Neutrophil Extracellular Traps Participate in Cardiovascular Diseases - Recent Experimental and Clinical Insights

Abstract

Neutrophil extracellular traps (NETs) have recently emerged as a newly recognized contributor to venous and arterial thrombosis. These strands of DNA extruded by activated or dying neutrophils, decorated with various protein mediators, become solid-state reactors that can localize at the critical interface of blood with the intimal surface of diseased arteries, and propagate and amplify the regional injury. NETs thus furnish a previously unsuspected link between inflammation, innate immunity, thrombosis, oxidative stress, and cardiovascular diseases. In response to disease-relevant stimuli, neutrophils undergo a specialized series of reactions that culminate in NET formation. DNA derived from either nuclei or mitochondria can contribute to NET formation. The DNA liberated from neutrophils forms a reticular mesh that resembles morphologically a “net,” rendering the acronym NETs particularly appropriate. The DNA backbone of NETs not only presents intrinsic neutrophil proteins (e.g. myeloperoxidase (MPO) and various proteinases), but can gather other proteins found in blood (e.g. tissue factor procoagulant). This review presents current concepts of neutrophil biology, the triggers to and mechanisms of NET formation, and the contribution of NETs to atherosclerosis and to thrombosis. We consider the use of markers of NETs in clinical studies. We aim here to integrate critically the experimental literature with the growing body of clinical information regarding NETs.

Neutrophil extracellular traps (NETs) have recently burst on the scene as a newly recognized contributor to venous and arterial thrombosis.¹ These strands of DNA extruded by activated or dying neutrophils, decorated with various protein mediators such as neutrophil elastase or azurocidine (a member of the serprocidine family which promotes monocyte adhesion and transmigration as well as a proinflammatory slant of macrophage functions), furnish a previously unsuspected link between inflammation, innate immunity, thrombosis, and cardiovascular diseases (for a detailed list of proteins decorating the NETs see Lim et al.²). In response to an expanding array of stimuli, neutrophils undergo a specialized series of reactions that culminate in NET formation. DNA derived from nuclei or from mitochondria can contribute to generation of NETs. Freed of ionic bond constraints that tether nuclear DNA to histones, the neutrophil unfurls linear DNA into the extracellular space that forms a reticular mesh that morphologically resembles a “net,” rendering the acronym NETs particularly appropriate.

The DNA that comprises the backbone of NETs not only presents in proteins present within the neutrophil itself, but can gather other proteins found in blood. Examples of neutrophil-derived proteins associated with NETs include myeloperoxidase (MPO), the serine proteinases characteristic of neutrophils (e.g. cathepsin G, neutrophil elastase, and proteinase 3), and the pro-inflammatory interleukin (IL)-1α. A prime example of a non-neutrophil protein associated with NETs, tissue factor procoagulant, clearly participates directly in thrombogenesis. This review will provide an update on this fast-moving field. We will reconsider basic principles of neutrophil biology, the triggers to and mechanisms of NET formation, and the contribution of NETs to atherosclerosis and to thrombosis. We consider the use of markers of NETs in clinical studies. We aim here to integrate critically the experimental literature with the growing body of clinical information regarding NETs.

Basic principles of neutrophil biology

Tradition views neutrophils as terminally-differentiated, short-lived phagocytes with a rather uncontrolled mode of action. Their prodigious turnover requires generating 10¹¹ neutrophils per day in the adult human body. Through a cascade of proliferation and differentiation steps, neutrophils develop and mature from hematopoietic stem and progenitor cells (HSPC) in the bone marrow, and reside as a rapidly mobilizable pool. During conditions of stress, however, neutrophil production can accelerate to adapt to additional demands. We now appreciate that hypercholesterolemia and hyperglycemia, two important modifiable risk factors of cardiovascular disease, alter inflammatory responses in part by reprograming HSPC function and boosting subsequent myelopoiesis including granulocyte generation. Indeed, accumulation of cholesterol in the cell membrane of HSPCs induces their proliferation, mobilization, and skews their maturation toward a myeloid bias.^{3, 4} Similarly, hyperglycemia alters granulopoiesis, a consequence of the release of S100A8/A9 from circulating neutrophils.⁵ For other atherosclerosis-related regulators of hematopoiesis see also the article by Schloss et al. in this compendium.⁶

In addition to accelerated production, metabolic risk factors also prime neutrophils, producing a state of heightened responsiveness. For example, hyperglycemia readies human and mouse neutrophils to form NETs, thus linking the overproduction of reactive oxygen species (ROS) in patients with diabetes to the release of neutrophil extracellular DNA, a process that requires ROS generation.⁷ Similarly, disrupted cholesterol efflux in neutrophils can also enhance inflammasome activation and promote arterial neutrophil infiltration and NET release and, as a consequence, accelerate atherosclerotic lesion formation.⁸

As we age, HSPCs accumulate somatic mutations in individuals who do not have hematologic malignancies. HSPCs carrying certain somatic mutations can expand to form clones in peripheral blood and associate remarkably with the risk of adverse cardiovascular outcomes.⁹ This condition, termed clonal hematopoiesis of indeterminate potential (CHIP), affects primarily cells of the myeloid lineage. Genes commonly mutated in CHIP include DNMT3a, TET2, ASXL1, and JAK2. In particular, a specific gain-of-function somatic mutation in JAK2 (Janus kinase 2, V617F) activates signal transducer and activator of transcription (STAT) signaling. In hyperlipidemic mice, Jak2^V617F in myeloid cells induced prominent erythrophagocytosis and neutrophil infiltration, leading to accelerated atherogenesis and increased characteristics of propensity to rupture.¹⁰ The Jak^V617F mutation also associates with enhanced spontaneous NET release^{10, 11} and thrombus formation.¹¹

