Endothelial plasminogen activator inhibitor-1 blocks the intrinsic pathway of coagulation, inducing the clearance and degradation of factor XIa.

Abstract

Objective:

Activation of coagulation factor (F) XI by activated coagulation FXII (FXIIa) is a prothrombotic process. The endothelium is known to play an antithrombotic role by limiting thrombin generation and platelet activation. It is unknown whether the antithrombotic role of the endothelium includes sequestration of FXIa activity. This study aims to determine the role of endothelial cells in the regulation of the intrinsic pathway of coagulation.

Approach and Results:

Using a chromogenic assay we observed that human umbilical vein endothelial cells (ECs) selectively blocked FXIa yet supported kallikrein and FXIIa activity. Western blotting and mass spectrometry analyses revealed that FXIa formed a complex with endothelial plasminogen activator inhibitor-1 (PAI-1). Blocking endothelial PAI-1 increased the cleavage of a chromogenic substrate by FXIa and the capacity of FXIa to promote fibrin formation in plasma. Western blot and immunofluorescence analyses showed that FXIa-PAI-1 complexes were either released into the media or trafficked to the early and late endosomes and lysosomes of ECs. When baboons were challenged with Staphylococcus aureus (S. aureus) to induce a prothrombotic phenotype, an increase in circulating FXIa-PAI-1 complex levels were detected by ELISA within 2–8 hours post-challenge.

Conclusions:

PAI-1 forms a complex with FXIa on ECs blocking its activity and inducing the clearance and degradation of FXIa. Circulating FXIa-PAI-1 complexes were detected in a baboon model of S. aureus sepsis. While ECs support kallikrein and FXIIa activity, inhibition of FXIa by ECs may promote the clearance of intravascular FXIa.

GRAPHIC ABSTRACT

INTRODUCTION.

Contact activation of blood is initiated by exposure of blood to negatively charged surfaces or macromolecules. The main contact activation system protease, activated factor XII (FXIIa), can initiate the intrinsic pathway of thrombin generation by activation of coagulation factors XI (FXI)¹ and prekallikrein (PK).² Yet the role of contact activation in (patho)physiology extends beyond eliciting thrombin generation to trigger platelet activation and blood clotting. For instance, contact activation promotes the generation of bradykinin after cleavage of high-molecular-weight kininogen (HK) by kallikrein. Bradykinin induces subsequent endothelial nitric oxide (NO) and prostacyclin production leading to local or systemic responses such as vasodilation, pain, increased vascular permeability, bronchospasm, hypotension, neutrophil chemotaxis, and others.²

While both the intrinsic pathway of thrombin generation and the kallikrein/kinin/angiotensin pathway can be initiated by contact activation, the contact system-initiated pathways are regulated independently of each other. Contact activation can generate thrombin, yet this pathway does not appear to be required for normal hemostasis, as people or animals deficient in FXII, PK or HK do not exhibit excessive bleeding.³ Since FXI deficiency may be associated with hemostasis impairment,⁴ hemostatic activation of FXI is likely independent of the contact system. However, experimental data suggest that all members of the contact system (FXII/PK/HK/FXI) may contribute to a variety of pathological processes or conditions, including, among others, inflammation, thrombosis, atherosclerosis, stroke, or sepsis,⁵ and may be up- or downregulated independently, depending on the specific condition. For example, deficiency of C1 esterase inhibitor (C1 inh) which is the plasma inhibitor of the contact system enzymes results in hereditary angioedema, a rare disorder in which excessive formation of bradykinin leads to aggressive attacks of tissue swelling.⁶ Surprisingly, C1 inh deficiency does not seem to cause thrombosis in these patients, suggesting that the vasoactive kallikrein-kinin-NO-angiotensin system can operate independently of the intrinsic pathway of thrombin generation.

Contact activation by negatively charged surfaces has been extensively studied, both in vitro and in vivo. Even though it is known that the members of the contact system can bind to receptors and cofactors on different cell types, it is less clear whether the cell surface regulates the activity of FXIIa, FXIa, and kallikrein. For instance, it is known that FXII and PK-HK can bind to urokinase receptor (uPAR), cytokeratin 1 (CK1) and the receptor for the complement protein C1q (gC1q-R) to form a multiprotein complex on endothelial cells (ECs), and that this assembly promotes the generation of FXIIa, kallikrein and bradykinin.⁷ Yet, whether the endothelium regulates the activity of FXIa is unclear. Several inhibitors of FXIa activity have been identified including the kunitz inhibitors protease nexin 2 and serine protease inhibitors C1 inh, protease nexin 1, α1-antitrypsin, α2-antiplasmin, antithrombin (AT) and plasminogen activator inhibitor-1 (PAI-1),^8–10 some of which are expressed on ECs.^11,12 It remains controversial as to whether or not FXIa even binds to ECs,^13–17 while endothelial regulation of FXIa activity remains an open question. This study suggests that ECs can selectively inhibit the intrinsic pathway of thrombin generation, thereby adding another so far unrecognized mechanism contributing to the antithrombotic phenotype of the endothelium.

METHODS

The data that support the findings of this study are available from the corresponding author on request. Details of the major resources can be found in the online-only Data Supplement.

Reagents.

Plasma-derived FXI, FXIa and thrombin were from Haematologic Technologies, Inc. (Essex Junction, VT, USA). Plasma-derived FXIIa, PK, kallikrein, HK and corn trypsin inhibitor (CTI) were from Enzyme Research Laboratories, Inc. (South Bend, IN, USA). FXIIa/kallikrein chromogenic substrate; Chromogenix S-2302, t-PA chromogenic substrate; Chromogenix S2288 and FXIa chromogenic substrate; Chromogenix S-2366 were from Diapharma Group, Inc. (West Chester, OH, USA). Tumor necrosis factor-α (TNFα) was from R&D System (Minneapolis, MN, USA). Heparinase I, II and III were from New England Biolabs (Ipswich, MA, USA). Elastase, Phe-Pro-Arg chloromethylketone (PPACK), hirudin and tissue plasminogen activator (t-PA) were from Sigma-Aldrich (St Louis, MO, USA). NeutrAvidin Agarose beads and sulfo-NHS-SS-biotin were from Thermo Fisher Scientific (Rockford, IL, USA). Recombinant PAI-1 (rPAI-1) was from Biovision (Milpitas, CA, USA). Hoechst 33342 was from Invitrogen (Carlsbad, CA, USA). Protein A/G PLUS-Agarose beads were from Santa Cruz Biotechnology (Dallas, TX, USA).

