Genetic deletion of platelet PAR4 results in reduced thrombosis and impaired hemostatic plug stability

Abstract

Background:

Protease-activated receptor 4 (PAR4) is expressed by a wide variety of cells, including megakaryocytes/platelets, immune cells, cardiomyocytes and lung epithelial cells. It is the only functional thrombin receptor on murine platelets. A global deficiency of PAR4 is associated with impaired hemostasis and reduced thrombosis.

Objective:

Generation of a mouse line with a megakaryocyte/platelet-specific deletion of PAR4 (PAR4^fl/fl;PF4^Cre+). Use the mouse line to investigate the role of platelet PAR4 in hemostasis and thrombosis in mice.

Methods:

Platelets from PAR4^fl/fl;PF4^Cre+ were characterized in vitro. Arterial and venous thrombosis was analyzed. Hemostatic plug formation was analyzed using a saphenous vein laser injury model in global PAR4 and megakaryocyte/platelet-specific deletion of PAR4, wild-type mice treated with thrombin or GPVI inhibitors.

Results:

PAR4^fl/fl;PF4^Cre+ platelets were unresponsive to thrombin or specific PAR4 stimulation but not to other agonists. PAR4^−/− and PAR4^fl/fl;PF4^Cre+ mice both exhibited a similar reduction in arterial thrombosis compared to their respective controls. More importantly, we show for the first time that platelet PAR4 is critical for venous thrombosis in mice. In addition, PAR4^−/− mice and PAR4^fl/fl;PF4^Cre+ mice exhibited a similar impairment in hemostatic plug stability in a saphenous vein laser injury model. Inhibition of thrombin in wild-type mice gave a similar phenotype. Combined PAR4 deficiency on platelets with GPVI inhibition did not impair hemostatic plug formation but further reduced plug stability.

Conclusion:

We generated a novel PAR4^fl/fl;PF4^Cre+ mouse line. We used this mouse line to show that PAR4 signaling in platelets is critical for arterial and venous thrombosis and hemostatic plug stability.

Introduction

Protease-activated receptor 4 (PAR4) has previously been studied in the context of thrombin-mediated platelet activation in humans and mice.^{1, 2} Human platelets express PAR1 and PAR4 whereas mouse platelets PAR3 and PAR4. Both, human PAR1 and PAR4 are druggable targets to reduce pathologic platelet activation in cardiovascular disease.^{1, 3} Since PAR4 on human platelets responds to higher thrombin concentrations, it is thought to contribute to pathologic thrombus formation but is not essential for platelet-mediated hemostasis in humans.¹ Currently, only global PAR4-deficient mice are available to investigate the role of PAR4 in vivo.⁴ PAR4 deficiency in mice results in platelets that are unresponsive to thrombin but respond normally to thromboxane receptor and glycoprotein (GP) VI stimulation.⁴ In addition, PAR4^−/− mice exhibited prolonged bleeding times and increased blood loss in a tail transection model compared to the PAR4^+/+ controls.^{4, 5} Moreover, PAR4^−/− mice or PAR4 inhibition in mice was associated with reduced thrombosis in mouse models of arterial thrombosis.^4–9

In addition to platelets, PAR4 is expressed by a variety of cells, including smooth muscle cells, endothelial cells (ECs), cardiomyocytes, immune cells, neuronal cells and lung epithelial cells.^10–16 Studies by us and others suggested that non-platelet PAR4 is protective in mouse models of myocardial infarction and virus infection.^{11, 17, 18} Thus, studies in PAR4^−/− mice have to be interpreted with caution as impaired function of more than one cell type may be responsible for the observed phenotype.

Here, we describe the generation of PAR4 floxed (PAR4^fl/fl) mice. These mice were crossed with mice expressing Cre recombinase under the platelet factor 4 (PF4) promoter to selectively delete PAR4 in megakaryocytes/platelets (PAR4^fl/fl;PF4^Cre+). Platelet function, hemostasis, and arterial and venous thrombosis were impaired in PAR4^fl/fl;PF4^Cre+ and PAR4^−/− mice, demonstrating that PAR4 on platelets is critical in hemostasis and thrombosis.

Methods

Mice

To generate mice with megakaryocyte/platelet-specific deletion of PAR4, female PAR4^fl/fl mice were crossed with male PAR4^fl/fl mice expressing the Cre recombinase under control of the PF4 promoter to generate PAR4^fl/fl;PF4^Cre+ mice.¹⁹ PAR4^−/−, PAR4^+/+ and wild-type (WT) mice on a C57Bl/6J background were also used.⁴ Adult mice (8–12 weeks old) of both sexes (equally distributed) were used for the study if not indicated otherwise. All animal experiments were performed in accordance with the guidelines of the animal care and use committee of the University of North Carolina at Chapel Hill and complies with National Institutes of Health guidelines.

