Research ArticleHypertension

Platelet-localized FXI promotes a vascular coagulation-inflammatory circuit in arterial hypertension

See allHide authors and affiliations

Science Translational Medicine  01 Feb 2017:
Vol. 9, Issue 375, eaah4923
DOI: 10.1126/scitranslmed.aah4923

Spotlight on factor XI

Hypertension, cardiovascular disease, and vascular inflammation are inextricably linked, often co-occurring. Kossmann et al. have now discovered a regulatory pathway linking these pathologies that could be inhibited to allow the control of treatment-resistant high blood pressure. In rats and mice with hypertension, the authors found that vascular disease is driven by an overactive thrombin-driven factor XI feedback loop on platelets. Inhibition of this feedback loop with an antisense molecule against factor XI reduced both the vascular pathology and hypertension. The authors show that this factor XI–dependent feedback loop also operates in patients with uncontrolled hypertension, raising the possibility that factor XI inhibition may prove a useful addition to our armamentarium for treating high blood pressure.


Multicellular interactions of platelets, leukocytes, and the blood vessel wall support coagulation and precipitate arterial and venous thrombosis. High levels of angiotensin II cause arterial hypertension by a complex vascular inflammatory pathway that requires leukocyte recruitment and reactive oxygen species production and is followed by vascular dysfunction. We delineate a previously undescribed, proinflammatory coagulation-vascular circuit that is a major regulator of vascular tone, blood pressure, and endothelial function. In mice with angiotensin II–induced hypertension, tissue factor was up-regulated, as was thrombin-dependent endothelial cell vascular cellular adhesion molecule 1 expression and integrin αMβ2– and platelet-dependent leukocyte adhesion to arterial vessels. The resulting vascular inflammation and dysfunction was mediated by activation of thrombin-driven factor XI (FXI) feedback, independent of factor XII. The FXI receptor glycoprotein Ibα on platelets was required for this thrombin feedback activation in angiotensin II–infused mice. Inhibition of FXI synthesis with an antisense oligonucleotide was sufficient to prevent thrombin propagation on platelets, vascular leukocyte infiltration, angiotensin II–induced endothelial dysfunction, and arterial hypertension in mice and rats. Antisense oligonucleotide against FXI also reduced the increased blood pressure and attenuated vascular and kidney dysfunction in rats with established arterial hypertension. Further, platelet-localized thrombin generation was amplified in an FXI-dependent manner in patients with uncontrolled arterial hypertension, suggesting that platelet-localized thrombin generation may serve as an inflammatory marker of high blood pressure. Our results outline a coagulation-inflammation circuit that promotes vascular dysfunction, and highlight the possible utility of FXI-targeted anticoagulants in treating hypertension, beyond their application as antithrombotic agents in cardiovascular disease.


Arterial hypertension is a risk factor for cardiovascular disease and death (1). Vascular inflammation is a hallmark of atherosclerosis (2) and arterial and venous thromboembolic disease (3). In this setting, vascular injury occurs with infiltration of lymphocyte antigen 6 complex locus C1high (Ly6Chi) monocytes into the vessel wall (4). Vascular inflammation and hypertension (5) induced by angiotensin II (ATII) similarly depends on the recruitment of ATII receptor type 1 (Agtr1)–expressing Ly6Chi monocytes to the vasculature (6), which cause NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) oxidase–dependent reactive oxygen species (ROS) production in arteries (7). The combined effects of the inflammation and oxidative stress triggered by infiltrating immune cells result in NADPH oxidase–dependent vascular dysfunction and high blood pressure (8, 9). ATII induces infiltration of immune cells into the vessel walls by promoting their adhesion to endothelial cells and transmigration (10). Vascular cellular adhesion molecule 1 (VCAM-1) is induced by ATII (11) and thrombin (12), which can be generated by coagulation activation through tissue factor (TF), another ATII-induced gene product (13). Platelets provide a pro-coagulant surface for amplified thrombin generation (TG) in hemostasis and thrombosis (14). They also promote leukocyte recruitment in arterial injury and angiogenesis models (15, 16) through the interaction of platelet glycoprotein Ibα (GPIbα) with the integrin αMβ2 (CD11b/CD18 or Mac-1) on leukocytes (17) and also regulate monocyte and neutrophil activation (18).

How platelets, coagulation factors, leukocytes, and the vessel wall interact to promote vascular inflammation in arterial hypertension remains unclear. We have therefore investigated the roles of factor XII (FXII), FXI, thrombin, and TF on inflammatory monocyte-driven vascular dysfunction and arterial hypertension in ATII-infused mice and rats, as well as in partially nephrectomized rats.


The hemostatic system contributes to ATII-induced leukocyte adhesion

We investigated mechanisms underlying leukocyte adhesion to the vascular endothelium of arteries by epifluorescence intravital video microscopy (IVM) imaging in normocholesterolemic C57BL/6 mice chronically infused with ATII (1 mg kg−1 day−1 for 7 days via osmotic minipumps). This model is characterized by robust arterial hypertension and accumulation of CD11b+ leukocytes in arterial vessels (8). ATII treatment induced extensive rolling and adhesion of leukocytes to the endothelium in the carotid artery (Fig. 1A and videos S1 and S2). Leukocyte integrin α4 [very late antigen 4 (VLA-4)] recognizes endothelial VCAM-1 to mediate leukocyte adhesion to inflamed endothelium (19). Short-term blockade of this interaction with anti–VLA-4 (10 mg kg−1) or anti–VCAM-1 (1 mg kg−1) antibody prevented both leukocyte rolling and adhesion in ATII-infused mice (Fig. 1, B and C, and videos S3 and S4). Short-term inhibition of TF with an anti-TF antibody (25 mg kg−1) or thrombin with lepirudin (1.7 mg kg−1) (20) immediately before IVM imaging also markedly reduced leukocyte rolling and adhesion (Fig. 1, D and E). Depletion of platelets with an anti-GPIbα antibody (3.4 mg kg−1) and antibody inhibition of the leukocyte integrin CD11b/CD18 (6.5 mg kg−1) similarly prevented leukocyte rolling and adhesion (Fig. 1, F and G), indicating that both coagulation and platelets contributed to ATII-induced leukocyte adhesion in high flow arterial vascular beds.

Fig. 1. A proinflammatory circuit of TF, thrombin, and platelets promotes ATII-induced endothelial leukocyte adhesion.

(A to G) Epifluorescence IVM of endothelial adherent and rolling leukocytes in the common carotid artery of ATII-infused mice. Nucleated cells were visualized with acridine orange (green fluorescence). (A) IVM of carotids at day 3 (d3) and day 7 (d7) of continuous ATII infusion compared to sham. Left: Representative image at day 3. Right: Quantification of adherent and rolling leukocytes. **P < 0.01; ***P < 0.001, one-way analysis of variance (ANOVA); n = 4 to 7 animals per group. (B to G) IVM of carotids at day 3 of ATII infusion, the time point of deflection of the blood pressure curve in response to ATII 60 hours after initial increase of blood pressure (8), before and 30 min after single intravenous injections of anti–VCAM-1 (B), anti–VLA-4 (C), anti-TF (D), the thrombin inhibitor lepirudin (Lepi) (E) of anti-GPIbα to acutely deplete platelets (F), and anti–Mac-1 (G). Injection of isotype control immunoglobulin G (IgG) antibodies had no significant effect on rolling (ATII versus ATII + control IgG: 39 ± 3.7 versus 44 ± 4.8 rolling cells/0.015 mm2) or adhesion (ATII versus ATII + control IgG: 1683 ± 377 versus 1486 ± 270 adherent cells per mm2). Left: Representative image at day 3. Right: Quantification of adherent and rolling leukocytes. *P < 0.05; **P < 0.01; ***P < 0.001, unpaired t test; n = 5 to 6 animals per group. Data are means ± SEM.

Platelet GPIbα is required for thrombin-dependent vascular inflammation

To further investigate this interaction of coagulation factors, platelets, and arterial vessels, we evaluated the contribution of coagulation to ATII-induced vascular inflammation and endothelial dysfunction by intervention with the anticoagulant and specific thrombin inhibitor lepirudin. In ATII-infused mice, simultaneous thrombin inhibition substantially improved endothelial dysfunction, as quantified by acetylcholine (ACh)–induced vascular relaxation (Fig. 2A). Consistent with previously established roles for thrombin in the induction of endothelial cell VCAM-1 (12) and monocyte chemoattractant protein 1 (MCP-1; encoded by Ccl2) (21), which is required for ATII-induced monocyte recruitment (22), lepirudin prevented expression of these genes in the vessel wall and attenuated Ly6c mRNA levels (Fig. 2D), indicating reduced vascular recruitment of inflammatory monocytes. This was paralleled by diminished vascular ROS production assessed by the superoxide-sensitive dye dihydroethidium (DHE) (Fig. 2G).

