Zn²⁺-containing protein S inhibits extrinsic factor X activating complex independently of tissue factor pathway inhibitor (TFPI)

Summary

Background

Protein S (PS) has direct anticoagulant activity, independent of activated protein C (APC). Mechanisms underlying this activity remain unclear because PS preparations differ in activity, giving rise to conflicting results. Some purification procedures result in loss of intramolecular Zn²⁺ that is essential for inhibition of prothrombinase.

Objective

To investigate inhibition of extrinsic FXase by Zn²⁺-containing PS.

Methods

Purified component extrinsic FXase assays were used to determine FXa generation in the presence/absence of PS and/or TFPI. Binding assays, immunoblots, and thrombin generation assays in plasma supported the FXase data.

Results

Zn²⁺-containing PS potently inhibited extrinsic FXase in the presence of saturating phospholipids, independently of TFPI, whereas inhibition of extrinsic FXase by Zn²⁺-deficient PS required TFPI. Immunoblots for FXa and functional assays showed that Zn²⁺-containing PS inhibited primarily the quantity of FXa formed by tissue factor (TF)/FVIIa, rather than FXa amidolytic activity. Zn²⁺-containing PS, but not Zn²⁺-deficient PS, bound to TF with high affinity (Kd_app=41 nM) and targeted TF function. Binding of PS to FVIIa was negligible, whereas PS showed appreciable binding to FX. Increasing FX concentrations by 10-fold reduced PS inhibition by 5-fold, suggesting that PS inhibition of FXase is FX-dependent. PS also exhibited TFPI- and APC-independent anticoagulant activity during TF-initiated thrombin generation in plasma.

Conclusions

PS that retains native Zn²⁺ retains anticoagulant functions independent of APC and TFPI. Inhibition of extrinsic FXase by PS at saturating phospholipids depends on PS retention of intramolecular Zn²⁺, interaction with FX, and particularly, interaction with TF.

Introduction

Protein S (PS) is a 75 kDa, vitamin K-dependent glycoprotein circulating in plasma partially in a complex with C4b-binding protein [1]. Heterozygous deficiency of PS is associated with increased risk of venous thrombosis and homozygous deficiency is potentially fatal in neonates [2,3]. PS knock-out mice die in utero with severe coagulopathy [4]. PS is an essential anticoagulant that acts as a cofactor in the proteolytic inactivation of factors Va and VIIIa by activated protein C (APC) [5]. In addition, PS exhibits direct anticoagulant activities that are APC-independent [6–8], and that are compromised in heterozygous PS-deficient mice [4]. Although the APC cofactor activity of PS has been well characterized, mechanisms by which PS exerts its direct activity have not been fully determined.

A confounding factor in assessment of molecular mechanisms for the direct anticoagulant activity of PS is the variation in activity depending on purification methods used. We showed that immunoaffinity-purified PS contains Zn²⁺ that is essential for its direct activity [9]. Zn²⁺-containing immunoaffinity-purified PS inhibits the prothrombinase activity of FXa/FVa in the presence of saturating phospholipids, while most, but not all, conventionally-purified PS preparations are Zn²⁺-deficient and inhibit prothrombinase poorly [9].

We hypothesized that Zn²⁺-containing PS is a multifunctional anticoagulant, and that some of its functions are TFPI-independent. Hackeng et al. [10] reported that PS did not inhibit extrinsic FXase but appeared to act as a cofactor for inhibition of FXase by TFPI. Here we report that Zn²⁺-containing PS inhibits FXa generation independently of TFPI, while PS that is Zn²⁺-deficient inhibits FXa generation only in the presence of TFPI. We further hypothesized that this inhibition was due to a specific interaction of PS with one or more FXase component.

Materials and methods

PS

Zn²⁺-containing PS was purified from citrated plasma by barium precipitation, followed by elution of the pellet with 33% saturated ammonium sulfate [11]. The eluate was dialyzed against Tris-buffered saline (TBS; 0.05 M Tris, 0.1 M NaCl, 0.02% NaN₃, pH 7.4). PS complexed with C4b-binding protein was removed by precipitation with 3.75% polyethylene glycol. Free PS was immunoaffinity purified [9] and subjected to SDS-PAGE and ELISA. PS was pooled, concentrated by membrane filtration, and dialyzed twice against Hepes-buffered saline (HBS; 0.05 M Hepes, 0.1 M NaCl, pH 7.4).

Zn²⁺-deficient conventionally-purified PS was obtained from Enzyme Research Laboratories (South Bend, IN, USA), or purified using MonoQ chromatography as described [12]. For some experiments, commercial PS was reconstituted with Zn²⁺ as described [9].

