Roles of PAD4 and NETosis in Experimental Atherosclerosis and Arterial Injury: Implications for Superficial Erosion

Abstract

Rationale

Neutrophils likely contribute to the thrombotic complications of human atheromata. In particular, neutrophil extracellular traps (NETs) could exacerbate local inflammation and amplify and propagate arterial intimal injury and thrombosis. Peptidyl arginine deiminase 4 (PAD4) participates in NET formation, but understanding of this enzyme’s role in atherothrombosis remains scant.

Objective

This study tested the hypothesis that PAD4 and NETs influence experimental atherogenesis and in processes implicated in superficial erosion, a form of plaque complication we previously associated with NETs.

Methods and Results

Bone marrow chimeric Ldlr deficient mice reconstituted with either wild type or Pad4 deficient cells underwent studies that assessed atheroma formation or procedures designed to probe mechanisms related to superficial erosion. PAD4 deficiency neither retarded fatty streak formation nor reduced plaque size or inflammation in bone marrow chimeric mice that consumed an atherogenic diet. In contrast, either PAD4 deficiency in bone marrow-derived cells, or administration of DNaseI to disrupt NETs, decreased the extent of arterial intimal injury in mice with arterial lesions tailored to recapitulate characteristics of human atheroma complicated by erosion.

Conclusions

These results indicate that PAD4 from bone marrow-derived cells and NETs do not influence chronic experimental atherogenesis, but participate causally in acute thrombotic complications of intimal lesions that recapitulate features of superficial erosion.

INTRODUCTION

Despite current therapies, acute coronary syndromes (ACS) remain one of the leading causes of mortality globally. For decades, pathological studies and fundamental research have focused primarily on the so-called “vulnerable plaque.” Such work has enabled tremendous advances in understanding the mechanisms of formation, development and complication of thin-capped atheromata, a morphology associated with plaque rupture and thrombosis^{1, 2}. Decades of therapies that have successfully lowered exposure to traditional risk factors such as low-density lipoprotein (LDL), hypertension, and smoking may have altered human atheromata, increasing the proportion of ACS caused by superficial erosion³. Lesions associated with superficial erosion in humans exhibit a very distinct morphology from thin-capped atheromata associated with plaque rupture^{4, 5}. In contrast to ruptured plaques, those complicated by erosion exhibit less lipid, macrophage/foam cell accumulation, and more smooth muscle cells (SMC) and collagen in the fibrous cap. Plaques associated with superficial erosion contain abundant SMC and extracellular matrix. Recent data obtained using intravascular optical coherence tomography indicate that up to one-third of ACS in the current era result from erosion rather than rupture⁶.

Neutrophils, key contributors to the acute phase of innate immunity, may contribute to atheromata formation, by promotion of inflammatory monocyte recruitment⁷, and may also participate in lesion evolution and complication. Yet the extent and timing of potential neutrophil involvement in human atherosclerosis remains unsettled^{8, 9}. While experimental atheromata in severely hypercholesterolemic mice contain neutrophils, the paucity of validated selective markers for human neutrophils has rendered their identification in pathological studies difficult. Furthermore, neutrophils may undergo phenotypic changes and modulate into antigen-presenting cells such as dendritic cells or phagocytic cells^{10, 11}. Our recent experiments have implicated neutrophils in processes postulated to pertain to superficial erosion. These results heightened the need to understand more fully mechanisms by which neutrophils might contribute to thrombotic complications of atherosclerosis prompted by superficial erosion. We previously reported that Toll-like receptor (TLR) 2 ligation can activate endothelial cells (ECs) through NFĸB, and potentiate neutrophil recruitment¹². Using a novel approach¹³, we also found that neutrophils play a pivotal role in intimal injury in superficial erosion. Human atheromata harvested from carotid arteries showed localization of cells bearing CD66b, myeloperoxidase (MPO) and neutrophil elastase (NE) near luminal ECs supporting the presence of granulocytes at the intimal surface of lesions that required clinically-indicated endarterectomy.

Activated neutrophils can release their cytoplasmic content and extrude their DNA in a process known as NETosis. Many effector mediators festoon this extracellular neutrophil DNA, including histones, multiple proteinases, and the pro-oxidant enzyme MPO. NETs exposed to blood also gather the potent procoagulant tissue factor. NETosis was originally described as a mechanism of defense against bacteria¹⁴, but also participates in pathological thrombosis, including deep vein thrombosis¹⁵ and atherothrombosis^{16, 17}. Mediators associated with NETs can stimulate inflammatory cells ranging from plasmacytoid dendritic cells¹⁸ to macrophages¹⁹. Recent insight suggests a role of NETs in macrophage activation, potentiating plaque formation in murine atherosclerosis¹⁹.