Platelet function analysis

Venous blood was collected into low molecular weight heparin (Lovenox, Sanofi, France).²⁰ Platelet counts were determined in whole blood by BD Accuri C6 flow cytometer (BD Biosciences, San Jose, CA) using anti-GPIX-AlexaFluor488 (clone Xia.B4, Emfret Analytics, Würzburg, Germany).²⁰ α-granule release (anti-P-selectin-AlexaFluor647, clone RB40.34, BD Biosciences) and αIIbβ3 integrin activation (JON/A-PE, Emfret Analytics) in the absence or presence of agonists (PAR4 agonist peptide [PAR4 AP, 800μM], ADP [10 μM], convulxin [Cvx, 100 ng/mL]) was analyzed.²⁰ For aggregometry studies, washed platelets (for α-thrombin [Enzyme Research Laboratories, South Bend, IN], PAR4-AP and Cvx) or platelet-rich plasma (for ADP) were resuspended in modified Tyrode’s buffer at a final concentration of 2.5 × 10⁸ platelets/mL in a Chrono-log 4-channel optical aggregation system (Chrono-log, Havertown, PA).^{20, 21}

Mouse arterial thrombosis model

Male mice were anesthetized with 2% isoflurane. Mice were placed in a supine position and the right carotid artery was isolated by blunt dissection.⁷ A one mm² piece of filter paper was soaked with FeCl₃ (8%) and applied to ventral surface of the exposed carotid artery for 3 min.²² A microvascular ultrasonic flow probe (MA0.5PSB, Transonic Systems Inc, Ithaca, NY, USA) was used to detect blood flow for 30 min after removal of the filter paper. Occlusion was defined as a cessation of blood flow (0 ml/min) for ≥2 min.^{5, 7}

Mouse venous thrombosis model

Male mice were anesthetized with isoflurane (1.5–3%) and body temperature maintained using a heated pad for the duration of the surgery procedure. All following procedures were performed under aseptic conditions. The abdomen was opened by midline laparotomy and the bowel externalized and wrapped in wetted gauze. The inferior vena cava (IVC) was stenosed as described.²³ At 48 hours post-stenosis induction, the IVC was inspected for the presence of thrombus with any resultant thrombus removed from the IVC and weighed using a microbalance.

Saphenous vein laser injury hemostasis model

Mice were anesthetized and the saphenous vein was exposed.²⁴ Mice were injected (intravenously, i.v.) with 2.5 μg of an anti-GPIX-AlexaFluor488 antibody to label platelets, and 2 μg of an anti-fibrin-AlexaFluor647 antibody to visualize fibrin formation. To test the effects of recombinant hirudin, mice were injected i.v. with lepirudin (resuspended in sterile saline, 50 mg/kg) immediately before laser injury. To test the effects of the direct thrombin inhibitor dabigatran etexilate, mice were fed custom-made AIN-93M chow (Dyets Inc., Bethlehem, Pa., USA) with peanut flavoring (2 g/kg chow) with or without dabigatran etexilate (10 mg/g chow) for 5 days.^25–27 To test the effects of inhibition of platelet GPVI signaling, mice were injected i.p. with ibrutinib (PCI-32765, SelleckChem, 12.5 mg/kg) 1 hour prior to experiment. The saphenous vein was injured by 3 532-nm laser pulses at 0.5, 5.5 and 10.5 mins using Ablate! photoablation system (Intelligent Imaging Innovations, Denver, CO, USA).²⁴ The 1st pulse generated an initial injury, and the 2^nd and 3^rd pulses reopened the existing platelet plug. If active bleeding was occurring at the injury site at the time of the next laser pulse, the pulse was omitted. Fluorescence images were captured on a Zeiss Axio Examiner Z1 microscope (Zeiss, Jena, Germany) with multicolor LED light source (Lumencor, Beaverton, OR, USA) and an Orca Flash 4.0 camera (Hammamatsu, Bridgewater, NJ, USA) for a total of 15.5 mins.²⁴ Data were analyzed with SLIDEBOOK 6.0 (Intelligent Imaging Innovations).

Statistics

Statistical analyses were performed with GraphPad Prism 9.2 (GraphPad Software Inc., La Jolla, CA). All data are represented as mean ± SEM. For 2-group comparison, the 2-tailed Student’s t test was used for normally distributed data, and the Mann Whitney test for non-normally distributed data. For multiple-group comparison, normally distributed data were analyzed by 1-way or 2-way ANOVA with Bonferroni correction for repeated measures. P-value ≤ 0.05 was regarded as significant.

