Coagulation factor Xa cleaves PAR1 and mediates signaling dependent on binding to the endothelial protein C receptor

Summary

Background and objective

Coagulation is intrinsically tied to inflammation and both pro- and anti-inflammatory responses are modulated by coagulation protease signaling through protease activated receptor-1 (PAR1). Activated factor X (Xa) can elicit cellular signaling through PAR1 but little is known about the role of cofactors in this pathway. Endothelial protein C receptor (EPCR) supports PAR1 signaling by the protein C pathway and in the present study we tested if EPCR mediates surface recruitment and signaling of Xa.

Methods and results

Here we show that Xa binds to overexpressed as well as native endothelial EPCR. PAR1 cleavage by Xa as analyzed by conformation sensitive antibodies and a tagged PAR1 reporter construct was strongly enhanced if EPCR was available. Anti-EPCR failed to affect the tissue factor dependent activation of factor X but high concentrations of Xa decreased EPCR-dependent protein C activation. Most importantly, the Xa mediated induction of Erk1/2 activation, expression of the transcript for connective tissue growth factor, and barrier protection in endothelial cells required binding to EPCR.

Conclusions

Our results demonstrate that EPCR plays an unexpected role in supporting cell surface recruitment, PAR1 activation, and signaling by Xa.

Introduction

The zymogen clotting factor X is catalytically activated into the active serine protease (Xa) upon binding to the cell surface complex formed by tissue factor and protease ligand activated factor VII (VIIa) [1]. If the ternary complex is not inhibited by tissue factor pathway inhibitor [2] Xa will dissociate, assemble with cofactor activated clotting factor V on cell surfaces, and proteolytically activate prothrombin into thrombin which ultimately will allow clot formation by cleaving fibrinogen and activating platelets. Thrombin in complex with the cell surface receptor thrombomodulin also activates the anticoagulant protein C pathway in a negative feedback loop [3].

Beyond directly controlling coagulation, clotting factors have non-hemostatic signaling functions that play important roles in physiology and disease. Coagulation factors induce signal transduction through protease activated receptors (PARs). These closely related G-protein coupled receptors allow cells to sense for proteolytic activity in their microenvironment [4]. Whereas thrombin, a major mediator of cell-signaling events, directly binds to, cleaves and activates its prototypical receptor PAR1, other clotting factors require co-receptor binding. VIIa has been shown to induce PAR2-mediated signaling dependent on binding to tissue factor [5]. Activated protein C (APC) proteolytically activates PAR1 and induces cytoprotective signaling in endothelial cells dependent on binding to endothelial protein C receptor (EPCR) [6, 7]. EPCR is a cell membrane glycoprotein that shares homology with the CD1 family of major histocompatibility complex class 1 molecules. Binding of protein C and APC to EPCR is Ca⁺⁺ dependent and involves protein C’s γ-carboxyglutamic acid (Gla) domain [8, 9].

Cellular signaling by Xa has been implicated in a variety of conditions including wound healing and tissue fibrosis, atherosclerosis and restenosis, airway remodeling, cancer dissemination, and angiogenesis (see [10] for a review). In view of these findings, therapeutic targeting of cellular signaling by Xa seems to be a promising concept even though many in vitro studies used relatively high Xa concentrations and a precise (patho-) physiological role of Xa signaling in vivo remains to be established. Accordingly a large number of studies addressed the question how cellular Xa signaling is mediated. Previous studies have shown that Xa can activate both PAR1 and -2 on a variety of different cell types including endothelial cells [10]. Xa in a ternary complex with tissue factor and VIIa can more efficiently signal through both PAR1 and -2 [11] but other cofactors have also been implicated [12].

Recent studies have shown that the Gla domains of not only APC but also of VIIa can bind to EPCR [13-15]. Given that Xa also contains a highly homologous Gla domain, we hypothesized that Xa, APC, and VIIa might share one or more coreceptors for endothelial cell surface binding and that EPCR might be one of them. Here, we show that Xa indeed binds to endogenously expressed EPCR and that cleavage of PAR1 by Xa is strongly enhanced in the presence of EPCR. Consistent with EPCR dependent activation of PAR1, induction of ERK1/2 phosphorylation and expression of connective tissue growth factor (CTGF) in an endothelial cell line were also found to depend on EPCR.