Sites of acute inflammation rapidly recruit neutrophils through a well-understood multistep recruitment cascade. This general recruitment scheme differs between tissues with important alterations in large arteries.¹² For example, arterial neutrophil adhesion seems to rely on chemokines derived from platelets, a mechanism much less important in the microcirculation.¹³ In particular, platelet-derived CCL5 and CXCL4 activate neutrophils for NET release, providing a link to NETs found on the luminal aspect of the developing atherosclerotic lesion.^{14, 15} Neutrophil behavior depends strikingly on circadian rhythms in mice. Not only do peripheral blood neutrophil counts oscillate over the day, the diurnal variation in neutrophil adhesion to large arteries exhibits a 12-hour phase shift compared to microvessels.¹⁶ In addition, recent data suggest that an intrinsic neutrophil clock regulates the activity of these cells including the ability to release NETs.¹⁷ Moreover, neutrophils can traffic into non-inflamed tissues such as the liver and the intestine through the microvasculature in a mechanism reminiscent of that described for non-classical, patrolling monocytes.¹⁸ Whether this phenomenon also occurs in large arteries remains unknown.

Neutrophils possess a large preformed toolkit poised to deploy upon arrival at sites of inflammation. These armaments include the production and release of ROS as well as of bioactive lipid mediators. In addition, neutrophils carry a large number of preformed granule proteins including several alarmins (e.g. cathelicidins, defensins) or serine proteases and matrix metalloproteinase (MMP)-8. Moreover, MPO, which generates the highly oxidant and chlorinating species hypochlorous acid (HOCl) comprises a major constituent of granulocytes. Thus, in addition to surface-associated nicotinamide adenine dinucleotide phosphate (NADPH) oxidase that produces superoxide anion (O₂⁻), myeloperoxidase can form HOCl extracellularly once released from the granulocyte. The other granular proteins can likewise undergo rapid release and act outside of cells. While most of these neutrophil products have recognized antimicrobial actions, recent data have identified important immune cell-stimulating effects of these granule proteins.¹⁹ When these granular enzymes bind to NETs they form a “solid state” reactor that may promote their local activity.²⁰ Understanding how the interaction of granular proteins with NET chromatin alters their function is under intense investigation.²¹

NET release pathways and NET composition

Both NADPH oxidase-dependent and NADPH-independent pathways can generate NETs. A variety of extrinsic (e.g. bacteria²²) and intrinsic (e.g. hydrogen peroxide²³) stimuli as well as other stimuli (e.g. concanavalin a²⁴, or phorbol myristate acetate,PMA²²) induce NADPH oxidase-dependent NETosis. These stimuli trigger activation of downstream signaling molecules (e.g. ERK), which activate NADPH oxidase.^{24, 25} O₂, CO₂, bicarbonate levels, and pH modulate NETosis. For example PMA, but not Staphylococcus aureus, requires normoxia to induce NETs.²⁶ After activation, NADPH oxidase converts molecular oxygen to superoxide. Therefore, pharmacological inhibition of NADPH oxidase or ROS scavengers block NET formation.²⁵ Superoxide dismutates to hydrogen peroxide and furnishes substrate for MPO catalyzed HOCl production. Indeed, inhibitors of MPO can also block NETosis.²⁷

Two other NET inducers, calcium ionophore A23187 (from Streptomyces chartreusensis), and the potassium ionophore nigericin (from Streptomyces hygroscopicus), remain independent from NADPH oxidase or MPO activity.^{28, 29}

In addition, nuclear peptidylarginine deiminase 4 (PAD4) converts positively charged arginyl residues abundant in histones to citrulline, an amino acid that lacks a charge. This action of PAD4 disrupts the ionic bonds that mediate tight association of negatively charged DNA with histones in the tightly packed nucleosomes around which DNA usually coils in close constraint. The action of this enzyme thus frees DNA to unwind and form strands that when extruded, form NETs. Li et al. showed a key role for PAD4 in NET formation and bacterial killing, and these mice also show a strong susceptibility to bacterial infections. The authors conclude that NET formation depends on histone hypercitrullination mediated via PAD4.³⁰ In contrast, Martinod et al. did not observe an increase in the susceptibility to bacterial infection induced by cecal ligation and puncture in PAD4-deficient mice. Thus, the anti-microbial role of PAD4 may depend on context.³¹ PAD4-independent mechanisms can also generate NETs³² and thus evade detection in PAD4-deficient mice when probed with anti-citrullinated histone antibodies. Accordingly, previous studies report diminished but not abolished NET abundance in PAD4-deficient mice detected by an anti-Histone H3 antibody.³³ Recent work suggests that the ability of neutrophils to discharge NETs depends on the time of day and that a neutrophil-intrinsic, CXCR2-dependent timer controls such rhythmicity. In conjunction with the circadian oscillation of the neutrophil proteome, NET composition and hence functionality differs with time of day.³⁴

NETs in early experimental atherosclerosis

Luminal endothelial cells activated by various stimuli initiate leukocyte recruitment during atherogenesis. Monocytes continually migrate into and accumulate within the atherosclerotic lesion where they mature into macrophages, proliferate, and form lipid-laden foam cells that can drive plaque progression.³⁵ Ongoing lesional inflammation can limit interstitial collagen production and augment the activity of collgenolytic enzymes that can render the fibrous cap thin and friable, and prone to rupture and provocation of thrombus formation.^36,37 In addition to monocytes, other inflammatory cells and their products populate atherosclerotic lesions. (Figure 1)

Figure 1: Emerging roles of NETs in atherosclerosis and atherothrombosis.