Results

Generation of conditional PAR4 (F2lr3) gene expressing mice

A PAR4 (F2rl3) floxed allele was generated in C57BL/6J mice by CRISPR-mediated genome editing using the UNC Animal Models Core. GuideScan and CCTop websites were used to identify potential Cas9 guide RNAs targeting the 5’ and 3’ flanking regions of the F2lr3 gene.^{28, 29} The entire F2rl3 gene was flanked with loxP sites, with the 5’ loxP site inserted ~1 kb upstream of exon 1 and the 3’ loxP site inserted ~1.6 kb 3’ downstream of exon 2. The floxed allele was generated by microinjecting embryos with Cas9 protein (400 nM), guide RNAs (25 ng/μL) targeting the two loxP insertion sites and a circular donor vector (20 ng/μL) designed to insert a loxP and a flippase (Flp) recognition target (FRT) site at the 5’ guide RNA cut site and a loxP site at the 3’ guide RNA cut site. Injected embryos were implanted in pseudo-pregnant B6D2F1/J recipient females. We obtained two founder animals with multicopy tandem integration of the donor vector at the target locus. The founders were mated to animals with a C57BL/6J-Rosa26-Flp recombinase transgene. The presence of a single FRT site in each copy of the knock-in event in a tandem array allows the array to be reduced to a single copy by Flp recombinase expression. F1 animals harboring the tandem integration conditional allele and the Flp transgene were subsequently mated to WT C57BL/6J animals (#000664, Jackson Laboratory, Bar Harbor, ME, USA) to obtain animals with a single-copy conditional allele without the Flp transgene.

Generation of PAR4^fl/fl,PF4^Cre+ mice

We generated PAR4^fl/fl;PF4^Cre+ mice in 3 steps: 1/ female PAR4^fl/fl mice were crossed with male WT mice expressing PF4 promotor driven Cre recombinase, 2/ female PAR4^+/fl were crossed with male PAR4^+/fl;PF4^Cre+ and 3/ female PAR4^fl/fl were crossed with male PAR4^fl/fl;PF4^Cre+.

Characterization of platelets from PAR4^fl/fl;PF4^Cre+ mice

Circulating platelet counts were similar between controls (PAR4^fl/fl) and PAR4^fl/fl;PF4^Cre+ littermates (Figure 1A). This is consistent with data from global PAR4^−/− mice that loss of PAR4 does not affect platelet count.⁴ Next, we assessed markers of platelet activation in whole blood by flow cytometry using antibodies recognizing the active conformation of αIIbβ3 integrin (JON/A-PE)³⁰ and P-selectin, a marker of α-granule secretion. No differences were observed in resting platelets (Figure 1B, C) from either genotype. As expected, PAR4^fl/fl;PF4^Cre+ platelets were unresponsive to thrombin or PAR4 activating peptide (PAR4-AP) (Figure 1B, C). However, no differences in integrin activation (Figure 1B) were observed between control and PAR4^fl/fl;PF4^Cre+ platelets when stimulated via P2Y₁/P2Y₁₂ (ADP) or GPVI (convulxin, Cvx), respectively. A small but significant increase in surface P-selectin expression was observed in PAR4^fl/fl;PF4^Cre+ platelets activated with Cvx compared to control platelets (Figure 1C). Next, platelet function was assessed by light transmission aggregometry. PAR4^fl/fl;PF4^Cre+ platelets were unresponsive to increasing concentrations of α-thrombin and high dose PAR4-AP but aggregated normally in response to ADP or Cvx when compared to control platelets (Figure 1D). These results demonstrate that platelets from PAR4^fl/fl;PF4^Cre+ mice are selectively unresponsive to stimulation via the PAR4 receptor in vitro.

Figure 1: Complete absence of activation response to PAR4 receptor stimulation in PAR4^fl/fl;PF4^Cre+ platelets.

(A) Platelet counts determined by flow cytometry in whole blood (n=3–4). (B,C) Flow cytometric analysis of αIIbβ3 integrin activation [(B); JON/A-PE] and α-granule secretion [(C); anti-P-selectin]. Diluted whole blood samples from PAR4^fl/fl;PF4^Cre+ and PAR4^fl/fl littermates were incubated for 10 minutes in the presence of antibodies, agonist (800 μM PAR4 activating peptide (PAR4-AP), 10 μM ADP, 100 ng/ml Cvx, PBS for resting controls) and Ca²⁺, before being further diluted in PBS for analysis of mean fluorescence intensity (MFI) by flow cytometry. Platelets were gated by GPIX expression. Data shown as mean ± SEM (n=3–4). (D) Aggregation responses of washed platelets (α-thrombin, PAR4-AP, Cvx) or PRP (ADP) from PAR4^fl/fl;PF4^Cre+ and PAR4^fl/fl littermates, representative of at least 2 experiments. Statistical significance was determined by unpaired t-test (A) or ANOVA with Bonferroni post-test (B,C).

Platelet PAR4 contributes to FeCl₃-induced arterial thrombosis.

