NET release (or NETosis), as first described by Brinkmann et al[18] in 2004, is an active form of neutrophil death that releases a web of chromatin and antimicrobial proteins into the extracellular space. At the core of NETs are chromatin fibers (about 17 nm in diameter) composed of DNA and histones, positively-charged proteins that normally function in the nucleus to package DNA and regulate gene expression. These fibers are further lined by granule-derived antimicrobial proteins such as neutrophil elastase, myeloperoxidase (MPO), cathepsin G, proteinase 3 (PR3), defensins, and cathelicidin LL-37. NETs target pathogens by a combination of sequestration (preventing their dissemination in the body) and highly-localized microbicidal activity[18]. Both Gram-negative (Shigella flexneri[18], Klebsiella pneumoniae[28]) and Gram-positive (Streptococcus aureus[29], Listeria monocytogenes[30]) bacteria can be targeted by NETs, as can fungi (Candida albicans[31], Aspergillus nidulans[32], Aspergillus fumigatus[33]). NETs have also been shown to be effective in killing particular protozoans and viruses[34-36]. Intriguing recent work has demonstrated that neutrophils are capable of sensing differences in microbe size such that NETs are preferentially released when the neutrophil is confronted by larger pathogens and microbial clusters that cannot be engulfed by phagocytosis[37].

It is also interesting to note that certain microbes have evolved mechanisms for evading NETs. For example, surface modification may dampen neutrophil activation and NET binding[38-40]. Also, pathogen-derived nucleases are well established as destabilizers of NETs[41-43]. That NETs form an important arm of antimicrobial innate immunity is exemplified by the fact that defects in NET generation, or experimental NET depletion, increase susceptibility to various kinds of infections in mice and humans[28,44-49].

NET release can be triggered by a variety of stimuli including microbes, pharmacological agents (phorbol 12-myristate 13-acetate and calcium ionophore[50]), inflammatory cytokines (interleukin 8[51], tumor necrosis factor α[52]), growth factors (granulocyte colony-stimulating factor[53]), activated endothelial cells[54], activated platelets[55], and immune complexes[56]. Following this initial trigger, various pathways intersect to facilitate the extrusion of NETs. For example, some think of NETosis as a variant of autophagy since netting neutrophils display characteristics of autophagy including the formation of autophagosomes[57]. Indeed, inhibition of autophagy-associated signaling prevents NETosis in some contexts[58]. Generation of reactive oxygen species (ROS) by the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase complex has also been considered by many as an absolute prerequisite to NET formation[48,59,60]. Mechanistically, protein kinase C activation[61] and RAF/MEK/ERK signaling[62] lead to phosphorylation of gp91^phox[63], p67^phox[64], and p47^phox[65], which results in assembly of the functional NADPH oxidase complex for ROS generation. However, recent evidence has also shown that activation of SK3 potassium channels, mediated by calcium influx, may lead to an alternative, NADPH oxidase-independent mechanism of NETosis[66].

Once activated, neutrophils preparing for NETosis flatten and adhere tightly to the substratum. ROS are generated and cytoplasmic granules disintegrate releasing their contents into the cytoplasm. Neutrophil elastase then migrates to the nucleus, where it degrades linker histone H1 and processes core histones, thereby promoting chromatin relaxation[28]. This is followed by the translocation to the nucleus of MPO, which also binds chromatin and promotes decondensation, albeit by an unknown mechanism[28]. In addition, the relaxation of chromatin is further promoted by post-translational modification of histone arginine residues to neutral citrullines by the enzyme peptidylarginine deiminase 4 (PAD4)[46,67-69]. Following dissolution of the nuclear membrane, the plasma membrane ruptures casting NETs into the extracellular space[48].

It should be noted that the above description is of what is sometimes called “suicidal” NETosis. However, NETs can also be released in more rapid fashion, in a manner that does not lead to neutrophil death. This concept of “vital” NETosis, which especially occurs in the context of the direct interaction between neutrophils and microorganisms, has been described in detail in a recent review article[70].

Coagulation factor XII, a plasma serine protease capable of activating factor XI and prekallikrein, is recognized as the traditional initiator of the intrinsic pathway. factor XII is well known to be activated by negatively-charged surfaces both in vitro and in vivo, and it turns out that the anionic backbone of NETs (i.e., DNA) is a capable activator of factor XII[80]. It has consequently been shown that factor XIIa (the activated form) can contribute to thrombus formation by both factor XI-dependent and independent mechanisms[81,82].