Tissue factor

Full-length lipidated TF (Innovin) was from Dade (Marburg, Germany) and full length nonlipidated TF was from Altor Biosciences (Miramar, FL, USA). TF cDNA (NM_001993) was obtained from Origene. Soluble (s) TF (residues 1-218) was prepared by PCR using primers (forward: 5′-CACCCTGGTGCCTCGTGGTTCAGGCACTACAAATACTG-3′ and reverse: 5′-CTATTATCTGAATTCACCTTTCTCCTGG-3′). The PCR fragment was cloned into pET151/D-TOPO (Invitrogen) containing an N-terminal His and V5 tag. Introduction of a Leu-Val-Pro-Arg-Gly thrombin cleavage sequence (underlined in forward primer) allowed for removal of the tags. Recombinant sTF was expressed in E. coli BL21 Star (DE3) cells and purified to >95% homogeneity from inclusion bodies on Ni-NTA Sepharose as described [13].

Other materials

FX and recombinant FVIIa were from Enzyme Research Laboratories. Chromogenic FXa substrate Pefachrome FXa 8595 and thrombin substrate pefaTH were obtained from Centerchem (Norwalk, CT, USA) and S2765 FXa substrate was from Chromogenix (Westchester, OH, USA). Full length TFPI activity standard was obtained from American Diagnostica (Stamford, CT, USA). 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC), and 1, 2-dioleoyl-sn-glycero-3-phosphoserine (DOPS) were obtained from Avanti Polar Lipids (Alabaster, AL, USA). Phospholipid vesicles, 80% DOPC/20% DOPS, were prepared by sonication [6]. Biotin conjugated-secondary antibodies and streptavidin-horseradish peroxidase conjugates were from Pierce (Rockford, IL, USA). Monoclonal anti-FX antibody was from Biodesign (Kennebunkport, ME, USA). o-Phenylenediamine substrate, porcine and fish skin gelatin, and calcium ionophore were from Sigma Aldrich, St. Louis MO, USA. Rabbit antibody against PS was obtained from Dako (Carpenteria, CA, USA). Rabbit neutralizing antibody against TFPI was a kind gift from Prof. Samuel L. Rapaport, Dr. L. Vijay Mohan Rao, and Dr. Bonnie Warn-Cramer, formerly of the UCSD School of Medicine.

PS-depleted plasma and pooled normal human plasma was obtained from Affinity Biologicals (Hamilton, Ontario, Canada). TFPI-depleted plasma was prepared as described [9].

Inhibition of TF/FVIIa activation of FX by PS and TFPI

Initial experiments (Fig. 1) were performed as described in Hackeng et al. [10]. 2 pM TF, 1 pM FVIIa, 15 μM phospholipid vesicles were preincubated in 0.5% BSA-HBS, 2 mM CaCl₂ with or without 80–100 nM PS for 10 min at 37°C in the absence or presence of 100 pM TFPI. FXa generation was initiated by addition of 160 nM FX. Aliquots were taken over time and quenched in buffer containing 20 mM EDTA. The rate of FXa generation was determined using S2765 (0.2 mM). In selected experiments, 20 μg mL⁻¹ of anti-TFPI antibody was used to neutralize inhibition by TFPI. Subsequent assays were modified by use of 23^oC, 2 pM FVIIa, 40 μM phospholipids, 5 mM CaCl_2, and Pefachrome FXa 8595 substrate.

Fig. 1.

Effect of PS and TFPI on extrinsic FXase activity. TF/FVIIa/phospholipids were incubated ± PS (80–100 nM) and/or TFPI (100 pM) as described in Methods. Following addition of FX, aliquots were removed over time and quenched in EDTA. FXa generation was measured with chromogenic substrate. % Inhibition was estimated by comparing curves at 100 s. (A) Zn²⁺-containing affinity-purified PS. Seven additional affinity-purified PS independently inhibited FXase, did not enhance inhibition by TFPI, and were unaffected by neutralizing antibody against TFPI. (B) PS conventionally-purified using anion exchange. (C) Commercial Zn²⁺-deficient PS before and after reconstitution of Zn²⁺. Zn²⁺-reconstituted PS inhibited FXase independently of TFPI and did not enhance TFPI inhibition. Curves are representative of two or more experiments, each performed in duplicate.

Binding Assays

Kd_app for the binding of PS to sTF (or to FX) was determined using a fluid phase assay [6,7]. sTF (or FX) at various concentrations was incubated with a fixed concentration of PS at 23^oC for 2 hr in the wells of siliconized low retention microtiter plates along with standards of PS. Free PS was determined by exposing the mixtures and standards to a plate coated with sTF (or FX) for a brief period of 10 min. Free PS bound to the sTF-coated (or FX-coated) plate was detected with anti-PS as described below. Free PS in the fluid phase mixtures was calculated by comparison of the signals to a curve constructed of signals of the PS standards. Bound PS in the mixtures was calculated by subtracting free PS from the initial concentration of PS in each mixture. 1 mol of PS was assumed to be bound to 1 mol of sTF (or FX). From a plot of sTF (or FX) bound versus initial concentration of sTF (or FX) in the mixtures, Kd_app was determined by nonlinear regression analysis for one site binding using Graph Pad Prism software (San Diego, CA, USA). Buffers used were identical to those for the solid phase binding experiments described below.