Global PAR4 deficiency was shown to be associated with reduced thrombosis in a FeCl₃-induced carotid artery thrombosis model in mice.^{5, 7} Here, we confirmed the findings that global PAR4 deficient mice had prolonged mean time of carotid artery occlusion of 22.6 vs. 6.7 min (Figure 2A) after FeCl₃ application. Next, mice with megakaryocytes/platelets PAR4 deletion were subjected to the arterial thrombosis model and we showed that platelet PAR4 contributes to occlusive thrombus formation. The mean occlusion time of the carotid artery was in PAR4^fl/fl;PF4^Cre+ longer compared to the PAR4^fl/fl control mice with 25.1 vs 11.5 min (Figure 2B).

Figure 2: Effect of a global PAR4 deficiency or megakaryocyte/platelet PAR4 deficiency on arterial thrombosis.

Mice with a global or megakaryocyte/platelet specific PAR4 deletion were subjected to a ferric chloride carotid artery thrombosis model. A) Mean time to occlusion in PAR4^+/+ and PAR4^−/− mice and B) PAR4^fl/fl and PAR4^fl/fl;PF4^Cre+ mice after 8% FeCl₃ for 3 min. Occlusion was defined as no blood flow for ≥2 min over a observational period of 30 min. One PAR4^+/+ mouse reopened approximal 3 min after our defined occlusion of no blood flow for 2 min. Data shown as mean. **P<0.01, ****P<0.0001. Statistical significance was determined by Mann-Whitney U test.

Platelet PAR4 contributes to venous thrombosis.

Platelets have been shown to contribute to venous thrombosis in the murine IVC stenosis model.³¹ Here, we investigated the effect of a global PAR4 deficiency or megakaryocytes/platelets PAR4 deficiency on venous thrombosis. PAR4^−/− mice exhibited reduced thrombus weight 48 hours after stenosis surgery compared to the WT controls (1.93±1.23mg vs. 12.29±2.57mg, Figure 3A). Global PAR4 deficiency was further associated with reduced overall thrombus incidence compared to the WT controls (Figure 3B). Similar data were obtained from PAR4^fl/fl;PF4^Cre+ mice which had thrombus weight of 0.40±0.31mg and the PAR4^fl/fl control of 6.89±2.06mg (Figure 3C). PAR4^fl/fl;PF4^Cre+ mice also had reduced thrombus incidence compared to the PAR4^fl/fl controls (Figure 3D).

Figure 3: Effect of a global PAR4 deficiency or and megakaryocyte/platelet PAR4 deficiency on venous thrombosis.

Mice with a global or megakaryocyte/platelet specific PAR4 deletion were subjected to IVC stenosis thrombosis model. Thrombus weight (A, C) and incidence of venous thrombus formation (B, D) in PAR4^+/+ and PAR4^−/− and PAR4^fl/fl and PAR4^fl/fl;PF4^Cre+ mice, respectively, at 48 hours postinduction. Data shown as mean. *P<0.05, **P<0.01. Statistical significance was determined by Mann-Whitney U test.

Effect of global or platelet-specific PAR4 deficiency on hemostatic plug formation in mice.

The saphenous vein laser injury model^{24, 32} was used to determine bleeding times and quantify platelet and fibrin accumulation during hemostatic plug formation in PAR4 mutant mice. First, we assessed hemostasis in mice with a global PAR4 deficiency. Bar graph representation of bleeding times shows that WT littermates and PAR4^−/− mice rapidly formed a hemostatic plug after injury (Figure 4A). All injuries in WT mice (7/7) formed a stable hemostatic plug after each successive injury (Figure 4A). In contrast, injuries in PAR4^−/− mice exhibited 2 distinct phenotypes; one third of the injury sites (4/12) formed a stable hemostatic plug after each successive laser ablation, like the WT controls, whereas the majority of the PAR4^−/− injury sites (8/12) formed unstable hemostatic plugs that re-opened several minutes after the first ablation leading to extended bleeding (Figure 4A). When bleeding times were calculated, WT littermates exhibited median bleeding times of 16, 20 and 20 secs after ablation (“shot”) #1, 2 and 3, respectively (Figure 4B). PAR4^−/− mice demonstrated severely prolonged bleeding (median bleeding time 731 secs after shot #1) despite rapidly forming an initial hemostatic plug (Figure 4C). Injury sites with prolonged bleeding (“bleeding injuries”) also had reduced platelet accumulation beginning around the time of hemostatic plug re-opening (~180 sec post-ablation; Figure 4D). Fibrin formation was comparable to PAR4^+/+ controls in PAR4^−/− “hemostatic injuries”, and also comparable at PAR4^−/− “bleeding injuries” after shot #1 (Figure 4E). Epifluorescence intravital microscopy still-frames from injury sites ~4 mins post-ablation are shown for PAR4^+/+ and PAR4^−/− mice demonstrating a stable platelet plug in the PAR4^+/+ and PAR4^−/− hemostatic injury, and active bleeding in the PAR4^−/− bleeding injury (Figure 4F). Corresponding movies for still-frames in figures 4–7 are available in the online supplement.