Materials and methods

Reagents

Human thrombin and PAR1 and PAR2 agonist peptides were as described [6, 11]. All other clotting proteases were from Haematologic Technologies (Essex Junction, VT) with the exception of Gla domainless APC (Enzyme Research Laboratories, South Bend, IN). All experiments involving stimulation with Xa, APC, or VIIa included hirudin (Calbiochem, La Jolla, CA) unless indicated otherwise. Control experiments demonstrated that hirudin alone had no effect in any of our assays. Monoclonal anti-PAR1 ATAP2 was as described [16] and SPAN12/5 was recloned from SPAN12 hybridoma cells that were kindly provided by Dr. Lawrence Brass [17]. Monoclonal rat anti-EPCR RCR92 (non-blocking) and RCR252 (blocking Gla-domain binding to EPCR) were kindly provided by Dr. Kenji Fukudome (Saga Medical School, Saga, Japan) and were used at 25 μg/ml [18]. Amidolytic assays for APC and Xa activity were as described previously [19]. The recombinant Xa inhibitor nematode anticoagulant protein 5 (NAP5) was provided by Dr. George Vlasuk [20].

Cell culture and transfection

EA.hy926 cells [21] and primary human umbilical vein endothelial cells (HUVEC; Cascade Biologics, Portland, OR) were cultivated as described previously [16, 22]. In experiments involving gene silencing, cells were plated together with complexes of small interfering RNA (siRNA; 30 pM final concentration) and Lipofectamine™2000 (Invitrogen) according to the manufacturer’s instructions. Cells were used for experiments 48 h after transfection and the tissue culture medium was replaced the day before the experiment. Chemically synthesized, double-stranded siRNA with 19-nucleotide duplex RNA and 2-nucleotide 3′ dTdT overhangs was obtained from Ambion (Austin, TX). The siRNA sequences were GGGAAUAUUGCCAAUGCUAtt (targeting PAR1), CAACCGCACUCGGUAUGAAtt (targeting EPCR), and GGAUCAAACUCUGCUUCCUtt (targeting PAR4 and used as a control). Real time PCR analysis of PAR1, PAR2, and EPCR mRNA levels was used to demonstrate efficiency of downregulation of the specific target and to rule out unspecific effects on other genes. CHO-K1 and HEK293 cells were obtained from the American Type Culture Collection (ATCC) and grown in DMEM/F12 (for CHO-K1 cells; Invitrogen) or DMEM (for HEK293 cells; Invitrogen). Both media were supplemented with 10% fetal bovine serum. The cells were transiently transfected with the pcDNA3.1/Zeo+ plasmid vector (Invitrogen) using Lipofectamine™2000. The expression constructs containing the human PAR1 or EPCR coding sequences were as described previously [6, 11]. To obtain a PAR1 cleavage reporter construct the coding sequence of PAR1 was cloned (EcoRI/XhoI restriction sites) into a modified pcDNA3.1/Zeo+ vector with a deleted ApaI site. The signal peptide of PAR1 was removed using EcoRI and PAR1’s native ApaI site and replaced by secretory alkaline phosphatase (SEAP). The SEAP coding sequence containing a 5′ EcoRI and a frame adjusted 3′ ApaI site was a kind gift from Dr. Laurent Mosnier [23]. HEK293 cells were used for experiments 24 h after transfection. To obtain CHO-K1 stably expressing EPCR the pcDNA3.1/Hygro+ plasmid containing the coding sequence for EPCR was linearized with SapI and transfected into CHO-K1 cells. Corresponding empty vector was used to generate control cells. Two days after transfection selection with hygromycin (800 μg/ml) was initiated. Untransfected CHO-K1 were found to be eliminated within 5 d if exposed to 400 μg/ml of hygromycin. After 10 d of selection, surviving cells were cloned and a clone with normal growth speed and high and stable EPCR expression was used for experiments.