Solid phase binding assays were also performed. 2 μg mL⁻¹ of sTF in 0.1 M carbonate, pH 9.0 was coated to the wells of Costar (Lowell, MA, USA) microtiter plates for 1 hr at 23^oC in a humidified chamber. Subsequently, plates were blocked with 3% fish skin gelatin. Control wells with no sTF were also blocked. Plates were washed once with TBS and incubated for 1 hr with various concentrations of PS in 0.1% porcine skin gelatin, 20 mM Tris, 150 mM NaCl, 2 mM CaC1₂, 0.03% Tween, pH 7.4. Bound PS was detected by incubation with 1 μg mL⁻¹ polyclonal anti-PS antibody in the same buffer, followed by 1 μg mL⁻¹ of biotin-secondary antibody, and 2 μg mL⁻¹ streptavidin-horseradish peroxidase. Each step was followed by three washes, except the last step, which was followed by two additional washes with TBS. Plates were developed with o-phenylenediamine, stopped with 1 N HCl and absorbance was read at 490 nm on a Molecular Devices (Sunnyvale, CA, USA) microplate reader using SoftmaxPro software. Binding constants (Kd_app) were determined by nonlinear regression analysis for one site binding.

Immunoblots for FXa generated by TF/FVIIa in the presence/absence of PS

Extrinsic FXase mixtures in the absence or presence of 100 nM PS were prepared as described above, except that BSA was reduced to 0.1% BSA. Following initiation of FXa generation by addition of FX, aliquots were quenched in EDTA or SDS after 50, 100, 125 and 150 sec. EDTA samples were assayed using chromogenic substrate to determine the amount of FXa generated. SDS samples were subjected to immunoblotting for FXa antigen. Electrophoresis was performed under nonreducing conditions in Tris-acetate buffer (Invitrogen, Carlsbad, CA, USA) on 7% polyacrylamide gels (Invitrogen). Proteins were transferred to PVDF membranes and blocked with 1% I-block (Applied Biosystems, Bedford, MA, USA). Membranes were cut horizontally to separate FXa (MW~49 kDa) from excess FX (MW~67 kDa) prior to development with 1 μg mL⁻¹ of monoclonal anti-FX antibody followed by a biotinylated secondary antibody, streptavidin–peroxidase, and Supersignal chemiluminescent substrate (Pierce). Blots were quantified by Image J software (National Institutes of Health, USA).

Effect of Zn²⁺-containing PS on thrombin generation in plasma

Thrombin generation assays in plasma in the presence of neutralizing antibodies against protein C were performed as described [9]. Various plasmas were tested after preincubation for 8 min at 37^oC with: neutralizing antibodies against PS or against TFPI; or PS (67 nM); or TFPI (100 pM); or HBS-0.5% BSA. After addition of TF/Ca/phospholipids and fluorogenic thrombin substrate (ZGGR-aminomethylcoumarin (Bachem, Torrance, CA, USA), fluorescence was monitored in a Spectramax plate reader (Molecular Devices, Sunnyvale, USA). Lag time was recorded when change in fluorescence reached 100 units/min, and change in fluorescence/min was calculated and plotted. Area under the curve, time to peak thrombin generation and peak height were recorded; lag time was the parameter most sensitive to the direct anticoagulant activity of PS. Lag times and times to peak thrombin generation were compared by student’s two tailed t-test using Prism software. Statistical significance was assigned to p values ≤ 0.05.

Results

Inhibition of extrinsic FXase by Zn²⁺-containing PS does not require TFPI

Immunoaffinity-purified PS contains Zn²⁺ and was shown to inhibit the generation of thrombin by the prothrombinase complex (FXa/FVa), while PS that was purified using anion exchange (e.g. MonoQ chromatography with a Ca²⁺ gradient) and commercial PS were found to be Zn²⁺-deficient and poor inhibitors of prothrombinase [9]. PS that was conventionally purified without use of anion exchange also contained Zn²⁺ and inhibited prothrombinase [9]. Previously, it was proposed that PS acts as a cofactor for TFPI during inhibition of extrinsic FXase and that PS does not convey direct anticoagulant activity in the absence of TFPI [10]. Those studies conflict with our own, however it is likely that the PS that was used in the afore mentioned studies was Zn²⁺-deficient. To resolve these conflicting results we determined the effect of Zn²⁺-containing PS versus Zn²⁺-deficient PS on the inhibition of FX activation by TF/FVIIa in the absence and presence of PS and TFPI.

Zn²⁺-containing PS inhibited extrinsic FXase in the absence of TFPI (Fig. 1A). TFPI contamination in the PS preparation was ruled out by using a neutralizing antibody to TFPI. Similar results were obtained using seven different immunoaffinity-purified PS preparations. Although TFPI alone inhibited FXa generation, the combination of TFPI and Zn²⁺-containing PS did not result in additional inhibition of FXa generation beyond that observed by Zn²⁺-containing PS in the absence of TFPI. In some experiments not shown, a modest additional effect was observed, but the combined effects were always less than additive. In contrast, Zn²⁺-deficient PS, either MonoQ-purified PS (Fig 1B) or commercially available PS (Fig 1C), had negligible effect on extrinsic FXase in the absence of TFPI. In the presence of TFPI, inhibition of extrinsic FXase was enhanced by Zn²⁺-deficient PS, in agreement with Hackeng et al [10]. 1 nM TFPI inhibited FXase >95% and less stimulation of inhibition by Zn²⁺-deficient PS occurred when inhibition by TFPI alone was >30% (not shown).