Yet, few studies have examined directly the effect of NETs on the formation, development, and the complication of atherosclerosis. The enzyme peptidyl arginine deiminase (PAD) 4 participates in NET formation by converting arginyl residues in histones of chromatin to citrulline (which lacks the positive charge of arginine), releasing the ionic bonds that usually constrain nuclear DNA to nucleosomes. Loss of the positive charge on the guanidino group of arginine due to PAD4 action frees the strands of DNA to unfurl and form the “solid state reactor” furnished by NETs. Indeed, NET formation requires PAD4²⁰. Inhibition of PADs with chloramidine prevented NETosis, reduced atheromata burden and retarded neutrophil and monocytes recruitment to arteries²¹. Yet, the lack of specificity of this inhibitor does not permit unambiguous probing of the role of PAD4 and NETs in atherothrombosis. Therefore, we undertook to evaluate the participation of PAD4 and NETs in atherothrombosis using mice with genetic deficiency of PAD4 in blood cells as a more rigorous tool.

This study demonstrates that PAD4 of bone marrow origin and NETs do not influence experimental atherogenesis but are involved in acute thrombotic complications recapitulating features of superficial erosion.

METHODS

Additional data are available on request from the lead author (gregory.franck@inserm.fr).

Human specimen

Plaques from human diseased carotid arteries were harvested during clinically-indicated endarterectomy according to protocols approved by the BWH Human Investigation Review Committee (Boston). Carotid samples were cryo-sectioned and classified according to their morphology (Supplemental Figure I). Human coronary arteries and aortas were harvested according to protocols approved by the INSERM ethic committee. Coronary arteries were isolated from hearts of cardiac transplant recipients with ischemic heart disease or hypertrophic cardiomyopathy. Coronary arteries were classified according to their localization, atheromatous status, nature of the donor, and then stored in a biobank (INSERM U1148, Bichat hospital, Paris, France).

Animals

C57BL/6 WT and Pad4^−/−, Apolipoprotein E (Apoe^−/−) and low-density lipoprotein receptor (Ldlr^−/−) deficient male mice (Jackson Laboratories, USA), 6-12 weeks of age, were housed in the Harvard Medical School Facilities at the New Research Building (Boston, MA, USA). Apoe^−/− mice consumed a normal chow diet while Ldlr^−/− mice consumed a high fat (40% kcal from lipids) and adjusted cholesterol diet (1.25% by weight) (Research Diets, USA), with water ad libitum. Mice were certified free of common pathogens by the suppliers and were monitored by the Harvard Medical Area Standing Committee on Animals. Sample size was determined according to pilot experiments.

Flow-mediated superficial erosion

Atheromata with features associated with superficial erosion were produced as previously described¹³. Briefly, eight-week-old Ldlr^−/− mice underwent electrical current injury of the left common carotid artery (LCCA) using a bipolar microcoagulator (Erbe ICC 200, USA) with a current pulse of 3W. This treatment resulted in reproducible neointima formation (Supplemental Figure IV). Four weeks later, local flow perturbation was induced proximal to the healed injury, using constrictive polyethylene cuffs manufactured to our specifications by 3D-stereolithography printing (Proto Labs, USA) (Supplemental Figure V). The non-constrictive proximal internal diameter (500 μm) of the cuff decreases gradually to become constrictive at its end (distal internal diameter: 250 μm). Previous work formally characterized the flow disturbance produced distal to the cuff using direct measurements and computational fluid dynamics. Animals that underwent a sham procedure (with no cuff placement) and those that had placement of a non-constrictive device served as controls for the surgical manipulations and for adventitial injury and inflammation.

Biological analysis

Flow perturbation continued for 1 or 6 hours and mice were euthanized by CO₂ inhalation and bilateral thoracotomy. After peripheral blood sampling, mice were perfused via left ventricular cannulation with cold PBS, and then fixed with 4% paraformaldehyde (PFA). Distal and proximal LCCA samples were collected and embedded in optimal cutting temperature medium. Six μm cryo-sections were prepared and stained for immunohistochemistry/immunofluorescence investigation.