Figure 4: Impaired hemostasis in large subset of global PAR4-deficient mice.

(A,B,C) Hemostasis was monitored by intravital microscopy following laser ablation of the saphenous vein. (A) Bleeding duration at individual injury sites represented in bar graph form. After 5 and 10 minutes, if hemostasis occurred, the platelet plug was disrupted by subsequent laser ablations (arrows/X’s; as in PAR4^+/+ and PAR4^−/− hemostatic injuries). If bleeding was ongoing at the 5 and/or 10 minute timepoint, subsequent laser ablation was omitted (as in PAR4^−/− bleeding injuries). Bleeding time was quantified as total bleeding time for each ablation (“shot”) (B) or time to formation of the first platelet plug (C). Data shown is for 7 injuries from 3 PAR4^+/+ mice and 12 injuries (4 hemostatic injuries, 8 bleeding injuries) from 4 PAR4^−/− mice. (D) Sum fluorescence intensity for platelets (anti-GPIX-AlexaFluor488) and (E) fibrin (anti-fibrin-AlexaFluor647) was quantified from intravital microscopy videos. PAR4^−/− intensity data was segregated between hemostatic and bleeding injuries. Data shown as mean. (F) Still frames from intravital microscopy videos at t ~4 minutes showing hemostasis at a PAR4^+/+ injury and PAR4^−/− hemostatic and bleeding injuries. Single channels in greyscale are shown for platelets and fibrin with the colored merged image on the far right. Scale bar represents 25 μm. Statistical significance was determined by Mann-Whitney U test for each shot (B) or unpaired t-test (C).

Figure 7: Further impaired hemostasis in PAR4^fl/fl;PF4^Cre+ mice treated with ibrutinib.

(A) Bleeding duration at individual injury sites represented in bar graph form. Bleeding time was quantified as total bleeding time for each ablation (“shot”) (B) or time to formation of the first platelet plug (C). Data shown is for 8 injuries from 4 PAR4^fl/fl ibrutinib-treated mice and 7 injuries from 3 PAR4^fl/fl;PF4^Cre+ ibrutinib-treated mice. (D) Sum fluorescence intensity for platelets (anti-GPIX-AlexaFluor488) and (E) fibrin (anti-fibrin-AlexaFluor647) was quantified from intravital microscopy videos. Data shown as mean. (F) Still frames from intravital microscopy videos at t ~4 minutes showing hemostasis at a PAR4^fl/fl injury and bleeding at a PAR4^fl/fl;PF4^Cre+ injury. Single channels in greyscale are shown for platelets and fibrin with the colored merged image on the far right. Scale bar represents 25 μm. Statistical significance was determined by Mann-Whitney U test for shot1 (B) or unpaired t-test (C).

Next, we subjected PAR4^fl/fl;PF4^Cre+ mice and their littermate controls (PAR4^fl/fl) to the laser injury hemostasis model. As seen for PAR4^+/+ mice, PAR4^fl/fl mice formed a stable hemostatic plug after each successive laser ablation (Figure 5A) with median bleeding times of 12, 21 and 22 secs after shot #1, 2 and 3, respectively (Figure 5B). In contrast, almost all of the injury sites in PAR4^fl/fl;PF4^Cre+ mice (7/8) demonstrated prolonged bleeding times (median bleeding time 737 sec) (Figures 5A, B), despite time to initial hemostasis being similar to that in PAR4^fl/fl mice (Figure 5C). Moreover, platelet accumulation after re-opening was significantly reduced at PAR4^fl/fl;PF4^Cre+ injuries (~150 sec post-ablation), while fibrin formation was similar to the control injuries after shot #1 (Figure 5E). Representative still-frames and movies from intravital video recordings show on-going bleeding at the injury site in a PAR4^fl/fl;PF4^Cre+ mouse and stable hemostatic plug formation in a PAR4^fl/fl control mouse (Figure 5F and online supplement).

Figure 5: Impaired hemostasis in mice with megakaryocyte/platelet-specific deficiency in PAR4.

(A) Bleeding duration at individual injury sites represented in bar graph form. Bleeding time was quantified as total bleeding time for each ablation (“shot”) (B) or time to formation of the first platelet plug (C). Data shown is for 6 injuries from 3 PAR4^fl/fl mice and 8 injuries from 3 PAR4^fl/fl;PF4^Cre+ mice. (D) Sum fluorescence intensity for platelets (anti-GPIX-AlexaFluor488) and (E) fibrin (anti-fibrin-AlexaFluor647) was quantified from intravital microscopy videos. Data shown as mean + SEM. (F) Still frames from intravital microscopy videos at t ~4 minutes showing hemostasis at a PAR4^fl/fl injury and bleeding at a PAR4^fl/fl;PF4^Cre+ injury. Single channels in greyscale are shown for platelets and fibrin with the colored merged image on the far right. Scale bar represents 25 μm. Statistical significance was determined by Mann-Whitney U test for shot1 (B) or unpaired t-test (C).