The direct anticoagulant activity of Zn²⁺-deficient PS is restored by incorporation of Zn²⁺ into PS through mild denaturation in the presence of Zn²⁺ [9]. Reconstitution of Zn²⁺ in Zn²⁺-deficient commercial PS resulted in >95% monomeric PS that inhibited extrinsic FXase in the absence of TFPI (Fig. 1C). Furthermore, the combination of TFPI and Zn²⁺-reconstituted PS did not result in additional inhibition of FXa generation beyond that observed for Zn²⁺-reconstituted PS in the absence of TFPI, indicating that Zn²⁺-deficient PS after Zn²⁺ reconstitution behaved similarly to Zn²⁺-containing PS.

Inhibition of generation of FXa protein by PS

In subsequent experiments, “PS” will refer to Zn²⁺-containing PS unless otherwise stated.

PS binds reversibly to FXa and inhibits FXa amidolytic activity, albeit much less efficiently than it inhibits FXa’s prothrombinase activity [7]. To determine whether PS inhibited the amount of FXa protein generated by TF/FVIIa, formation of FXa in the presence and absence of PS was followed over time by Western blot. In the presence of PS, significantly less FXa was formed than in the absence of PS (Fig. 2A). Quantitative densitometry indicated a 50% reduction in FXa generation after 150 sec in the presence of PS. In parallel experiments, PS inhibited FXase activity by 58% after 150 sec (Fig. 2B). In separate experiments, the same concentration of PS inhibited the amidolytic activity of FXa by only 18% in 150 sec. These results show that inhibition of extrinsic FXase by PS is caused predominantly by inhibition of FXa generation and not by inhibition of FXa amidolytic activity.

Fig. 2.

PS inhibition of the generation of FXa protein by FXase. TF/FVIIa/phospholipids were incubated ± 100 nM PS prior to addition of FX as described in Methods. Aliquots were quenched in parallel in SDS or EDTA after 50–150 s. (A) Blot for FXa generated by the extrinsic FXase complex. SDS samples were subjected to immunoblotting for FXa antigen. After transfer, membranes were cut horizontally to separate FXa (MW~49 kDa) from excess FX (MW~67 kDa) prior to development with anti-FX antibodies and chemiluminescent detection. (B) EDTA-quenched samples were assayed for FXa activity. Results were confirmed in several experiments.

Effect of phospholipid concentration on the inhibition of extrinsic FXase by PS

When phospholipids are limiting, PS can inhibit prothrombinase by competing with FXa/FVa for limiting membrane surface [14]. Therefore, inhibition of extrinsic FXase by PS was tested at variable phospholipid concentrations. It should be noted that even when no phospholipids were added, the TF preparation itself contributed a small amount of phospholipids. At lower concentrations of exogenous phospholipids (<20 μM), Zn²⁺-containing PS appeared to inhibit FXase partly via competition for phospholipid (Fig. 3). However, at higher lipid concentrations (25–50 μM), inhibition by PS persisted, and was independent of lipid concentration. Thus, competition for negatively charged phospholipid binding sites may be one of the mechanisms of inhibition of FXase by PS but is not responsible for all of the inhibition observed. Subsequent experiments were performed at 40 μM phospholipids to avoid PS competition for phospholipid.

Fig. 3.

Phospholipid dependence of inhibition of extrinsic FXase by PS. TF/FVIIa with varying phospholipid concentrations were incubated ± 100 nM Zn²⁺-containing PS or ± 333 nM Zn²⁺-deficient PS as described in Methods. FX was added. Aliquots were removed between 40–120 s and quenched in 20 mM EDTA. FXa generation was measured with chromogenic substrate. % inhibition relative to the controls without PS at each phospholipid concentration was calculated by comparing activity progress curves at 120 s. The TF preparation contributed a small amount of phospholipids, such that inhibition by PS in the absence of exogenous phospholipids was similar to or greater than inhibition in the presence of 5 μM exogenous phospholipids. Results are representative of several experiments.

Even at limiting phospholipids, 333 nM Zn²⁺-deficient PS did not inhibit FXase as well as 100 nM Zn²⁺-containing PS (Fig. 3). As phospholipid concentrations were increased, inhibition of FXase by Zn²⁺-deficient PS gradually fell to <10%.

PS binds FX and also competes with FX for binding to the FXase complex

In order to determine if one mechanism of inhibition of extrinsic FXase by PS involves binding of PS to the substrate FX, several fluid phase assays were performed to measure binding (Fig. 4A). PS bound FX (Kd_app=127±28 nM) ~12-fold less avidly than FXa (Kd_app=9–13 nM) [7,9], suggesting that PS inhibition of FXase may be at least partly due to binding of PS to FX substrate. Increasing FX concentration (40–400 nM) in FXase assays at fixed PS concentration (100 nM) reduced PS inhibition by 5-fold (Fig. 4B), suggesting that PS inhibition is FX-dependent.