Antibodies.

The anti-FXI/XIa light chain (LC) antibody (10C9) was generated as previously described.¹⁸ The anti-FXI antibody (1A6) used for immunoprecipitation and fibrin generation assays was generated as previously described.^19,20 The anti-FXI antibody (14E11) used for immunofluorescence assays was generated as previously described.^19,20 A biotin-conjugated anti-FXI antibody was from Affinity Biological, Ancaster, ON, Canada. The anti-PAI-1 antibody used for western blotting, anti-mouse IgG-k, anti-C1 inh antibody and anti-AT antibody were from Santa Cruz Biotechnology (Dallas, TX, USA). Rabbit anti-human PAI-1 used for ELISA was from American Diagnostica, Stamford, CT. A blocking anti-PAI-1 antibody was from GeneTex (Irvine, CA, USA). A blocking anti-tissue factor (TF) antibody was from Sekisui Diagnostics (Stamford, CT, USA). Rabbit anti-Rab7 and anti-EEA1 antibodies were from Abcam (Cambridge, UK). Rabbit anti-LAMP1 and anti-Rab 11 antibodies was from Cell Signaling Technology (Danvers, MA, USA). Alexa Fluor 488 Goat anti-mouse IgG and Alexa Fluor 546 Goat anti-rabbit IgG were from Invitrogen (Carlsbad, CA, USA).

Endothelial cells.

Human umbilical vein endothelial cells (HUVECs, ATCC, Manassas, VA, USA) were grown to confluence on 0.1% gelatin with endothelial basal medium-2 enriched with supplements (Lonza, Walkersville, MD, USA). Primary baboon arterial endothelial cells (BAECs) were isolated using collagenase digestion and mechanical disruption of baboon carotid arteries as described.²¹

Measurement of FXI or PK activation by FXIIa.

HUVECs were grown to confluence in 96-well plates and incubated for 1hr with either FXI (100nM) or PK (100nM) in the presence of HK (100nM) and FXIIa (1nM) in a PBS/CaCl₂/MgCl₂ buffer with 0.3% BSA and 10μM ZnCl₂. Reactions were stopped with CTI (40μg/ml); FXIa and kallikrein activity was determined by adding the chromogenic substrates Chromogenix S-2366 (0.8mM) or Chromogenix S-2302 (0.6mM), respectively. The rate of substrate hydrolysis was measured at 405nm. Selected experiments were performed with HUVECs incubated for 16hrs in serum-free medium with 0.3% BSA in the presence of TNFα (0.5ng/ml) or incubated for 4hrs with elastase (50nM).

FXIa and kallikrein activity assay.

HUVECs were grown to confluence in 96-well plates and incubated for 2hrs with increasing concentrations of FXIa (0–250pM) or kallikrein (0–250pM) in a PBS/CaCl₂/MgCl₂ buffer with 0.3% BSA and 10μM ZnCl₂. FXIa and kallikrein activity was determined by adding the chromogenic substrate Chromogenix S-2366 (0.8mM) or Chromogenix S2302 (0.6mM), respectively and the rate of substrate hydrolysis was measured at 405nm.

rPAI-1 activity assay.

Increasing concentrations of rPAI-1 (0–40nM) were incubated with 5nM FXIa, kallikrein and t-PA at 37°C for 30 min. Samples were diluted 20-fold and FXIa, kallikrein or t-PA activity was determined by adding the chromogenic substrate Chromogenix S-2366 (0.8mM), Chromogenix S2302 (0.6mM) or Chromogenix S2288 (0.6mM), respectively and the rate of substrate hydrolysis was measured at 405 nm.

Blood collection and preparation of plasmas.

Human venous blood was drawn by venipuncture from healthy male and female adult volunteers into sodium citrate (in 0.32%w/v sodium citrate unless otherwise noted) in accordance with the OHSU Institutional Review Board (IRB #1673). Informed consent was received from all human blood donors. Platelet-poor plasma (PPP) was prepared by centrifugation of citrated whole blood from three separate donors at 2150×g for 10 min. Further centrifugation of the plasma fractions at 2150×g for 10 min yielded PPP, which was pooled and stored at −80°C until use.

Preparation of supernatant from activated platelets.

Human venous blood was drawn in accordance with an IRB-approved protocol from healthy donors and platelets purified as previously described.²² 3×10⁷ platelets/ml were stimulated with 1nM thrombin for 15 min at 37°C. Subsequently, 10 U/ml hirudin was added to neutralize thrombin. Platelets were then removed from the suspension by centrifugation and the supernatant was used to measure FXIa activity.

Fibrin generation assay.

HUVECs were grown to confluence in 96-well plates and incubated for 2hrs with FXIa at 100 pM in a PBS/CaCl₂/MgCl₂ buffer with 10μM ZnCl₂. HUVECs were then incubated with a 10μg/ml blocking anti-tissue factor (TF) antibody before a solution of citrated PPP (33% final) in the presence of CTI (40μg/ml) was added and fibrin formation initiated with 8.3mM CaCl₂. Fibrin formation was measured as change in turbidity at 405nm. In selected experiments HUVECs were grown to confluence in 96-well plates and incubated with TNFα for 16hrs; fibrin generation of PPP in the presence of CTI (40μg/ml) or FXI-depleted plasma was quantified as change in turbidity at 405nm.

Immunoprecipitation and Western blotting.

HUVECs or BAECs were grown to confluence in 6-well plates and incubated with vehicle or FXIa (30nM) in serum free medium with 0.3% BSA for 2hrs at 4°C. Cell media were collected and cells were washed three times in a PBS/CaCl₂/MgCl₂ buffer followed by the addition of a lysis/immunoprecipitation (IP) buffer (10mM Tris/HCl, pH 7.4, 150mM NaCl, 2mM EDTA, 1% (v/v) Triton X-100). HUVECs or BAECs lysates were pre-cleared with Protein A/G Sepharose and then incubated with 2μg of an anti-FXI antibody or non-specific IgGs overnight at 4°C. Antibody-protein complexes were then captured with Protein A/G PLUS-Agarose beads (4hrs, 4°C) and washed three times in an IP buffer. FXI/FXIa precipitates were then eluted through the addition of 50μl of Laemmli sample buffer (Bio Rad, Hercules, CA) containing 200mM DTT. Protein samples were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membranes (PVDF), blotted with an anti-FXI/XIa light chain (LC) antibody or an anti-PAI-1 antibody and horseradish peroxidase-conjugated secondary antibodies. Protein was detected using ECL (Thermo Scientific).