Our data demonstrate that PAR4 signaling is required for hemostatic plug stability following a small penetrating injury to blood vessels in mice, and that the bleeding phenotype is comparable between mice with either global or megakaryocyte/platelet-specific PAR4 deficiency.

Interestingly, our results showed that platelet PAR4 was not required for initial hemostatic plug formation but was required for maintaining hemostatic plug stability.

Effect of thrombin inhibition on hemostatic plug formation in mice

Recently, we reported reduced expression of tissue factor (TF) in TF^low mice was associated with reduced platelet accumulation and fibrin generation but minimally impacted bleeding in the saphenous vein laser injury model.²⁴ Because complete loss of TF^33–35 or prothrombin^{36, 37} is embryonic lethal, we used a pharmacological approach to inhibit thrombin activity by treating mice with either dabigatran etexilate or recombinant hirudin. Thrombin inhibition prior to the saphenous vein injury was performed to test if the observed phenotype in global PAR4 and megakaryocyte/platelet deficient mice can be reproduced by an inhibition of thrombin.

We first determined if anticoagulation with dabigatran etexilate (10 mg/g chow)^25–27 influenced hemostatic plug formation. Dabigatran-fed WT mice had relatively normal bleeding times (Figure 6A–C) and platelet accumulation (Figure 6D), and reduced fibrin generation after shot #1 although fibrin intensity increased following subsequent injuries (Figure 6D). We then determined if more robust thrombin inhibition with i.v. recombinant hirudin would prolong bleeding to a greater extent. In WT mice injected with hirudin, only a third of the injury sites (3/9) demonstrated severely prolonged bleeding, and platelet accumulation was comparable to injury sites in placebo-treated mice despite the near complete loss of fibrin generation at sites of ablation (Figure 6A–D). Representative still frames and movies from intravital video recordings show comparable hemostatic plug formation and reduced or absent fibrin generation at the injury site of a dabigatran- or hirudin-treated mouse, respectively, compared to placebo-treated mice (Figure 6F and online supplement).

Figure 6: Partially impaired hemostasis in mice treated with thrombin inhibitors.

(A) Bleeding duration at individual injury sites represented in bar graph form. Bleeding time was quantified as total bleeding time for each ablation (“shot”) (B) or time to formation of the first platelet plug (C). Data shown is for 5 injuries from 2 placebo-treated mice, 11 injuries from 4 dabigatran etexilate chow-fed mice and 9 injuries from 4 hirudin-injected mice. (D) Sum fluorescence intensity for platelets (anti-GPIX-AlexaFluor488) and (E) fibrin (anti-fibrin-AlexaFluor647) was quantified from intravital microscopy videos. Data shown as mean. (F) Still frames from intravital microscopy videos ~4 minutes showing hemostatic plug formation in all mice and absent and reduced fibrin formation in hirudin- and dabigatran-treated mice, respectively. Single channels in greyscale are shown for platelets and fibrin with the colored merged image on the far right. Scale bar represents 25 μm. Statistical significance was determined by 2-way (B) or 1-way (C) ANOVA with Bonferroni post-test.

The data show that direct inhibition of thrombin, the main PAR4 activating protease, was able to partly reproduce the finding observed with the genetic approach of platelet-specific deletion of PAR4.

Effect of combined loss of PAR4/GPVI signaling on hemostatic plug formation

The ability of PAR4-deficient mice to rapidly form an initial hemostatic plug suggests that other ligand/receptor partners can mediate initial platelet plug formation after small vessel injuries. We have previously demonstrated that mice lacking GPIbα binding to von Willebrand Factor (vWF) have severely prolonged initial hemostatic plug formation.²⁴ The collagen receptor GPVI is an additional candidate to mediate rapid platelet adhesion and activation. We thus treated PAR4^fl/fl;PF4^Cre+ mice or littermate controls with the Bruton’s tyrosine kinase (Btk) inhibitor, ibrutinib, which inhibits signaling downstream of GPVI (Supplemental Figure 1).³⁸ Ibrutinib-treated PAR4^fl/fl mice demonstrated normal bleeding times (median bleeding times 18, 20, and 18 sec after shots #1, 2 and 3, Figure 7A–C) and platelet/fibrin accumulation (Figure 7D,E), in agreement with historical data demonstrating that GPVI plays a minimal role in hemostatic plug formation.³⁹ However, despite retaining the ability to form an initial hemostatic plug, ibrutinib-treated PAR4^fl/fl;PAR4^Cre+ mice demonstrated faster plug re-opening and longer bleeding times than PAR4^fl/fl;PAR4^Cre+ mice (median bleeding time 876 secs, Figure 7B). Platelet accumulation was also severely impaired although fibrin formation was increased, likely due to excessive bleeding beginning earlier on (Figure 7D,E).²⁴ Still-frames demonstrate stable plug formation in ibrutinib-treated control mice and active bleeding in ibrutinib-treated PAR4^fl/fl;PF4^Cre+ mice (Figure 7F). These results suggest that GPVI activation contributes to early plug stability in the absence of PAR4, but even with loss of both PAR4 and GPVI signaling, platelets are able to transiently bridge small injuries likely via GPIbα/vWF.