Luminally netting neutrophils activate leukocytes, platelets and endothelial cells (EC) creating a pro-inflammatory milieu presumably resulting in endothelial dysfunction, the initial trigger of lesion development. Lesional NETs may be induced by CCL7 released from activated vascular smoothe muscle cells (VSMCs) and NETs initiate a IL-1beta/TH17 and/or type I interferon response, which leads to further activation of lesional leukocytes, releasing more pro-inflammatory mediators. Eventually NET-driven pro-inflammatory responses will cause an inflammatory environment that favors plaque destabilization and rupture. Inlay: Enlarged NETting neutrophil.

Presence of NETs in atherosclerotic lesions

Megens et al. first reported the presence of NETs, stemming from neutrophils, in mouse and human atherosclerotic lesions.¹⁴ Subsequent in-depth analysis of 56 categorized human atherosclerotic plaques revealed neutrophils and NETs near clusters of apoptotic endothelial and smooth muscle cell-rich regions, implying that NETs contribute to plaque disruption and distribute differentially depending on plaque classification.^{38, 39} Studies examined the presence, extent, and distribution of neutrophils and NETs in plaques of differing morphologies. 64 paraffin-embedded human coronary plaque segments comprising 44 complicated plaques (intraplaque haemorrhages, erosions and ruptures) and 20 intact plaques were further categorized by immunohistochemistry to visualize neutrophils (myeloperoxidase, neutrophil elastase and CD177) and NETs (citrullinated histone-3 (cH3) and PAD4). Neutrophils and NETs localized in all types of complicated lesions, with no significant differences in their extent between ruptures, erosions, and intraplaque hemorrhages. Intact plaques, however, did not contain NETs while adjacent perivascular tissue of complicated, but not of intact plaques, also contained high numbers of neutrophils and NETs.⁴⁰ Taken together, the presence of NETs in human lesions points to an important role of these structures in plaque destabilization. The presence of NETs in human lesions supports their involvement in atherothrombosis, yet despite strong experimental evidence, proof of causality of NETs in human disease, and whether their formation precedes or follow initial plaque disruption remains inconclusive.

Substantial evidence supports a role for NETs in experimental atherosclerosis

Oxidative stress mediates NETosis

Murine atherosclerotic lesions contain NETs, but not macrophage extracellular traps, as early as 3 weeks following consumption of a western-type diet (WD), and the prevalence of NETs increased over time.⁴¹ Moreover, modified LDL induces (luminal) NETs^14,42 which can propagate endothelial dysfunction at all stages of lesion development.^{38, 42, 43} Aged mice also show enhanced abundance of lesional NETs as described in a study that compared LDL receptor (Ldlr)-deficient mice transplanted with bone marrow from either wild-type or mitochondrial catalase (mCAT) mice. mCAT transgenic mice contain ectopically expressed the human catalase gene in mitochondria, yielding reduced mitochondrial oxidative stress (mitoOS). The mice with limited mitoOS had smaller lesions and diminished appearance of NET structures compared with age-matched controls, resembling the character of lesions in young animals.⁴⁴ As mitoOS represents an important trigger in human aging and cardiovascular disease this study suggests a causative link between endogenous neutrophil mitoOS and subsequent NETosis during aging with enhanced atherosclerotic lesion formation.

Mechanistically similar results also pertain to mice with lupus-like disease.⁴⁵ Exosomes released from macrophages treated with LDL that contain miR-146a can also provoke oxidative stress. This activation then induces NET formation by increasing ROS production in neutrophils through the inhibition of superoxide dismutase 2 expression, augmenting lesion formation. The authors concluded that exosomal miR-146a-dependent oxidative stress induction represents a new link between pro inflammatory immune responses mediated by macrophages and neutrophils.⁴⁶ Not only oxidative stress, but also repetitive psychosocial stress induced by housing of eight-week-old male Apolipoprotein E-deficient (Apoe^−/−)mice with a larger CD-1 mouse in a shared home cage accelerated atherosclerosis. Socially defeated Apoe^−/− mice, after consuming an atherogenic diet for 6 weeks, exhibited depressive-like behaviours and showed enhanced lesion development compared to the non-stressed group. Augmented NET formation within the atherosclerotic plaques accompanied these findings, while treatment with DNase I (a NET disruptor) abrogated lesion formation in socially defeated Apoe^−/− mice, indicating that NETosis can contribute to the development of atherosclerosis aggravated by social stress. This observation, if it applies to humans, might contribute to the link between depression and cardiovascular disease.^{47, 48}

Inflammasomes and NET formation

Westerterp et al. showed that myeloid cell-specific NLRP3 inflammasome activation triggered by reduced cholesterol efflux capacity fosters neutrophil recruitment, NET formation, and subsequent lesion development in female LDLR-deficient mice. Yet, these effects are more pronounced after 8 weeks of WD compared to 12 weeks WD feeding.⁴⁹ In addition, extracellular double-stranded (ds)DNA, a prominent NET component, co-localizes in mature atherosclerotic lesions with the non-canonical inflammasome absent in melanoma 2 (AIM2) in lesional macrophages.⁵⁰ In turn, AIM2-deficient Apoe^−/− mice displayed a diminished production of IL-1β and reduced features related to plaque destabilization, suggesting a novel role for the AIM2 inflammasome in controlling plaque stability but not lesion size.⁵¹ Yet, Gaidt et al. describe AIM2 as being dispensable for DNA-mediated inflammasome activation in human myeloid cells while activation of the NALP3 pathway in the same cells seems to contribute importantly to inflammatory conditions associated with cytosolic DNA sensing.⁵² Unraveling the role of AIM2 in cardiovascular disease in humans will require further research.