Immunohistochemistry / Immunofluorescence

Immunostaining of frozen cross sections used a rat monoclonal anti-mouse Ly6G (clone 1A8), rabbit polyclonal anti-citrullinated Histone H3 (H3cit, Abcam) and anti-citrullinated Histone H4 (H4cit, Millipore), a rat monoclonal anti-CD31 (BD Pharmingen), a rabbit polyclonal anti-CD66b (Abcam) or a rabbit polyclonal anti-myeloperoxidase (DAKO). Immunostaining was performed on human coronary artery endothelial cells (HCAEC) using a rabbit anti-C5b-9 complement (Abcam) or a rabbit anti-VE-Cadherin (Abcam). Apoptotic ECs were visualized using fluorescent in situ DNA strand breaks detection kit (TUNEL, Roche). Immunostaining was amplified using peroxidase-conjugated streptavidin complexes (Vector Laboratories) and peroxidase was detected using AEC (Vector Laboratories) substrate. Sections were lightly counterstained with hematoxylin, mounted in gelatin-glycerol and examined with a bright field microscope (Nikon Optiphot-2 equipped with a Nikon digital camera DXM 1200F). For double immunostaining studies, cross sections were incubated with primary antibodies followed by incubation with fluorophore-coupled anti-species antibody (Life Technologies), stained with DAPI, and mounted with fluorescent mounting medium (DAKO). Slides were kept in the dark at 4°C.

H3cit sandwich-ELISA assay

Ninety-six-well plates were coated with appropriate dilutions of commercially available monoclonal mAb 11D1 mouse anti-Histone H3 (Citrullinated R2 + R8 + R17, N-Terminal sequence) (Cayman #17939), washed, and incubated with mouse plasma samples. After washing, the wells were incubated with biotinylated rabbit monoclonal antibody to the C-terminal sequence of Histone H3, followed by incubation with streptavidin covalently conjugated to horseradish peroxidase (HRP). After washing and addition of appropriate fluorogenic or chromogenic HRP substrates, fluorescence or absorbance were determined using a plate reader. A truncated H3cit peptide containing both the N- and C-terminal sequences linked by a poly-GC flexible linker (Supplemental Figure VI) was used as standard.

Endothelial permeability

Endothelial permeability in vivo was investigated by monitoring extravasation of Evans Blue dye (EBD, Sigma). Mice with previously tailored intimas were injected retro-orbitally with 50μl of 7% EBD 6 hours after flow perturbation. After 10 min, mice were euthanized by CO₂ inhalation, perfused by left ventricular apical cannulation with PBS at 4°C supplemented with 10U/ml heparin, and were then fixed by a 10% formalin perfusion. LCCA were embedded in OCT and kept at -80°C for histologic examination. LCCA were harvested en bloc, opened longitudinally from the aortic arch to the bifurcation and stabilized on glass with hardening mounting medium (Vector, Vecta Mount H-5000). Arteries were examined by bright field imaging.

En face microscopy

Mouse vasculature was rinsed with perfusion of cold PBS followed by a fixation with cold 4% PFA. Carotids (n=8 per group) were dissected and opened longitudinally on a Sylgard dish and fixed in 4% PFA for 48-72h. Fixation was quenched with 100 mM glycine (pH 7.4). After extensive washing with PBS, arteries were permeabilized and treated with blocking buffer containing 5% native goat serum, 1% bovine serum albumin (BSA) and 0.05% Triton X-100 during 2 h at room temperature. Primary antibodies were diluted in blocking buffer and incubated with carotid artery specimens for 48 h at 4°C. After washing in PBS and in blocking buffer for 2 hours, the carotids were incubated with fluorophore-conjugated antibodies diluted in blocking buffer for 1 hour at room temperature. Finally, tissues were washed in blocking buffer for 1 h at room temperature, then washed in PBS, mounted with Vectashield HardSet Antifade Mounting Medium containing DAPI for DNA staining (Vector Laboratories) and stored in the dark at room temperature.

Bone marrow transplantation

Recipient Ldlr^−/− male C57B6 mice (6-8-weeks-old) received 1,050 rads of split-dosed lethal irradiation (5.25 Gy, twice) 4 h apart to reduce gastrointestinal toxicity, and antibiotics (sulfamethoxazole and trimethoprim, HiTech Pharmacal) until the end of the experiment. Total bone marrow cells from WT or Pad4^−/− donor mice were prepared from femurs and tibias collected aseptically in RPMI 1640 medium supplemented with 2% FBS, 10 U/ml heparin, penicillin and streptomycin. Bones were punched with a needle at both extremities and centrifuged. Cells were resuspended and washed twice in serum-free RPMI 1640 medium supplemented with 20mM HEPES, penicillin and streptomycin at pH 7.4. Five million total bone marrow cells were injected i.v into each 6-week-old recipient mouse for reconstitution. Mice without bone marrow reconstitution served as sentinels. Mice were monitored until the final experiments, and survival curves were established for each group (Supplemental Figure IIA).

In vitro adhesion experiment

Neutrophils from both WT and Pad4^−/− mice from bone marrow were isolated after flushing whole bone marrow cells and lysing red blood cells, neutrophil granulocytes were isolated by negative selection using a neutrophil isolation kit (Miltenyi Biotec). Isolated neutrophils were then labeled with calcein (Invitrogen) according to manufacturer’s instructions and incubated with resting or TNFα-activated mouse aortic endothelial cells (20 ng/ml; 6h at 37°C) for 15 minutes at 37°C.