Fig. 2. Blockade of thrombin and GPIbα protects against endothelial dysfunction and preserves regulation of vascular tone.

(A to C) Concentration-relaxation curves in response to ACh of isolated aortic segments from C57BL/6 mice treated for 7 days with either lepirudin (Lepi) (A) or anti-GPIbα (B) or hIL-4R/1bα chimeric mice and C57BL/6 controls (C) treated with or without ATII for 7 days. *P < 0.05 versus C57BL/6; #P < 0.05 versus C57BL/6 + ATII, one-way ANOVA and Bonferroni’s multiple comparison test of maximal relaxation; n = 5 to 8 animals per group. (D to F) Aortic mRNA expression of Ccl2, Vcam-1, and Ly6c. (G) Oxidative fluorescence microtopography. Top: Representative photomicrographs of isolated aortic segments incubated with DHE (1 μM, 30 min at 37°C). Green, laminae (autofluorescence); red fluorescence, superoxide formation; E, endothelium; M, media; A, adventitia. Bottom: Densitometric analysis, integrated optical density (IOD). *P < 0.05; **P < 0.01; ***P < 0.001, one-way ANOVA and Bonferroni’s multiple comparison test; n = 4 to 6 animals per group. Data are means ± SEM.

Because platelets promote localized thrombin formation by providing a pro-coagulant surface, we evaluated their role in vascular dysfunction in ATII-infused mice. Platelet depletion with anti-GPIbα antibody (3.4 mg kg−1) during chronic ATII administration significantly reduced or prevented endothelial dysfunction (Fig. 2B); Vcam-1, Ccl2, and Ly6C mRNA expression in the vessel wall (Fig. 2E); and vascular ROS production (Fig. 2G). Thus, depletion of platelets largely recapitulated the effects of thrombin inhibition, suggesting a crucial role for platelets in promoting TG or mediating its proinflammatory effects.

To better understand thrombin interactions with platelets, we analyzed human interleukin-4 receptor (hIL-4R)/Ibα mice with a defective platelet GPIbα, a well-characterized thrombin receptor (23). These mice lack the extracellular ligand-binding domains of GPIbα but retain GPIbα cytosolic receptor interactions, surface expression of additional subunits of the GPIb-IX complex, GPIbβ and GPIX, and normal platelet size (24). ATII-infused hIL-4R/Ibα mice were completely or largely protected from ATII-induced vascular endothelial dysfunction (Fig. 2C), oxidative stress (Fig. 2G), and increased expression of Ccl2, Vcam-1, and Ly6c mRNA (Fig. 2F), as also seen in platelet-depleted mice. These findings identified platelet GPIbα as a relevant platelet receptor for ATII-induced endothelial dysfunction and vascular inflammation.

FXI promotes vascular dysfunction by amplifying thrombin formation involving platelet GPIbα

We next determined how coagulation is initiated in ATII-dependent vascular dysfunction. As expected from diminished leukocyte adhesion seen in ATII-infused mice after acute anti-TF treatment (Fig. 1D), long-term blockade of TF during ATII administration attenuated endothelial dysfunction (Fig. 3A) and reduced oxidative stress within the vessel wall [DHE staining: C57BL/6 + ATII versus C57BL/6 + ATII + anti-TF: 163 ± 16% versus 87 ± 5% IOD (% of C57BL/6 ± SEM)]. We analyzed ATII-induced endothelial dysfunction in FXII−/− mice or wild-type mice treated with an antibody (14E11) that interferes with FXII-dependent activation of FXI and found that contact-phase FXII-mediated FXI activation played no major role in mediating ATII-induced endothelial dysfunction (Fig. 3B). This result indicated that initiation of TG by the TF pathway was necessary and sufficient to cause endothelial dysfunction in response to chronically elevated levels of ATII in mice.

Fig. 3. FXI promotes ATII-induced vascular dysfunction initiated by TF.

(A to D) Concentration-relaxation curves in response to ACh (endothelium-dependent) of isolated aortic segments of mice treated with or without ATII (1 mg kg−1 day−1 for 7 days). (A) C57BL/6 mice treated for 7 days with anti-TF or control IgG (n = 8 to 9 animals per group). (B) C57BL/6 ± 14E11 (antibody blocking FXII-dependent activation of FXI) and FXII−/− mice (n = 4 to 12 animals per group). (C) FXI−/− and C57BL/6 mice (n = 4 to 7 animals per group). (D) C57BL/6 mice with continuous in vivo inhibition of FXI synthesis by FXI ASO or scrambled control ASO (FXI ASO and Scr ASO, respectively; n = 4 to 10 animals per group). *P < 0.05 versus C57BL/6 or C57BL/6 + Scr ASO; #P < 0.05 versus C57BL/6 + ATII or C57BL/6 + Scr ASO + ATII; $P < 0.05 versus FXII−/−; P < 0.05 versus C57BL/6 + 14E11, one-way ANOVA and Bonferroni’s multiple comparison test of maximal relaxation. n.s., not significant. (E) Experimental protocol of FXI ASO studies. To control for nonspecific effects of ASO injections, mice that were not injected with FXI ASO were control-injected with Scr ASO in all respective experiments. iv, intravenously; sc, subcutaneously. (F) Hepatic mRNA expression of F11 measured by real-time reverse transcription polymerase chain reaction (RT-PCR). Unpaired t test; n = 4 to 10 animals per group. Ctr, control. (G) FXI activity in platelet-poor plasma (PPP) of mice assessed after 1, 2, or 3 weeks of treatment with FXI ASO (25 or 50 mg kg−1) in vivo. Kruskal-Wallis and Dunn’s multiple comparison test; n = 4 to 25 animals per group. (H) FXI protein in plasma. Top: Densitometry. Bottom: One representative Western blot of four independent experiments. Kruskal-Wallis and Dunn’s multiple comparison test; n = 4 animals per group. (I) Concentration-relaxation curves in response to ACh of isolated aortic segments of FXI-depleted, ATII-infused C57BL/6 mice with or without continuous in vivo reconstitution with hFXI (hemoleven; unpaired t test of maximal relaxation; n = 8 animals per group). Data are means ± SEM. *P < 0.05; **P < 0.01.

Thrombin can amplify coagulation by feedback activation of FXI (25, 26). In addition to binding thrombin, GPIbα can recruit FXI via binding to its apple 3 domain (2729). We found that endothelial dysfunction in ATII-infused mice was significantly improved in FXI−/− mice (Fig. 3C) and in wild-type mice after pharmacologic inhibition of FXI synthesis by FXI-specific antisense oligonucleotides (FXI ASOs; Fig. 3D). FXI ASO interferes with F11 mRNA and leads to F11 mRNA degradation (30). FXI ASO treatment dose-dependently reduced FXI plasma pro-coagulant activity, hepatic F11 mRNA levels, and FXI protein expression in plasma (Fig. 3, E to H). Specificity of the effects of FXI depletion was confirmed by continuous in vivo supplementation with human FXI (hFXI) (Hemoleven). This treatment restored the reduction plasma FXI activity caused by FXI ASO application (Fig. 4A) and reestablished vascular endothelial dysfunction in FXI ASO–treated mice infused with ATII (Fig. 3I).

Fig. 4. A platelet localized thrombin-FXI loop promotes ATII-induced vascular inflammation.