Fig. 4.

Binding of Zn²⁺-containing PS to FX and competition between FX and PS for binding to the FXase complex. (A) Fluid phase assay for binding of FX to Zn²⁺-containing PS. Assay described in Methods. Binding constant (Kd_app) was determined using nonlinear regression analysis for one site binding. Results from two experiments are combined. (B) Inhibition of extrinsic FXase in the presence of varying [FX]. TF/FVIIa/phospholipids were incubated ± PS as described in Methods except that varying concentrations of FX were used. Aliquots were removed over time and quenched in EDTA. FXa generation was measured with chromogenic substrate. Results were confirmed in a separated experiment.

PS preferentially targets tissue factor in the extrinsic FXase complex

To determine which of the components of the extrinsic FXase is targeted preferentially by PS to inhibit FXa generation, a series of extrinsic FXase assays were performed in which each of the individual extrinsic FXase components were preincubated with PS before initiation of the reaction by addition of the remaining components. Preincubation of PS with FX/phospholipid and initiation with TF/FVIIa/phospholipid resulted in 20% inhibition of FXa generation (Fig. 5A). In contrast, preincubation of PS with TF/phospholipid and initiation with FX/FVIIa resulted in 80% inhibition of FXa generation (Fig. 5B), whereas preincubation of PS with FVIIa/phospholipid followed by initiation with relipidated TF/FX inhibited FXa generation by 30% (Fig. 5C). Thus, interaction of PS with TF is an important determinant by which PS inhibits the extrinsic FXase. Preincubation of PS with a mixture of TF/FVIIa/phospholipids followed by initiation with FX resulted in significantly less inhibition of FXa generation (30%) compared to PS preincubation with TF/phosphlipids alone (Fig. 5D). This implies that FVIIa might obstruct the interaction between PS and TF, or that the FVIIa-bound conformation of TF is less favorable for PS binding.

Fig. 5.

Functional assay to identify the most significant target for PS inhibition of FXase. The FXase assay described in Methods was modified such that different FXase components were incubated for 10 min with 100 nM PS in the presence of saturating phospholipids prior to addition of the remaining FXase components to start FXa generation. Aliquots were removed over time, quenched in 20 mM EDTA, and FXa generation was measured using a chromogenic substrate. (A) FX/phospholipid was preincubated ± PS before TF/FVIIa were added. (B) TF/phospholipid ± PS were preincubated before FVIIa/FX were added. (C) FVIIa/phospholipid were preincubated with PS before TF/FX were added. (D) TF/FVIIa/phospholipid ± PS were preincubated before FX was added. All data were collected within 1 h using the same reagents in each set, and confirmed in a separate set of experiments performed in duplicate on a different day. Lag times were taken as the time when thrombin generation reached a threshold signal of 100 fluorescence units/min.

Binding of TF to PS

Since interaction of PS with TF appeared to be an important determinant by which PS inhibited the extrinsic FXase, binding of PS to full-length nonlipidated TF and sTF was examined. In fluid-phase binding assays, Zn²⁺-containing PS bound sTF with Kd_app=41 nM (Fig. 6A). In solid-phase assays, Zn⁺²-containing PS bound sTF and TF with relatively high affinity, while negligible binding was observed for commercial Zn⁺²-deficient PS (Fig. 6B). Overall, PS bound to sTF and TF with Kd_app=41±22 nM (n=9). To confirm that the ability of PS to bind to sTF was dependent on the intramolecular incorporation of Zn²⁺ in PS, commercial Zn⁺²-deficient PS was reconstituted with Zn²⁺. Reconstitution of Zn²⁺ restored the ability of PS to bind sTF (Kd_app=39 nM, Fig. 6B). In experiments not shown, 10–20 μg/ml of sheep antibodies against TF (Haematologic Technologies, Essex Junction, VT, USA) or monoclonal antibody 7 (Corvas, San Diego, CA, USA) blocked binding of PS to immobilized sTF. Monoclonal antibody 5 (Corvas) did not block binding of PS to sTF, and served as a control. These results confirm that TF is the preferred target for Zn²⁺-containing PS and that binding of Zn²⁺-containing PS to TF is an important determinant for inhibition of extrinsic FXase by PS.

Fig. 6.

Binding of sTF to Zn²⁺-containing PS. (A) Fluid phase assay described in Methods. Data from two experiments are combined. (B) PS binding to immobilized sTF as described in Methods. Binding constants were determined using nonlinear regression analysis for one site binding. Kd_app=41±22 nM, n=9.