As mentioned above, histones are positively-charged proteins that normally function to package DNA in the nucleus; they are also the most abundant proteins in NETs. Histones trigger platelet activation and thrombin generation in a dose-dependent manner[83,84]. Indeed, upon treatment with histones, platelets exhibit several activation-associated characteristics such as aggregation, exposure of phosphatidylserines, and surface expression of P-selectin[83,84]. The ability of histones to activate platelets seems to be at least partially dependent on signaling through platelet Toll-like receptor 2 (TLR2) and TLR4[83,84], with a further contribution from several notable intracellular pathways including ERK, Akt, p38, and nuclear factor-κB. Importantly, when histones complex with DNA (as is observed in NETs), their ability to promote platelet activation and thrombin generation is further amplified[84]. Intersecting with coagulation pathways, histone-activated platelets release polyphosphates, which potently promote thrombin activation[84]. Independent of platelets, it has also been suggested that histones contribute to the activation of thrombin by sequestering thrombomodulin and protein C (a natural anticoagulant), and thereby preventing thrombomodulin-dependent activation of protein C[85]. These varied experiments (primarily done with purified components in vitro) have been supported by work in animal models, where infusion of histones promotes DVT formation in mice in the context of inferior vena cava flow restriction[76].

Granule-derived serine proteases, which are among the most abundant non-histone proteins in NETs, potentially engage with blood coagulation in a number of ways. For example, mice deficient in neutrophil elastase and cathepsin G exhibit defects in tissue factor activation, fibrin formation, and thrombus stabilization[86]; in this system, at least one function of the proteases is to degrade an antagonist of coagulation, tissue factor pathway inhibitor (TFPI). Interestingly, the DNA component of NETs is required for the coassembly of TFPI and the proteases, thereby inactivating TFPI at the point of thrombosis[86].

Other mechanisms have also been described. Neutrophil elastase promotes the proteolytic cleavage of the anticoagulant antithrombin[87]. Elastase (in cooperation with matrix metalloproteinase 9) also degrades the proteoglycan network of the arterial media, thereby exposing collagen for VWF binding and platelet adhesion[88]. Further, cathespin G can promote a procoagulant state by cleaving and activating platelet protease activated receptor 4 signaling, thereby enhancing thrombus formation and fibrin deposition under flow conditions[89,90].

In 2012, von Brühl et al[80] showed that the combination of intravascular NET formation and tissue factor are essential for development of thrombi in a mouse model of DVT. The NETs were not only decorated with tissue factor, but also with protein disulfide isomerase, which can activate it. In this system, the tissue factor was felt to originate especially from monocytes, before migrating to, and activating on, the NETs[80]. However, in neutrophils isolated from patients with sepsis, neutrophils themselves seem to be the source of tissue factor, utilizing the machinery of autophagy to deliver tissue factor to the NETs[91]; indeed, in this context, tissue factor-bearing NETs can stimulate both thrombin generation and platelet activation ex vivo.

In one clinical study, 150 patients with symptomatic DVT were compared to controls who had clinical suspicion for DVT, but negative objective testing[96]. As compared to controls, patients with DVT had higher levels of both circulating nucleosomes and activated neutrophils, with elevated levels of either suggesting an approximately three-fold risk of DVT[96]. Another group obtained venous thromboembolism specimens from 11 patients and classified these into various stages of thrombus organization based on morphological characteristics[97]. Immunochemical staining suggested that NETs were especially present in organizing venous thrombi, indicating that they play an important role in thrombus maturation[97].

Experimentally, restriction (stenosis) of blood flow in the iliac vein of baboons[75] or the inferior vena cava of mice[76,80], results in elevation of plasma DNA levels and development of NET-containing venous thrombi. Further, in this model, infusion of histones increases both thrombus size and plasma levels of VWF, with the latter potentially contributing to platelet activation and recruitment[76]. Importantly, neutrophil depletion results in comparatively smaller thrombi[80], as does treatment with DNase[76]. Thrombus formation is also abrogated in PAD4-knockout mice, which are deficient in NET production[79]. In the PAD4 knockouts, thrombosis could be rescued by infusion of wild-type neutrophils[79], arguing that PAD4’s role in thrombosis is at the level of neutrophils (and presumably NETosis).

