Neutrophil extracellular traps and long COVID

We present early evidence linking the persistence of NETs to pulmonary fibrosis, cardiovascular abnormalities, and neurological dysfunction in long COVID. 



長期COVID,也被稱為後急性COVID-19後遺症,包括一系列COVID-19康復者經歷的系統性症狀。長期COVID的潛在病理機制已成為熱門的研究話題。雖然長期COVID的慢性發炎反應已受到相當的關注,但作為對發炎反應的主要應答者且是所有免疫細胞中最豐富的中性粒細胞的角色卻常被忽視,這可能是由於它們短暫的壽命所致。本文探討了長期COVID患者中觀察到的持續發炎反應中中性粒細胞外陷阱(NETs)的新興作用。詳細內容可參考《Front Immunol》2023年9月27日發表的研究:Front Immunol. 2023 Sep 27;14:1254310. doi: 10.3389/fimmu.2023.1254310. eCollection 2023. Neutrophil extracellular traps and long COVID. 閱讀全文

2. NETs與長期COVID中的肺纖維化、心血管異常和神經功能障礙


3. 未解決的問題與未來研究方向


4. 針對NETs的治療策略開發


5. Neutrophil extracellular traps in long covid

5.1. The theorized role of NETs in long COVID

Sawadogo et al., drawing upon knowledge of NETs and their interaction with the adaptive immune system, hypothesized a link between prolonged neutrophil activation, NET release, and the development of LC (181). They highlighted the fact that NET components—including double-stranded DNA, histones, citrullinated peptides, MPO, and proteinase-3—are unbeknownst to the adaptive immune system and, hence, constitute neoantigens. These neoantigens have the potential to initiate and sustain autoimmune processes by triggering the production of autoantibodies (14, 182). Compelling evidence from conditions like systemic lupus erythematosus and rheumatoid arthritis shows that the presence of autoantibodies can induce NETosis and subsequently protect them from degradation, supporting the notion that NETs can foster autoimmunity by harboring neoantigens (182). This phenomenon would fuel a chronic pro-inflammatory response, activation of the coagulation cascade, and fibrosis (40, 82). Similar to SLE, COVID-19 exhibits the presence of LDNs in circulation, which release NETs enriched with oxidized nucleic acids that possess heightened immunostimulatory capabilities, further enhancing autoimmunity and IFN responses (27). Our group similarly postulated a role for NETs-related autoimmune vasculitides as a mechanism of LC vascular disease (183).

The discussion below focuses on the role of NETs in autoimmunity, lung disease, cardiovascular disease, and neurologic/neuropsychiatric complications seen in LC (Figure 2). We also present emerging data showing cellular senescence-associated inflammation to play a role in LC pathogenesis.

Figure 2

The potential role of NETs in long COVID may involves multiple pathways. NETs have been shown to drive pro-fibrotic responses in the alveolar epithelium and lung fibroblasts, leading to lung fibrosis. An array of autoantibodies detected in long COVID, including anti-NET IgM and IgG, which have been shown to protect NETs against degradation, contributing to downstream inflammation, vascular damage, and thrombosis. NETs are known to contribute to atherosclerosis and myocardial inflammation, perhaps explaining the cardiovascular sequelae in long COVID patients. Neuropsychiatric abnormalities in long COVID are associated with neuroinflammation and blood-brain barrier disruption, which can be induced by neutrophil-endothelial cell interactions and NET production. This Figure was created using

5.2. NETs and autoantibodies in long COVID

Studies have demonstrated that circulating neutrophils isolated from individuals with LC display higher levels of NET induction compared to those of healthy controls (184, 185). Longitudinally following surrogate NET markers for at least 6 months in patients previously hospitalized for COVID-19 demonstrates sustained elevations in serum concentrations of NE, MPO, and cell-free DNA, although they were lower compared to acute COVID-19 patients but still higher than non-COVID controls (186). These findings indicate an incomplete resolution of NETs after acute COVID-19.

Severe COVID-19 features the emergence of autoantibodies against a wide array of proteins including type I IFNs, numerous interleukins, and self-antigens associated with autoimmune diseases, such as anti-nuclear antibodies (ANAs), anti-histone antibodies, and anti-neutrophil cytoplasmic antibody (ANCA), which are components of NETs (187190). Most of the autoantigens identified exist complexed to extracellular nucleic acids—and such would be the case within NETs—which are then recognized by nucleic acid PRRs such as Toll-like receptors (e.g., TLR7) (189). Although the exact mechanism of their pathogenicity is unclear, their presence is positively correlated with disease severity. However, a large cohort study comparing non-ICU and ICU-admitted COVID-19 patients demonstrated that ANAs are not associated with disease severity but rather reflect a dysregulated immune response due to extensive cell death (191).

Importantly, the detection of IgG and IgM autoantibodies against NETs is frequent in hospitalized COVID-19 patients, their levels tracking with increased circulating NET markers and worse disease outcomes (192). The levels of anti-NET IgG inversely correlate with the ability to clear NETs, suggesting that autoantibodies protect NETs from degradation by DNase-1 (192). Similar observations have been made in SLE, where anti-dsDNA and anti-histone antibodies are thought to protect NETs against degradation by circulating DNase (193).