NETs may not only induce inflammasome activation but vice versa. Growing evidence suggests that cytosolic detection of DNA in neutrophils results in non-canonical inflammasome activation and gasdermin D cleavage by caspase 11, which induces NET formation.^{53, 54} Moreover, NETs and component proteins (e.g. the antimicrobial peptide cathelicidin) not only induce NLRP3 inflammasome activation, but mature IL-18 generated by inflammasomes can induce NET formation.⁵⁵

The importance of PAD4 for NETosis in experimental atherosclerosis

Other studies investigated the impact of systemic or cell-specific PAD4-deficiency, an enzyme important in histone hypercitrullination as detailed above, on NET formation and subsequent lesion formation. Knight et al.⁴³ reported that inhibition of PADs by chloramidine treatment prevents NET formation and thereby decreases atherosclerotic lesion size and delays carotid artery thrombosis in Apoe^−/− mice receiving a WD. Inhibition of NETosis by chloramidine yielded a decrease in arterial interferon (IFN)α expression, less neutrophil recruitment to the arterial wall, and a decline in intimal macrophages. Furthermore, combining administration of chloramidine into Apoe^−/− mice with a neutrophil depletion antibody or injection of chloramidine into mice lacking a functional type I interferon receptor did not alter lesions size.⁴³ These data suggest that NET formation and the subsequent type I IFN responses augment atherogenesis, consistent with earlier work showing that protein-DNA complexes (putatively NET-derived) stimulate a plasmacytoid dendritic cell-mediated type I interferon response and increase lesion formation.⁵⁶ Yet treatment with the non-specific PAD inhibitor chloramidine did not alter advanced atherosclerosis, suggesting the participation of neutrophils in earlier phases of atherogenesis.⁴³

Cholesterol crystals, a sterile danger signal in plaques, induce lesional NET formation in Apoe^−/− mice-fed a WD for 8 weeks. These NETs primed (lesional) macrophages for production of IL-1β and IL-6 resulted in activation of a T helper cell (T_h) 17 response, which amplifies immune cell recruitment into atherosclerotic lesions.⁵⁷ T_h17 responses, however, may be a double-edged sword depending on the mediators stimulating the response. IL-6 and transforming growth factor (TGF)-β, for example, induce a subtype of T_h17 cells that produce both IL-17 and IL-10, which together mediate an atheroprotective immune response. Conversely, IL-23 and IL-6 participate in the differentiation of pathogenic T_h17 cells that produce IL-17 and IFN-γ, boosting atherosclerosis through the induction of proinflammatory cytokines.⁵⁸ To limit NETosis in Apoe^−/− mice, the authors used different approaches; however, in their hands, only inhibition of NADPH oxidase with diphenylene iodonium or blocking of the neutrophil elastase or proteinase 3 limited NET generation.⁵⁷ In contrast to other reports^{33, 43}, in this study chloramidine treatment did not inhibit NET release.⁵⁷ Nevertheless, in mice with myeloid PAD4-deficiency (LysM^Cre Pad4^flox) on an Apoe^−/− background fed WD for 6 or 10 weeks, atherosclerosis burden significantly decreased in association with diminished NET formation and reduced inflammatory responses in the aorta attributed to a lessened activation of macrophages. Lesion size in myeloid PAD4-deficient mice did not decrease further with DNAse I treatment consistent with the NET-directed effect.⁴¹

In contrast, examining PAD4-deficiency in hematopoietic cells in LDLR-deficient mice transplanted with PAD4-deficient bone marrow, Franck et al. did not report any changes in lesion size or composition after 5 or 10 weeks of WD feeding despite decreased lesional NETs in these mice with hematopoietic PAD4- deficiency.³⁹ Yet, diminished NETosis limited intimal injury and subsequent thrombus formation under conditions that mimic aspects of superficial erosion (see also NETs in atherothrombosis). These findings agree with another recent study that investigated pharmacologic inhibition (chloramidine) and genetic ablation of PAD4 in mice with plaque destabilization. These studies revealed that the inhibition of NET release reduces characteristics associated with atherosclerotic plaque vulnerability in humans, putatively by prevention of smooth muscle cell apoptosis. Intima thickness, however, did not differ between controls and mice with reduced NET abundance.³³

Taken together, the above-mentioned studies used differing methodologies and present certain limitations, which may account for divergent findings. Chloramidine, for example, not only inhibits PAD4 but also other PADs.⁵⁹ Moreover, PAD4 itself mediates not only hypercitrullination of histones but can also affect smooth muscle cell activation directly⁶⁰ or regulate the proliferation of hematopoietic stem cells⁶¹. Uncertainty prevails regarding the extent to which NET formation requires histone citrullination, as some stimuli induce NETosis without PAD4 activity.⁶² In addition, as always, the details of the mouse experiments may influence results. For example, Liu et al. used LysM^Cre Pad4^flox mice, which not only exhibit myeloid PAD4-deficiency but also a non-functional LysM⁶³. The latter could already modify atherosclerosis in general without an effect mediated by the lack of PAD4 in LysM-expressing cells. Also, differences in atheroma development may vary with the experimental preparation. Knight et al.,⁴³ for example, studied systemic PAD4-deficient mice on an Apoe^−/− background while Franck et al.³⁹ transplanted PAD4-deficient bone marrow into LDLR-deficient animals. In mice, lesional NET presence seems to increase over time. The current literature collectively suggests that if NETs contribute to growth of experimental atheromata, they do so particularly during earlier plaque formation. As lesions advance, NETs seem to influence features associated with thrombotic complication rather than atheroma volume per se.