Discussion

Here, we present the generation of a mouse carrying a floxed PAR4 allele. Crossing these mice with mice expressing Cre recombinase under control of the PF4 promoter (megakaryocyte/platelet-specific gene)⁴⁰ allowed us to investigate, for the first time, the specific role of platelet PAR4 in arterial and venous thrombosis as well as hemostatic plug formation in mice. We show that platelet PAR4 is critical for arterial thrombus formation. We also showed that a lack of PAR4 on platelets reduced thrombosis in a murine IVC stenosis model. In addition, we found that platelet-specific PAR4 deletion did not impact initial hemostatic platelet plug formation after saphenous vein laser ablation, but the hemostatic plugs were unstable and reopened, leading to significantly prolonged bleeding times. Inhibition of GPVI signaling further impaired plug stability in PAR4^fl/fl;PF4^Cre+ mice, and inhibition of thrombin, the main activator of PAR4 on platelets, reproduced, in part, the bleeding phenotype of mice with global or megakaryocyte/platelet-specific PAR4 deficiency.

The role of PAR4 for thrombin-mediated platelet activation in mice was first shown by Sambrano et al..⁴ In that study, global PAR4 deficiency resulted in platelets unresponsive to thrombin or PAR4-AP, prolonged bleeding times/increased blood loss and partial protection from arterial thrombosis.⁴ Thrombin-mediated platelet activation at sites of vascular injury contributes to thrombosis that can cause myocardial infarction and a subset of stroke.^{41, 42} We confirmed that a lack of thrombin-mediated platelet activation prolonged time to occlusion in an artery injury model. In addition, we showed that specific platelet PAR4 was responsible for the increased arterial thrombosis in the model. Platelets were indicated to contribute to venous thrombosis in mice³¹, we show here that platelet PAR4 is one of the important mediators of venous thrombosis in mice. Global PAR4 deficient mice and mice with a megakaryocyte/platelet PAR4 deficiency showed similar results in the used arterial and venous thrombosis models suggesting that the major contributor for thrombus formation is platelet PAR4.

Using the saphenous vein laser injury model²⁴, which combines a small injury hemostasis model with intravital fluorescence microscopy, we confirmed that PAR4 on platelets is critical for hemostatic plug formation in mice. Owing to the small injury size (~50 μm diameter) and the use of real-time intravital fluorescence microscopy, we are able to detect more subtle roles for platelet receptors and signaling proteins in hemostatic plug formation. In the large injury tail bleed assay, PAR4 deficiency results in injuries that are entirely unable to close.⁴ However, in a small laser injury thrombosis model, PAR4 deficiency resulted in reduced platelet accumulation with no impact on fibrin generation⁶, similar to our findings. Interestingly, we observed that initial plug formation was normal in the absence of platelet PAR4 (a finding also observed by Vandendries et al.⁶), likely mediated by platelet adhesion and activation by GPIbα/vWF and GPVI/collagen interactions. Indeed, mice with platelets lacking the vWF-binding domain of GPIbα had drastically prolonged bleeding times caused by an inability to form an initial hemostatic plug.²⁴ Additionally, treatment of PAR4^fl/fl;PF4^Cre+ mice with the Btk inhibitor ibrutinib, to inhibit GPVI-mediated platelet activation, led to more rapid re-opening of hemostatic plugs and further prolonged bleeding times compared to untreated PAR4^fl/fl;PF4^Cre+ mice. While loss of expression or function of GPVI by itself does not significantly impair hemostasis⁴³, we also observed prolonged bleeding in P2Y₁₂-deficient mice treated with ibrutinib³⁸, suggesting GPVI plays a supporting role in hemostasis but becomes more critical in the absence of other major platelet signaling receptors. Our results also confirm our previous findings²⁴ that in small injuries, platelets are critical for hemostatic plug formation while fibrin plays a less important role, as exemplified by hemostatic injuries in hirudin-treated mice where fibrin was nearly undetectable. Normal hemostasis in dabigatran etexilate-fed mice could be explained at least in part by incomplete inhibition of thrombin activity. The dose used was shown not to cause spontaneous bleeding due to impairment of hemostasis.^{26, 27, 44} In addition, we believe the level of inhibition with hirudin was higher than dabigatran etexilate because it is injected rather than delivered via the chow. Moreover, the partial phenotype in hirudin-treated mice could be explained by a role for additional proteases in PAR4 cleavage such as cathepsin G.⁴⁵ Indeed, neutrophils, a primary source of cathepsin G in blood, are recruited to the hemostatic plug in our model (unpublished observation). Importantly, our data demonstrate that for small injuries, PAR4-mediated platelet activation is not required for initial formation of the hemostatic plug but instead plays a critical role in platelet activation for plug stability. These findings may suggest that the initial TF-dependent thrombin burst plays a lesser role in small injuries and the cell surface-generated second wave of thrombin mediates the majority of PAR4 cleavage.⁴⁶ Finally, we have previously demonstrated that mice lacking the Rap1 GTPase activator CalDAG-GEFI are unable to form a hemostatic plug at least five minutes post-ablation⁴⁷, suggesting platelet activation is required to bridge even small injuries. Therefore, it was surprising to observe initial plug formation with platelets lacking both PAR4 and GPVI signaling. Additional candidates to induce rapid platelet signaling are ADP/P2Y₁₂ and TxA₂/TP, and GPIbα/vWF (or other ligands). In fact, GPIbα is not only critical for platelet tethering but can also induce platelet activation and release of second wave mediators.⁴⁸ The contribution of GPIbα activation signaling during initial plug formation in our model will need to be explored.