Effect of varying FVIIa, FVII, TF, and phospholipid composition on inhibition of FXase by PS

In our hands, the functional Kd for binding of FVIIa to TF is ~50 pM. Therefore we tested PS inhibition of FXase at a range of FVIIa and TF concentrations and in the presence of FVII (Table 1). Compared to the 1–2 pM FVIIa used in most of our experiments, inhibition of FXase by PS in the presence of 25–200 pM FVIIa was similar in extent (Table 1, Exp. 1–4). In our standard protocol used here, FVIIa and TF were premixed, and this preformed complex of FVIIa/TF appeared to be more resistant to PS inhibition than TF alone, as seen in Fig. 5 experiments. Compared to the 2 pM of TF used in most of our experiments, inhibition of FXase by PS in the presence of 4–9 pM TF (supranormal concentrations) was similar in extent (Exp. 5 and 1). At 18 pM TF/25 pM FVIIa, there was a diminution of PS inhibition (Exp. 6 versus 1), possibly because of approaching the Kd for FVIIa/TF interaction, and thus more resistance of the FVIIa/TF complex to inhibition by PS. The presence of FVII at the plasma concentration of 10 nM with FVIIa/TF at 100 pM/10 pM did not affect PS inhibition of FXase (Exp. 7 versus 8). Interestingly, incorporation of 20% phosphatidylethanolamine into phospholipid vesicles containing 20% phosphatidylserine (DOPS) did not enhance PS inhibition of FXase (Exp. 9 versus 10), however, incorporation of 20% phosphatidylethanolamine into vesicles containing 5% DOPS did enhance PS inhibition of FXase compared to our standard DOPS/DOPC vesicles (Exp. 9 versus 11). Phosphatidylethanolamine is thought to be more abundant on activated cell surfaces than is DOPS.

Table 1.

Inhibition of Extrinsic FXase by Zn²⁺-containing PS under various conditions

	PS, nM	FVIIa, pM	TF, pM	FVII, nM	PL* at 40 μM	% Inhibition at 120 sec
1	100	25	9	0	DOPS/DOPC	41 ± 5
2	100	50	9	0	DOPS/DOPC	37 ± 3
3	100	100	9	0	DOPS/DOPC	34 ± 4
4	100	200	9	0	DOPS/DOPC	36 ± 2
5	100	25	4	0	DOPS/DOPC	48 ± 6
6	100	25	18	0	DOPS/DOPC	25 ± 1
7	120	100	10	0	DOPS/DOPC	36 ± 6
8	120	100	10	10	DOPS/DOPC	38 ± 3
9	100	100	10	0	DOPS/DOPC	36 ± 6
10	100	100	10	0	DOPS/DOPE/DOPC 20/20/60	19 ± 13
11	100	100	10	0	DOPS/POPE/DOPC 5/20/75	65 ± 1

^*vesicles were in the molar ratios of 20%/80% for DOPS and DOPC except where noted.

DOPE denotes dioleolylphosphatidylethanolamine.

Anticoagulant effect of Zn²⁺-containing PS on thrombin generation in plasma

Findings above using purified systems were extended to plasma. Lag times were compared from TF-initiated thrombin generation assays using various plasmas with additions of PS, TFPI, or antibodies against PS or TFPI (Fig. 7A). A low concentration of TF was chosen in order to increase sensitivity to PS. Lag times were 5 or 7 min shorter when TF (0.3 pM) was added to PS-depleted plasma or normal plasma, respectively (data not shown). TF-induced lag times in normal and PS-depleted plasmas were insensitive to corn trypsin inhibitor (30 μg/ml), while in the absence of TF, corn trypsin inhibitor greatly increased lag times and decreased peak thrombin generation (not shown). Under the conditions used, we found that 67 nM Zn²⁺-containing PS had a significant anticoagulant effect in normal, PS-depleted, and TFPI-depleted plasmas (Fig. 7A). Neutralizing antibodies against PS also significantly decreased the lag times in both normal and TFPI-depleted plasma, showing that at least a portion of the anticoagulant activity of plasma PS is TFPI-independent (Fig. 7A). Times to peak thrombin generation were significantly changed when lag times were changed (data not shown).

Fig. 7. Lag times for thrombin generation in normal plasma and depleted plasmas.

(A) Thrombin generation in pooled normal human plasma (NHP, light grey bars), PS-depleted plasma (PSdP, striped bars), or TFPI-depleted plasma (TFPIdP, dark grey bars). Coagulation was initiated with 0.3 pM TF, 15 mM CaCl₂, and 10 μM phospholipids as described in Methods. Additions prior to initiation of coagulation, where indicated, were 67 nM PS, 500 pM TFPI, or sufficient anti-TFPI (αTFPI) to neutralize all TFPI. From thrombin generation profiles, the parameter most sensitive to PS was lag time, as plotted here. Experiments were performed on each of 4 days, and each bar represents 4 to 9 assays. Where statistical significance (p<0.05) was achieved in a particular plasma, the p value is given above the relevant comparison with the parent plasma. (B) Lag times for thrombin generation in PSdP reconstituted with Zn²⁺-containing PS (vertical striped bars) or Zn²⁺-deficient PS (cross-hatched bars) at the concentrations indicated (in nM). αTFPI, where indicated, was incubated in the PSdP + PS for 8 min before procoagulants were added.