NETs may aggravate thrombotic complications of atherosclerotic plaques

Atherosclerotic plaques that have produced fatal events characteristically have large lipid cores, prominent macrophage accumulation, and thin fibrous caps typically depleted of smooth muscle cells and collagen. As noted above, thinning of the fibrous cap as a consequence of heightened collagen degradation or reduced collagen production, e. g. resulting from smooth muscle cell death, may predispose to plaque rupture, a frequent cause of fatal myocardial infarction (Figure 2).³⁶ While macrophages and T cell subsets dominate the immune cell landscape in advanced atherosclerotic lesions,⁶⁴ neutrophils appear at much lower frequency. Yet low counts at any given time may simply reflect the typically short lifespan of neutrophils in tissues. Even though advanced atheromata contain only rare neutrophils, clinical data suggest that they may contribute to plaque complications. Specifically, peripheral blood neutrophil counts or neutrophil/lymphocyte ratios consistently associate with cardiovascular risk and outcome.^{65, 66} Similarly, plasma levels of MPO correlate positively with the risk of coronary artery disease.⁶⁷ In addition, plasma nucleosome levels associate with an increased risk of coronary stenosis, and MPO–DNA complexes correlate with the occurrence of major adverse cardiac events. These results suggest that NET-derived markers can predict atherosclerotic disease burden and events (see below.)⁶⁸

Figure 2: Emerging roles of NETs in atherothrombosis.

During atherothrombosis then NETs may trigger activation of the coagulation cascade and increase thrombus stability thus orchestrating arterial occlusion.

NETs in plaque destabilization

Histological analyses of human carotid plaque specimens furnished important insights regarding the possible contribution of neutrophils to plaque destabilization. Abundant lesional neutrophils localized in plaques with a large lipid core, high macrophage numbers, and low collagen levels and smooth muscle cell numbers, implicating neutrophils in the core processes of plaque destabilization, i. e. necrotic core growth and fibrous cap thinning.⁶⁹ In agreement with this notion, a recent study established the contribution of neutrophils and in particular NETs to plaque destabilization by promoting precisely these two features.³³ Mechanistically, modulated smoot muscle cells resident in the fibrous cap interact physically with neutrophils, resulting in their activation characterized by ROS production and the release of NETs, the latter being mediated by smooth muscle cell-borne CCL7. Inhibition of NET release by treatment with a PAD inhibitor in mice with pre-existing lesions or those lacking Pad4 exhibited lesions with fewer characteristics of vulnerability, including smaller necrotic cores and higher smooth muscle cell content compared to control mice. The striking correlation between neutrophils and NETs on the one hand and dying smooth muscle cells, necrotic core sizes and thin fibrous caps in mice on the other hand suggest a direct cytotoxic action of NETs. In fact, ex vivo studies revealed that NETs kill smooth muscle cells. This effect did not depend on neutrophil proteins derived of granules or the cytoplasm, but rather on nuclear histones, specifically histone H4, in agreement with earlier reports on cytotoxic activities of NET-borne histones.^{70, 71}

Mechanistically, histone H4, a strongly cationic protein, can interact with negatively charged smooth muscle cells surfaces, thereby exerting membrane activity with membrane bending and ultimately pore formation, leading to lytic cell death. Smooth muscle cell depletion may promote thinning of the fibrous cap by removing a source of interstitial collagen synthesis. The importance of histone H4 in promoting features associated with plaque disruption was further corroborated by the use of a histone H4 neutralizing antibody and cyclical histone interference peptides (HIPE) that target the N-terminus of histone H4. Both therapeutic strategies reinforced aspects of plaques associated with plaque stability. In addition, in mice with advanced atherosclerosis, activation of dsDNA sensing AIM2 inflammasome results in a strong production of the pro-atherogenic cytokines IL-1β and IL-18 and lack or neutralization of AIM2 generates plaques with fewer characteristics of propensity to rupture, notably a smaller necrotic core and a thicker fibrous cap.⁵¹ However, as mentioned above, the role of AIM2 in human plaque stability requires further study.

NETs in arterial thrombosis

Considerable recent evidence implicates neutrophils and neutrophil-derived inflammatory mediators in thrombosis.⁷² For example, NETs promote thrombin generation,⁷³ and activated platelets can trigger NET formation⁷⁴. In addition, NETs present associated prothrombotic molecules such as tissue factor, factor XII, histones H3 and H4, and fibrinogen, all of which favor thrombosis. NETs and the fibrin mesh that they can generate also furnish a scaffold that entraps platelets and red blood cells.^{75, 76} Still, most evidence on NETs-mediated thrombosis stems from animals, and in vitro work mostly focuses on ‘isolated’ constituents, which may exert stronger single procoagulant effects compared to a fuller NET structure.⁷⁷ For this section we have focused on arterial thrombosis while venous thrombosis has recently been reviewed elsewhere.¹