Determination of surface binding of proteases

Cell surface protease binding was quantified using a modified method described by Ghosh et al [14]. Briefly, confluent cell layers were washed once with calcium free HBSS containing 5 mM EDTA followed by HBSS containing 1% bovine serum albumin, 1 mM MgCl₂ and 5 mM CaCl₂ and kept in this buffer for all subsequent steps. Cells were chilled and kept on ice throughout the experiment. Following incubation with biotinylated proteases for 3 h the cells were washed twice, incubated for 10 min with horseradish peroxidase (HRP)-coupled streptavidin (Invitrogen) and washed again for 5 times before tetramethylbenzidine was added for spectrophotometric quantification. Negatively charged phospholipid rich cell membrane domains were quantified using biotinylated annexin V (Calbiochem) instead of biotinylated protease in the same assay with a 15 min incubation time.

Cell surface immunoassays, Western blotting, and real time PCR

Cell surface PAR1 and EPCR were quantified by cell surface ELISA as described previously [16]. PAR1 was detected with biotinylated (Mini-Biotin-XX Protein Labeling Kit, Invitrogen) mouse monoclonal SPAN12/5 (2 μg/ml) or ATAP2. EPCR was detected with biotinylated rat monoclonal RCR252. Streptavidin-coupled HRP and tetramethylbenzidine were used for spectrophotometric quantification of cell surface antibody binding. Immunoblotting was carried out as described [16]. Mitogen-activated protein kinase phosphorylation was detected with rabbit anti-phospho-ERK1/2 (#9101; Cell Signaling Technology, Beverly, MA) as described [24]. Optical density of immunoreactive bands was quantified using Scion Image Alpha 4.0.3.2 software. Quantification of mRNA encoding EPCR or connective tissue growth factor (CTGF) by real time PCR was as described [25] and normalized to GAPDH transcript levels.

Permeability assay

Macromolecular monolayer permeability was analyzed in a dual chamber system using Evans blue-labeled bovine serum albumin (BSA) as described previously [22, 26]. Briefly, EA.hy926 cells were plated on Transwell polycarbonate membranes of 3-μm pore size and 12-mm diameter (#3402; Corning Inc., Corning, NY). The upper and lower chambers were filled with 500 μl and 1500 μl growth medium, respectively. Cells were grown for 2 days, agonists were added in serum-free medium containing 0.4% BSA and permeability was assayed 3 h later using 0.67 mg/ml Evans blue diluted in growth medium containing 4% BSA. Fresh growth medium was added to the lower chamber and the medium in the upper chamber was replaced with Evans blue/BSA. After 10 min the optical density at 650 nm was measured in a 1:3 diluted 50 μl sample from the lower chamber.

PAR1 cleavage reporter assay

HEK293 cells transiently expressing alkaline phosphatase (AP) tagged PAR1 were washed two times and incubated with agonists for 20 min. The supernatants were removed, separated from cell debris by passing through a cellulose ester filter (pore size 0.45 μm) and AP activity was quantified using the colorimetric substrate para-Nitro Phenyl Phosphate (1-Step™ PNPP, Thermo Scientific, Rockford, IL).

Statistical analysis

Data analysis was performed using the NCSS Statistical & Power Analysis or SigmaStat 3.5 (Systat Software Inc.) software. A two-sample two-tailed homoscedastic t-test was used to calculate the indicated p-values.

