TFPI from the other reactions of TFPI. We therefore sought a method of appropriate sensitivity that would permit measurements of VIIa TF inhibition by preformed Xa TFPI without requiring further factor Xa formation by the catalyst. Preliminary experiments established that the rate of hydrolysis of the fluorescent peptidyl substrate Z-VVR-AMC was substantially enhanced by the interaction of factor VIIa with TF as has been documented previously for a variety of synthetic peptidyl substrates (36, 50). The K i * for the inhibition of VIIa TF by TFPI or the preformed Xa TFPI binary complex was therefore determined following prolonged incubation and assessing the VIIa TF-directed cleavage of the fluorescent peptidyl substrate Z-VVR-AMC. Representative data (Fig. 5) illustrate that the preformed Xa TFPI binary complex is a potent inhibitor of VIIa TF/PCPS with K i * pm. In contrast, in the absence of factor Xa, the direct effects of TFPI on VIIa TF activity are relatively minor (Fig. 5, inset). These observations are in good agreement with previous findings (15) and provide the experimental basis for ignoring direct effects of TFPI alone on VIIa TF/PCPS in the overall regulation of the extrinsic Xase complex at the concentrations used in this study (Scheme I). The thermodynamic contribution of the ability of Xa TFPI to bind reversibly to PCPS membranes was investigated by repeating these experiments using TF reconstituted in pure PC membranes. The K i * calculated for the inhibition of VIIa TF/PC FIG. 6. Kinetics of the inhibition of VIIa TF by Xa TFPI. Progress curves for the hydrolysis of Z-VVR-AMC by 0.2 nm VIIa TF (1 nm TF/PCPS, 0.2 nm VIIa) are illustrated in the absence (E) or presence ( ) of 1.6 nm Xa TFPI (1.6 nm Xa, 5.5 nm TFPI preincubated 4 h). The line is drawn following analysis as described under Experimental Procedures to yield k obs ( ) 10 3 s 1. Inset, dependence of k obs on increasing concentrations of Xa TFPI. The line is drawn following weighted linear regression analysis to yield k INH, VIIa TF M 1 s 1 and k,viia TF ( ) 10 3 s 1. by Xa TFPI in several experiments was maximally 3-fold greater than the K i * observed for the inhibition of VIIa TF/ PCPS. Since membranes composed purely of PC cannot support high affinity binding of factor Xa (51), the minor difference between the K i * determined using PCPS relative to PC suggests that the ability of factor Xa within the Xa TFPI binary complex to interact with membranes does not contribute extensively to the stability of the inhibited complex. The rate constant for the inhibition of VIIa TF/PCPS by the preformed Xa TFPI complex was determined by monitoring the exponential decay in the rate of Z-VVR-AMC hydrolysis to a final steady state rate. Typical progress curves in the presence or absence of TFPI (Fig. 6) indicate that inhibition of the peptidyl hydrolytic activity by VIIa TF/PCPS occurs slowly, over 5 min, prior to the achievement of a limiting steady state rate representing the equilibrium distribution between the inhibited and active forms of VIIa TF. The observed rate constant for this process (k obs ) was found to increase with increasing concentrations of Xa TFPI (Fig. 6, inset). Although data sets from individual experiments yielded largely consistent results, significant variability in k obs was noted between data sets that could not be resolved by alternate experimental approaches. This likely arises from the difficulty in assessing the relatively short time constant for this process in open cuvette experiments and the requirement for high sensitivity settings to monitor a small fluorescence change with high background and/or contributions from some experimental variable that we have been unable to identify. We have presented and analyzed the results of four separate experiments simultaneously, assuming that deviations from the line result from experimental variability rather than systematic deviation from Equations 3 and 4. By applying the same logic as used in the analysis of the inhibition of factor Xa by TFPI (above), analysis of the slope and intercept determined by weighted regression yielded k INH, VIIa TF M 1 s 1 and k,viia TF ( ) 10 3 s 1. Both values are in reasonable agreement with rate constants for this reaction determined previously from studies of factor X activation (21). Quantitative Description of the Regulatory Behavior of TFPI

8 Regulation of Factor X Activation by TFPI 4385 TABLE I Kinetic constants for the regulation of the extrinsic pathway by TFPI Kinetic constant a Experimentally determined b Numerical analysis (Fig. 2A) c Numerical analysis (Fig. 3B) Numerical analysis (Fig. 2B) K m (nm) d k cat (s 1 ) (F) 7 (F) 0.7 (F) k INH,Xa ( M 1 s 1 ) k,xa ( 10 4 s 1 ) (F) 3.6 (F) 3.6 (F) k INH,VIIa TF ( M 1 s 1 ) k,viia TF ( 10 4 s 1 ) (F) 11 (F) 11 (F) a Kinetic constants correspond to those illustrated in Scheme I. b The K m and k cat terms determined under comparable conditions with TF/PCPS were taken from Ref. 27. Kinetic constants are listed 1 S.D. c Numerical analyses were performed according to the ordinary differential equations accounting for Scheme I using the terms denoted by (F) as constants. The errors in the other terms represent 95% confidence limits derived from iterative least squares error minimization of the data. d The reduced rate of factor X activation using TF/PC was adequately accounted for by fixing the value for k cat at 1/10th that observed with TF/PCPS and fitting the value of K m. Since the numerical analyses only require the adequate prediction of the initial rate of factor Xa formation, alternate values for k cat and K m yielded adequate descriptions of the data provided they could account for the reduced rate of factor Xa formation on PC membranes. on Factor Xa Formation The rate constants determined for the component inhibition reactions (Scheme I), summarized in Table I, indicate that the forward rate-limiting