While we showed here the effect of PAR4-mediated signaling on mouse platelets and its role in arterial and venous thrombosis and the saphenous vein laser injury model for hemostatic plug formation and stability, the generation of a mouse carrying the floxed PAR4 allele will help to delineate the cell-specific role of PAR4 in different disease models in vivo. For instance, we showed that PAR4 limits RNA virus infection in mice.¹⁷ However, by using PAR4^−/− mice we were unable to distinguish between platelet PAR4, immune cell PAR4 or other resident cell PAR4 in their specific contribution to the anti-viral responses.¹⁷ The here described PAR4^fl/fl mice will be an important tool to investigate those cell-specific roles of PAR4. Others indicated a role for PAR4 in cardiomyocyte, neuron, neutrophil, or regulatory T cell functions.^{11, 12, 14, 18, 49}

In conclusion, we described the generation of PAR4^fl/fl mice. We showed that PAR4 on platelets contribute to both arterial and venous thrombosis in mice. In addition, platelet PAR4 contributes to the stability of the hemostatic plug but not initial hemostatic plug formation in the murine saphenous vein after laser injury. Further, we showed that lack of PAR4 in combination with GPVI inhibition did not change initial hemostatic plug formation but further destabilized the hemostatic plug.

Supplementary Material

fS1

Supplemental Figure 1: Ibrutinib inhibits platelet GPVI signaling. Blood samples were collected from control mice or mice 1 hour after ibrutinib treatment. Platelets in diluted whole blood were activated for 10 mins with Cvx (50 or 200 ng/ml) or PAR4-AP (400 μM) in the presence of JON/A-PE and anti-P-selectin-AlexaFluor647 antibodies, and samples were then diluted in PBS for flow cytometric analysis. Platelets were gated by GPIX expression.

vS1

Supplemental video 1: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a wild-type (PAR4^+/+) mouse. See figure 4A+F, injury #3 in main text.

vS2

Supplemental video 2: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a PAR4^−/− mouse. See figure 4A+F, hemostatic injury #1 in main text.

vS3

Supplemental video 3: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a PAR4^−/− mouse. See figure 4A+F, bleeding injury #7 in main text.

vS4

Supplemental video 4: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a PAR4^fl/fl (control) mouse. See figure 5A+F, injury #6 in main text.

vS5

Supplemental video 5: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a PAR4^fl/fl;PF4^Cre+ mouse. See figure 5A+F, injury #3 in main text.

vS7

Supplemental video 7: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a wild-type mouse treated with dabigatran etexilate chow. See figure 6A+F, injury #6 in main text.

vS6

Supplemental video 6: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a wild-type mouse treated with placebo chow. See figure 5A+F, injury #2 in main text.

vS8

Supplemental video 8: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a wild-type mouse treated with hirudin. See figure 6A+F, injury #7 in main text.

vS10

Supplemental video 10: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a PAR4^fl/fl;PF4^Cre+ mouse treated with ibrutinib. See figure 7A+F, injury #2 in main text.

vS9

Supplemental video 9: Platelet (green) and fibrin (red) accumulation after saphenous vein laser injury of a PAR4^fl/fl (control) mouse treated with ibrutinib. See figure 7A+F, injury #5 in main text.

Essentials.

• Global PAR4 deficiency was shown to reduce thrombosis and hemostasis in mice.

• Generation of a new PAR4^fl/fl;PF4^Cre+ mouse line.

• Platelet PAR4 contributes to arterial and venous thrombosis.

• Platelet PAR4 and GPVI are required to maintain hemostatic plug stability.