Interestingly, neutralizing antibodies against TFPI and concentrations of TFPI that had significant effects on lag times in normal plasma did not have significant effects on lag times in PS-depleted plasma (Fig. 7A, striped bars). Thus, exogenous TFPI is dependent on PS for its anticoagulant activity under these conditions, and functional TFPI in plasma may be diminished when PS is depleted from plasma. The latter notion is supported by data from Castoldi et al., who found that functional plasma TFPI is reduced when PS is immunodepleted [15].

Varying doses of Zn²⁺-containing and Zn²⁺-deficient PS were used to reconstitute PS-depleted plasma in TF-initiated thrombin generation assays (Fig. 7B). Zn²⁺-containing PS had a dose-dependent anticoagulant effect that approximated that of the PS in normal plasma when [PS] was 150 nM. For comparison, plasma contains 120–130 nM free PS and 170–200 nM PS-C4BP. In contrast, Zn⁺²-deficient PS had negligible anticoagulant effects at concentrations up to 200 nM. Antibodies against TFPI did not affect the behavior of PS. Previous studies showed that >12-fold higher concentrations of Zn²⁺-deficient PS were required to achieve similar anticoagulant effects as Zn²⁺-containing PS in TFPI-depleted plasma [9].

Discussion

These studies provide insight into roles of native Zn²⁺-containing PS in regulation of coagulation that are independent of APC and TFPI. Recently, we showed that immunoaffinity-purified PS and PS conventionally-purified without using anion exchange contain Zn²⁺ that is essential for high-affinity binding of FXa and for inhibition of prothrombinase activity [9]. In contrast, most conventionally-purified PS preparations were Zn²⁺-deficient and did not appreciably inhibit thrombin generation. In studies presented here, Zn²⁺-containing PS inhibited extrinsic FXase independently of TFPI, at varying concentrations of FVIIa and TF and in the presence and absence of plasma levels of FVII. TFPI contamination of the PS preparations was ruled out by preincubating PS preparations with neutralizing antibody against TFPI. No enhancement of TFPI inhibition was seen with Zn²⁺-containing PS. However, results obtained with Zn²⁺-deficient PS in the extrinsic FXase were similar to those published by Hackeng et al. [10]. Zn²⁺-deficient PS did not inhibit extrinsic FXase but appeared to enhance inhibition by TFPI. Zn²⁺-deficient PS that was reconstituted with Zn²⁺ in the presence of urea and then dialyzed extensively regained its ability to inhibit FXase independently of TFPI.

We hypothesized that inhibition of extrinsic FXase by Zn²⁺-containing PS is due to binding of PS to one or more of the extrinsic components: FX; FVIIa; TF; the complex TF/FVIIa; or the product FXa, or that PS competes with extrinsic pathway components for limiting phospholipid.

PS reversibly binds FXa and inhibits the amidolytic activity of FXa, although much less efficiently than it inhibits FXa prothrombinase activity [7]. However, the same concentration of PS that inhibited 58% of FXase activity inhibited only 18% of FXa amidolytic activity. Reduction in generation of FXa protein by TF/FVIIa in the presence of Zn²⁺-containing PS was confirmed in immunoblots for FXa. Thus, PS primarily inhibited generation of FXa, while inhibition of the amidolytic activity of the FXa product was a minor effect.

PS can compete with procoagulant factors for limiting phospholipids [14]. At lower concentrations of phospholipid (5–20 μM), Zn²⁺-containing PS did compete with FXase components for limiting phospholipid. However, at higher, saturating phospholipid concentrations (25–50 μM), inhibition by PS was still observed and was independent of lipid concentration. Competition for limiting phospholipid may be one mechanism of inhibition by PS but is not responsible for inhibition in the presence of saturating phospholipids.

Since PS binds FXa it is plausible that PS may bind and sequester the FX substrate. Zn²⁺-containing PS bound FX with Kd_app~130 nM, suggesting ~12-fold lower affinity than PS binding to FXa (Kd_app=9–13 nM) [7,9], while binding of Zn²⁺-deficient PS to FX was not measurable. Increasing the FX concentration 10-fold in FXase assays at fixed PS concentration reduced PS inhibition 5-fold, suggesting that PS competed with FX for binding to the TF/FVIIa complex. A Lineweaver-Burke plot of FXa generation in the presence and absence of PS and at variable [FX] revealed a competitive inhibition pattern (not shown). Thus, Zn²⁺-containing PS can partially inhibit FXa generation by binding FX substrate and by competing with FX for binding to the extrinsic TF/FVIIa complex.

Inhibition of FXase was not due to binding of PS to FVIIa, since PS bound FVIIa very weakly. Zn²⁺-containing PS and Zn²⁺-deficient PS that was reconstituted with Zn²⁺ bound soluble TF with high affinity (Kd_app=41 nM), making TF the likely primary target of PS. Notably, Zn²⁺-deficient PS did not bind to soluble TF.