Results

Overexpressed EPCR supports Xa binding to the cell surface

In order to investigate whether EPCR can recruit Xa to the cell surface we analyzed binding of active site blocked biotinylated human Xa and APC to chinese hamster ovary (CHO-K1) cells. Cells were incubated at 4 °C to prevent internalization of the bound proteases. Binding to native CHO-K1 cells and mock transfected control cells was very low indicating that CHO-K1 cells do not significantly express high affinity receptors for the human proteases. Unexpectedly, CHO cells stably expressing human EPCR (CHO/EPCR cells) bound Xa and APC with comparable efficiency (Fig. 1A). Similar results were obtained in transiently transfected HEK293 cells (not shown). Confluent monolayers of CHO/EPCR cells had approximately three times higher EPCR expression levels compared to the expression of native EPCR on human endothelial EA.hy926 cells (Fig. 1B). Xa and APC binding to CHO/EPCR cells was Ca⁺⁺ dependent (Fig. 1C). The Gla domain containing coagulation proteases can bind to negatively charged phospholipids in the cell membrane in a Ca⁺⁺ dependent manner. We therefore tested whether CHO/EPCR cells express higher levels of these binding sites using annexin V, which specifically binds to anionic phospholipids. As shown in Figure 1D, annexin V binding was comparable in CHO-K1, mock transfected, and CHO-EPCR cells. Furthermore, active site blocked biotinylated thrombin bound similarly to control and EPCR expressing cells (Fig. 1E). Taken together, these data indicate that human EPCR can support the calcium dependent cell surface recruitment of Xa. Active site blocked APC, zymogen protein C, zymogen factor X, and VIIa all competed for Xa binding to CHO/EPCR cells consistent with the conclusion that Xa shares binding sites with APC and VIIa (Fig. 1F). In contrast, protein S, factor Va, and active site blocked thrombin did not compete with Xa-EPCR binding to CHO/EPCR cells. The absence of competition by the Gla domain containing protein S indicates that unspecific competition by the Gla domain does not explain the results using protein C/APC, factor X, and VIIa.

Fig. 1.

Binding of biotinylated active site blocked Xa to overexpressed EPCR. (A) CHO-K1 cells either non transfected or stably expressing empty vector or human EPCR were incubated for 3 h at 4°C with the indicated concentrations of biotinylated active site blocked Xa (Xa-bEGR) or APC (APC-bEGR) and the amount of surface-associated proteases was determined. (B) Quantification of monoclonal rat anti-EPCR (RCR252) binding to the indicated cell lines. (C) Binding of 50nM of Xa-bEGR or APC-bEGR was quantified in the absence and presence of calcium. (D) Binding of biotinylated Annexin V to the indicated cell lines was analyzed. (E) Binding of biotinylated active site blocked thrombin was quantified. (F) In CHO-K1 stably expressing EPCR the surface binding of 10 nM Xa-bEGR was analyzed in the absence or presence of 1 μM of unlabeled active site blocked APC (APC-DEGR), active site blocked thrombin (Thrombin-FPRCK), and other coagulation factors as indicated. Results in all panels are expressed as mean±SEM, n=6 (A, B, C and F), 5 (D), or 12 (E). **p<0.005 compared to control.

Reduced Xa binding to the endothelial cell surface in the presence of APC or anti-EPCR

Next, we analyzed Xa binding to human endothelial cell lines. Binding of biotinylated Xa and APC to EA.hy926 cells was time (not shown) and dose dependent (Fig. 2A) and half maximal binding was reached for both proteases at concentrations of approximately 30 nM. Xa binding to primary HUVECs was comparable (Fig. 2B). Binding was again calcium dependent because almost no surface binding was detected in the absence of calcium in the binding buffer or in the presence of EDTA (data not shown). To determine whether the endothelial surface binding sites for Xa are shared with other clotting factors we analyzed Xa binding in the presence of a 100-fold molar excess of unlabeled zymogen factor X, active site blocked APC or active site blocked thrombin. Xa binding was significantly reduced in the presence of factor X or APC whereas thrombin did not compete for Xa binding (Fig. 2C). The finding that competition by factor X was more efficient than by APC may indicate that affinities for a shared receptor on endothelial cells are different or that one or more receptors that support factor X/Xa binding do not bind APC. To test whether EPCR constitutes binding sites that are shared by Xa and APC we used the RCR-252 monoclonal anti-EPCR that is known to block the interaction of APC with EPCR. RCR-252 indeed reduced surface binding of both APC and Xa to a similar degree (Fig. 2D), suggesting that EPCR is a major binding receptor for Xa on the EA.hy926 cell surface.