Mechanistically, after an infection, the germinal center B-cell response is crucial for the development of high-affinity antibodies through the processes of class-switching, somatic hypermutation, and affinity maturation. Impaired germinal center B-cell responses, as observed in severe and critical COVID-19 cases, lead to the emergence of extrafollicular B-cell responses (194, 195). This aberrant activation pathway induces the production of polyreactive autoantibodies with limited somatic hypermutation, similar to patterns observed in SLE (196). These polyreactive antibodies, along with an abundant supply of self-antigens from dead or dying cells and NETs, may drive the development of autoantibodies in COVID-19 (197).

In the context of LC, COVID-19 survivors exhibit higher detectable levels of ANAs 3 months and 12 months post-infection compared to age- and sex-matched healthy controls (198). Persistently positive ANA titers are associated with LC symptoms of fatigue, dyspnea, and cough (198). ANA positivity also correlates with higher levels of TNF-α (198). The presence of autoantibodies associated with antiphospholipid syndrome, including anti-cardiolipin, anti-phosphatidylserine/prothrombin, and anti- β2 glycoprotein, is evident in approximately half of hospitalized COVID-19 patients (199). Moreover, these IgG autoantibodies induce NETosis in neutrophils isolated from healthy patients and accelerate venous thrombosis when injected into mice (199). More recently, it was demonstrated that a significant fraction of LC patients exhibit positivity for IgM/IgG anti-cardiolipin and anti-β2 glycoprotein autoantibodies (186). In the same cohort, LC patients demonstrated higher levels of NET markers than healthy controls, indicating incomplete resolution of NETs after recovery from SARS-CoV-2 (186).

These findings draw intriguing parallels between COVID-19 and autoimmune diseases like SLE, highlighting the presence of NETs as a potential source of neoantigens. Dysregulated humoral responses, coupled with the sustained presence of NETs, contribute to the initiation and perpetuation of autoimmunity. However, discerning the pathogenicity of specific autoantibodies in LC remains largely unchartered territory, as existing data does not establish causal relationships.

5.3. NETs in long COVID lung disease

Respiratory symptoms such as cough and dyspnea are frequently observed in individuals with LC (200, 201). A meta-analysis of over 257,000 COVID-19 patients reported that dyspnea persisted for > 12 months after the initial COVID-19 infection in 31% of cases (202). Radiologically, a combination of persistent inflammation (ground glass opacities and consolidation) and fibrosis (fibrotic bands, interlobular septal thickening, and honeycombing) can be observed (203, 204). Persistent inflammation dominates the early post-acute phase, with fibrosis becoming more prevalent during the follow-up period (203). Significantly, approximately 45% of severe COVID-19 survivors will develop pulmonary fibrosis (205). Risk factors for the development of pulmonary fibrosis as LC include various indicators of more severe disease, such as ICU admission, mechanical ventilation, longer hospitalization, and steroid/immunoglobulin treatment (205).

One study by George et al. study has provided substantial insight into the role of NETs in the pathogenesis of LC lung disease, particularly by including a comparison group of patients who experienced complete clinical and radiologic resolution of acute COVID-19 symptoms (206). The authors demonstrated that a subset of severe COVID-19 survivors developed interstitial lung disease (ILD)-related changes on chest computed tomography scan at 3-6 months post-SARS-CoV-2 infection, accompanied by a restrictive pattern on pulmonary function testing. Patients with persistent interstitial changes exhibited significantly elevated neutrophil counts and serum MPO concentrations compared to controls, which positively correlated with the radiologic extent of pulmonary disease. Comparing the plasma proteome of these patients revealed that the neutrophil chemoattractant IL-17 was the only protein significantly associated with persistent ILD on multivariate analysis (OR 3.72, 95% CI 1.20-16.84, p=0.0403) (206). Furthermore, neutrophil chemokines CXCL1 and CXCL8 positively correlated with the degree of restrictive disease on pulmonary function testing, while CXCL8 and the inflammasome-related cytokine IL-18, along with its receptor IL-18R1, were directly associated with the radiological extent of interstitial disease (206). These differences were also reflected in nasal brushing samples taken to model mucosal immunological changes in the upper airways. In vitro experiments using alveolar epithelial cell lines showed that purified NETs increased the expression of fibronectin-1, vascular endothelial growth factor, and alpha-smooth muscle actin, while reducing E-cadherin expression, indicating that NETs drive epithelial-to-mesenchymal transition (EMT) and subsequent extracellular matrix deposition, which are established processes in pulmonary fibrosis (206). These changes were primarily driven by the host response to infection, such as through NETs, rather than direct viral infection of alveolar epithelial cells (206).