NETs and their components localize in human coronary^{78, 79} and ischemic stroke thrombi⁸⁰ regardless of the type of plaque, but seem to dominate numerically in fresher thrombi rather than older, organized thrombi.⁸¹ A large-scale multicenter study found NETs (defined by colocalized extracellular DNA and neutrophil elastase) in 25% of coronary stent thrombi remain consistent with participation of NETosis in this process.⁷⁸ De Boer et al. investigated specimens obtained from patients after acute myocardial infarction, examining thrombi in various stages of evolution. Neutrophils, NETs, and immunoreactive IL-17A localized in fresh but not organized thrombi, suggesting their contribution to thrombus stabilization and growth.⁸² Histologic analysis of 64 human coronary artery plaque segments comprising 44 complicated plaques (intraplaque hemorrhages, erosions and ruptures) and 20 intact plaques showed more frequent presence of neutrophils and NETs in non-organized (fresh) thrombi and in on-going intraplaque hemorrhages compared to older (more organized) thrombi. These observations implicate NETs in distinct types of atherothrombosis, particularly, once again, in “fresher” arterial thrombi.⁴⁰ 111 thrombi retrieved from patients with ST elevation acute coronary syndrome revealed higher NET burden and correlated negatively with ST resolution but positively with infarct size. These data imply that NETs in culprit lesions propagate thrombosis and inflammation distally into the infarcted myocardium and contribute to myocyte death during atheroembolism. Activity of DNAse, an enzyme that acts to clear NETs by digesting the DNA strands, in these lesions correlated negatively with infarct size but positively with ST-segment resolution. Ex vivo addition of DNAse to these thrombi accelerated their lysis.⁸³ These and many other observational studies in human specimens generate intriguing hypotheses, but do not permit conclusions regarding causality.

The procoagulant tissue factor triggers many acute coronary syndromes, and can decorate NETs. Thrombi retrieved from culprits of ST-segment–elevation myocardial infarction (STEMI) and samples from the non-infarct-related artery of 18 patients revealed local accumulation of tissue factor–bearing NETs at culprit but not non-culprit sites. The authors concluded that interaction of neutrophils and platelets at sites of plaque rupture promote NET formation and increase active tissue factor abundance, stimulating prothrombotic events.⁸⁴ High-mobility group box 1 protein of platelet origin may also induce NET formation, as apparent from histology analysis on 26 thrombi from patients after acute myocardial infarction.⁸⁵ Novotny et al. compared 81 human arterial thrombi harvested during percutaneous coronary intervention to arterial thrombi collected from mice who underwent experimental provocation of thrombosis using ferric chloride (FeCl₃) or wire injury of the carotid artery. The thrombi underwent detailed histological analysis. The findings revealed that murine arterial thrombi induced by FeCl₃ and those from human coronary thrombi had similar immune cell composition. Both types of experimental thrombi contained primarily neutrophils correlating with NET abundance and coagulation factors. Treatment with the PAD inhibitor chloramidine in thrombosis abrogated NET formation and reduced FeCl₃-induced arterial thrombosis.⁸⁶ These findings agree with those from other mouse experiments describing reduced thrombosis after interference with NET formation as detailed below.

Accumulating evidence also supports a role for NETs in stroke thrombi. Laridan et al. queried 68 thrombi from ischemic stroke patients for neutrophil and NET (citrullinated H3, cH3, and extracellular DNA) markers. The authors found more abundant cH3 in thrombi of cardioembolic origin compared to other etiologies and older thrombi contained significantly more neutrophils and cH3 compared to fresh thrombi, in contrast to findings summarized above for coronary arterial clots. These disparities may reflect the type and time of intervention (thrombus aspiration procedure) but suggest that optimal therapeutic approaches may depend on the site of and time elapsed during vascular thrombosis.

The addition of DNase 1 to tissue plasminogen activator facilitated ex vivo lysis of patient-derived thrombi.⁸⁰ Another study that analyzed 108 thrombi from patients with acute ischemic stroke supported this finding. Histological analysis revealed NET structures in all thrombi concentrating in their outer layers. Quantitative measurement of NET content in thrombi, however, did not associate with clinical outcome but correlated significantly with endovascular therapy procedure length and device number of passes performed to achieve successful recanalization. These results further highlight the potential clinical importance of NETs in thrombosis. Again, ex vivo, recombinant DNAse 1 accelerated tissue plasminogen activator-induced thrombolysis. DNAse 1 alone, however, did not.⁸⁷ This finding may relate to the results of Farkas et al. who compared NET content and related platelet occupancy and fibrin structure of thrombi extracted from patients with acute ischemic stroke, myocardial infarction, or peripheral artery disease. They found the least DNA in thrombi from patients with acute ischemic stroke, perhaps explaining why DNase 1 alone did not disrupt thrombi effectively. DNA/fibrin ratio was significantly lower in acute ischemic stroke thrombi compared to those from peripheral arteries, while the peripheral artery disease thrombi contained less platelets. In contrast, thrombi from all 3 locations contained comparable cH3.⁷⁹ Hence, a combination of tissue plasminogen activator and DNAse I appears most effective in lysing thrombi and addition of DNAse I to DNA-poor thrombi (e.g. stemming from ischemic stroke patients) does not suffice to lyse the thrombus. In addition, presence and involvement of NETs in thrombosis may vary depending on the type of artery involved. But the latter may also reflect on the type and time of intervention (thrombus age and aspiration procedure) and indicate that potential therapeutic targeting will require tailoring depending on to the site of and the timing of treatment.