Human NET isolation

Human neutrophils were isolated from EDTA-treated fresh blood obtained from Bichat Hospital (Paris, France). NETs were generated by stimulating neutrophils with 50 nM phorbol myristate acetate (PMA) for 4 hours and isolated as previously described with slight modifications²². After 4 hours, neutrophils were gently washed with PBS and NETs were harvested by pipetting. Neutrophils that did not form NETs were discarded. Solid phase DNA was washed and sonicated for 5 min at 37°C to allow for homogenization. DNA quantification was performed using Sytox green nucleic acid labeling (Thermo Fisher Scientific).

Cell culture and incubation with NETs

HCAECs (Lonza) were cultured on plates coated with 100μg/ml type 1 collagen (Thermo Fisher Scientific) in EGM-2 MV (Lonza) supplemented with 15% FBS (Gibco). After reaching confluence, HCAECs were conditioned either in 15% FBS or 15% baby rabbit serum (Cedarlane Corporation), decomplemented or not by treatment at 56°C for 30 minutes, and incubated with 1.5 mg/ml NETs for 6 hours. The supernatants were harvested for detached cells count and the attached cells were washed and fixed in 2% PFA for further immunocytofluorescence analysis.

Image acquisition and morphometric analysis

In situ images were captured digitally using a BX61WI microscope coupled with a Fluoview FV1000 confocal unit (Olympus) equipped with an Olympus DP72 camera and running Fluoview 10-ASW Software (Olympus) for confocal microscopy or with a Nikon Optiphot-2 equipped with a Nikon digital camera DXM 1200F for bright field microscopy. Macroscopic images were captured using the confocal apparatus and Cell Sens software (Olympus). The number of Ly6G-positive cells, CD31-positive cells, CD31⁺TUNEL⁺ cells, endothelium continuity, and EBD coverage were quantified by computer-assisted immunostaining (Image-Pro Plus, Media Cybernetics) using a double-blind randomized approach.

Statistics

Animals were randomly allocated to treatment or control groups. Investigators were blinded to the group allocation during the experiment. Data were analyzed to assess normality of distribution. A non-parametric test was used for skewed data. For normally-distributed variables, data were expressed as mean ± SEM and analyzed by the t-test, Mann-Whitney or Kruskal–Wallis one-way analysis of variance and Dunn’s post hoc test (GraphPad Prism, USA). Differences were considered statistically significant at the p<0.05 level (unadjusted p values). Correlation was estimated with Pearson product-moment.

RESULTS

NETs localize differentially in human carotid plaques with erosion-prone vs. rupture-prone characteristics

Our recent study classified atherosclerotic lesions from our human carotid endarterectomy tissue collection by morphologic criteria as “erosion-prone” or “rupture-prone”¹². Localization of NET markers in plaques classified by the same criteria (Supplemental Figure I), revealed a differential presence of NETs in lesions with rupture-prone vs. eroded characteristics, as assessed by staining for the membrane-associated neutrophil marker CD66b, NE, an enzyme released by activated neutrophils implicated in vascular damage, and the NET marker H4cit (Figure 1A and 1B). As opposed to rupture-prone plaques, plaques with erosion-prone characteristics demonstrated colocalization of NE and H4cit, mainly in contact with the luminal surface. NETs were also qualitatively localized (n=8) in longitudinally-sectioned human coronary arteries isolated freshly from explanted hearts from patients undergoing transplantation (Figure 1C). NETosing neutrophils could be observed within atheromata (Figure 1D and 1E) and in contact with the luminal surface in areas with EC denudation (Figure 1F) or exhibiting a luminal thrombus (Figure 1G) as shown in a representative specimen. In both cases, further characterization in space of NET structure used a 3D-reconstruction approach (Figure 1H and 1I).

Figure 1. Localization of NETs in human atheromata.

(A) A region of carotid artery plaque with a “rupture-prone” morphology contains neutrophils (CD66b) colocalizing with neutrophil elastase (NE) and citrullinated histone H4 (H4cit). The panels on the right show H4cit, NE and DAPI immunofluorescence within the intima. (B) Carotid artery plaques with erosion-prone morphology stained by immunohistochemistry for neutrophils (CD66b), neutrophil elastase (NE), and citrullinated histones (H4cit). The panels on the right show H4cit, NE and DAPI positive NET structures by immunofluorescence at the intima surface. Scale bar: 150 μm. (C) Shows a representative longitudinal section of a human left coronary artery stained for CD31, H4cit and DNA (DAPI). Elastin autofluorescence appears in white. The four insets show specific areas containing NETosing neutrophils in both plaques (D and E) and near eroded luminal endothelium (F and G), as shown with CD31, H4cit and DAPI staining, and an adjacent section stained for CD31, CD66b and DAPI. Arrows show flow orientation. Arrowheads show the limit of local endothelial denudation. Stars indicate lumen. NET visualization in space after 3D reconstruction from plaques (H) and eroded endothelium (I).