Fig. 2.

Binding of Xa-bEGR to the endothelial cell surface. EA.hy926 cells (A) or HUVECs (B) were incubated for 3 h at 4°C with the indicated concentrations of Xa-bEGR or APC-bEGR followed by analysis of surface binding. (C) Surface binding of 10 nM Xa-bEGR to EA.hy926 cells was analyzed in the absence or presence of 1 μM of the indicated competitors. (D) Cells were preincubated in the absence or presence of 25 μg/mL of EPCR blocking antibody RCR252 before the 3 h incubation at 4°C in the absence or presence of APC-bEGR or Xa-bEGR. Results are expressed as mean±SEM, n=3 (A and D), 5 (B), or 6 (C), *p<0.05, **p<0.005 compared to control.

Gla-domain dependent cleavage of endothelial PAR1 by Xa

Cell surface immunoassays were used to quantify cleavage of endogenous endothelial PAR1 by Xa. We have previously established that ATAP2 is conformation sensitive and does not bind to cleaved PAR1 in the immunoassay [16]. SPAN12/5 anti-PAR1 was raised against a peptide spanning the cleavage site and it is expected to be cleavage sensitive [17]. Small interfering RNA (siRNA) targeting PAR1 reduced PAR1 message and antigen to less than 5% (data not shown). Cell surface binding of SPAN12/5 was reduced close to baseline if cells were pretreated with siRNA targeting PAR1 or if the SPAN12/5 epitope on PAR1’s N-terminus was removed by the specific agonist thrombin, demonstrating that SPAN12/5 specifically binds to full length uncleaved PAR1 (Fig. 3A). A 3 h incubation with both Xa and APC significantly reduced binding of monoclonal anti-PAR1 ATAP2 and SPAN12/5 (Fig. 3B). As expected, VIIa had no effect on anti-PAR1 binding. The downregulation of SPAN12 staining after a short 30 min incubation with Xa as well as APC was dose dependent with highly significant effects already at 12.5 nM protease concentration. The Gla-domain deficient corresponding proteases had amidolytic activity similar to the wildtype (data not shown) but did not downregulate SPAN12/5 staining at up to 50 nM concentration (Fig. 3C). The Xa inhibitor NAP5 blocked Xa but not APC responses demonstrating specificity and the dependence on Xa’s proteolytic activity (Fig. 3D). These results for the first time directly demonstrate that Xa leads to cleavage of native endothelial PAR1 dependent on its Gla-domain and proteolytic activity.

Fig. 3.

Quantification of cleavage of endogenous endothelial PAR1 by Xa using a SPAN12/5-based immunoassay. (A) EA.hy926 endothelial cells were pretreated with control siRNA or PAR1 silencing RNA followed by a 30 min incubation with control or thrombin. Binding of monoclonal anti-PAR1 ATAP2 and SPAN12/5 was quantified in the absence and presence of the respective immunization peptide (blocking peptide). (B) ATAP2 and SPAN12/5 monoclonal anti-PAR1 binding was analyzed after a 3 h incubation with the indicated agonists. Background-subtracted normalized data are shown. (C) Cells were incubated for 30 min with the indicated concentrations of wildtype or Gla-domain deleted (desGla) proteases followed by analysis of SPAN12/5 binding. (D) Cells were incubated (30 min) with the indicated agonists in the absence and presence of the Xa inhibitor NAP5 (1 μM) followed by analysis of SPAN12/5 binding. Data are given as mean±SEM, n=4 (A), 5 (B), or 6 (C and D), **p<0.005 compared to control.