Biotinylation of cell surface proteins.

HUVECs were incubated with 0.5mg/ml sulfo-NHS-SS-biotin for 30 min at 4°C. Cells were washed three times with ice-cold quenching buffer (50mM Glycine in PBS/CaCl₂/MgCl₂, pH 7.4). Biotinylated cells were lysed in lysis/IP buffer and precipitated using NeutrAvidin Agarose beads (4hrs, 4°C). Beads were washed three times in an IP buffer and the bound complexes were dissolved in 50μl of Laemmli’s sample buffer. The precipitates were probed using an anti-PAI-1 antibody or anti-PECAM-1 antibody.

Identification of potential FXIa serpin inhibitors.

HUVECs were grown to confluence in a 75cm² flask and incubated with vehicle or FXIa (30nM) in serum-free medium for 2hrs at 4°C. Cells were washed three times in a PBS/CaCl₂/MgCl₂ buffer follow by the addition of a lysis/IP buffer (10mM Tris/HCl, pH 7.4, 150mM NaCl, 2mM EDTA, 1% (v/v) Triton X-100). HUVECs lysates were incubated with an anti-FXI antibody at 4°C. Antibody-protein complexes were then captured with Protein A/G PLUS-Agarose beads (4hrs, 4°C) and washed three times in an IP buffer. FXI/FXIa precipitates were then eluted through the addition of 50μl of a Laemmli sample buffer containing 200mM DTT. Two bands of ~80 kDa and ~70 kDa were detected by Coomassie blue. Gel bands from each sample were cut, sliced into small pieces, and in-gel digested with trypsin in 0.01% ProteaseMAX (Promega, Madison, WI) at 50°C for 3hr. Peptides were extracted using the Promega protocol and samples were filtered with Millipore 0.22μm filters, dried, dissolved in 20μl of 5%FA and 20μl/sample was injected into the QExactive HF.

LC-MS/MS Analysis.

Protein digests were separated using liquid chromatography with a NanoAcquity UPLC system (Waters) and analyzed using a QExactive HF Mass Spectrometer (Thermo Fisher). Survey mass spectra were acquired over m/z 375−1400 at 120,000 resolution (m/z 200) and data-dependent acquisition selected the top 10 most abundant precursor ions for tandem mass spectrometry. Data were analyzed using COMET (version 2015.01, revision 1)²³ to search MS2 Spectra against an October 2016 version of the sprot_human_both.fasta (40598 entries) FASTA protein database, with concatenated sequence-reversed entries to estimate error thresholds and potential contaminant sequences as previously described.²⁴

Fluorescence microscopy.

HUVECs were grown to confluence on glass coverslips in 24-well plates. Cells were fixed for 1hr at 4°C in PBS containing 4% paraformaldehyde. Following fixation, HUVECs were incubated in a blocking solution (PBS with 3% BSA and 2% FBS) for 2hr, and stained with a mouse anti-FXI antibody (10μg/ml) overnight at 4°C in PBS. Alexa Fluor 488 anti-mouse IgG (4μg/ml) and Hoechst 33342 (10μg/ml) were added in PBS for 2hrs. For the detection of internalized FXIa, following 4% PFA fixation HUVECs were blocked and stained with an anti-FXI antibody. An unlabeled anti-mouse IgG-k (2μg/ml) was added in PBS to block any surface FXIa signal. HUVECs were then permeabilized with 0.2% Triton X-100 and blocked for 2hrs with a blocking solution. HUVECs were stained with a mouse anti-FXI antibody together with 1μg/ml rabbit anti-Rab7, 5μg/ml rabbit anti-EEA1, or 4μg/ml rabbit anti-LAMP1 antibodies, followed by washing and addition of Alexa Fluor 488 anti-mouse IgG, Alexa Fluor 546 anti-rabbit IgG and Hoechst 33342 in PBS. Coverslips were mounted with Fluoromount G on glass slides. HUVECs were imaged using a Zeiss ×63 oil immersion 1.4 NA lens on a Zeiss Axio Imager M2 microscope. For two-channel colocalization analyses, immunofluorescence overlap was quantified using an in-house Matlab application to calculate the Pearson’s coefficient.

Baboon model of S.aureus sepsis.

The pilot study was approved by the Interfaculty Animal Ethics Committee of the University of the Free State (UFS), Bloemfontein, South Africa and the Institutional Care and Use Committee of Oklahoma Medical Research Foundation. Healthy Papio ursinus baboons (8–20.2 kg body weight) with a leukocyte count less than 13000/μL and hemoglobin greater than 10 g/dL were included in the study. Staphylococcus aureus (S.aureus) subsp. aureus Rosenbach (ATCC® 12598™) was from American Type Culture Collection (Manassas, VA). Heat-inactivated S.aureus was prepared by heating an exponential-phase culture for 60min at 70°C. Baboons were challenged with 3×10¹⁰ bacteria/kg (lethal dose) by intravenous infusion over a 2hr period. Plasma samples were collected before challenge (T0) and at 2, 4, 6, 8 and 24hrs post-challenge.

FXIa-PAI-1 ELISA.

FXIa-PAI-1 complexes in EDTA-plasma were measured by sandwich ELISA. 96-well plate were coated with rabbit anti-human PAI-1 (1μg/ml in PBS) overnight at 4°C. After each step, the plate was washed 3× using 300μL wash buffer (PBS containing 0.1% Tween-20). Next, the plate was blocked with 1% BSA in PBS containing 0.1% Tween-20 for 90 min. Plasma samples were diluted in PBS containing 1% BSA, 0.1% Tween-20, 5mM benzamidine hydrochloride and 5mM EDTA and incubated for 60 min at RT. Then, the plate was incubated with a biotin-conjugated anti-FXI antibody (100ng/ml) for 60 min at RT, followed by streptavidin-peroxidase conjugate (250ng/ml). Orthophenylenediamine was used as a peroxidase substrate and the reaction was stopped by adding sulfuric acid. Absorbance was recorded at 492nm. The FXIa-PAI-1 complex concentration was calculated using a FXIa-PAI-1 complex standard made by incubating human FXIa (1μM) with human recombinant PAI-1 (4μM) at 37°C for 60 min. This standard was considered 1μM FXIa-PAI-1 complex. In the blank well, all steps were the same except we added sample diluent instead of plasma. To verify the specificity of assay plasma samples were added in uncoated wells (without PAI-1 antibody coating) as a control. In parallel, FXIa-AT (70pM), FXIa-C1INH (70pM) or tPA-PAI-1 (80nM) complexes were added to the ELISA plate and the absorbance was compared to the FXIa-PAI-1 complex standard (62.5pM). Our data show that these complexes did not interfere with the detection of FXIa-PAI-1 complexes (Figure III in the online-only Data Supplement). Further, plasma (T0, before E. coli infusion at a 1:25 dilution) was spiked with recombinant PAI-1 (0.1μg/ml) to test if increased PAI-1 levels enhanced the signal. Our results show that addition of exogenous PAI-1 did not increase the signal; the slight decrease likely reflects a competitive effect of exogenously-added free PAI1 with the endogenous FXIa-PAI-1 complexes (Figure IV in the online-only Data Supplement).