PAD4 deficiency confined to bone marrow-derived cells does not impede fatty streak formation or prevent plaque formation

Investigation of the impact of PAD4 deficiency limited to bone-marrow-derived cells on atherogenesis used chimeric mice. Eight-week-old Ldlr^−/− mice underwent lethal irradiation, followed by reconstitution with bone marrow either WT or Pad4^−/− animals, and subsequent consumption of high fat diet (HFD) for 5 or 10 weeks (Figure 2A). The groups of bone marrow chimeric mice had similar blood cholesterol concentrations (Supplemental Figure IIC). Likewise, the lipid content of aortae from these groups did not differ significantly, as assessed by Oil Red O staining (Figure 2B, 2C, 2E and 2F). Furthermore, PAD4 deficiency altered neither plaque size (Figure 2C and 2D) nor macrophage accumulation after either 5 or 10 weeks of HFD (Figure 2G and 2H). Aortic root fatty streaks did not contain smooth muscle cells (SMCs) after 5 weeks of HFD in either group of mice (Figure 2I and 2J). SMCs accumulated in the fibrous cap region after 10 weeks of HFD, but the SMC area and the collagen content did not differ between the two groups of bone marrow chimeric mice (Figure 2I through 2L). Altogether, these data indicate that PAD4 deficiency limited to hematopoietic cells did not modulate plaque burden or features implicated in “stability.”

Figure 2. Bone marrow PAD4 deficiency does not influence plaque formation and development in LDLR deficient mice.

(A) Experimental design summarizing the procedure. (B) En face visualization of Oil red O stained aortas isolated from Ldlr^−/− mice with WT or Pad4^−/− deficient bone marrow, after 5 or 10 weeks of high fat diet, and aortic root plaque size was quantified (C). Quantification of root plaque size (D). Staining and quantification of lipids (Oil red O, E,F), macrophages (Mac3, G,H), smooth muscle cells (αSMA, I,J), and collagen (Sirius red, K,L). Each data point represents one mouse. Data are expressed as mean (red bar) ±SEM.

PAD4 deficiency abrogates NETosis in experimental atheromata

Brachiocephalic artery and aortic root plaques of mice lethally irradiated and reconstituted with wild-type bone marrow (Figure 3A) contained NETs, as assessed by the colocalization of diffuse DNA and H4cit in neutrophil-rich areas (Figure 3B and 3C). In contrast, mice reconstituted with Pad4^−/− bone marrow never showed diffuse extracellular H4cit/DNA, indicating that plaque NETosis depends on PAD4 in bone marrow-derived cells. Quantification of H3cit area within atheromata showed lack of NETosis in Pad4^−/− mice (Figure 3D and 3E). Mice with Pad4^−/− bone marrow showed a non-significant trend to decreased neutrophil accumulation within atheromata (Figure 3F and 3G). Yet, the number of circulating neutrophils (Supplemental Figure III), and their ability to adhere to endothelial cells in vitro (Figure 3H) was similar in WT mice and Pad4^−/− mice.

Figure 3. Impaired NETosis in atheromata of mice lacking PAD4 in bone marrow-derived cells.

(A) Experimental design summarizing the procedures. (B) Immunofluorescent staining for H4cit, neutrophils (ly6G), and DNA (DAPI) in the brachiocephalic artery (top), and in the aortic root (bottom), in mice reconstituted either with WT (left) or PAD deficient bone marrow (right). Arrows show neutrophils. Dashed lines delimit the intima in aortic roots. (C) 3D reconstruction of H4cit/DAPI positive NETs in aortic root from WT bone marrow mice. Quantification of H3cit (D and E) and Ly6G (F and G) in the brachiocephalic artery (D and F) or the aortic root (E and G). (F) or the aortic root (G). (H) Adhesion of neutrophils isolated from WT or Pad4^−/− mice to endothelial cells resting or activated with TNFα. The right panel shows the quantification of the neutrophil-related fluorescence. Each point represents one mouse. *P<0.05. ****P<0.0001. Mann–Whitney U test.