EPCR supports PAR1 cleavage by Xa

After having established an assay that allows the specific and sensitive quantification of PAR1 cleavage by Xa and in view of our evidence for substantial binding of Xa to EPCR, we tested if Xa requires EPCR coreceptor binding in order to efficiently cleave PAR1. We first investigated if monoclonal anti-EPCR prevents PAR1 cleavage. Both Xa and APC were found to efficiently cleave PAR1 in the absence of anti-EPCR and in the presence of the non-blocking anti-EPCR RCR92. In contrast, the blocking anti-EPCR RCR252 completely prevented the downregulation of SPAN12/5 binding by Xa and APC (Fig. 4A). Cleavage of PAR1 by low-dose thrombin was not affected by anti-EPCR. Two different complementary approaches were used to independently confirm the surprising finding that EPCR is required for PAR1 cleavage by Xa in endothelial cells. Firstly, we used siRNA to downregulate EPCR expression in EA.hy926 cells. Transcript levels for EPCR in the transfected cells were below 20% and surface protein was approximately 40% of control transfected cells under our experimental conditions (Fig. 4B). This moderate downregulation was associated with a highly significant decrease in the efficiency of PAR1 cleavage by Xa and APC whereas the effects of thrombin and plasmin on SPAN12/5 binding were unchanged (Fig. 4C). Importantly, downregulation of EPCR did not affect native PAR1 expression.

Fig. 4.

Role of EPCR in the Xa mediated downregulation of SPAN12/5 binding to endothelial cells. (A) EA.hy926 cells were preincubated for 15 min in the absence or presence of 25 μg/mL of monoclonal anti-EPCR RCR92 (non-blocking) or RCR252 (blocking) followed by a 30 min incubation with the indicated agonists and quantification of cleavage-sensitive anti-PAR1 SPAN12/5 binding. (B) EA.hy926 cells were transfected with control siRNA or siRNA targeting EPCR. EPCR transcript levels were analyzed by real time PCR and normalized to GAPDH (left panel). Cell surface expressed EPCR protein was quantified by immunoassay using monoclonal anti-EPCR RCR252 (right panel). (C) SPAN12/5 binding was quantified after a 30 min incubation with the indicated agonists in EA.hy926 cells either transfected with control or siRNA targeting EPCR. Data are given as mean±SEM, n=9 (A+B right panel+C) or 3 (B left panel), *p<0.05, **p<0.005.

Secondly, we used a reporter construct that encodes alkaline phosphatase (AP) fused to the N-terminus of human PAR1 to directly analyze PAR1 cleavage in HEK293 cells. Co-expression of EPCR led to significantly more efficient release of AP by Xa and APC from the reporter construct (Fig. 5A). This finding was not caused by increased surface expression of AP tagged PAR1 in the EPCR cotransfected cells as shown in Figure 5B. AP release by both proteases was dose dependent with significant effects already at 12.5 nM (Fig. 5C). Xa and to a lesser extent APC also led to AP release in cells only transfected with PAR1 but both proteases were significantly more efficient if EPCR was also expressed. Thus, EPCR supports Xa-mediated cleavage of both endogenous endothelial PAR1 and overexpressed tagged PAR1 in the HEK293 cell system.

Fig. 5.

Cleavage of overexpressed alkaline phosphatase tagged PAR1 by Xa. (A) HEK293 cells were transiently transfected with human PAR1 tagged with alkaline phosphatase (AP) at the N-terminus either alone or together with human EPCR. Cells were incubated with the indicated agonists for 20 min and released AP activity was quantified in the conditioning medium. (B) Cell surface AP activity was quantified in HEK293 cells transfected as indicated. (C) HEK293 cells were transfected with AP-PAR1 and control or EPCR followed by incubation (20 min) with various concentrations of Xa or APC as indicated. Released AP activity was analyzed. Data are given as mean±SEM, n=8 (A), 12 (B), or 5 (C), **p<0.005.

EPCR coreceptor binding does not affect factor X activation

The endothelial cell surface is highly organized and compartmentalized and we tested whether the interaction with EPCR might play a role modulating activation of factor X by the tissue factor-VIIa complex. Quiescent EA.hy926 only express minimal amounts of tissue factor and were found to marginally support Xa generation whereas tumor necrosis factor-α (TNFα)-induced cells supported substantial Xa generation. Blocking of EPCR had no effect on Xa generation in quiescent or TNFα-stimulated cells (Fig. 6A). EPCR binding of zymogen protein C increases activation efficiency by the thrombin-thrombomodulin complex and it was recently described that high concentrations of VIIa can decrease protein C activation slightly by competing for EPCR binding [14]. We found that high concentrations of Xa, VIIa (Fig. 6B), or APC-DEGR (not shown) indeed all similarly interfere with APC generation, providing additional evidence that APC, VIIa and Xa all compete for EPCR binding on the endothelial surface.