(A to D) ATII- and sham-infused C57BL/6 mice were treated with Scr ASO or FXI ASO ± hFXI in vivo, as described in Fig. 3E. (A) FXI activity in PPP. Left: One-way ANOVA and Bonferroni’s multiple comparison test; n = 4 to 12 animals per group. Right: Mann-Whitney test, n = 4 animals per group. (B) IVM showing endothelial adherent leukocytes in the common carotid artery. One-way ANOVA and Bonferroni’s multiple comparison test; n = 3 to 5 animals per group. (C) Flow cytometry of circulating CD41+CD11b+GR-1intermF4/80+ platelet-monocyte conjugates in blood. Kruskal-Wallis and Dunn’s multiple comparison test; n = 3 to 5 animals per group. (D) Enzyme-linked immunosorbent assay (ELISA) of circulating TAT complexes. One-way ANOVA and Bonferroni’s multiple comparison test; n = 6 to 8 animals per group. (E) Thrombin-evoked ETP in PRP of ATII- and sham-infused C57BL/6 and hIL-4R/Ibα chimeric mice treated with Scr ASO or FXI ASO in vivo; hFXI was administered ex vivo into PRP, as indicated. In pilot studies, addition of excess hFXI to the PRP of C57BL/6 mice ex vivo increased ETP (C57BL/6 versus C57BL/6 + hFXI versus C57BL/6 + ATII versus C57BL/6 + ATII + hFXI: 100.0 ± 11.6% versus 127.8 ± 12.2% versus 134.9 ± 18.4% versus 152.5 ± 20.2% of C57BL/6, respectively). One-way ANOVA and Bonferroni’s multiple comparison test; n = 5 to 11 animals per group. (F) Endothelium-dependent relaxation of isolated aortic segments of C57BL/6 mice treated with or without M1/70 (anti–Mac-1 antibody) ± ATII for 7 days. *P < 0.05 versus C57BL/6; #P < 0.05 versus C57BL/6 + ATII, one-way ANOVA and Bonferroni’s multiple comparison test of maximal relaxation; n = 5 to 8 animals per group. (G) Aortic mRNA expression of Ccl2, Vcam-1, and Ly6c. One-way ANOVA and Bonferroni’s multiple comparison test; n = 6 animals per group. (H) Flow cytometry of circulating CD41+CD11b+GR-1intermF4/80+ platelet-monocyte conjugates in blood. Kruskal-Wallis and Dunn’s multiple comparison test; n = 3 to 5 animals per group. (I) Thrombin-evoked ETP in PRP. One-way ANOVA and Bonferroni’s multiple comparison test; n = 6 animals per group. Data are means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

IVM showed that ATII-induced leukocyte rolling, as well as adhesion in the carotid artery, was also reduced by FXI ASO treatment. Supplementation of FXI-depleted mice with hFXI restored ATII-induced leukocyte adhesion (Fig. 4B and videos S5 to S8). These results demonstrate that FXI activity is required for the ATII-induced and coagulation-dependent vascular pathology and indicate a role for thrombin-FXI feedback activation in the vasculature in vivo.

ATII also increased association of CD11b+ monocytes with CD41+ platelets; this process may involve interaction of GPIbα on platelets with leukocyte-expressed Mac-1 (15). Treatment of ATII-infused mice with the Mac-1–inhibiting antibody M1/70 significantly attenuated endothelial dysfunction (Fig. 4F) as well as vascular Ccl2 and Vcam-1 mRNA expression (Fig. 4G) and circulating platelet-monocyte conjugates (Fig. 4H). The number of platelet-monocyte conjugates was also normalized in FXI ASO–treated mice (Fig. 4C), indicating that coagulation activation was crucial for the inflammation-promoting interaction of Mac-1 with platelets. Circulating levels of thrombin-antithrombin (TAT) complexes, a marker for intravascular thrombin formation, were neither increased by ATII infusion nor altered by FXI ASO (Fig. 4D), excluding the possibility that the platelet-monocyte interaction was promoted by overt intravascular coagulation in the circulation.

We therefore hypothesized that thrombin-FXI feedback activation was localized to platelets in the inflammatory cross-talk of ATII-induced vascular dysfunction. In platelet-rich plasma (PRP), ex vivo addition of thrombin promotes additional TG [thrombin-evoked endogenous thrombin potential (ETP)] (31). PRP from ATII-treated mice showed increased TG compared to control mice, but Mac-1 inhibition did not affect platelet-localized TG (Fig. 4I). In contrast, PRP from mice with pharmacological inhibition of FXI production in the liver failed to amplify TG after chronic ATII exposure (Fig. 4E). Addition of hFXI to PRP from these mice was sufficient to restore the platelet-dependent TG. Consistent with the proposed role of GPIbα as a binding site for both thrombin and FXI (27, 28), endogenous TG was not increased in ATII-treated hIL-4R/Ibα mice and remained unchanged after addition of hFXI to PRP in vitro (Fig. 4E). Thus, the FXI-dependent TG feedback loop amplifies TF-initiated coagulation and requires platelet GPIbα, but not Mac-1, to cause vascular inflammation.

Targeting FXI attenuates ATII-induced vascular inflammation

We further evaluated the therapeutic potential of interrupting FXI synthesis and function (30, 32) to attenuate vascular dysfunction in arterial hypertension. Aortas of ATII-infused mice showed an expansion of CD45+ leukocytes, CD11b+GR-1+ myelomonocytic cells (C57BL/6 versus C57BL/6 + ATII versus C57BL/6 + FXI ASO versus C57BL/6 + FXI ASO + ATII: 24.3 ± 5.5 versus 329.7 ± 248.2 versus 15.9 ± 4.0 versus 140.3 ± 42.9 cells per aorta), and CD11b+Ly6GLy6Chi monocytes (Fig. 5A). Vascular injury was also macroscopically visible as inflamed patches at predilection sites of aortic aneurysm formation, a late sequel of ATII challenge (fig. S1). Aortas of ATII-exposed animals showed increased Ccl2, Vcam-1, and Nos2 [encoding for inducible nitric oxide synthase (iNOS)] mRNA expression (Fig. 5B). Aortic protein expression of nox2, reflecting the presence of phagocyte-type NADPH oxidase and heme oxygenase-1 (HO-1), a classical anti-inflammatory and antioxidant response gene, was increased by ATII (fig. S2). Lowering FXI levels with FXI ASO treatment normalized all these parameters of vascular inflammation and leukocyte infiltration. In addition, vascular ROS formation and fibrotic remodeling were blocked by FXI ASO administration in ATII-infused mice (Fig. 5, C and D).

Fig. 5. Inhibition of FXI protects against vascular inflammation.

C57BL/6 mice were treated with FXI ASO or Scr ASO and infused with ATII or sham for 7 days (see Fig. 3E for experimental protocol). (A) Flow cytometry of CD45+ leukocytes and CD11b+Ly6Chi cells in aortic lysates and absolute numbers of viable CD45+ and CD45+CD11b+Ly6ChiLy6GNK1.1 cells. Left: Representative original plots of CD11b+Ly6ChiLy6GNK1.1 myelomonocytes. Right: Quantification. One-way ANOVA and Bonferroni’s multiple comparison test; n = 5 animals per group. (B) Aortic mRNA expression of Ccl2, Vcam-1, and Nos2. One-way ANOVA and Bonferroni’s multiple comparison test; n = 4 to 13 animals per group. (C) Oxidative fluorescence microtopography. Left: Representative photomicrographs of aortic cryosections; superoxide formation appears in red. Right: Densitometry, IOD. One-way ANOVA and Bonferroni’s multiple comparison test; n = 5 animals per group. (D) Sirius red staining of aortic sections. (E) Summary of 192 hours of telemetrically recorded systolic blood pressure (in millimeters of mercury). Two-way ANOVA and Bonferroni post test; n = 4 to 5 animals per group. Data are means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001. In (E), *P < 0.05 versus C57BL/6 + Scr ASO and C57BL/6 + FXI ASO; #P < 0.05 versus C57BL/6 + Scr ASO + ATII.

Targeting FXI prevents blood pressure increase in arterial hypertension

On the basis of our observation that FXI was critical for ATII-induced vascular dysfunction, we continuously recorded arterial blood pressure by telemetry. Reducing FXI levels by FXI ASO significantly attenuated the ATII-induced blood pressure increase in mice (Fig. 5E).

To exclude species limitations, we confirmed the efficacy of this therapeutic intervention with FXI inhibitors in Wistar rats. Compared to Scr ASO, rat-specific FXI ASO very effectively reduced F11 mRNA levels in the liver (control versus FXI ASO: 100 ± 4% versus 7 ± 1% of control). ATII infusion induced endothelial dysfunction (Fig. 6A), which was significantly attenuated by FXI ASO treatment. ATII infusion increased aortic mRNA expression of Ccl2 and Vcam-1 as well as of Ccr2, which encodes the MCP-1 receptor, and of Spn, which encodes sialophorin (CD43) and is involved in leukocyte adhesion (33) and a marker for monocytes in rats (Fig. 6B) (34). Accordingly, ATII infusion induced vascular accumulation of CD45+ leukocytes (Fig. 6C) and increased vascular oxidative stress in the aorta (Fig. 6D) and systolic blood pressure (Fig. 6E). FXI ASO application attenuated these alterations: neither endothelial function, vascular mRNA expression of inflammatory genes, accumulation of CD45+ cells, vascular oxidative stress, nor blood pressure was significantly different in ATII-infused FXI ASO–treated rats compared to sham-infused Scr ASO– or FXI ASO–treated rats. This recapitulated the findings obtained in the ATII-infused C57BL/6 mice.