Statistical analysis.

Data are shown as means ± SEM.

All statistical analyses were performed using R (R foundation for statistical computing, version 3.6.0). Assessment of parametric model assumptions was carried out using Levene’s test of equal variances, Shapiro’s test of normal distribution, and visual inspection of model diagnostic normal quantile-quantile and residual versus fitted value plots. In order to better approximate a normal distribution, the data for figure 1A was natural log transformed prior to analysis, and the data for figures 1C, 1E, 3A, 3C, 3D, 3F, and 6 were rank-transformed prior to analysis. Experiments comparing groups against concentration gradients of FXIa, rPAI-1, or kallikrein were analyzed using two-way ANOVAs calculating coefficients for gradient concentration and group. Gradient coefficients were statistically significant for every experiment described here and are not reported. Analyses of multiple groups were performed using one or two-way ANOVAs as appropriate. Post-hoc Tukey’s tests were performed as necessary, except for figure 3H which utilized a Dunnett’s test. Differences between two groups were calculated using unpaired t-tests, except in figure 5C where unbalanced data necessitated the use of the nonparametric Mann-Whitney U tests. Figure 3D was analyzed with separate Spearman’s correlations. Data from the pilot experiment using non-human primates was not evaluated statistically in order to avoid overpowering the analysis of data from a small animal cohort. Unless otherwise indicated, experiments had an N of 3 per group, which we acknowledge was underpowered in some cases.

Figure 1. Detection of FXIa or kallikrein generation by FXIIa on the surface of HUVECs.

(A) HK (100 nM) and FXI (100 nM) or (B) PK (100 nM) and HK (100 nM) were incubated with FXIIa (1 nM) for 1 hr in the presence (middle panel) or absence (left panel) of HUVECs. Reactions were stopped with CTI (40 μg/ml). FXIa and kallikrein activity was determined by adding S-2366 (0.8 mM) or S-2302 (0.6 mM), respectively; the rate of substrate hydrolysis was measured at 405 nm. Empty wells or HUVECs were incubated with increasing concentrations of (A, right panel) FXIa or (B, right panel) kallikrein for 2 hrs at 37°C FXIa and kallikrein activity was determined by adding S-2366 (0.8 mM) or S2302 (0.6 mM), respectively and the rate of substrate hydrolysis was measured at 405 nm. (C) HUVECs were incubated for 16 hrs in the absence (□) or presence of TNFα (0.5 ng/ml) (○) or incubated for 4 hrs with elastase (50 nM) (△). Subsequently, cells were incubated with increasing concentrations of FXIa. FXIa activity was determined by adding S-2366 (0.6 mM) and the rate of substrate hydrolysis was measured at 405 nm. (D) HUVECs were incubated for 3 hrs in the absence or presence of heparinase I, II and III (1 U/ml) followed by cell surface detection of heparan sulfate using an anti-10E4 epitope antibody. (E) HUVECs were incubated for 3 hrs in the absence (□) or presence (○) of heparinase I, II and III (1 U/ml). Subsequently, cells were incubated with increasing concentrations of FXIa. FXIa activity was determined by adding S-2366. (F) HUVECs were incubated for 4 hrs in the absence (□) or presence of thrombin (20 nM) (○). Subsequently, cells were incubated with increasing concentrations of FXIa. FXIa activity was determined by adding S-2366. Data are mean ± SD (n = 3). * Indicates between-groups differences with P < 0.05, ** with P < 0.01.

Figure 3. Detection of FXIa activity in the presence of a blocking anti-PAI-1 antibody.

(A) Empty wells (●, ○) or HUVECs (■, □) were incubated with FXIa in the absence (□, ○) or presence (■, ●) of a blocking anti-PAI-1 antibody (20 μg/ml) for 2 hrs. at 37°C. FXIa activity was determined by adding S-2366. (B) HUVECs were incubated for 16 hrs with TNFα (0.5 ng/ml). Subsequently, cells were incubated with FXIa (500 pM) in the absence (black bars) or presence (white bars) of an anti-PAI-1 antibody (20 μg/ml) for 2 hrs at 37°C. FXIa activity was determined by adding S-2366. (C) Empty wells (○) or HUVECs (□, ●) were incubated with FXIa in the absence (□) or presence (●) of t-PA (30 nM) for 2 hrs at 37°C. FXIa activity was determined by adding S-2366 (D) Increasing concentrations of rPAI-1 were incubated with 5 nM FXIa, kallikrein or t-PA at 37°C for 30 min and activity was determined by adding S-2366, S2302 or S2288 respectively. (E) PAI-1 detection by western blot in ECs or platelets supernatant. (F) FXIa (150 pM) were incubated with vehicle or supernatant from activated platelets in the presence (white bars) or absence (black bars) of an anti-PAI-1 antibody. (G) HUVECs were incubated with 0.1 nM FXIa in the presence or absence of a blocking anti-PAI-1 antibody (20 μg/ml) for 2 hrs at 37°C. Fibrin generation was determined in recalcified PPP in the presence of CTI and a blocking anti-TF antibody. (H) HUVECs were incubated with TNFα. Fibrin generation was determine in PPP in the presence of CTI or FXI depleted plasma (FXI-dep) in the presence or absence of an anti-TF, an anti-FXI or an anti-PAI-1 antibody. Data are mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 between groups (A, B, C) or with respect to vehicle (F, G, H).

Figure 6. Detection of FXIa-PAI-1 complexes in vivo.

Time course change of (A) FXIa-PAI-1 complexes, (B) PAI-1 and (C) FXI levels in baboons challenged intravenous 2 h (T0–T120 min) infusion with a lethal dose (LD100) (1–2×10¹⁰ colony forming units/kg) of Staphylococcus (S.) aureus. Data are mean ± SEM (n = 3).