In mice, neutrophil-derived externalized nucleosomes, which associate with NETs, have undergone study in arterial thrombosis induced by FeCl₃ application. Infusion of an antibody that neutralizes histones into FeCl₃–treated wild-type mice led to prolonged time to occlusion and lower thrombus stability in carotid arteries. Yet, antibody infusion in neutrophil elastase/cathepsin G–deficient mice after induction of vessel injury did not do so. This finding may result from externalized nucleosomes enabling the assembly of neutrophil elastase and its substrate tissue factor pathway inhibitor on the surface of activated neutrophils, actions that may favor thrombosis. Hence, in sterile inflammation, neutrophil-derived serine proteases and nucleosomes may contribute to large-vessel thrombosis, leading to myocardial infarction and stroke.⁸⁸ Single components may exert stronger procoagulant effects than complete NETs.⁷⁷ Yet in vivo, Knight et al. documented enhanced NET formation in New Zealand mixed mice, which have a systemic lupus-like condition driven by type I IFNs, and manifest accelerated vascular dysfunction and prothrombotic risk. Chloramidine-treated New Zealand mixed mice showed reduced NET formation, improved endothelium-dependent vasorelaxation, and prominently delayed time to arterial thrombosis.⁸⁹ In addition, PAD4 accelerates thrombus formation and stability by reduction of clearance of von Willebrand factor (vWf)-platelet string structures in the circulation. This effect might result in part from inhibition of the vWf-degrading proteinase ADAMTS-13 (A disintegrin and metalloprotease with thrombospondin-1-like domain 13) through PAD4.⁹⁰ In contrast, neutrophils from mice treated with the specialized pro-resolving mediator resolvin D4 showed less susceptibility to ionomycin-induced NETosis, suggesting that delivery of such specialized resolution mediators can modulate the severity of thrombo-inflammatory diseases in vivo by interference with NET formation.⁹¹ NETs may also aggravate aspects of thrombosis related to superficial erosion. In mouse experiments that introduce flow disturbance induced by a periarterial cuff in arteries previously injured to mimic characteristics of human lesions complicated by superficial erosion PAD4-deficiency in myeloid cells or DNAse I treatment reduced NET abundance and acute thrombosis. Under these conditions, NETs impair endothelial cell survival, ability to adhere to the basement membrane, and barrier function.³⁹

Potential clinical cardiovascular applications of biomarkers of NETs

A number of studies have examined the relationship of potential biomarkers of NETs and clinical variables, including some discussed above. The validation of such biomarkers remains incomplete, but various investigations have adopted double stranded DNA, MPO-bound to DNA, citrullinated histones, neutrophil elastase among other NET constituents in blood or aspirated thrombi as a way of assessing NET involvement in cardiovascular diseases (Table). Developing and validating more specific biomarkers of NETs presents a major challenge for clinical translation. Such studies can explore mechanistic hypotheses, possibly provide prognostic information, and eventually might serve to allocate specific therapies based on the degree of NET involvement, in the spirit of personalized or precision medicine.⁹²

Table 1:

NET-derived biomarkers in cardiovascular disease

Disease	Biomarker(s)	Correlation*	Study
	In Blood Plasma
STEMI	dsDNA	positive	Shimony et al. 201093
coronary artery disease	dsDNA, cH4, MPO-DNA comlexes	positive	Borissoff et al. 201368
acute coronary syndrome	dsDNA	positive	Cui et al.201394
STEMI	nucleosomes, dsDNA	positive	Mangold et al. 201583
coronary artery disease	dsDNA, nucleosome	positive	Helseth et al. 201695
acute ischemic stroke	extra.DNA, nucleosomes, cH3	positive	Valles et al. 201796
acute ischemic stroke	extra.DNA, nucleosomes, cH4	positive	Hirose et al. 201497
	In Thrombi
in stent thrombosis	extra. DNA, NE	positive	Riegger et al. 201678
acute ischemic stroke, coronary & peripheral artery disease	Extra. DNA, cH3	positive	Farkas et al. 201979
ischemic stroke	NE, cH3, extra.DNA	positive	Laridan et al. 201780
STEMI	extra. DNA, cH3	positive	Mangold et al. 201583
STEMI	extra. DNA, MPO, cH3, TF	positive	Stakos et al. 201584
acute myocardial infarction	extra. DNA, MPO, cH3, cH4, HMGB1	positive	Maugeri et al. 201485
acute myocardial infarction	extra. DNA, NE	positive	Novotny et al. 201886
acute ischemic stroke	extra. DNA, MPO, cH4	positive	Ducroux et al. 201887

^*positive correlation between NET markers and disease severity

Abbreviations: STEMI = ST-elevation myocardial infarction; ds = double stranded; DNA = deoxyribonucleic acid; cH = citrullinated Histone; MPO = Myeloperoxidase; extra. = extracellular; NE = neutrophil elastase; TF = Tissue Factor; HMGB1 = High-Mobility-Group-Protein B1

Borisoff et al. found a relationship between NETs markers and the presence of greater degrees of coronary atherosclerosis in stable coronary artery disease as assessed by various imaging modalities.⁶⁸ Patients with acute coronary syndromes have elevated NET markers. Compared to controls, patients with acute coronary syndromes, particularly those with STEMI, have increased cell free DNA and other NET constituents.^{93, 94,95} Similarly, acute ischemic stroke patients display an increase of NET markers in blood.^{96, 97} Aspirated thrombi from patients with acute coronary syndrome, stent thrombosis, ischemic stroke, and acute peripheral arterial disease also contain NET markers.^{78–80, 83, 85, 87, 98} Clinical variables relate to NET marker concentrations in several studies on patients with atherosclerosis. NET burden associates with infarct size as determined by magnetic resonance imaging,^{83, 99} and worsened outcomes in 2 year follow-up in patients with stable coronary artery disease.^{100, 101}