Fig. 6. FXI ASO administration prevents vascular dysfunction and arterial hypertension in ATII-infused and 5/6Nx rats.

(A to E) Wistar rats were treated for 1 week with ATII or sham and with FXI ASO (Wistar + FXI ASO ± ATII) or Scr ASO (Wistar + Scr ASO ± ATII) starting 2 weeks before ATII infusion to be analogous to the mouse treatment scheme (see Fig. 3E). (A) Endothelium-dependent relaxation of isolated aortic segments. *P < 0.05 versus Wistar + Scr ASO; #P < 0.05 versus Wistar + Scr ASO + ATII, one-way ANOVA and Bonferroni’s multiple comparison test of maximal relaxation; n = 5 to 9 animals per group. (B) Aortic mRNA expression of Ccl2, Vcam-1, sialophorin (Spn), and chemokine receptor 2 (Ccr2). Kruskal-Wallis and Dunn’s multiple comparison test; n = 3 to 6 animals per group. (C) Immunostaining of aortic cryosections. Green, autofluorescence; red, CD45; blue, 4′,6-diamidino-2-phenylindole (DAPI). Red fluorescence was quantified in whole aortic rings. One-way ANOVA and Bonferroni’s multiple comparison test; n = 4 to 5 animals per group. (D) Superoxide formation in aortic cryosections. Left: Representative DHE photomicrotopographs of aortic cryosections; superoxide formation appears in red. Right: Quantification. One-way ANOVA and Bonferroni’s multiple comparison test, n = 6 animals per group. (E) Systolic blood pressure after 1 week of ATII infusion or sham treatment. Kruskal-Wallis and Dunn’s multiple comparison test; n = 3 to 14 animals per group. (F to I) Wistar rats underwent 5/6 nephrectomy (5/6Nx) or sham operation (sham) and were administered for 3 weeks with FXI or Scr ASO, starting 1 week after surgery. (F) Endothelium-dependent relaxation of isolated aortic segments. One-way ANOVA and Bonferroni’s multiple comparison test of maximal relaxation; n = 5 to 12 animals per group; *P < 0.05 versus Wistar + Scr ASO; #P < 0.05 versus Wistar + Scr ASO + 5/6Nx. (G) Oxidative fluorescence microtopography. Left: Representative photomicrographs. Right: Densitometry. One-way ANOVA and Bonferroni’s multiple comparison test; n = 6 animals per group. (H) ETP in PRP. Kruskal-Wallis and Dunn’s multiple comparison test; n = 4 to 6 animals per group. (I) Systolic blood pressure after nephrectomy. #P < 0.05 versus Wistar + Scr ASO + 5/6Nx, two-way ANOVA and Bonferroni’s post tests; n = 4 (sham) and 8 to 10 (5/6Nx) animals per group. Data are means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001 (B to D, F, and G).

To investigate an alternative model of arterial hypertension that reflects human pathology, we next studied 5/6 nephrectomized (5/6Nx) compared to sham-operated Wistar rats. This rat model is an established model of chronic kidney disease characterized by endogenous up-regulation of the renin-angiotensin-aldosterone system (RAAS) with increased systemic ATII levels and progressive and sustained arterial hypertension and vascular dysfunction (35). Hepatic F11 mRNA levels were not increased by the surgery and were effectively reduced by FXI ASO in both sham-operated and 5/6Nx rats (sham versus 5/6Nx versus sham + FXI ASO versus 5/6Nx + FXI ASO: 100 ± 6% versus 92 ± 8% versus 2 ± 1% versus 1 ± 0% of sham). Four weeks after surgery, 5/6Nx rats had endothelial dysfunction, as compared to sham-operated animals, which was prevented by concomitant FXI ASO administration (Fig. 6F). FXI ASO decreased the vascular oxidative stress in the aorta elicited by the increased RAAS activation in the 5/6Nx rats (Fig. 6F). TG was increased in PRP of 5/6Nx rats and was markedly diminished by FXI ASO application, mimicking the results from hypertensive mice (Fig. 6H). Together, these results indicate that attenuation of the inflammatory pro-coagulant milieu in this rat model prevented blood pressure increase (Fig. 6I), as seen in ATII-infused mice and rats.

Treatment of arterial hypertension with FXI ASO reduces blood pressure, vascular dysfunction, and kidney injury

To extend our finding that FXI ASO could attenuate arterial hypertension when given preventively, we next aimed to treat established hypertension. Eight weeks after 5/6 nephrectomy, 5/6Nx rats were characterized by vascular endothelial dysfunction (Fig. 7A) as well as increased vascular mRNA expression of Ccr2 and Spn (Fig. 7B), vascular accumulation of CD45+ leukocytes (Fig. 7C), and ROS formation (Fig. 7D). In rats that had been nephrectomized for 8 weeks, Pai-1 mRNA (encoding for plasminogen activator inhibitor-1) expression in the renal cortex was drastically increased (Fig. 7E) and associated with kidney fibrosis, structural kidney damage (Fig. 7, F and G), and impaired kidney function, as indicated by retention of urea nitrogen (urea-N) and creatinine in plasma (Fig. 7H) and proteinuria (Fig. 7I). Blood pressure significantly increased 2 weeks after 5/6 nephrectomy and plateaued between weeks 4 and 8 (Fig. 7J). Continuous treatment of 5/6Nx rats with FXI ASO starting 3 weeks after surgery significantly improved vascular injury, renal damage, and expression of Pai-1 mRNA and significantly and persistently reduced arterial hypertension. Thus, the coagulation-inflammation circuit involving the thrombin-FXI loop is a major regulator of arterial hypertension, and interruption of this pathway reduces established vascular dysfunction and arterial hypertension and can treat long-term complications of kidney injury associated with hypertension.

Fig. 7. FXI ASO treatment of established arterial hypertension attenuates vascular and kidney injury and reduces blood pressure in 5/6Nx rats.

Wistar rats underwent 5/6 nephrectomy (5/6Nx) or sham operation (sham) and were treated for 5 weeks with FXI ASO or Scr ASO, starting 3 weeks after surgery. (A) Endothelium-dependent relaxation of isolated aortic segments. *P < 0.05 versus Wistar sham, one-way ANOVA and Bonferroni’s multiple comparison test of maximal relaxation; n = 4 to 6 animals per group. (B) mRNA expression levels of Spn and Ccr2 in aortic lysates. Kruskal-Wallis and Dunn’s multiple comparison test; n = 4 to 6 animals per group. (C) Immunostaining of aortic cryosections. Green, autofluorescence; red, CD45; blue, DAPI. Red fluorescence was quantified in whole aortic rings. One-way ANOVA and Bonferroni’s multiple comparison test; n = 4 animals per group. (D) Oxidative fluorescence microtopography. Left: Representative photomicrographs. Right: Densitometry. One-way ANOVA and Bonferroni’s multiple comparison test; n = 4 to 6 animals per group. (E) mRNA expression levels of Pai-1 in renal cortex. Kruskal-Wallis and Dunn’s multiple comparison test; n = 4 to 6 animals per group. (F) Sirius red staining and (G) periodic acid–Schiff (PAS) staining of kidney sections (left) and score for PAS staining (right). One-way ANOVA and Bonferroni’s multiple comparison test; n = 4 to 6 animals per group. Quantification of (H) plasma urea-N and creatinine and (I) proteinuria. Kruskal-Wallis and Dunn’s multiple comparison test; n = 4 to 6 animals per group. (J) Systolic blood pressure. #P < 0.05 versus Wistar + 5/6Nx; §P < 0.05 versus Wistar + Scr ASO and Wistar + FXI ASO; *P < 0.05 versus Wistar + Scr ASO + 5/6Nx, two-way ANOVA and Bonferroni’s post tests; n = 8 (sham) and 12 (5/6Nx) animals per group. Data are means ± SEM. *P < 0.05; **P < 0.01 (B to I).