Flow perturbation at sites of intimal expansion promotes accumulation of NETosing neutrophils

8-week-old Apoe^−/− mice underwent LCCA injury to produce intimal lesions that replicate aspects of plaques associated with human eroded lesions, followed 4 weeks later by induction of local flow perturbation with a constrictive cuff (CC) during 1hour (Figure 4A). These sequential manipulations yielded deposition of H3cit on the LCCA lumen (Figure 4B and 4C), which colocalized with and correlated positively with the number of adherent neutrophils (Ly6G⁺ cells, Figure 4D and 4E). In contrast, neither sham arterial cut down nor placement of a non-constrictive cuff (NC) elicited H3cit deposition. Sites of flow perturbation contained granular structures in contact with a disrupted luminal ECs layer and platelets, as revealed by scanning electron microcopy (Figure 4F), and neutrophils extruding DNA observed by en face microscopy (Figure 4G). These structures contacted the endothelium closely, and co-localized with H3cit and MPO (Figure 4H). Finally, the mice that underwent flow perturbation had higher plasma concentrations of H3cit than the control groups (Figure 4I). Collectively, these results indicate that flow perturbation at sites of intimal expansion fosters neutrophil recruitment and NETosis.

Figure 4. Arterial flow perturbation in experimentally expanded intimas promote the recruitment of neutrophils undergoing NETosis.

(A) Experimental design summarizing the procedure to examine the effects of introduction of flow disturbance over an expanded intima “tailored” to replicate features of eroded human plaques to influence acute neutrophil recruitment to the luminal surface. In this and panel E the scale bars indicate 100μm. (B) Immunostaining showing luminal citrullinated Histone H3 (H3cit) and (C) quantification of the signal 1 hour after introduction of the flow disturbance. (D) Positive correlation between the number of adherent neutrophils and the H3cit signal 1 hour after flow perturbation. (E) Immunofluorescent staining localizing H3cit in recruited Ly6G⁺ cells. Scale bar: 25μm. (F) Transmission electron microscopy shows extracellular granulocytic content adjacent to luminal endothelial cells 1h after flow perturbation. Arrow shows disrupted endothelial junction, ni indicates neointima; ec, endothelial cell, and lum, lumen. (G) En face immunofluorescent staining for Ly6G and DNA (DAPI) shows extruding DNA from Ly6G⁺ cells on the luminal surface after flow perturbation. (H) Comparison between areas upstream and downstream of the partial stenosis showing NETs (delineated by H3cit/DAPI co-staining) in the vicinity of the endothelium (left), their release by Ly6G⁺ cells (center) and the colocalization of MPO with ly6G⁺ cells (right). Arrow shows NETs filaments. Scale bar: 10μm. (I) Circulating H3cit concentrations in mice from groups NC, CC, or from mice undergoing a sham procedure. *P<0.05 and **P<0.01. Mann–Whitney U test. Each point represents one mouse.

PAD4 deficiency in bone marrow-derived cells or DNaseI administration abrogated NETosis and prevented endothelial injury and thrombus formation

Further experiments to test the hypothesis that NETs promote EC injury in experimental superficial erosion, used Pad4 deficiency in bone-marrow-derived cells to halt the ability of neutrophils to undergo NETosis. Eight-week-old Ldlr^−/− male mice underwent electrical injury to the LCCA and then consumed a modified HFD (1.25% cholesterol). Four weeks after the procedure, mice underwent lethal irradiation and reconstitution with either WT or Pad4^−/−-deficient bone marrow cells (Figure 5A). Survival curves, intima:media ratio in harvested LCCA samples, blood cholesterol (Supplemental Figure IIA, IIB and IIC), and the number of circulating monocytes and neutrophils (Supplemental Figure II) did not vary significantly between groups, whereas the levels of circulating H3cit tended to decrease in mice reconstituted with Pad4^−/− cells (Figure 5B). Two weeks after bone marrow reconstitution, CC placement produced local flow perturbation in the artery with the tailored neo-intima during 6h. LCCA in mice reconstituted with Pad4^−/− cells exhibited less intimal permeability after 6h of flow perturbation as assessed by Evans blue staining (Figure 5C), and decreased neutrophil accumulation in the intima (Supplemental Figure IID). While regions of flow perturbation in the carotid arteries of mice with WT bone marrow showed NETs in contact with luminal ECs (Figure 5D, top), hematopoietic Pad4 deficiency abrogated intimal NETosis (Figure 5D, bottom) and protected intimal integrity as gauged by preservation of endothelial continuity (Figure 5E). The preservation of luminal ECs in mice with Pad4^−/− BM associated with smaller intraluminal thrombus (Figure 5F), decreased luminal tissue factor (Figure 5G) and decreased number of intimal TUNEL⁺ cells (Supplemental Figure IIE). Flow perturbation associated with the accumulation of MPO⁺ cells and Tissue factor in the subendothelial layer (Supplemental Figure IIF). An alternative strategy to address the role of NETs on luminal EC injury involved the administration of soluble DNaseI or vehicle before flow perturbation for 6 hours (Figure 5H). DNaseI treatment protected the endothelium downstream of flow disturbance against denudation (Figure 5I and 5J) and decreased the number of CD31⁺TUNEL⁺ luminal cells (Figure 5K). As previously observed, the number of adherent ly6G⁺ cells decreased in group DNAseI, suggesting decreased local inflammation (Supplemental Figure IIG). Together these results using independent methods that perturb NET formation or NET degradation demonstrate in vivo that NETs jeopardize endothelial cell survival, adhesion and barrier function.