step in the inhibition pathway corresponds to the initial reaction between TFPI and factor Xa rather than the subsequent inhibition of VIIa TF by preformed Xa TFPI. This finding is consistent with the enhancing effects of factor Xa intentionally premixed with TFPI on factor Xa generation (Fig. 3B). Numerical simulations were therefore performed according to the ordinary differential equations accounting for Scheme I to test the ability of the newly determined rate constants (Table I) to account for the behavior of TFPI during factor X activation. This approach did not yield simulated progress curves in satisfactory agreement with the data (not shown). Preliminary calculations indicated that large changes in the simulated progress curves could be achieved by changes to the forward rate constants (k INH,Xa, k INH,VIIa TF ) while the outcome was largely insensitive to adjustments in the reverse rate constants (k,xa, k,viia TF ). This finding is consistent with the fact that the overall equilibrium constants for the individual reactions lie far to the right. We combined iterative nonlinear least squares error minimization with numerical solution of the relevant ordinary differential equations to further investigate the problem. The resulting fits provided an adequate description of entire families of progress curves (Fig. 2, A and B, and Fig. 3B). This assertion is based on the fact that the final root mean square deviation approximated experimental error. The fitted kinetic parameters for factor X activation and k INH,Xa were in reasonable agreement with experimentally determined terms (Table I). However, in each case, adequate description of the empirical data required major increases in k INH,VIIa TF over the determined rate constant for this reaction. All three sets of analyses consistently required values of k INH, VIIa TF in excess of 10 8 M 1 s 1 for adequate description (Table I). Interestingly, the values for k INH,VIIa TF determined by numerical analyses are in remarkable agreement with the same rate constant inferred from kinetic studies of VIIa TF inhibition during X activation using picomolar concentrations of Xa and TFPI (22). These observations suggest that the rate constant for the inhibition of VIIa TF by the Xa TFPI binary complex determined from kinetic studies of the isolated half-reaction in the absence of factor X very significantly underestimates the rate constant for the inhibition of the catalyst during Xa formation. This discrepancy either implies that the presumed mechanism of action of TFPI (Scheme I) does not adequately account for the behavior of the inhibitor or that the kinetic constants derived from studies of the individual inhibitory half-reactions cannot be extrapolated to describe the kinetics of the same processes during factor Xa formation by VIIa TF. Previous studies as well as the present work on the individual inhibition reactions with TFPI are largely consistent with the prevailing idea that the initial reaction between Xa and TFPI is requisite for the efficient inhibition of VIIa TF. Thus, there is no reason, a priori, to justify drastic modifications to Scheme I. The large value for the estimated rate constant for the inhibition of VIIa TF by Xa TFPI during X activation inferred from the numerical analyses indicates that the inhibition of VIIa TF proceeds at a near-diffusion limited rate and is therefore undetectably rapid during ongoing factor X activation. Potential explanations may lie in the possibility that the predominant pathway for the action of TFPI during X activation by VIIa TF involves the initial inhibition of factor Xa either in the near vicinity of or bound to the catalyst. These possibilities would convert an otherwise bimolecular inhibition reaction between VIIa TF and Xa TFPI (step 2, Scheme I), readily approximated by experimental approaches illustrated in Figs. 5 and 6, to a unimolecular process or a bimolecular reaction confined to the membrane surface with dimensional restrictions. Either possibility would yield rate constants substantially greater than those directly measured for the inhibition of VIIa TF by preformed Xa TFPI. Numerical analyses of factor X activation with pure PC membranes also yield large values for the rate constant for inhibition of VIIa TF (Table I). Since PC membranes cannot support the high affinity binding of factor Xa (51), it seems less probable that the unpredictably large rate constant for the inhibition of VIIa TF arises from membrane delimited reactions with the Xa TFPI complex. Consequently, the modified reaction pathway illustrated in Scheme II may provide one reasonable explanation for the findings. We speculate that the predominant pathway for the action of TFPI during factor Xa formation may involve the inhibition of product in a bimolecular reaction prior to its dissociation from the enzyme followed by the formation of the inhibited quaternary complex through a unimolecular step. It follows that the kinetics of this second step cannot be adequately approximated by studies of the bimolecular reaction describing the inhibition of VIIa TF by preformed Xa TFPI in the absence of the substrate, factor X. Conclusions At potentially physiological reactant concentrations, TFPI is a potent product-activated feedback inhibitor of factor Xa formation by the extrinsic pathway. The empirical studies of this process indicate that the action of TFPI during factor X activation by VIIa TF yields blunted progress curves where both the initial rate of factor Xa formation as well as the extent of product formed is linearly proportional to the concentration of the extrinsic Xase complex. As a result, the active concentration of factor Xa that escapes regulation by TFPI to participate in the subsequent steps of coagulation might be expected to be linearly proportional to the magnitude of the initiating stimulus. These findings indicate that TFPI