FXI-dependent TG is increased in humans with uncontrolled hypertension

To assess the applicability of our findings to human disease, we explored platelet-localized TG in an all-comer population of patients with arterial hypertension who came to the emergency room and outpatient clinic of the University Medical Center Mainz. We observed a positive correlation between mean arterial blood pressure and ETP, which corresponds to the overall thrombin generated over time, as well as peak (maximum amount of thrombin generated) and velocity of TG in PRP (Fig. 8A). According to current guidelines (36), we divided our study population into three groups. Except for systolic, diastolic, and mean arterial blood pressure, there were very few demographic differences between the groups (Table 1). Individuals with hypertension grade II or higher showed significantly higher TG than controls, recapitulating our findings from mouse and rat models (Fig. 8B). Addition of an FXI anti–apple 3 domain antibody to block FXI activation by thrombin markedly reduced ETP, peak TG, and velocity of TG in PRP from control subjects and completely abrogated the increase of these markers in PRP from patients with uncontrolled hypertension (Fig. 8C). To exclude the possibility that these results were due to alterations in plasma components and make sure that they were related to platelets, we resuspended washed platelets in PPP from healthy donors. When the resuspended platelets were derived from individuals with uncontrolled hypertension, but not when they were from healthy controls, we measured significantly increased peak TG and velocity of TG, confirming that platelets from patients with hypertension are sufficient and essential to support TG in plasma (Fig. 8D).

Fig. 8. Blood pressure correlates with platelet-localized, FXI-dependent TG in humans.

(A) In all participants of the FACTO-RR study, linear regression analysis was performed between mean arterial pressure (MAP) and ETP and between peak TG and velocity of TG; n = 71, Pearson correlation coefficient; exact two-tailed P values are given. (B) On the basis of current guidelines (36), the population was stratified into three groups, for example, controlled hypertension (control, n = 19), arterial hypertension grade I (HT°I; n = 16), and arterial hypertension grade II or higher (HT≥°II; n = 36). ETP in PRP was analyzed for each individual group. Kruskal-Wallis test with Dunn’s multiple comparison test was used. (C) Participants of the FACTO-RR study were split into two groups: controlled hypertension (C; n = 19) and uncontrolled (HT°I and HT≥°II) hypertension (HT; n = 52). ETP, peak TG, and velocity of TG were measured before and after addition of 1A6, an anti–apple 3 domain antibody blocking the feedback activation of FXI by thrombin. Data are means ± SEM. Kruskal-Wallis test with Dunn’s multiple comparison test was used. (D) In selected patients, washed platelets of controlled (n = 10) and uncontrolled (n = 13) hypertensives were resuspended in PPP of healthy donors and triggered with exogenous TF to evoke ETP. Mann-Whitney test. Data are means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001. Each measurement was performed in triplicate, yielding one data point. (E) Scheme illustrating the main findings. Blue color marks the most relevant interaction partners of the newly identified mechanism. Green, positive feedback loop; red, inhibitory action.

Table 1. Patient characteristics of all participants of the FACTO-RR study.

On the basis of the current guidelines on arterial hypertension, the population was stratified into three groups (36), for example, controlled hypertension (control), arterial hypertension grade I, and arterial hypertension grade II or higher (summarized as uncontrolled hypertension in some analyses). *P < 0.05 versus control; #P < 0.05 versus hypertension grade I, Kruskal-Wallis test with Dunn’s multiple comparison test or Fisher’s exact test. bpm, beats per minute; ACE, angiotensin-converting enzyme; NSAIDs, nonsteroidal anti-inflammatory drugs; BMI, body mass index; MI, myocardial infarction; CAD, coronary artery disease; PAD, peripheral artery disease; COPD, chronic obstructive pulmonary disease; CVD, cardiovascular disease.

View this table:


Our findings provide insight into the pathogenic role of the thrombin-FXI loop in inflammation and show that this pathway makes a crucial contribution to arterial hypertension in the absence of overt vascular thrombotic occlusion. We delineate an unexpected sequence of events that specifically fosters vascular inflammation in RAAS-driven arterial hypertension. ATII triggers TF up-regulation in the vasculature, providing initial TG that is amplified by an FXI feedback loop mediated through GPIbα. Additional thrombin generated on platelets leads to leukocyte adhesion and vascular inflammation, which depends on VCAM-1, VLA-4, and Mac-1. We demonstrated that FXI inhibition blocks ATII-driven vascular immune cell infiltration and attenuates arterial hypertension (see scheme in Fig. 8E). FXI inhibition introduces a disruption at the crossroads of an orchestrated inflammatory response that involves platelets, leukocytes, and the vessel wall. Localized platelet-amplified thrombin formation mediates the recruitment of proinflammatory monocytes to the vasculature in a process that leads to arterial hypertension. Given that arterial hypertension is a multifactorial disease, which requires combined medical treatment targeting multiple pathways [salt/water retention, increased sympathetic tone, and activated RAAS (36)], it is interesting that targeting FXI alone was effective in reducing not only vascular inflammation but also blood pressure.

Platelet inhibitors, such as aspirin (acetylsalicylic acid), which are among the most prescribed drugs worldwide, are very effective in reducing mortality in people who have experienced adverse cardiovascular events such as myocardial infarction and are also being considered for primary prevention in the general population (37, 38). Aspirin has an anti-inflammatory preventive potential in individuals with increased high-sensitivity C-reactive protein, who are at risk for future cardiovascular events (39). The pathway we have described here is driven by ATII and critically depends on platelet-expressed GPIbα, a known receptor for both thrombin and FXI. Anti-platelet drugs such as abciximab, a GPIIb/IIIa antagonist, and aspirin only partially influence TF-triggered TG in PRP (40). Furthermore, there is no evidence that aspirin or GPIIb/IIIa inhibitors reduce blood pressure. Therefore, our findings suggest that inhibiting the GPIbα-dependent, thrombin-FXI–amplifying loop may provide added cardiovascular benefits that are synergistic with those of established platelet inhibitors.

Anticoagulants blocking vitamin K–dependent synthesis of coagulation factors (vitamin K antagonists) and newer, protease-specific direct oral anticoagulants such as thrombin inhibitors could both be of clinical utility to interrupt the proinflammatory TF-thrombin pathway uncovered in our study. Our data unexpectedly show that pharmacological targeting of FXI can achieve beneficial effects comparable to thrombin inhibition in preventing vascular inflammation in our model. FXI has previously been implicated in the development of thrombosis (26, 41), and more recent evidence suggests that FXI may also be involved in atherogenesis through its ability to promote inflammation. Although reduced numbers of lesional macrophages were described in FXI-deficient ApoE−/− mice, the mechanism remains unclear (42); it may involve a pathogenic mechanism similar to that described in our study.

In humans, FXI deficiency is associated with a decreased risk of deep vein thrombosis (43) and stroke; however, because hereditary FXI deficiency is rare, there is no conclusive evidence that low FXI levels are linked to decreased prevalence of arterial hypertension in humans (44). The pathway that we have described here for FXI-dependent TG is associated with clinically manifest arterial hypertension in humans, and furthermore, platelet-dependent amplified TG may be a candidate surrogate marker to stratify patients for therapeutic intervention with FXI inhibitors. A wide range of antihypertensive drugs are available to control blood pressure; however, we unexpectedly discovered that established arterial hypertension can be attenuated by FXI ASO in rats. Inhibition of FXI was sufficient to prevent PAI-1 expression in the kidney. PAI-1 is induced by thrombin in both the vasculature (45) and the kidneys (46) and is a surrogate marker for kidney injury (47), and PAI-1 activity leads to kidney fibrosis induced by ATII (48). Thus, targeting of FXI does not simply modulate a redundant pathway in blood pressure regulation but may substantially improve hypertension-associated renal dysfunction, kidney fibrosis, and injury.

Preclinical (30) and clinical (49) evidence further indicate that inhibition of FXI is associated with a lower risk of bleeding complications than is warfarin or enoxaparin. The described surrogate marker evaluation of this pathway (Fig. 8), together with the reported safety in avoiding bleeding complications (30, 49), suggests that the pleiotropic effects of FXI inhibitors to prevent vascular inflammation, dysregulation of vascular tone, and ensuing kidney damage should be exploited in medical care. Because our mechanistic results are largely limited to animal models, randomized clinical trials are needed to test whether these effects hold true in humans. At first, the anti-inflammatory and blood pressure–lowering effects of FXI inhibitors should be monitored once this type of drug enters clinical use as antithrombotic medication. FXI inhibitors may prove more therapeutically useful than anticipated in individuals with cardiovascular disease and activated RAAS.


