Despite recent advances in medicine, sepsis remains one of the leading causes of death in critically ill people and animals (1–3). Sepsis is defined as life-threatening organ dysfunction caused by a dysregulated host response to infection (4). Multiple organ dysfunction, associated with increased mortality and morbidity, is a common manifestation of sepsis (1, 5). Factors such as virulence of the invading organisms, co-morbidities, and the host immunocompetence dictate the progression and outcome of sepsis (3, 6, 7).

Neutrophils are short-lived granulocytes that play a pivotal role in the initial defense against invading pathogens in mammals. Neutrophils, recruited to the site of infection, effectively kill microorganisms by phagocytosis, degranulation, and generation of reactive oxygen species (ROS) (8). In certain conditions, neutrophils enhance their antimicrobial properties by releasing neutrophil extracellular traps (NETs), composed of extracellular chromatin decorated with histones and numerous granular proteins (9). Many of these granular components like myeloperoxidase (MPO), α-defensins, elastase (NE), cathepin G, and lactoferrin, have bactericidal activities capable of eliminating microorganisms and/or their virulence factors. Uncontrolled inflammatory response during sepsis is the proposed underlying cause of excessive NET formation (10, 11). Increasing experimental and clinical evidence indicates that overzealous NET formation during sepsis can lead to the development of multiple organ dysfunction highlighting the pathophysiological role of NETs in sepsis (12–15). This review aims to summarize the recent knowledge on the underlying mechanisms of NET formation in varied species, as well as, the beneficial and detrimental effects of NETs found in various septic animal models.

As sentinel cells of innate immunity, neutrophils can respond to many pathogens or their associated molecular patterns by releasing NETs. “NETosis” is the term commonly used to describe the sequence of cellular events leading up to the active release of NETs (9, 16). Similar to other forms of cell death such as apoptosis or programmed cell death and necroptosis, a regulated form of necrosis, NETosis is a highly regulated process. Dysregulation of NETosis found in many disease states like sepsis, can result in collateral damage to the host. The cellular mechanisms mediating the release of NETs, however, remain poorly understood. Brinkmann et al. and Fuchs et al. first documented in vitro NETosis in human neutrophils using the potent protein kinase C activator, phorbal 12-myristate 13-acetate (PMA) (17, 18). Following PMA stimulation, human neutrophils undergo morphological changes including chromatin decondensation, loss of nuclear envelope, mixing of nuclear contents and cytoplasmic granular proteins, loss of membrane integrity and, ultimately, release of cell free DNA (cfDNA) (18). A recent ex vivo study by the authors documented similar morphological changes in PMA-activated canine neutrophils indicating that dog neutrophils may undergo suicidal NETosis (19). Cell death is inevitable in neutrophils undergoing suicidal NETosis as they are unable to maintain a constant intracellular environment without an intact cell membrane. For that reason, some investigators describe this type of NETosis as “lytic” or “suicidal” (16). Table 1 is a summary of microorganisms known to induce NETosis in various species.

The generation of reactive oxygen species (ROS) by NAPDH oxidase is an integral, but not essential, cellular process in NETosis. Suicidal NETosis induced by PMA and pathogens such as Aspergillus fumigatus and Toxoplasma gondii are dependent on ROS generation, which occurs upstream of p38 mitogen-activated protein kinase (MAPK) and extracellular signal regulated kinase (ERK) phosphorylation (45, 46) (Figure 2). Interestingly, although PMA-activated neutrophils in horses undergo ROS generation, PMA is considered a weak trigger of ex vivo NETosis, suggesting that the role of ROS in NETosis may vary among species (24, 47) (Figure 1A). It is not yet clear how ROS generation and its downstream effects ultimately lead to chromatin decondensation and the release of cfDNA. Papyannopoulos et al. found that the translocation of both neutrophil elastase (NE) and myeloperoxidase granules from the cytoplasm to the nucleus is essential for chromatin decondensation during PMA-mediated NETosis. This process appears to be independent of the enzymatic activities of NE but the exact molecular mechanism responsible for this translocation is unclear (48).