Fig. 6.

Effect of Xa-EPCR interaction on Xa and APC generation. (A) EA.hy926 cells were serum starved for 5 h in the absence or presence of 1 nM TNFα and incubated with VIIa (20 nM) and zymogen factor X (100 nM) in the absence or presence of RCR252 or an isotype matched unspecific control antibody (25 μg/ml). The generated Xa was quantified by amidolytic assay. (B) Cells were incubated with zymogen protein C (80 nM) and thrombin (1 nM) for 1 h in the presence of the indicated concentrations of competitors for EPCR binding followed by quantification of the generated APC. Mean±SEM, n=3, *p<0.05, **p<0.005 compared to no competitor.

EPCR binding is required for downstream signaling by Xa

We analyzed if EPCR coreceptor binding is also required for downstream signaling responses to Xa in endothelial cells. Endothelial cells express both PAR1 and PAR2 and Xa has been shown to elicit signaling through both of these PARs. Xa and APC as well as the PAR1 and -2 specific agonist peptides substantially induced phosphorylation of ERK1/2 MAP kinase in EA.hy926 cells. If the cells were preincubated with EPCR blocking antibodies phospho-MAP kinase induction by Xa and APC was abolished whereas responses to the agonist peptides were not affected demonstrating that both PAR1 and PAR2 were still available for signal transduction (Fig. 7A). Xa signaling has been shown to induce a pro-fibrotic response including the induction of connective tissue growth factor (CTGF) [27]. Induction of the CTGF transcript by Xa was strongly reduced in the presence of blocking anti-EPCR (Fig. 7B). We have previously shown that Xa can mediate endothelial barrier protective signaling through either PAR1 or PAR2 signaling [26]. Blocking anti-EPCR significantly reduced barrier enhancement by Xa in a dual chamber assay (Fig. 7C). Thus, downstream signaling by Xa requires binding to EPCR in our assays.

Fig. 7.

Effect of blocking EPCR binding on downstream signaling by Xa in endothelial cells. (A) EA.hy926 cells were serum starved in the absence or presence of an unspecific isotype matched rat antibody or anti-EPCR RCR252 (both at 25 μg/ml) followed by a 7 min incubation with the indicated proteases and agonist peptides (AP; PAR1 AP was TFLLRNPNDK at 20 μM, PAR2 AP was SLIGRL at 25 μM). The upper part of the panel shows a representative immunoblot and the lower part shows quantitative analyses of another set of 4 experiments. β-Actin was used as a loading control to normalize the quantitative results. (B) CTGF transcript expression was analyzed by real time PCR after a 3 h incubation of EA.hy926 cells with the indicated agonists. Results were normalized to GAPDH transcript levels. (C) Subconfluent cells in a Transwell chamber were incubated for 3 h with the indicated agonists and permeability was determined as described in Materials and methods. Means±SEM, n=4, *p<0.05.