Study design

The overall objective of the study was to test the hypothesis that leukocytes, platelets, and coagulation factors interact to promote vascular injury in arterial hypertension. We designed a translational study, involving experimentation with laboratory animals (mice, in part genetically modified, and rats) and blood plasma as well as an observational analysis of a monocentric nonrandomized unblinded all-comer clinical cohort. In the experimental studies, samples were measured at least in duplicate and replicated at least three times (n ≥ 3 animals per group, specified in each figure legend). Experimentators were blinded to the treatment arm by pseudonymization of the samples, whenever feasible.


All chemicals were of highest analytical grade and obtained from either Sigma-Aldrich or Merck. FXI ASOs [mouse: lot nos. 404071 (FXI ASO) and 421208 (control); rat: lot nos. 404088 (FXI ASO) and 141923 (control)] were produced by Ionis Pharmaceuticals. They were designed to suppress hepatic mRNA levels of FXI in mice and rats. The ASOs had a length of 20 nucleotides, and phosphorothioate and 2′-O-methoxyethyl were added (30).

Animals and in vivo treatment

FXII−/− (50), FXI−/− (51), and hIL-4R/Ibα mice (24) were generated and backcrossed more than 10 times to the C57BL/6J background, as described previously. C57BL/6J mice were purchased from the Jackson Laboratory and used as control mice in all experiments. Male mice (10 to 12 weeks old) were used as experimental animals. Mice were infused subcutaneously with ATII (1 mg kg−1 day−1 for 7 days) via miniosmotic pumps (model 1007D, ALZET) versus sham. In selected experiments, mice were injected subcutaneously with FXI ASO (50 mg kg−1) or Scr ASO solved in 0.9% NaCl twice per week for a period of 3 weeks, starting 2 weeks ahead of the ATII regimen (see Fig. 3E). After 3 weeks, mice were euthanized by exsanguination under isoflurane anesthesia, and aorta, liver, and blood were collected. In selected experiments, C57BL/6 mice were intraperitoneally or intravenously injected with platelet-depleting anti-GPIbα antibody [5A7 Ab (52); 3.4 mg kg−1 intraperitoneally three times per week or acutely 3.4 mg kg−1 intravenously], TF-inactivating antibody (21E10; 20 mg kg−1 intraperitoneally three times per week or acutely 25 mg kg−1 intravenously), anti–VCAM-1 antibody (clone 429 MVAM.A, eBioscience; 1 mg kg−1, intravenously), anti–VLA-4 antibody (clone PS2α; 10 mg kg−1,intravenously), anti–Mac-1 antibody (clone M1/70, BioXCell; acutely 6.5 mg kg−1), and an antibody blocking the FXII-dependent activation of FXI (14E11, provided by A. Gruber; 5 mg kg−1 intraperitoneally three times per week) or respective control IgGs resolved in phosphate-buffered saline (PBS). In some experiments, mice were reconstituted intravenously with highly purified plasma-derived hFXI (30 U kg−1) (Hemoleven, Swedish Orphan Biovitrum) every second day for 1 week or sham (0.9% NaCl). Thrombin inhibition was achieved by continuous infusion of lepirudin via miniosmotic pumps (model 1007D, ALZET) implanted intraperitoneally (0.3 mg kg−1 hour−1) or by acute intravenous injections (1.7 mg kg−1). In addition, male Wistar rats (8 to 10 weeks old; 450 to 500 g; Charles River Laboratories) were used. In analogy to the mouse experiments, rats were treated subcutaneously with ATII (1 mg kg−1 day−1 for 7 days) via miniosmotic pumps (model 2001, ALZET) versus sham. Rats were injected subcutaneously with FXI ASO (50 mg kg−1) or Scr ASO solved in 0.9% NaCl twice per week for a period of 3 weeks, starting 2 weeks ahead of the ATII regimen.

As an additional hypertension model, 5/6Nx rats were used. Studies were carried out on male Wistar rats from Janvier Labs, aged 8 weeks old. Rats underwent one-step 5/6 subtotal nephrectomy, as described previously (53), or sham operation performed by Janvier Labs as the commercial provider. Briefly, surgery consists of right nephrectomy and surgical ablation of the lower and upper poles of the left kidney. Depending on the study protocol, 1 week (preventive protocol) or 3 weeks (treatment protocol) after operation, subcutaneous injection of FXI ASO (50 mg kg−1) or Scr ASO twice per week for a period of 3 weeks (or 5 weeks, respectively) was started. Animal treatment was approved by the board of examinations of the state of Rhineland-Palatinate (authorization number 23 177-07/G12-1-002).

Blood pressure recordings

For blood pressure recordings of mice, carotid catheters were implanted into C57BL/6 animals [TA-PA11C10, Data Science International (DSI)]. For anesthesia and analgesia, mice received intraperitoneal injections of midazolam (5 mg kg−1; Ratiopharm GmbH), medetomidine (0.5 mg kg−1 body weight), and fentanyl (0.05 mg kg−1 body weight; Janssen-Cilag GmbH). After the surgical procedure, the animals were administered atipamezole (0.05 mg kg−1), flumazenil (0.01 mg kg−1), and naloxone (0.024 mg kg−1) subcutaneously to antagonize anesthesia. Postoperative analgesia was carried out with buprenorphine (0.075 mg kg−1). The implantation of the catheters was performed under sterile conditions. After surgery, mice recovered for 1 to 2 weeks until the recording was started. Blood pressure was continuously recorded to receiver platforms (DSI), and the data were saved using DataQuest system (DSI). Rat blood pressure measurements were performed by tail cuff with the Coda Monitor System (Kent Scientific). Blood pressure in ATII versus sham-infused rats was measured 1 week after pump implantation and in 5/6Nx versus sham-operated rats once a week for a period of 4 or 8 weeks starting 1 week after surgery.

Preparation of PPP and PRP from mice and rats

Blood samples were collected from experimental animals under isoflurane anesthesia by puncture of the inferior caval vein. Complete blood counts and hematocrit were determined with an automatic cell counter KX-21N (Sysmex Corporation). The obtained blood (700 μl; 10:1 with 3.2% sodium citrate) was gently mixed for proper anticoagulation. After centrifugation for 15 min at 2500g at 20°C, PPP was transferred into new tubes, snap-frozen, and stored at −80°C until further use. PRP from experimental animals was prepared by differentiated centrifugation of citrate-anticoagulated whole blood, taken by retro-orbital vein puncture using silanized glass capillaries.

FXI activity assay and activated partial thromboplastin time

PPP was defrosted at 37°C, every sample was gently stirred, and assays were performed using the Siemens BCS II according to the manufacturer’s protocol. Results are compared to human pooled values.

Preparation of PPP and PRP from human subjects

Blood was obtained by venipuncture via a 21-gauge needle and collected into 0.129 M sodium citrate (9:1, v/v). Complete blood counts and hematocrit were determined with an automatic cell counter KX-21N (Sysmex Corporation). PRP was prepared within 1 hour of blood sampling with a first centrifugation of 190g for 10 min at 20°C. The supernatant was removed, and the platelet count was adjusted to 150 × 109 per liter with PPP prepared with a 10-min centrifugation at 1750g at 20°C. In platelet resuspension experiments, PRP was centrifuged for 4 min at 1750g at 20°C, supernatant was removed, platelets were resuspended in a pool of PPP from healthy volunteers, and platelet count was adjusted at 150 × 109 per liter.

TG assay in PRP

Platelet-dependent TG was assessed in PRP upon stimulation with α-thrombin (0.1 U/ml) by fluorogenic calibrated automated thrombography in vitro in a microtiter plate fluorometer (Fluoroskan Ascent) using the thrombinoscope and Synapse BV software program (31, 5456). Where indicated, an anti–apple 3 domain of FXI (1A6; generated by A. Gruber) was added at a concentration of 5 μg/ml. The parameters calculated by the software were the lag time, thrombin peak, time to peak, velocity, and ETP, which corresponds to the area under the curve.

Vascular reactivity studies

To assess vasodilator properties of isolated aortic segments (~4 mm), they were mounted to force transducers in organ chambers to test their response to ACh. The aortic rings were preconstricted with phenylephrine (0.15 μM) or prostaglandin F2α (3 nM) to reach 50 to 80% of the tone induced by KCl. Concentration-relaxation curves were recorded in response to the endothelium-dependent vasodilator ACh (1 nM to 3 μM) (8).