Many critics question the physiological relevance of PMA-mediated NETosis (49, 50). First, NETosis induced by PMA in vitro requires hours to occur, whereas neutrophils in vivo normally undergo phagocytosis and degranulation within minutes after encountering microorganisms. Second, some investigators consider PMA-induced NETosis “suicidal,” given that PMA-activated neutrophils can no longer maintain normal cell function following NETosis. To better understand the relevance of NETosis in vivo, Yipp et al. directly visualized the behaviors of neutrophils within Staphylococcus aureus (S. aureus) skin infections in mice and human beings using intravital microscopy. They found that neutrophils undergo chromatin decondensation and DNA release within minutes after encountering S. aureus while maintaining an intact cell membrane. Interestingly, the NETosing neutrophils, now anuclear, continued to chemotax to and phagocytize nearby bacteria (20). This unique mechanism of NETosis, termed “vital” NETosis, was later confirmed in septic murine models using either lipopolysaccharides (LPS) or E. coli. Besides maintaining their functional capacity following vital NETosis, neutrophils must somehow release their DNA while preserving the integrity of the cell membrane. The Kubes laboratory answered this question by demonstrating that in the presence of S. aureus human neutrophils form budding DNA-containing vesicles from the nuclear envelop, which later, fuse with the plasma membrane to release DNA to the extracellular space (22) (Figure 2). In horses, ex vivo exposure of neutrophils to endometritis causing bacteria, also results in rapid NET formation but the viability and functional capacity of those neutrophils remain unknown (24). To date, how neutrophils commit to one form of NETosis over the other remains unclear, but some investigators believe that it may be stimulus-dependent (11, 16).

Studies in murine, human, and canine neutrophils demonstrated that NETosis and the release of histones and DNA require histone post-translational modification (30, 51, 52). Histone citrullination, catalyzed by the enzyme, peptidylarginine deiminase 4 (PAD4), results in a net loss of positive charge of histones. This, in turn, obliterates electrostatic interactions between DNA and histones causing chromatin decondensation and release of cfDNA during NETosis (53–55). Induction of PAD4 activation in neutrophils appears to be stimuli dependent. The proinflammatory cytokine, Tumor Necrosis Factor-α, hydrogen peroxide and molecular patterns like LPS and lipoteichoic acid have been shown to induce PAD4 activation and histone citrullination (19, 30, 32, 55). The elevation of citrullinated histones in clinical septic patients suggest that ongoing activation of PAD4 occurs during sepsis.

In addition to being the primary effector cells of hemostasis, recent evidence indicates that murine, human and canine platelets play a direct role in innate immunity by directly interacting with pathogens or recognizing pathogen-associated molecular patterns (PAMPs). Platelets also augment innate immunity by facilitating NETosis via platelet-neutrophil interaction (26). In ex vivo systems, human, murine, and canine neutrophils undergo a limited degree of NETosis in response to LPS (19, 31, 56). But in the presence of LPS-activated platelets, murine and human neutrophils release substantial amounts of NETs, indicating that sepsis-associated NETosis is highly dependent on platelet-neutrophil interaction. Clark et al. first discovered that platelet-mediated NETosis induced by LPS in humans is dependent on platelet Toll-like receptor 4 (TLR4) (26). Subsequent studies in mouse sepsis models showed that intravascular NETosis under shear conditions is highly dependent on platelet-neutrophil interaction. McDonald et al. found that LPS-treated mice not only had increased platelet-neutrophil aggregates but also NETs within their liver sinusoids. However, this process does not occur in platelet-depleted mice treated with the same doses of LPS (27). In addition to LPS, platelets activated by the agonists, thrombin, ADP, and arachidonic acid also stimulate NETosis in humans and mice (57). The exact mechanism of platelet-mediated NETosis, however, is not clearly understood and, likely, differs among species.