Figure 5. Immunofluorescence of FXIa trafficking in endothelial cells.

(A) HUVECs were incubated with vehicle or FXIa (30 nM) at 4°C for 2 hrs and then were transferred to 37°C for 0, 15, or 30 min. Cells were washed and then fixed and immunostained for FXIa and visualized by fluorescence microscopy. (B) HUVECs were incubated with vehicle or FXIa (30 nM) at 4°C for 2 hrs. and then were transferred to 37°C for 0, 15, 30, 60 or 120 min. Cells were washed, fixed and incubated with an anti-FXIa antibody, followed by incubation with an unlabeled goat-anti-mouse antibody to block FXIa bound to the surface before permeabilization. After permeabilization, cells were immunostained for FXIa to detect internalized FXIa, which was visualized by fluorescence microscopy. Selected experiments were performed in the presence of PPACK (0.1 mM) or using FXI zymogen. (C) HUVECs were incubated with vehicle or FXIa (30 nM) at 4°C for 2 hrs. and then were transferred to 37°C for 1 hr. Cells were washed, fixed and incubated with an anti-FXIa antibody, followed by incubation with an unlabeled goat-anti-mouse antibody to block FXIa bound to the surface before permeabilization. After permeabilization, cells were immunostained for FXIa to detect internalized FXIa (green) as well as the endosomal and lysosomal markers EEA1, Rab 7, LAMP-1 or Rab 11 (red) which were visualized by fluorescence microscopy. Data are mean ± SD (n = 3). Pearson’s correlation reveals significantly increased colocalization of FXIa with EEA1, Rab 7 and LAMP-1. * indicates P < 0.05 with respect to vehicle.

RESULTS

HUVECs inhibit FXIa but not kallikrein or FXIIa activity.

Activation of FXI or PK by FXIIa was determined in the presence or absence of HUVECs. As shown in Figure 1A, FXIIa was able to generate FXIa activity in the absence of ECs. In contrast, the capacity of FXIIa to generate FXIa activity was reduced by the presence of HUVECs. FXIIa generated robust kallikrein activity beyond the innate ability of HUVECs to generate kallikrein activity alone (Figure 1B). These results suggest that HUVECs either inhibit the activation of FXI by FXIIa or directly inhibit FXIa activity following generation by FXIIa. We next designed experiments to validate whether HUVECs directly inhibit FXIa activity by incubating increasing concentrations of FXIa on HUVECs and measuring the rate of cleavage of a chromogenic substrate. As shown in Figure 1A–B, HUVECs significantly reduced the enzymatic activity of purified FXIa but not kallikrein. The presence of HK (100 nM) did not affect the capacity of HUVECs to block the activity of FXIa (Figure I in the online-only Data Supplement).

We next studied whether activation of ECs changed their ability to inhibit FXIa, as it has been shown that TNFα stimulation of ECs upregulates the expression of serpin inhibitors including antithrombin and C1-inhibitor,^11,25 which are known to inhibit FXIa in plasma. Our results show that FXIa activity was further suppressed following stimulation of HUVECs with TNFα; in contrast, treatment of HUVECs with elastase,^26,27 which cleaves and inactivates serine protease inhibitors, reversed the ability of HUVECs to inhibit FXIa activity (Figure 1C). Specific removal of anticoagulant proteoglycans heparan sulfate with heparinase I, II and III had a significant but minimal effect on the inhibitory phenotype of HUVECs toward FXIa, while stimulation of HUVECs with thrombin had no effect whatsoever (Figure 1D–F). The supernatant from quiescent or TNFα-stimulated HUVECs did not affect the cleavage of the chromogenic substrate by FXIa (Figure I in the online-only Data Supplement).

Identification of endothelial cell plasminogen activator inhibitor-1 (PAI-1) as a serpin that forms a complex with FXIa.

We next studied whether FXIa formed a complex with specific serpin inhibitor(s) present on the EC surface. FXIa was incubated with HUVECs at 4°C for 2 hours to permit binding yet not allow internalization. FXIa present in the cell media or the cell lysate after washing was immunoprecipitated and analyzed by western blot under reducing conditions using an anti-FXI antibody which detects the catalytic domain-containing light-chain (LC) of FXI(a). Parallel experiments were performed with zymogen FXI.

As expected, the LC of FXIa (~30 kDa) and FXI (~ 80 kDa) were detected in the cell media following incubation of FXIa or FXI with ECs, respectively (Figure 2A). As expected the IgG used for FXI/FXIa immunoprecipitation was detected in the western blot and served as a loading control. When the cell lysate of ECs that had been incubated with FXI was probed with the LC-specific anti-FXI antibody, we again observed a band at 80 kDa (Figure 2B). Interestingly, when the cell lysate of ECs incubated with FXIa was probed with the anti-LC FXI antibody, we observed a band at 30 kDa (FXIa LC) and two bands of higher molecular weight ~80 kDa and ~70 kDa, presumably due to FXIa being in complex with an inhibitor(s) (Figure 2B). These bands were not observed when FXIa were incubated with ECs in the presence of the serine protease inhibitor PPACK; under those conditions only a single ~30 kDa band was detected by the LC anti-FXI mAb. These results suggest that the catalytic domain of the LC of FXIa is forming a complex with at least two inhibitors present in or on ECs.

Figure 2. Identification of a FXIa-endothelial inhibitor complex.

HUVECs were incubated with FXI, FXIa (30 nM) or vehicle (−) at 4°C for 2 hrs in the presence or absence of PPACK (0.1 mM). Supernatant (A) or cells lysates after washing (B) were incubated with an anti-FXI/FXIa mAb for 16 hrs at 4°C. Protein A/G Plus Agarose beads were added to the samples and incubated for 4 hrs at 4°C. Beads were washed followed by western blotting with an anti-FXI light-chain (LC) antibody. (C) HUVECs were incubated with FXI, FXIa (30 nM) or vehicle (−) at 4°C for 2 hrs. Cell were washed and incubated with PPACK (0.1 mM) for 30 min, lysed in the presence of PPACK followed by immunoprecipitation and western blotting with an anti-FXI LC antibody or (D) with an anti-PAI-1 antibody. (E) HUVECs were incubated with FXI, FXIa (30 nM) or vehicle (−) at 4°C for 2 hrs; cells were washed and lysed followed by western blotting with an anti-PAI-1 antibody. (F) BAECs were incubated with FXI, FXIa (30 nM) or vehicle (−) at 4°C for 2 hrs. Cell were washed and incubated with PPACK (0.1 mM) for 30 min, lysed in the presence of PPACK followed by immunoprecipitation and western blotting with an anti-FXI LC antibody. Representative images and blots for at least n=3