Fluorescence oxidative microtopography

To quantify vascular ROS production, the aortas were cut into rings of 4 mm in length and then incubated in Krebs-Hepes buffer plus protease inhibitors [NaCl (5.78 g liter−1), KCl (0.35 g liter−1), CaCl2 (0.37 g liter−1), MgSO4 (0.30 g liter−1), NaHCO3 (2.1 g liter−1), K2HPO4 (0.14 g liter−1), Hepes (5.21 g liter−1), and d-glucose (2.0 g liter−1)] for 10 min at 37°C. The aortic rings were placed in aluminum cups filled with OCT (optimal cutting temperature) resin (Sakura Finetek) and slowly frozen with liquid nitrogen. ROS production was quantified with DHE (1 μM)–derived fluorescence after incubation for 30 min at 37°C. A Zeiss Axiovert 40 CFL microscope and an AxioCam MRm camera (Zeiss) were used for detection, and images were analyzed using the AxioVision data acquisition software (Zeiss) and quantified as IOD.

Reverse transcription polymerase chain reaction

FXI-mRNA expression of hepatic tissue was analyzed by quantitative RT-PCR using 7900HT Fast Real-Time PCR System (Applied Biosystems). Hepatic mouse mRNA was isolated following the manufacturer’s instructions of the RNeasy Fibrous Tissue Mini Kit (Qiagen).

For isolation of aortic RNA from snap-frozen mouse or rat aortas, rat kidneys or rat liver tissues were homogenized with the TissueLyser II (Qiagen), and for RNA isolation, the modified guanidine isothiocyanate method of Chomczynski and Sacchi (57) was used. RT-PCR was performed with the CFX96 Real-Time PCR Detection System (Bio-Rad). For RT-PCR analysis, total RNA (0.125 μg) was used with the QuantiTect Probe RT-PCR kit (Qiagen). TaqMan Gene Expression assays were used as probe and primer sets (Applied Biosystems) for TATA-box binding protein (mouse: Tbp, Mm00446973_m-1; rat: Tbp, Rn01455646_m1) and FXI (mouse: F11, Mm01194987_m1; rat: F11, Rn01767420_m1), Vcam-1 (mouse: Mm00449197_m1; rat: Rn00563627_m1), MCP-1 (Ccl2; mouse: Mm00441242_m1; rat: Rn00580555_m1), Ly6c (mouse: Mm03009946_m1), iNOS (Nos2; mouse: Mm00440485_m1), Spn (rat: Rn02061804_s), Ccr2 (rat: Rn01637698_s1), and Pai-1 (mouse: Serpine1, Mm00435858_m1; rat: Serpine1, Rn01481341_m1). Results were quantified with the relative Ct method and normalized to TATA-box binding protein as the endogenous control. mRNA levels were expressed relative to levels of control.

Western blot analysis

PPP and isolated aortic tissue were used for Western blot analysis. Under nonreducing conditions, proteins of PPP were separated by 7.5% SDS–polyacrylamide gel electrophoresis (PAGE). Sample volume of 1 μl was diluted in sample buffer. Immunoblotting was performed with an antibody against FXI (1:250; Nordic Immunology). Protein suspension of homogenized aortic tissue was separated by 12% SDS-PAGE under reducing conditions. After blotting on a nitrocellulose membrane, immunoblotting was accomplished with antibodies against α-actinin as a loading control (1:1000; Sigma-Aldrich), nox2 (gp91phox, 1:500; BD Biosciences), and HO-1 (1:2000; Epitomics). Detection of specific bands was performed with peroxidase-conjugated secondary antibodies (1:10,000; Vector Laboratories) and enhanced chemiluminescence according to the manufacturer’s instruction. These bands were analyzed by densitometry.

Enzyme-linked immunosorbent assay

TAT complexes in PPP were quantified by ELISA. Plasma samples were diluted 1:10 with sample diluent. The ELISA Kit for TAT complexes (USCN Life Science, E90831Mu) was performed according to the manufacturer’s instructions. The assay was performed with a Millenia Kinetic Analyzer (Diagnostic Products Corporation), and results were analyzed with SoftMax Pro (Molecular Devices).

Picro-sirius red staining

Isolated aortic rings and decapsulated left kidneys were fixed in paraformaldehyde (4%) directly after removal and embedded afterward in paraffin. After deparaffinization, nuclei were prestained with hemealaun. Samples were stained in picro-sirius red solution (0.1% with 1.2% picric acid). Finally, specimens were dehydrated with ethanol and coverslipped with Entellan. To determine the presence of early fibrosis and collagen accumulation, 10 images per animal were taken using an Olympus IX73 microscope and an Olympus SC30 camera and were examined.

Immunofluorescence staining of rat aorta

Aortic cryosections (5 μm) were blocked with 1% bovine serum albumin in PBS with 0.05% Tween 20. Tissue sections were stained with the primary anti-CD45 antibody (1:150, ab10558; Abcam) at 4°C overnight. Slides were washed three times with PBS and incubated with the secondary antibody Alexa 594 donkey anti-rabbit (1:2000, A21207; Life Technologies) for 1 hour at room temperature in the dark. After three washes in PBS for 5 min each, slides were counterstained with DAPI and mounted using ProLong Diamond Antifade Mounting medium with DAPI (Life Technologies). Sections were imaged with an Olympus IX73 microscope. Negative controls were performed for every set of experiments by omitting the primary antibodies from the procedure. For quantification, red fluorescence of whole aortic rings was quantified by digital image analysis using ImageJ. For publication images with DAPI staining, green autofluorescence and red fluorescence were taken using a 10× objective.

Kidney histological evaluation

The left kidneys were decapsulated, fixed in phosphate-buffered 4% paraformaldehyde (pH 7.4), embedded in paraffin wax, and cut longitudinally to a thickness of 5 μm. Renal tissue sections were stained with PAS and hematoxylin and eosin. All histology stainings were performed by the Histology Core Facility of the Institute of Molecular Biology (University of Mainz, Germany). To determine kidney injury, 10 consecutive cortical and juxtamedullary images per animal were taken using an Olympus IX73 microscope and an Olympus SC30 camera and were examined. Quantitative analysis of kidney injury was determined by semiquantitative injury scoring (PAS score: 0, ≤5%; 1, 5 to 15%; 2, 15 to 25%; 3, 25 to 50%; 4, ≥50%). These examinations were performed blinded by two investigators, and the mean values were calculated.

Protein excretion

Creatinine (milligram per deciliter; Jaffe reaction) in spontaneous urine samples was measured by an Abbott ARCHITECT c8000 Clinical Chemistry Analyzer (Abbott) using Abbott reagents. The total protein content was determined by Bradford assay using Roti-Quant (Bio-Rad) following the manufacturer’s instructions. Proteinuria was calculated as milligram of protein per milligram of creatinine.

Plasma analysis of kidney function

Heparinized blood was collected by right ventricular heart puncture. Creatinine (milligram per deciliter; Jaffe reaction) and urea-N (milligram per deciliter; urease method) were measured in mouse heparinized plasma by an Abbott ARCHITECT c8000 Clinical Chemistry Analyzer using Abbott reagents.

Flow cytometric analysis of aortic lysates and of platelet-leukocyte conjugates in peripheral blood

Aortic vessels were cleaned of perivascular fatty tissue and adventitia, minced, and digested by liberase (1 mg/ml; Roche Diagnostics), as described previously (8, 58). Single-cell suspensions were stained with CD45–allophycocyanin (APC)–eFluor 780, CD11b–phycoerythrin (PE), Ly6G– or GR-1–fluorescein isothiocyanate (FITC), Ly6C–peridinin chlorophyll protein (PerCP)–Cy5.5, NK1.1-PE-Cy7, F4/80-APC, and Viability Dye eFluor 506 monoclonal antibodies. At least 2.5 × 105 to 4.0 × 105 cells were treated with Fc-block, washed, and surface-stained. On the basis of a live gate, events were acquired and analyzed using a BF FACSCANTO II flow cytometer (Becton Dickinson) and FACSDiva software (Becton Dickinson), respectively. Monocyte-platelet conjugates were analyzed in citrate-anticoagulated whole blood (59), taken by retro-orbital vein puncture using heparinized glass capillaries. Platelets were stained with anti-CD41–FITC antibody; monocytes were stained with anti-CD11b–PE, anti–GR-1–V450, and F4/80-APC antibodies and gated from CD45+ viable cells.