In general, platelet-neutrophil interactions can be broadly divided into 2 mechanisms: adhesive and soluble. Regardless of the mechanisms involved, platelet activation induced by pathogens, damage-associated molecular patterns, or classic agonists must occur in order for secretion of soluble mediators and expression/activation of adhesion molecules to occur. Platelet activation by stimulants results in inside-out signaling that leads to the expression of P-selectin and a conformational change in the extracellular domains of the integrin, α_IIbβ₃, increasing its affinity for ligands. The binding of platelet P-selectin to its neutrophil receptor, P-selectin glycoprotein ligand-1 (PSGL-1) is shown to be essential in inducing NETosis in mice with sepsis and transfusion-induced acute lung injury (58, 59). However, treatment of activated human platelets with anti-P-selectin function blocking antibodies does not attenuate NETosis suggesting that P-selectin/PSGL-1-mediated NETosis may be species dependent (60, 61). In mice and humans, platelet-neutrophil interaction and NETosis also are mediated by the binding of α_Mβ₂ (MAC-1), a neutrophil integrin, to its counterreceptor, glycoprotein 1bα (GP1bα), a heterodimeric glycoprotein on platelets (62–64).

Upon activation, platelets secrete a variety of soluble mediators like high mobility group box-1 (HMBG-1) and platelet factor 4 (PF4), known to modulate NETosis. Platelet factor 4 or CXCL4 and CCL5 (RANTES), released by LPS-activated platelets, are platelet-derived chemokines that can activate neutrophils to undergo adhesion. In mice, thrombin-stimulated platelets release both CXCL4 and CCL5, which induce NETosis via the neutrophil G-protein coupled receptors. The heterodimerization of platelet-derived CXCL4 and CCL5 are found to enhance NETosis (64). LPS or the synthetic lipopeptide, Pam3CSK4, also stimulates platelets via TLR 4 or 2, respectively, to release CXCL4 enhancing NETosis in human neutrophils (61). Platelet-derived HMGB-1, a damage-associated molecular pattern, secreted by activated platelets, stimulates human and mouse neutrophils to undergo NETosis by binding to receptor for advanced glycation endproducts (RAGE). The role of HMGB1-RAGE signaling pathway in the formation of NETs is unclear but downstream pathways of RAGE have been shown to facilitate autophagy, an essential cellular event during NETosis (60, 65–67). Activated neutrophils also contribute to this cellular cross-talk by shuttling arachidonic acid containing extracellular vesicles to platelets, in which arachidonic acid is synthesized to thromboxane A₂ by cyclooxygenase-1 for further platelet activation (68). It is important to note that the adhesive and soluble interactions between platelets and neutrophils synergistically mediate NETosis since inhibition of either type of interaction inhibits NET formation. Table 2 summarizes the molecular interactions between neutrophils and platelets during NETosis.

While NETs protect the host by limiting microbial growth and dissemination, excessive NETosis during sepsis can be detrimental to the host. Recent discoveries in in vitro experiments and animal models demonstrated the crucial role of NETs in the pathogenesis of intravascular thrombosis, disseminated intravascular coagulation (DIC), and multiple organ dysfunction, all of which can increase morbidity and mortality in sepsis (1, 5).