Correlation studies have hinted at a relationship between DNA, NETosis, and atherosclerotic/atherothrombotic disease[98]. In a cohort of 282 patients with well-characterized coronary artery disease, severity of disease was predicted by levels of circulating cell-free DNA as well as a number of NET markers (nucleosomes, citrullinated histone H4, and MPO-DNA complexes)[98]. Further, these markers also correlated with evidence of active coagulation (soluble CD163 and thrombin-antithrombin complexes)[98]. In mice, NETs can be detected in close association with plaques in the carotid lumen of atherosclerosis-prone ApoE(-/-) mice[99], while the PAD4 inhibitor Cl-amidine (which also blocks NETosis) prevents NET formation and decreases atherosclerotic lesion area in this model[77]. Mechanistically, the cathelicidin-derived proteins LL-37 (human) and CRAMP (mouse), which are abundant in NETs, seem to promote atherosclerosis[100,101]. For example, Döring et al[102] demonstrated that CRAMP-DNA complexes stimulate plasmacytoid dendritic cells (pDCs) to produce type I interferons that promote plaque growth, a phenotype that could be reversed by either CRAMP deficiency or degradation of the DNA backbone of NETs.

Regarding arterial thrombosis, coronary thrombi can be rich in NETs as detected by immunochemical staining[103]; the authors of this study were particularly interested in the role of neutrophil interleukin-17A/F, and indeed both cytokines were present in not only neutrophils, but NETs themselves[103]. It has also been suggested that NETs present in the thrombi of acute myocardial infarctions expose tissue factor, which is functional in activating both thrombin generation and platelets when studied ex vivo[104]. However, that functionality was lost with digestion of the DNA backbone of NETs[104]. Finally, in a mouse model of arterial wall injury by ferric chloride, NET nucleosomes, as well as neutrophil serine proteases (elastase and cathepsin G), are essential for thrombus formation[86].

An important intersection between NETs and the vasculature involves the ability of SLE NETs to engage TLRs and thereby promote the formation of type I interferons by pDCs[111-113]. Type I interferons then play a multifaceted role in endothelial dysfunction, accelerating foam cell formation and impairing endothelial progenitor numbers and function[124,125]. Further, given the abundance of neutrophils in circulation (especially relative to rare cells like pDCs), it is noteworthy that netting neutrophils may themselves be a source of type I interferons in SLE[113,126].

In addition to promoting the production of potentially anti-vascular cytokines like type I interferons, NETs may also play a direct role in endothelial damage in SLE[113]. For example, SLE NETs contain matrix metalloproteinase-9, which activates endothelial matrix metalloproteinase-2, and thereby triggers endothelial cell death[127]. Endothelial damage may be further compounded in SLE by the NET- and MPO-mediated oxidation of high-density lipoprotein (HDL), which causes HDL to lose its normally vasculoprotective properties[128].

The best evidence for a role of NETs in not just the vascular damage, but also the prothrombotic diathesis of SLE, comes from mouse models of the disease. Indeed, NETs play an important role in pathogenesis of some[78,129], but not all[130], SLE models. In the NZM2328 model, NET release can be prevented by treatment with an inhibitor of PAD4 that prevents histone citrullination and consequently NETosis[78]. Over time, PAD inhibition protects against endothelial damage as measured by an acetylcholine-dependent vascular relaxation assay[78]. NZM2328 mice are also prothrombotic at baseline, rapidly forming carotid thrombi after photochemical injury of the endothelium. These carotid thrombi are rich in neutrophils and NETs, and can be prevented by treatment with either DNase or a PAD inhibitor[125]. These findings are reminiscent of work in models of atherosclerosis, where NETs are important in not just vascular damage[100,102], but also thrombosis[77]. Whether prevention of NETosis can protect against thrombotic disease in patients with SLE remains to be determined, although it is noteworthy that antimalarial drugs like chloroquine both block NETosis[128], and track with a reduced risk of thrombosis in patients[131].

Mechanistically, ANCA likely promote NETosis by engaging granule proteins that have migrated to the cell surface in primed neutrophils[134]. Indeed, one study found that ANCA are more potent than SLE IgG in this regard, and further that ANCA-associated NETosis correlates well with vasculitic disease activity[135]. The mechanism of NET induction by a nontraditional ANCA, anti-lysosomal membrane protein-2 (LAMP-2), has recently been investigated in detail. It appears that anti-LAMP-2 directs neutrophils away from apoptosis and toward NETosis by activating the vacuolization typically seen in autophagy[136]. Whether autophagy machinery is also required for NETosis mediated by traditional ANCA (anti-PR3 and anti-MPO) remains to be elucidated.

When NETs form in ANCA patients, they are relatively resistant to degradation by plasma DNase, an effect that is not explained by a direct effect of ANCA on DNase itself[135]. Along similar lines, the anti-thyroid drug propylthiouracil (PTU) is a recognized inducer of ANCA production in humans; in an animal model, PTU leads to the formation of NETs that are particularly resistant to DNase-mediated degradation, thereby exacerbating both pulmonary capillaritis and glomerulonephritis[137]. It was recently shown that ANCA-induced NETs appear to be relatively potent activators of the alternative complement cascade[138], and can also promote both platelet activation and conversion of pentameric C-reactive protein (CRP) into prothrombotic monomeric CRP[139].