Concordant with a growing experimental literature cited above,¹⁰² NET-derived products also relate to cardiovascular aspects of other diseases. For example, they associate with atherosclerosis in patients with rheumatoid arthritis.¹⁰³ A cross-sectional study performed in 60 maintenance hemodialysis patients, 30 age- and sex-matched healthy individuals (negative control), and 30 patients with acute infection (positive control) revealed uremia-associated-increased NET formation correlating with an increased burden of atherosclerosis.¹⁰⁴ Considerable evidence links NET formation to the vexing clinical problem of the thrombotic diathesis associated with cancer.¹⁰⁵ For example, a study of cancer-related stroke in 138 subjects showed increased abundance of NET structures.¹⁰⁶

Therapeutic implications of NETs in cardiovascular conditions

The recognition of the contributions of NETs to thrombosis and as an amplifier of inflammation and vascular injury highlights the potential of NET formation and NETs as a therapeutic target. A number of experimental studies cited above have used an inhibitor of PAD4, chloramidine. While useful for demonstrating principles, chloramidine inhibits several isoforms of PAD and therefore lacks specificity that would be desirable for a clinically useful therapeutic. Studies with other inhibitors of PAD4 could provide an approach to limiting NET formation. Beyond limiting NET generation, dissolution of these structures provides another therapeutic avenue. For example, DNase-1 can dissolve NETs by breaking up the DNA strands. DNase has seen applications to a similar end in clearing bronchial mucous rich in neutrophil products, for example in cystic fibrosis. Experimental and initial clinical studies have shown improvements in myocardial infarction following ischemia reperfusion injury by infusion of DNase.^{107, 108}

Myeloperoxidase inhibitors might quell one of the NET-associated generators of ROS. Other pharmacologic interventions to forestall NET formation or limit its consequences proposed include colchicine, inhibitors of complement or phosphodiesterase 4.^109–111 Indeed, the Colchicine Cardiovascular Outcomes Trial (COLCOT)¹¹² showed a significantly lower risk of recurrent ischemic cardiovascular events in patients with a recent myocardial infarction in patients treated with low-dose colchicine vs. placebo. A similar trial, (Colchicine for Prevention of Vascular Inflammation in Non-cardio Embolic Stroke - CONVINCE) will examine the effect of colchicine on stroke outcomes.^{113, 114} Colchicine may act in part by inhibition of inflammasome activity,¹¹⁵ which may also reduce NET burden in recent myocardial infarction or stroke and could contribute to the outcome of the COLCOT trial. On the other hand, some evidence suggests that anticoagulants and anti-platelet therapies may not influence NET accumulation in acute coronary syndrome.¹¹⁶ Moreover, some experimental evidence suggests that NETs can promote macrophage polarization toward reparative functions, and might exert beneficial effects on myocardial healing.¹¹⁷ Thus, NET inhibition could be a double-edged sword, sometimes promoting inflammation, and limiting such responses under some circumstances. This consideration underscores the need for properly powered and conducted clinical trials to evaluate any intervention that targets NETs and their downstream consequences.

Conclusions

Since the initial description of NETs in 2004, the field has burgeoned. Experimental and clinical literature links NETs to numerous cardiovascular conditions, opening up new avenues for mechanistic exploration of pathophysiology. Moreover, the use of biomarkers of NETs may provide a step towards personalized precision medicine by identifying groups of individuals particularly susceptible to certain therapies. In this regard, the recognition of the role of NETs in cardiovascular disease identifies a new set of therapeutic targets currently under intense consideration. The importance of NETs has nonetheless generated some controversy, reinforcing the need for rigor in basic and clinical research in this domain.³²

The experimental and clinical insight into pathophysiologic mechanisms offered by the exploration of NETs has already proven fruitful. The clinical translation of NET biology will require overcoming a number of obstacles. The development, standardization, and validation of biomarkers of NETs requires more work. The field should move beyond small observational studies to include rigorous clinical trials using standardized essays and targeted therapies to evaluate the added value of biomarkers that reflect NET activity and of NET-directed therapies. Despite these challenges, the insights into novel aspects of innate immune and inflammatory contributions to cardiovascular disease afforded by NET research promises to provide new inroads into improved diagnosis and therapy of cardiovascular diseases.

Nonstandard Abbreviations and Acronyms:

AIM2

absent in melanoma 2

ADAMTS-13

A disintegrin and metalloprotease with thrombospondin-1-like domain 13

Apoe

Apolipoprotein E

cH3

citrullinated histone-3

CHIP

clonal hematopoiesis of indeterminate potential

COLCOT

Colchicine Cardiovascular Outcomes Trial

CONVINCE

Colchicine for Prevention of Vascular Inflammation in Non-cardio Embolic Stroke

ds

double-stranded

FeCl₃

ferric chloride

HSPC

hematopoietic stem and progenitor cells

HOCl

highly oxidant and chlorinating species hypochlorous acid

HIPE

histone interference peptides

IFN

interferon α

IL

interleukin

Ldlr

LDL receptor

MMP

matrix metalloproteinase

mCAT

mitochondrial catalase

mitoOS

mitochondrial oxidative stress

MPO

myeloperoxidase

NETs

Neutrophil extracellular traps

NADPH

nicotinamide adenine dinucleotide phosphate

PAD4

peptidylarginine deiminase 4

PMA

phorbol myristate acetate

ROS

reactive oxygen species

STAT

signal transducer and activator of transcription

SMCs

smooth muscle cells

STEMI

ST segment elevation myocardial infarction

T_h

T helper cell

TGF

transforming growth factor

vWf

von Willebrand factor

WD

western-type diet