Mass spectrometry analysis of the ~80 kDa and ~70 kDa bands formed by incubating FXIa on ECs identified plasminogen activator inhibitor 1 (PAI-1) and serpin B6, which have molecular weights of ~50 kDa and ~40 kDa, respectively, as candidates for the serpins that form a complex with FXIa. Serpin B6 is a cytosolic protein which is not secreted by ECs²⁸, suggesting that this complex formed as an artifact following lysis of ECs. Indeed, the presence of PPACK during the lysis step eliminated the ~70 kDa band corresponding to FXIa LC-serpin B6 complex, while the ~80 kDa band corresponding to FXIa LC-PAI-1 complex remained (Figure 2C). The identity of PAI-1 as a FXIa-inhibiting serpin on ECs was confirmed by the presence of a single ~80kDa band detected by an anti-PA1–1 mAb following immunoprecipitation of FXI(a) from the cell lysate (Figure 2D). Conversely, ~50 kDa and ~80 kDa bands were detected following blotting of the whole cell lysate with the anti-PAI-1 mAb (Figure 2E). Equivalent results were obtained when we incubated FXIa with BAECs (Figure 2F). It should be noted that despite HUVECs expressing C1 inh and a small amount of AT under resting and serum-free medium conditions (Figure II in the online-only Data Supplement), we were not able to detect FXIa-AT or FXIa-C1 inh complexes under the conditions tested herein. These results indicate that FXIa predominantly forms a complex with the serpin PAI-1 on the surface of cultured ECs.

PAI-1 present on endothelial cells inhibits the procoagulant activity of FXIa.

Our data show that ECs inhibit the activity of FXIa, perhaps due to inactivation of FXIa by the serpin PAI-1 on the EC surface. To test this hypothesis, we measured the activity of FXIa in the presence of a blocking antibody against PAI-1. We observed that the inhibitory action of HUVECs towards FXIa was reversed when PAI-1 was blocked (Figure 3A), even when ECs were stimulated with TNFα (Figure 3B). Likewise, the addition of excess molar tissue-plasminogen activator (tPA; Figure 3C) eliminated the ability of HUVECs to inactivate FXIa. The anti-PAI-1 Ab did not exhibit any inhibitory activity toward FXIa alone, while tPA alone did not cleave the chromogenic substrate.

Biochemical studies in purified systems have shown that PAI-1 is able to inhibit FXIa with a second-order rate constant of inhibition of 2.1×10⁵ M⁻¹ s⁻¹.⁹ We tested the effect of recombinant PAI-1 (rPAI-1) on the amidolytic activity of FXIa, tPA and kallikrein. We observed that rPAI-1 inhibited FXIa in a concentration-dependent manner; a similar result was observed for tPA, while rPAI-1 was able to only marginally inhibit kallikrein (Figure 3D).

Platelets are able to release PAI-1 after activation.²⁹ In order to determine whether platelet-derived PAI-1 was capable of inhibiting FXIa activity, we activated platelets with thrombin to induce degranulation and PAI-1 release (Figure 3E). We observed that while the releasate from activated platelets inhibited FXIa activity, the presence of an anti-PAI-1 Ab did not block the inhibitory effect of the platelet releasate on FXIa activity (Figure 3F). These results are in line with the notion that the ability of platelet releasate to inactivate FXIa is largely dependent upon the potent FXIa inhibitor protease nexin 2 rather than PAI-1.^29,30

We next tested whether EC PAI-1 played a functional role in limiting the procoagulant activity of FXIa to generate fibrin in plasma. For these experiments, FXIIa activity was blocked with CTI, while potential TF activity on ECs was blocked with a function blocking anti-TF Ab. Under these conditions, HUVECs alone were not able to promote fibrin generation in plasma (Figure 3G). Fibrin formation on ECs was induced by the addition of 100 pM FXIa after a lag time of ~1000 sec; this lag time was significantly reduced by ~250 sec when EC PAI-1 was blocked with an antibody (Figure 3G). The anti-PAI-1 Ab did not exhibit any inhibitory activity toward FXIa in the absence of HUVECs. We next examined whether endothelial PAI-1 blocked the procoagulant activity of FXIa generated by thrombin. HUVECs were preincubated with TNFα to induce TF expression, and fibrin generation was measured. In all experiments, plasma was pretreated with CTI in order to block the activity of FXIIa. The presence of a blocking anti-TF mAb and an antibody that blocks the activation of FIX by FXIa (1A6) delayed fibrin formation (Figure 3H), indicating that indeed under these conditions fibrin generation was dependent of the generation of FXIa downstream of TF. The addition of a blocking anti-PAI-1 Ab decreased the initiation time for fibrin generation from ~1000sec to ~700sec (Figure 3H). Importantly, the effect of blocking PAI-1 was lost when FXI-depleted plasma was used (Figure 3H). These results indicate that PAI-1 plays a functional role in sequestering the procoagulant activity of FXIa at the blood-endothelium interface.

The FXIa-PAI-1 complex is released into the ECs culture media.

We next examined whether FXIa in complex with PAI-1 was released from the surface of ECs into the media. HUVECs were incubated with FXIa at 4°C for 2 hrs to induce binding. Cells were then washed and transferred to 37°C for up to an additional 2 hrs. At select times, the cell media was collected and immunoprecipitated for FXIa. Samples were analyzed with the anti-FXIa LC antibody or with an anti-PAI-1 antibody. As shown in Figure 4A–B, a ~80kDa band indicative of a FXIa-PAI-1 complex began to appear in the cell media after only 5 min of incubation at 37°C, and was detected by both the anti-FXI and anti-PAI-1 Abs. Conversely, the ~80 kDa band corresponding to the FXIa-PAI-1 complex was initially present in the cell lysate yet decreased with time and was undetectable by 120 min (Figure 4C–D), suggesting that once formed, the FXIa-PAI-1 complex is released by ECs into the media.

Figure 4. Detection of the FXIa-PAI-1 complex.