Discussion

Using several approaches, i.e., overexpression of EPCR in different cell lines, EPCR blocking antibodies, and reducing endothelial cell surface expression of EPCR we demonstrate that EPCR supports surface binding, PAR1 cleavage, and signaling by Xa. EPCR was recently shown to also bind VIIa [13-15] and our results add to accumulating evidence that this receptor is able to support cell recruitment and positioning of Gla-domain containing proteins other than protein C. Previous studies analyzing binding of APC to the endothelial cell surface indicated that factor X and Xa compete only very inefficiently for binding [28]. In contrast, recent results from Ghosh et al. indicate that on the endothelial cell surface factor X and protein C are similarly efficient in competing for EPCR binding of VIIa and co-competitions of protein C plus factor X did not additionally affect VIIa binding, suggesting that APC, VIIa, and factor X share binding sites [14]. Interestingly, however, factor X failed to compete for VIIa binding to EPCR-transfected CHO-K1 cells in the study by Ghosh et al., whereas factor X did efficiently compete for Xa binding in our studies in stably transfected CHO-K1 cells (Fig. 1F). The Gla domains of both human protein C and factor VII contain a leucine at position 8 and this residue has been shown to be crucial for EPCR recognition using purified proteins in surface plasmon resonance studies [13]. Given that human factor X contains a methionine at this position it is possible that EPCR is necessary but not sufficient for Xa binding and that cell type specific differences in the expression of other cofactor(s) that contribute to EPCR-dependent and -independent binding may explain the conflicting results in competition studies. One such other cofactor may be annexin 2 which was recently implicated as an endothelial cell receptor that supports PAR1-dependent signaling by Xa [12]. Another possibility is that different forms of EPCR itself have specific relative affinities for the various ligands. Extensive studies in purified systems using EPCR from different sources will be required to quantitatively establish binding affinities and competition efficiency of the coagulation proteases for EPCR binding. Analysis of these data together with results using cell surface expressed EPCR may reconcile the discrepant results in the literature.

EPCR supports PAR1 activation by APC [6] and is known to colocalize with PAR1 in lipid rafts in the cell membrane [29]. Consistent with the concept that EPCR can recruit and position coagulation proteases in close proximity to PAR1 for efficient activation, our results directly demonstrate that cleavage of endothelial PAR1 by Xa depends to a large extent on the availability of EPCR binding. However, endothelial cells express high levels of both PAR1 and PAR2 and one fundamental difference between APC and Xa is that APC signaling seems completely PAR1-dependent whereas Xa can activate both PAR1 and PAR2 on endothelial cells. Specifically, we have previously shown that Xa can mediate the induction of MAP kinase phosphorylation and barrier protection in EA.hy926 cells through either PAR1 or PAR2 [26]. Our present finding that both these downstream responses to Xa in the same cell system depend on EPCR thus indicates that PAR2 activation by Xa also requires EPCR binding and suggests that EPCR also colocalizes with PAR2 in the cell surface. If this turns out to be the case another question will be why endothelial PAR2 supports Xa signaling but not APC signaling given that APC can activate PAR2 in an overexpression system [6]. One possibility is that posttranslational modifications, e.g. glycosylation [30] of PAR2 play a role in restricting activation to certain proteases in endothelial cells. Clearly, further studies and novel tools will be required to directly analyze cleavage of endogenous PAR2 in endothelial cells in order to address these questions.

The physiological or pathophysiological role of Xa signaling remains to be established even though a variety of cellular responses of potential relevance have been described in cell culture [10]. Our current study shows that EPCR binding plays an important role in PAR activation by Xa and both Xa’s profibrotic effects (CTGF induction) and the potentially cytoprotective and anti-inflammatory barrier enhancement. It is possible that other cellular responses to Xa are EPCR independent. However, PAR activation has been implicated in most Xa responses in cells and it is therefore likely that expanding the scope of the present studies will lead to the discovery of other downstream effects of Xa that are EPCR dependent. Relatively high Xa concentrations of unknown physiological relevance were used for most in vitro studies. However, these concentrations may be reached locally, resulting in Xa dependent signaling, e.g. in the microenvironment of endothelial and tumor cells [31]. Many cancer cell lines are known to express not only PARs but also EPCR [32]. EPCR binding may not only play a key role for Xa signaling but create a pool of factor X/Xa on the cellular surface and affect clearance and trafficking of this coagulation factor across cell barriers [33]. In conclusion, our results demonstrate that EPCR plays a key role in supporting cell surface recruitment, PAR1 cleavage, and signaling by factor Xa in endothelial cells. Thus, EPCR constitutes an unexpected link where the protein C and factor X pathways intersect. Further studies are needed to address the implications of this crosstalk for potential approaches to therapeutically target pro- and anti-inflammatory cellular signaling by Xa and APC.