We also used functional assays to determine which component of extrinsic FXase is the preferred target for PS. A fixed concentration of PS was preincubated with saturating phospholipids and with each of the components TF, FVIIa, or FX before the reaction was initiated by the other two. Preincubation of PS with TF yielded 80% inhibition of FXase activity, while preincubation of PS with FX or FVIIa yielded 20–30% inhibition. Thus, binding of PS to TF is likely the principal mechanism by which PS inhibits extrinsic FXase, consistent with binding data. Interestingly, preincubation of PS with TF/FVIIa complex followed by initiation with FX resulted in only ~30% inhibition. Therefore, FVIIa may partially shield TF from binding of PS, either by blocking a PS binding site on TF, or by inducing a less favorable conformation of TF for PS binding. Thus, Zn²⁺-containing PS may inhibit extrinsic FXase by several mechanisms: competing for limiting phospholipids; partially sequestering FX substrate; and most importantly, binding and inhibiting TF. These proposed functions are illustrated in Fig. 8, left and center. The high affinity of PS for TF may be tempered by the shielding effect of FVIIa toward TF and the even higher affinity of FVIIa for TF.

Fig. 8.

Schematic of mechanisms by which PS may inhibit extrinsic FXase. (Left) Competition for phospholipids was suggested in Fig. 3. (Center) TFPI-independent interaction of Zn²⁺-containing PS with TF and FX was suggested in Figs. 4–6. (Right) TFPI-dependent interaction of Zn²⁺-deficient PS was suggested in Fig. 1B,C and in published data [10]. This interaction may lead to enhanced FXa inhibition, based on published data [10,18].

Inhibition of FXase by Zn²⁺-containing PS in combination with TFPI was always less than additive, that is, inhibition by one of the anticoagulants seemed to interfere with inhibition by the other. The most likely explanation is that Zn²⁺-containing PS and TFPI compete for the same target(s). This could be TF and/or FXa, since both anticoagulants have now been reported to bind to these targets.

It was proposed that direct inhibition of prothrombinase by PS was due to PS multimers that were an artifact of purification [16]. However we showed that PS monomers and multimers exist in plasma and have similar ability to inhibit prothrombinase [17]. Note that Zn^2+-deficient PS that was reconstituted with Zn²⁺ was almost entirely monomeric and inhibited extrinsic FXase as efficiently as other Zn²⁺-containing preparations that contained both monomers and multimers.

Our findings on inhibition of extrinsic FXase activity by PS were borne out in dilute TF-initiated thrombin generation assays in plasma. Purified Zn²⁺-containing PS significantly increased the lag times in TFPI-depleted plasma, and neutralizing antibodies against PS significantly decreased the lag times in TFPI-depleted plasma. The latter is not in agreement with Hackeng et al [10], possibly because we used a higher dose of PS antibodies, or because we used less procoagulant conditions. Another consideration is that the antibodies against PS used by both our lab and Hackeng’s may not totally neutralize the ability of plasma PS to inhibit extrinsic FXase, as suggested by the shorter lag time in PS-depleted plasma than in normal plasma treated with the antibodies. Also, in purified component FXase assays using these antibodies, we could not achieve as much as 50% neutralization of PS ability to inhibit FXase (not shown). To rule out the presence of TFPI in the TFPI-depleted plasma, in some experiments we added neutralizing antibodies against TFPI to the TFPI-depleted plasma, but they did not cause a decrease in the lag time in the absence of PS (Fig. 7). The fact that antibodies against PS caused a greater decrease in lag time in normal plasma than in TFPI-depleted plasma (though both effects were significant) suggests that plasma PS may have both TFPI-independent and TFPI-dependent anticoagulant activities in normal plasma that are independent of APC. Since some Zn²⁺-containing PS preparations have as little as 0.7 moles of Zn/mole PS [9], it is possible that a minor fraction of PS in plasma is Zn²⁺-deficient and can stimulate TFPI activity.

Zn²⁺-containing PS also downregulated procoagulant activity on activated platelets, and antibodies against TFPI and protein C had no effect on this downregulation (unpublished studies in progress). Together with the plasma data and FXase data, this further demonstrates that some anticoagulant functions of PS are independent of both TFPI and APC.

It was previously reported that PS regulates coagulation in the absence of APC by enhancing the interaction between TFPI and FXa and that PS alone does not inhibit extrinsic FXase [10,18]. Here we show that Zn²⁺-containing PS inhibits FXase even in the absence of TFPI and does not enhance TFPI activity. On the other hand, we reported that Zn²⁺-containing and Zn²⁺-deficient PS bind TFPI with similar affinities (Kd_app~21 nM) [9], and here we confirmed that Zn²⁺-deficient PS appears to enhance inhibition of FXase by TFPI. Thus, there may be APC-independent anticoagulant effects of PS that are either TFPI-independent (Fig. 8, schemes on left and center) or TFPI-dependent (Fig. 8, scheme on right). We previously reported TFPI-independent inhibition of prothrombinase by Zn²⁺-containing PS [6,9], and here we demonstrate TFPI-independent inhibition of extrinsic FXase by Zn²⁺-containing PS that is largely due to PS binding to TF. PS emerges as a multifunctional anticoagulant that can act as a cofactor to APC, and inhibit both FXa generation and the prothrombinase activity of FXa.