NETs have been found in close proximity to inflamed glomeruli in vasculitic kidneys[134], as well as in vasculitic skin lesions[140], arguing that NETs play a role in tissue toxicity. It has also been suggested that NETs play a particular role in ANCA-associated thrombotic events, especially venous. For example, thrombi obtained from ANCA vasculitis patients are particularly rich in both NETs[141] and histone citrullination[142].

An intriguing mechanistic role has also been suggested for tissue factor[143]. Specifically, Kambas et al[143] demonstrated tissue factor-positive NETs in sera, bronchoalveolar lavage fluids, and renal biopsies of ANCA vasculitis patients. Further, tissue factor-positive NETs and microparticles correlated with higher disease activity (similar to thrombosis), and could be induced when control neutrophils were treated with ANCA in vitro[143]. How unique these phenotypes are to ANCA-associated NETs, as compared to NETs that form in other infectious and inflammatory diseases, remains to be determined.

Despite the name of the syndrome (anti-phospholipid), the best understood antigen in APS is not a phospholipid, but rather a lipid-binding protein that circulates at high levels in blood (100-200 μg/mL) called beta-2 glycoprotein I (β₂GPI). Autoantibodies to β₂GPI activate various types of cells in vitro[146-149], and promote both thrombosis and pregnancy loss when injected into mice[150,151]. Currently, three assays are used to diagnose APS clinically. These include tests for (1) anti-cardiolipin antibodies; (2) anti-β₂GPI antibodies; and (3) a group of coagulation assays collectively referred to as “lupus anticoagulant” - functional testing that takes advantage of the fact that antiphospholipid antibodies paradoxically prolong phospholipid-dependent clotting assays in vitro. It is interesting to note that ELISAs for anti-cardiolipin are often actually detecting anti-β₂GPI, with the reactivity to cardiolipin mediated by β₂GPI protein present in the patient’s serum. Antibodies to thrombin may also sometimes cause APS, although testing has not been standardized, and anti-thrombin is therefore not routinely assessed in clinical practice. In summary, this group of antibodies is (despite the inaccuracy) referred to as antiphospholipid antibodies, with anti-β₂GPI being the best characterized and the most likely to be pathogenic.

While antiphospholipid antibodies are recognized to be pathogenic, the origin of these antibodies, and the reason that lupus patients are especially at risk for their development, are not well understood. Further, there are currently no targeted treatments for APS. Instead, therapy focuses on masking the prothrombotic effects of antiphospholipid antibodies with anticoagulant medications like warfarin and heparin. These drugs often need to be taken for life, and at the same time predispose to catastrophic bleeding complications[152]. While anticoagulants are somewhat effective in preventing APS-associated blood clotting, they often have no bearing on the neurologic and renal complications of APS, which can progress to organ failure[145].

Our group has recently made a number of important observations about APS neutrophils[153]. First, NETs circulate at high levels in the plasma of APS patients, even between thrombotic episodes[153]. Indeed, freshly isolated neutrophils from APS patients are primed to undergo spontaneous NETosis when cultured ex vivo (Figure 2). Mechanistically, anti-β₂GPI IgG promotes NETosis by engaging β₂GPI protein on the neutrophil surface; this process is independent of the Fc receptor, but does require ROS production and TLR4 signaling[153]. Further, and pointing to disease relevance, anti-β₂GPI-stimulated NETs promote thrombin generation in vitro[153]. In addition to our work, Leffler et al[154] have shown that some patients with APS have a defect in DNase-mediated NET degradation. This potentially sets up a vicious prothrombotic cycle, in which the threshold for NETosis is reduced in APS neutrophils, followed by the exaggerated persistence of the NETs that do form. A final interesting point is that antiphospholipid antibodies seem to engage not just neutrophils[153], but NETs themselves[154]. This observation deserves further exploration as to its potential role in APS pathogenesis.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Antiphospholipid syndrome neutrophils are prone to “spontaneous” neutrophil extracellular trap release. Freshly-isolated neutrophils from a healthy control (A) or antiphospholipid syndrome (APS) patient (B) were seeded onto poly-lysine-coated coverslips and incubated in serum-free media for 2 h. Samples were then fixed with paraformaldehyde and stained with Hoechst 33342 (DNA = blue) and anti-neutrophil elastase (Abcam, green). Cells were not specifically permeabilized and neutrophil elastase staining is therefore primarily extracellular. These representative micrographs show more neutrophil extracellular trap release in the APS neutrophils, as determined by overlapping DNA and neutrophil elastase staining.