Figure 5. Abrogation of the capacity to form NETs decreases endothelial injury, death and thrombosis in arteries with expanded intimas exposed to flow disturbance.

(A) Experimental protocol and time points studied: Mice underwent a LCCA injury and were lethally irradiated 4 weeks later. Mice were immediately reconstituted with either WT or Pad4^−/− bone marrow cells. After 2 weeks, flow perturbation was induced for 6 hours in the previously injured LCCA using a constrictive cuff (CC). (B) Circulating H3cit levels were quantified in mice from both groups. (C) LCCA isolated from Ldlr^−/− mice reconstituted with either WT or Pad4^−/− bone marrow cells were probed for endothelial permeability using Evans blue intravital staining after 6 h of flow perturbation. (D) Visualization by immunofluorescence of luminal NETosis using an anti-H3Cit and a DAPI staining with either CD31 (left panel) or Ly6G (right panel) immunolocalization, in Ldlr^−/− mice reconstituted with WT (top) or Pad4^−/− bone marrow (bottom). White arrows show accumulated neutrophils undergoing NETosis. (E) CD31 immunoreactivity shows endothelium. The graph (right) quantitates endothelial continuity. The star shows a neutrophil-rich fresh thrombus. (F) Fibrinogen staining shows preferential thrombus formation on the intimal surface in mice reconstituted with WT bone marrow compared to those with Pad4^−/− bone marrow. The graph (right) shows the quantification of intimal fibrinogen staining. *P<0.05. (G) Tissue factor (TF) staining shows higher content in tissue factor in the lumen of the LCCA subjected to flow perturbation in WT>Ldlr^−/− vs Pad4^−/−>Ldlr^−/−. (H) Experimental procedure showing the use of DNAseI in mice subjected to LCCA injury and flow perturbation. (I) Endothelial lining, neutrophil and DNA accumulation was probed in mice treated with DNAseI or vehicle, using a 3D reconstruction, and endothelial continuity was quantified in (J). (K) IF and quantification showing TUNEL+ luminal endothelial cells in DNaseI or vehicle treated mice. Each point represents one mouse. Arrows show TUNEL+ endothelial cells. Dashed lines show internal elastic laminae. *P<0.05 and **P<0.01. Mann–Whitney U test.

NETs promote endothelial cell death and detachment in a complement-dependent manner

Flow perturbation associated with complement accumulation at the vicinity of experimentally eroded lesions in mice (Supplemental Figure VIIA). We next explored the contribution of complement activation to NET-induced damage of cultured human ECs. Exposure of primary HCAEC to NETs prepared from PMA-stimulated neutrophils increased VCAM-1 expression (Supplemental Figure VIIB) and augmented EC apoptosis (Supplemental Figure VIIC). Pre-treatment of NETs with DNase I eliminated NET-induced apoptosis (Supplemental Figure VIIC and VIID). EC culture using complement-rich baby rabbit serum (BRS) increased complement deposition on cells assessed by C5b-9 staining (Supplemental Figure VIIE). BRS markedly enhanced apoptotic cell count when added with NETs, as compared to BRS alone (Supplemental Figure VIIF and VIIG). HCAEC apoptosis declined significantly in cells exposed to BRS decomplemented by heat treatment compared to untreated BRS (Supplemental Figure VIIF and VIIG). In addition, the combination of BRS and NETs augmented HCAEC detachment, an effect mitigated by heat decomplementation of BRS (Supplemental Figure VIIH).

DISCUSSION

NETosis in mice depends on PAD4 activity. In addition to mediating NETosis, PAD4 activity can also affect SMC activation²³. A previous study reported that Cl-amidine, a pan-PAD inhibitor administered systemically, reduced experimental atherosclerosis in mice²¹. Yet, the lack of selectivity of Cl-amidine considerably limits the interpretation of experiments that used this small molecule inhibitor. We therefore studied mice with selective genetic deficiency of PAD4 in bone marrow-derived cells, abolishing PAD4 activity in leukocytes but not in intrinsic vascular wall cells and other tissues.