Since the outcome in sepsis depends heavily on early recognition and interventions, clinical assessment of NETs may serve as a valuable biomarker for the early diagnosis of sepsis. Direct visualization and quantification of NETs using immunofluorescence microscopy in clinical samples can be challenging since this technique is labor intensive and requires advanced training in microscopy. Preliminary studies using flow cytometry and advanced sequencing to quantify surrogates of NETosis like cfDNA, histones and nucleosomes in serum or plasma have shown promising results as these methods are more objective, reliable and repeatable than microscopy(96). Because overzealous production of NETs causes organ dysfunction and mortality, therapeutic interventions that target NET production or individual NET components present novel treatment strategies for sepsis. As mentioned previously, elevated levels of cfDNA released from NETosing neutrophils can have detrimental effects by activating the coagulation system and inflammation. One study showed that delayed systemic treatment of recombinant DNase in a mouse sepsis model reduces organ damage and bacterial dissemination, while early administration (2 h after cecal ligation) yields the opposite effects suggesting that NET-targeted therapy may be time-dependent (14). Interestingly, findings from a different study showed that early and concurrent treatment with DNase and antibiotics resulted in improved survival, reduced bacteremia and organ dysfunction(15). This indicates that combined therapies that incorporate conventional treatments such fluid therapy, antibiotics and NET-targeted drugs can potentially optimize treatment efficacy and outcome in clinical septic patients. To the authors' knowledge, there are no clinical trials to date evaluating the use of systemic DNase in clinical sepsis.

Treatment of PAD4 is a suitable therapeutic target because of its essential role in sepsis-mediated NETosis. In a mouse model of lupus, systemic treatment with the PAD4 inhibitor, BB-Cl-amidine, protects mice from developing NET-mediated vascular damage, endothelial dysfunction and kidney injury (97, 98). Sepsis models utilizing PAD4 knockout mice demonstrated that PAD4 deficiency improves survival and decreases the severity of organ dysfunction without exacerbating bacteremia (99, 100). Nonetheless, since the long-term physiologic consequences of systemic PAD4 inhibition are unknown, developing a suitable targeted therapy for PAD4 can be challenging.

Given its proinflammatory, cytotoxic and prothrombotic properties, citrullinated histone H3 (citH3) is a potential molecular target in sepsis. By neutralizing circulating citH3 in a septic mouse model, Li et al. found that blockade of citH3 significantly improves survival (101). Non-anticoagulant heparin, which binds to extracellular histones with minimal affinity to antithrombin, has been shown to reduce histone-mediated cytotoxicity, attenuate endotoxin-mediated lung injury and improve survival in septic mice (102). Although citrullinated histones have been found in canine and equine NETs, further studies are needed to characterize their effects in these species (Figure 1) (30). Activated protein C, a natural anticoagulant, cleaves extracellular histones abolishing the ability of histones to induce ex vivo platelet activation and potentially its cytotoxic and proinflammatory properties (69). However, large-scale randomized clinical trials evaluating the efficacy of recombinant protein C for human septic patients did not demonstrate any clinical benefits of protein C, ultimately, leading to its withdrawal (103).

Because platelet activation and platelet-neutrophil interaction are crucial for NETosis to occur, antiplatelet therapy may attenuate NETosis and its detrimental effects in sepsis. Recent studies in human and canine platelets have shown that endotoxin-mediated platelet activation requires excessive production of eicosanoids like thromboxane A₂ (104, 105). In a mouse model of endotoxin-triggered acute lung injury, pretreatment of mice with acetylsalicylic acid or aspirin, which prevents thromboxane A₂ generation, decreases intravascular NET formation and the degree of lung injury (57). Inhibition of the platelet ADP receptor, P2Y12, may also attenuate platelet-neutrophil interaction and NETosis, as endotoxin-mediated platelet activation is dependent on ADP in canine and equine platelets (19). Interestingly, prehospital administration of antiplatelet therapy has been shown in several observational studies, to be associated with improved outcome in human clinical patients with sepsis(106, 107). A large-scale clinical trial evaluating the benefits of aspirin therapy in sepsis is currently underway (108).

Evidence in the literature indicates that NETosis is a highly conserved mechanism of innate immunity among numerous species. Research in mice and people demonstrates that the dysregulation of NETosis caused by sepsis can have detrimental effects resulting in inflammation, thrombosis and multiple organ failure. NETs may play a similar pathophysiological role in other species. NETosis and NET components are potential therapeutic targets for the treatment of sepsis.

All authors contributed substantially to the conception of this review. RL drafted it and FT and RL revised it critically. All authors approved the final version of this review.