HUVECs were incubated with vehicle (−) or FXIa (30 nM) at 4°C for 2 hrs. Cells after washing were incubated at 37°C for 5, 15, 30, 60 or 120 min. Supernatant (A and B) or cell lysates (C and D) were incubated with an anti-FXI/FXIa antibody for 16 hrs at 4°C. Protein A/G Plus Agarose beads were added to the samples and incubated for 4 hrs at 4°C. Beads were washed followed by western blotting with an anti-FXI LC antibody (A and C) or anti-PAI-1 antibody (B, D). HUVEC cell surfaces were biotinylated and incubated with vehicle (−) or FXIa (30 nM) in the absence (E) or presence of PPACK (F) at 37°C for 5, 15, 30, 60 or 120 min. Cell lysates were precipitated with NeutrAvidin Agarose beads. The precipitates were probed with anti-PAI-1 or anti-PECAM-1 antibodies. (G) HUVECs were incubated with vehicle (−) or FXII (100 nM), PK (100 nM), HK (100 nM) and FXI (30 nM) for 1 hrs at 37°C in the presence or absence of CTI (40 μg/ml). Supernatants were inmunoprecipitated with an anti-FXI/FXIa antibody followed by western blotting with an anti-PAI-1 antibody. (H) HUVEC cell surfaces were biotinylated and incubated with vehicle (−) or FXII (100 nM), PK (100 nM), HK (100 nM) and FXI (30 nM) for 1 hrs at 37°C in the presence or absence of CTI (40 μg/ml). Cell lysates were precipitated with NeutrAvidin Agarose beads. The precipitates were probed with anti-PAI-1 or anti-PECAM-1 antibodies. Representative images and blots for at least n=3.

We next determined whether formation of FXIa-PAI-1 complexes affected the expression level of PAI-1 on the EC surface. Cell surface proteins were biotinylated and cell lysates were precipitated using NeutrAvidin Agarose beads. Using a biotinylation assay we found that the surface expression of PAI-1 on ECs decreased in a time-dependent manner after incubation with FXIa (Figure 4E); the addition of the serine protease inhibitor PPACK eliminated this effect (Figure 4F).

Lastly, we studied whether EC PAI-1 would sequester and inactivate FXIa that had been formed by the contact activation pathway. Indeed, FXIa that had been formed as a result of incubating FXII, PK, HK and FXI together was found to be in complex with PAI-1 in the cell media (Figure 4G). Moreover, the expression level of PAI-1 on the EC surface was diminished when FXIa was generated by the contact pathway (Figure 4H). These effects were completely blocked by the FXIIa inhibitor CTI, suggesting that once FXI is activated by FXIIa, FXIa is inhibited by PAI-1 present on ECs and released as a complex to the cell media.

FXIa is internalized and degraded by endothelial cells.

The binding of FXIa to ECs was analyzed by immunofluorescence. HUVECs were incubated with FXIa at 4°C for 2 hrs to induce binding before the cells were transferred to 37°C for increasing times. As shown in Figure 5A, robust binding of FXIa to the surface of ECs was detected at 4°C. Of note, this binding was lost upon transfer of ECs to 37°C. We next used an unlabeled anti-FXI Ab to bind any EC-surface bound FXIa prior to permeabilization of ECs and staining with a labeled anti-FXI Ab to probe for intracellular FXIa. Using this strategy we were able to detect an increase in intracellular FXIa over time following transfer of ECs to 37°C, mainly around the perinuclear region of the cells (Figure 5B). The level of intracellular FXIa decreased with time, indicative of a degradation process. FXIa binding and internalization was eliminated by PPACK. To determine whether intracellular FXIa was localized in endosomes, we next quantified the subcellular distribution of FXIa in ECs by dual immunofluorescence microscopy. HUVECs were immunostained with an anti-FXIa antibody and antibodies specific to select endosome-specific marker proteins. We observed a fraction of FXIa localized with endosomal markers EEA1 and Rab 7 and the lysosomal protein marker, LAMP1, 30 min after transferring the ECs to 37°C (Figure 5C). We did not observe any colocalization of FXIa with the endosome protein marker Rab 11 (Figure 5C), suggesting that FXIa is not recycled to the cell surface, but is degraded in the lysosomes upon internalization by ECs.

Detection of FXIa-PAI-1 complex in vivo.

As a proof-of-concept pilot experiment, we next examined whether FXIa-PAI-1 complexes could be detected in vivo in the circulation of baboon following a bacterial challenge to induce a prothrombotic phenotype (Figure 6). In this model, baboons were challenged with a lethal dose of heat-inactivated S.aureus, which we have recently shown induces activation of the contact activation pathway of coagulation.³¹ Within the first 2–8 hours following the bacterial challenge, a 10-fold increase in FXIa-PAI-1 complex level was detected (Figure 6A). The level of FXIa-PAI-1 complexes returned to baseline after 24 hrs. This pilot experiment provides the first proof-of-concept data that FXIa-PAI-1 complexes are detectable in the bloodstream following a prothrombotic or procoagulant challenge.

DISCUSSION

Coagulation factor XI may play a significant role in thrombosis, with FXI antigen levels having been shown to be an independent risk factor for deep vein thrombosis, ischemic stroke, and myocardial infarction.^32–34 While some FXI deficient patients may experience mild trauma-induced bleeding symptoms, they are typically restricted to particular vascular beds, suggesting that the local endothelial microenvironment plays a role in regulating FXI activation or activity. We designed the current study to determine whether endothelial cells interact with and inactivate FXIa.

The procoagulant activity of FXIa is not restricted to the generation of thrombin through activation of FIX. FXIa can activate FX, FV and FVIII,^35–37 and proteolytically inactivate TFPI¹⁸ in a manner enhanced by short chain polyphosphates secreted by platelets.^38,39 Together, these reactions enhance coagulation, at least in vitro. FXIa can also cleave ADAMTS13,⁴⁰ inducing platelet adhesion and aggregation on ECs under flow conditions in whole blood in vitro. Thus, to maintain endovascular thromboprotection while ensuring extravascular hemostasis under physiological conditions, we posit that the endothelium employs several inhibitory mechanisms to selectively block the procoagulant activity of FXIa. The present study suggests that inhibition of FXIa by the endothelium may induce the clearance of thrombogenic FXIa from the circulation. On the surface of endothelial cells, uPAR, CK1 and gC1q-R are responsible for the binding and activation of members of the kallikrein-kinin system (FXII/HK/PK).⁷ While platelets have been shown to catalyze FIX activation by FXIa, the same has not been observed for endothelial cells, leading to the conclusion that platelets have more binding sites for FXI/FXIa than the endothelium.^13–15,17 Whether FXI binds to the endothelium at all is a controversial topic in itself. Perhaps missing in prior hypotheses and studies addressing this question was the possibility that the endothelium produces a protease inhibitor that selectively blocks the intrinsic pathway of coagulation yet not the kallikrein-kinin system.