PAD4 deficiency in hematopoietic cells did not significantly alter the formation and progression of atheromatous plaques in hypercholesterolemic mice. Yet, it critically attenuated the response to flow perturbation of arterial regions with experimentally expanded intimas. These observations implicate NETosis in this form of atherothrombotic complication which recapitulates features of superficial erosion. Conversely, the lack of effect on atherosclerosis of PAD4 deficiency in bone marrow-derived cells in the present experiments indicates that PAD4 does not play an essential role in the earlier stages of lesion formation. The slight reduction in neutrophil number in mice with PAD4-deficient bone marrow is surprising. In this regard, while PAD4 reportedly regulates the proliferation of hematopoietic stem cells²⁴, our results show that Pad4^−/− mice have similar levels of circulating neutrophils and monocytes as their WT counterparts. Additionally, PAD4 deficiency did not alter neutrophil adhesion to endothelial cells in vitro, suggesting that decreased neutrophil number in superficial erosion in vivo likely reflects locally decreased inflammation, due to impaired NETosis.

Yet this study has certain limitations. Various other hematopoietic cell types, including mast cells and macrophages, may undergo NETosis²⁵. The experiments presented here do not distinguish the effects of NETs derived from granulocytes from those induced by such other hematopoietic cells. Also, the flow perturbation used in these experiments to trigger intimal dysfunction was acute and of relatively brief duration compared to conditions that may prevail in human superficial erosion. Yet, the use of this preparation contributes to a growing body of evidence that NETs participate in the propagation of plaque erosion.

Our experimental approach to study intimal functions related to superficial erosion showed that PAD4 deficiency in hematopoietic cells significantly preserved endothelial barrier function following acute perturbation of blood flow, protected endothelial cells from desquamation, and decreased their death. Therefore, NETs appear to jeopardize normal endothelial functions and thus represent as novel target for the treatment and / or prevention of thrombotic complications of atherosclerosis including superficial erosion. Systemic treatment with DNaseI merits consideration as a therapeutic approach. Indeed, formulations of DNase I (Pulmozyme®) approved for the treatment of cystic fibrosis²⁶ exert beneficial effects in mice with experimental inflammation and thrombosis. In addition, levels of circulating markers associated with NET (citrullinated histones, free DNA, MPO) could help to diagnose superficial erosion and to triage the treatment of acute coronary syndromes in a more personalized approach than currently practiced²⁷, a prospect supported by an early biomarker study in individuals with atherothrombosis¹⁷. In this regard, a recent preliminary clinical study reported the efficacy of an anti-thrombotic strategy without intravascular intervention for the treatment of ACS provoked by superficial erosion²⁸.

Our data suggest a role for the complement pathway in NET-induced EC death and detachment. NETs can activate the alternative complement pathway²⁹ and promote endothelial damage affecting glomeruli in antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis³⁰. NETs could constitute a critical scaffold promoting the local activation of the complement pathway in the vicinity of vascular ECs, exacerbating EC death, detachment and thrombosis. Thus, strategies that limit complement activation also merit consideration as an adjunct to treatment of the subset of ACS due to superficial erosion³¹.

Overall, this study using genetic approaches to interfere with NET formation provides experimental evidence for a stage-specific contribution of NETs to atherosclerosis and its thrombotic complications. These findings have important translational implications for diagnosis, therapy, and managements of patients with or at risk for atherothrombosis.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Atherogenesis nd the complications of atherosclerosis may involve neutrophil extracellular traps (NETs.)

• NET formation requires the enzyme peptidy arginoine deiminase 4 (PAD4.)

What New Information Does This Article Contribute?

Genetic loss of PAD4 function does not affect atherogenesis in mice, but does protect against endothelial desquamation and thrombus formation in experiments that replicate in mice aspects of superficial erosion.

Previous studies have reported the presence of NETs in both human and mouse atheromata. We had previously localized NETs to human plaques with features of superficial erosion. Yet, few studies have examined directly the effect of NETs on the formation, development, and the complication of atherosclerosis. We evaluated the role of PAD4 and NETs in atherothrombosis using mice with genetic deficiency of PAD4 in blood cells. We found that PAD4 derived from cells of bone marrow origin; and hence, NETs do not influence experimental atherogenesis, although NETs do participate in acute arterial thrombosis under conditions that recapitulate features of superficial erosion.

Nonstandard Abbreviations and Acronyms

ACS

Acute coronary syndromes

ApoE

Apolipoprotein E

EC

Endothelial cells

HCAEC

Human coronary artery endothelial cells

LCCA

Left common carotid artery

LDL

Low-density lipoprotein

LdlR

Low-density lipoprotein receptor

MPO

Myeloperoxidase

NE

Neutrophil elastase

NETs

Neutrophil extracellular traps

PAD

Peptidyl arginine deiminase

SMC

Smooth muscle cells

TLR

Toll-like receptor

TUNEL

Terminal deoxynucleotidyl transferase dUTP nick end labeling