Intravital fluorescence microscopy

Mice were anesthetized with midazolame, medetomidine, and fentanyl injected intraperitoneally as described previously. Animals were fixed on a custom-built stage to maintain a physiological temperature. The right and left common carotid arteries were dissected free. For the quantification of leukocyte adhesion, 100 μl of acridine orange (0.5 mg ml−1; Sigma-Aldrich) was injected via jugular vein catheter (inside diameter, 0.28 mm; outer diameter, 0.61 mm; Smiths Medical Deutschland GmbH) to stain circulating leukocytes in vivo. Measurements were performed with a high-speed wide-field Olympus BX51WI fluorescence microscope using a long-distance condenser and a 10× (numerical aperture, 0.3) water immersion objective with a monochromator (MT 20E; Olympus Deutschland GmbH) and a charge-coupled device camera (ORCA-R2, Hamamatsu Photonics). For image acquisition and analysis, a Realtime Imaging System eXcellence RT (Olympus Deutschland GmbH) software was used. Cell recruitment was quantified in four fields of view (100 × 150 μm) per carotid artery. Adherent cells were defined in each vessel segment as cells that did not move or detach from the endothelial lining within an observation period of 10 s and presented per square millimeter.

The FACTO-RR study

We enrolled 71 all-comer patients with arterial hypertension admitted to either the outpatient clinic or the emergency room of the Center for Cardiology, University Medical Center Mainz into the “Interaction of coagulation FACtors, Thrombocytes and leukOcytes in patients with aRterial hypeRtension study” (FACTO-RR study, DRKS00011232). Inclusion criteria were age (≥18 years old) and history of essential arterial hypertension of at least 6 months. Exclusion criteria were age (≥90 years old), anticoagulant therapy (unfractionated or low–molecular weight heparin, vitamin K antagonist, FXa inhibitors, and thrombin inhibitors), acute coronary syndromes, exacerbated disease requiring critical care medicine (for example, hypertensive crisis with pulmonary edema and respiratory failure), or immediate surgery (for example, aortic dissection or rupture). The study protocol was approved by the local ethics committee of the state of Rhineland-Palatinate, Germany, reference number 837.354.12 (8455-F). Written informed consent was obtained from every participant. Medical history was taken, and body weight, height, and heart rate were obtained. Office blood pressure was measured in each study participant in an upright sitting position (room temperature, 22°C) using the Omron 705CP-II device, and upper arm cuffs were adapted to upper arm circumference (17 to 22 cm, 22 to 32 cm, and 32 to 42 cm, respectively). The upper part of the body was undressed, and measurements were taken beginning 5 min after application of the blood pressure cuff. The first measurement was taken on both the left and right arm, and both second and third measurements were taken 3 min after another. If in the first measurement differences between systolic and diastolic blood pressure between the two arms were below 20 or 15 mmHg, the second and third measurements were taken only on the left upper arm. For each individual, the mean of the second and third measurements was calculated and included in the study. According to current guidelines (36), grade I hypertension was defined as systolic blood pressure (SP) of 140 to 159 mmHg and/or diastolic blood pressure (DP) of 90 to 99 mmHg; grade II hypertension or higher was defined as SP ≥160 mmHg and/or DP ≥100 mmHg. MAP was calculated as follows: MAP = ((2 × SP) + DP)/3. Here, the groups hypertension grade I and hypertension grade II or higher were called “uncontrolled hypertension”; all others were termed “controlled hypertension.” A maximum of 10 ml of venouscitrated blood was drawn from the cubital vein of the right arm to prepare PPP and PRP and to assess blood count using the Sysmex XP300 analyzer (Sysmex Corporation). In a subset of the study participants, in which preparation of PRP exceeded a minimum volume of greater than 3 ml after the TG assays were performed, as described previously (31, 5456), we prepared washed platelets by centrifuging the PRP at 2500g for 4 min at room temperature. We resuspended the platelets in aliquots of control PPP that we had obtained from six healthy volunteers (mean age, 28.5 ± 1; 50% female; no cardiovascular risk factors), pooled, aliquoted, frozen at −80°C, and thawed. In those samples, we performed the TG assays as described previously for PRP.


Data are expressed as means ± SEM. Statistical calculations were performed with GraphPad Prism 5 (GraphPad Software Inc.). D’Agostino-Pearson normality test was first performed, and Pearson’s correlation, Fisher’s exact test, Mann-Whitney test, paired or unpaired t test, Kruskal-Wallis test, and one- or two-way ANOVA with post hoc Bonferroni’s or Dunn’s multiple comparison test were used as appropriate. P values of <0.05 were considered significant and marked by asterisks (*P < 0.05; **P < 0.01; ***P < 0.001).


Fig. S1. Therapeutic targeting of FXI protects from vascular inflammation.

Fig. S2. Protein expression of gp91phox and HO-1.

Video S1. C57BL/6.

Video S2. C57BL/6 + ATII.

Video S3. C57BL/6 + ATII + control IgG.

Video S4. C57BL/6 + ATII + anti–VCAM-1 antibody.

Video S5. C57BL/6 + Scr ASO.

Video S6. C57BL/6 + ATII + Scr ASO.

Video S7. C57BL/6 + FXI ASO.

Video S8. C57BL/6 + ATII + FXI ASO.


  1. Acknowledgments: We acknowledge the expert technical assistance of A. Conrad, K. Perius, K. Schwierczek, and J. Schreiner. This work contains results that are part of the doctoral theses of M. Ehlken and Y. Weihert. We acknowledge D. Gailani (Vanderbilt University, Nashville) for providing the FXI−/− mice. We thank D. Sollinger and J. Lutz (1st Medical Clinic, Department of Nephrology, University Medical Center Mainz) for expert assistance in analyzing the kidney histology. Funding: This work was supported by a grant from the Stiftung Pathobiochemie und Molekulare Diagnostik to P.W. and K.L. and grants from the Federal Ministry of Education and Research (BMBF 01EO1003 and 01EO1503) to P.W., S.K., M.K., M.B., T.M., C.R., K.J., S.J., U.W., and W.R. T.R. acknowledges funding by the German Research Society (SFB841, TP B8) and a European Research Council grant (ERC-StG-2012-311575_F-12). A.G. was supported by NIH grants HL 128016, 106919, and 101972. Z.M.R. is supported by NIH grant HL-42846. Z.M.R. and W.R. are funded by NIH grant HL 31950. W.R. is funded by the Alexander von Humboldt Foundation. In conducting the study, P.W. was supported by further funds of the German Research Foundation (DFG WE 4361/3-1 and WE 4361/4-1), the Stiftung Mainzer Herz, and the Center for Translational Vascular Biology, University Medical Center Mainz. Author contributions: S.K. and J.L. performed experiments; collected, analyzed, and discussed data; performed statistical analysis; and contributed to writing the manuscript. S.J., K.J., M.E., T.S., M.K., M.B., N.X., H.L., S.H.K., Y.W., and M.O. performed experiments and collected, analyzed, and discussed data. A.D., C.R., U.W., and T.R. conceived experiments and discussed results and strategy. K.L. discussed results and strategy. A.G. and B.M. made essential methodological contributions. Z.M.R. contributed essential mouse lines and discussed results and strategy. W.R. and T.M. conceived experiments, analyzed data, discussed results and strategy, and contributed to writing the manuscript. P.W. managed and designed the study, conceived and performed experiments, analyzed data, performed statistical analysis, discussed results and strategy, and wrote the manuscript, which was revised and approved by all authors. Competing interests: B.M. is an employee and shareholder of Ionis Pharmaceuticals Inc. A.G. and Oregon Health and Science University (OHSU) are equity holders in Aronora Inc. and may have financial interest in the findings of this research. W.R. reports paid consulting relationships with Iconic Therapeutics Inc. A.G. is the inventor on U.S. patents 8236316, 8388959, 8399648, 8940883, and 9125895, which are owned by OHSU and cover the FXI inhibitor 14E11 used in this work. Data and materials availability: The data for this study have been deposited at smb://c5-s4/cth$. The FACTO-RR study has been registered at the Deutsches Register für Klinische Studien (DRKS00011232). FXI ASOs and respective controls ASOs are available from Ionis Pharmaceuticals Inc.

Stay Connected to Science Translational Medicine

Navigate This Article