Here we show that while the endothelium catalyzes the activation of PK by FXIIa, the serpin PAI-1 selectively binds and inhibits FXIa. Upon inhibition, FXIa is then destined for one of two fates: internalization and eventual degradation by lysosomes or release as a complex with PAI-1. Traditionally thought of as the primary inhibitor of tissue-type and urokinase-type plasminogen activator, PAI-1 is known to inhibit several serine proteases involved in blood coagulation, including thrombin.⁴¹ Yet, PAI-1 is not a universal serine protease inhibitor, as suggested by the inability of PAI-1 to inhibit FXa or FIXa even in the presence of PAI-1 cofactors such as heparin or vitronectin.⁴² Berrettini et al were the first to show that PAI-1 binds and neutralizes the amidolytic activity of FXIa in a purified system. This adds PAI-1 to a list of serpin family members that act as physiological inhibitors of FXIa that includes C1 inh, protease nexin 1, α1-antitrypsin, α2-antiplasmin, and AT. While C1 inhibitor is the major FXIa inhibitor in plasma, the second order rate constant of C1 inhibitor for FXIa of 1.8 × 10³ M⁻¹ s⁻¹ is more than 3 orders of magnitude lower than the calculated rate for the inhibition of FXIa by PAI-1 (2.1 × 10⁵ M⁻¹ s⁻¹),⁹ while the rate constant of antithrombin for FXIa is only 4.4 × 10³ M⁻¹ s⁻¹ even in the presence of saturating amounts of heparin.⁸

The plasma concentration of PAI-1 is at least 3 to 4 orders of magnitude lower than the circulating concentrations of antithrombin and C1 inhibitor, respectively; yet, the endothelial cell surface expresses all three of these serpins along with protease nexin 1.^11,12,43 Herein PAI-1 and serpin B6 were the only serpins we found in complex with FXIa, with the later a result of lysis of ECs to release serpin B6 into solution to inactivate FXIa. It has been suggested that the kallikrein-kinin system and the intrinsic pathway of coagulation are differentially regulated and can be activated independently of each other. For instance, activation of both FXII and PK, but not FXI, has been observed in the plasma of systemic amyloidosis patients.⁴⁴ Similarly, mast cell-heparin has been shown to activate FXII, which initiates bradykinin formation. Yet, FXI activity was not observed under these conditions,⁴⁵ suggesting that the kallikrein-kinin system and the intrinsic pathway of coagulation, which are both activated by FXIIa, can be regulated by distinct mechanisms. Perhaps the sequestration of FXIa activity by endothelial PAI-1 provides a mechanistic explanation for these in vivo observations.

Increased thrombin generation has been observed in plasma as a consequence of thrombolytic therapy with t-PA or u-PA,^46,47 implicating an involvement of the fibrinolytic system in inciting coagulation. We observed that the incubation of endothelial cells with t-PA abrogated the capacity of PAI-1 to inhibit the activity of FXIa, suggesting that thrombolytic therapy with t-PA may increase thrombin generation by decreasing the inhibition of FXIa by PAI-1. Further, analogous to our observation that t-PA enhanced the activity of FXIa, we found that incubation of HUVECs with neutrophil elastase enhanced the amidolytic activity of FXIa. This makes sense as neutrophil elastase is known to cleave and inactivate PAI-1.⁴⁸ Neutrophils are among the first cells to accumulate at sites of thrombus formation and have been reported to promote venous thrombosis in a FXII/FXI-dependent manner.⁴⁹ Our findings implicating neutrophil elastase in the removal of the inhibitory pathway of FXIa inactivation by PAI-1 may provide a possible mechanisms by which neutrophils promote the intrinsic pathway of coagulation.

The physiologic significance of PAI-1 inhibition of FXIa is at present unknown. PAI-1 deficiencies are rare yet surprisingly result in a mild to moderate bleeding diathesis, with bleeding episodes occurring after trauma, surgery or dental extraction. Many individuals with low PAI-1 activity are asymptomatic.⁵⁰ Expression of PAI-1 is elevated in a number of pathologic situations including cardiovascular diseases. PAI-1-deficient mice breed and grow normally, yet their response to disease includes both worsening and preventing various pathologies.⁵¹ It is also possible that increased expression of endothelial PAI-1 at sites of vascular injury could be protective, particularly if PAI-1 downregulates FXIa-mediated coagulation. In support of this notion, in a non-human primate sepsis model the treatment of baboons with an anti-PAI-1 antibody resulted in massive fibrin deposition in the lung, decreasing the amount of lung-associated TFPI.⁵² Also, as a proof-of-concept we were able to detect for the first time FXIa-PAI-1 complexes in the circulation following a procoagulant challenge in a baboon model of S.aureus sepsis. In conclusion our study demonstrates that the endothelium inactivates FXIa activity in a PAI-1-dependent manner without affecting the kallikrein-kinin system. This study supports the concept that endothelium is anticoagulant and also suggests that PAI-1 may play a role in the regulation of the intrinsic pathway of coagulation.

HIGHLIGHTS:

• Endothelial cells support the generation of kallikrein activity by FXIIa but specifically block the generation of FXIa activity by FXIIa.

• Endothelial cells express PAI-1, which forms a complex with FXIa, blocking the procoagulant activity of FXIa and inducing the clearance and degradation of FXIa.

• Endothelial cells promote the activation of the kallikrein-kinin system, while inhibiting the activation of the intrinsic pathway of thrombin generation.

• FXIa-PAI-1 complexes are detected in the circulation of baboons challenged with heat-inactivated S. aureus.

ABBREVIATIONS

FXI/FXIa

zymogen/activated coagulation factor XI

FXII/FXIIa

zymogen/activated coagulation factor XII

HK

high molecular weight kininogen

PK

prekallikrein

PAI-1

plasminogen activator inhibitor-1

AT

antithrombin

C1 inh

C1 esterase inhibitor

CTI

corn trypsin inhibitor, inhibits FXIIa

PPACK

Phe-Pro-Arg chloromethylketone

t-PA

tissue plasminogen activator

HUVECs

human umbilical vein endothelial cells

BAECs

Primary baboon arterial endothelial cells

S. aureus

Staphylococcus aureus

PPP

human platelet-poor plasma

LC

light-chain

HC

heavy-chain

TF

tissue factor

NO

nitric oxide

TNFα

tumor necrosis factor-α

uPAR

urokinase receptor

CK1

cytokeratin 1

TNFα

Tumor necrosis factor-α