These findings are consistent with lung autopsies of deceased COVID-19 patients revealing the presence of alveolar epithelial cells co-expressing mesenchymal markers (207). In vitro experiments demonstrate that co-culturing alveolar epithelial cells with neutrophils, alveolar macrophages, and SARS-CoV-2 virus results in the production of factors such as TGF-β, IL-8, and IL-1β by alveolar macrophages, along with NET production, ultimately leading to a complete EMT signature. Notably, removing either neutrophils or alveolar macrophages resulted in an incomplete EMT phenotype (207). These findings support the notion of an alveolar macrophage/neutrophil/NETosis axis, whereby factors released by alveolar macrophage-derived factors induce NETosis, which, in turn, promotes EMT in pneumocytes.

Another study explored the role of Kruppel-like factor 2 (KLF2) in pulmonary sequelae of LC (208). KLF2 is a fibroblast protein, and its downregulation has been implicated in fibrosing disorders (209, 210). Lung fibroblasts stimulated with plasma from severe COVID-19 patients downregulate KLF2 and acquire a pre-fibrotic phenotype (208). Treating lung fibroblasts with a combination of DNAse-1 (to degrade NETs) and JAK/IL-6 inhibitors baritinib/tocilizumab (to attenuate inflammation) normalized KLF2 expression. Significantly, COVID-19 patients treated with this combination showed better outcomes compared to those receiving standard-of-care therapy. Furthermore, exposing lung fibroblasts to the plasma of treated patients resulted in higher KLF2 expression (208).

Together, these findings suggest that NETs and the inflammatory environment in the circulation and lung parenchyma of COVID-19 patients, particularly in severe cases, induce fibrotic phenotypes in alveolar epithelial cells and lung fibroblasts, which may explain the development of pulmonary fibrosis observed in a significant proportion of severe COVID-19 survivors.

5.4. NETs in the cardiovascular manifestations of long COVID

Several prospective and retrospective studies have consistently demonstrated a higher incidence of vascular pathologies, such as arterial thrombosis, venous thrombosis, atherosclerosis, vasculitis, and hypertension, in patients with LC (211214). Notably, a prospective study of 153,760 COVID-19 patients revealed that convalescent individuals had a significantly higher risk of developing future cardiovascular disease, cerebrovascular disease, thromboembolic events, and ischemic heart disease compared to healthy contemporary and historical controls (211). This highlights a distinctive vascular feature of LC, characterized by widespread activation of pro-coagulant pathways (215218). Indeed, elevated levels of pro-inflammatory and pro-thrombotic mediators have been observed in LC patients compared to healthy controls (219). It has recently been shown that NETosis persists at a greater level in LC patients compared to convalescent recovering patients (184). Persistent activation of pathways related to immunothrombosis and neutrophil activation has also been observed in COVID-19 survivors 6 months after the initial SARS-CoV-2 infection (220).

NETs are critical mediators of immunothrombosis and endotheliitis, exerting their function through various mechanisms (221, 222). NETs have also been implicated in the development of atherosclerosis, where endothelial cells promote NET formation. In turn, NETs lead to an increase in pro-inflammatory signaling and subsequent recruitment of immune cells to atherosclerotic plaques (223228). NETs have also been shown to participate in the pathophysiology of ANCA-associated vasculitides (229), suggesting a potential link between autoimmunity, NETs, and vasculopathy. Promising results have been observed with NET-targeted therapies in the treatment of vascular pathologies, further supporting the involvement of NETs (230233). Additionally, a larger burden of NETs has been observed in the coronary thrombi of COVID-19 convalescent patients with ST-elevation myocardial infarction when compared to historical controls (12, 234). These findings provide evidence of plausible evidence of NETs in the pathogenesis of vascular disorders in LC.

Cardiac involvement is an archetypal feature of LC and encompasses a wide range of presentations, including cardiac inflammation, cardiac fibrosis, dysrhythmias, ischemic heart disease, and cardiac impairment (235, 236). Cohort studies utilizing cardiac magnetic resonance imaging in patients with LC have demonstrated evidence of impaired ventricular function in addition to increased cardiac edema and inflammation (237239). Additionally, patients with LC demonstrated a heightened incidence of myocardial injury particularly in the form of myocardial fibrosis (240). The interplay between NETs and cardiac disorders in several viral and bacterial infections was recently reviewed (241). Clinical evidence for the involvement of NETs in COVID-19-induced cardiac inflammation has recently been established through autopsy reports from 21 SARS-CoV-2 infected individuals, showing the presence of NETs in all patients and its association with myocarditis and cardiac injury (242). Additionally, targeting NETs in a mouse model of COVID-19 through DNase I therapy resulted in the attenuation of cardiac injury (243). Neutrophils recruited to the heart via the cytokine midkine (MK), which then induces NETosis, contribute to the pathogenesis of myocarditis and cardiac inflammation (244). Targeting MK attenuates neutrophil infiltration and NET formation, associated with a reduction in ventricular systolic dysfunction and myocardial fibrosis (244). In vitro studies have highlighted a potential role for NETs in the development of cardiac fibrosis through enhancing fibroblast migration and promoting cardiac myofibroblast differentiation (245). These findings suggest a potential role for NETs in non-ischemic cardiac injury in LC.

5.5. NETs in neurologic manifestations of long COVID

Another prominent feature of LC is the array of neurological complications the sizeable chronic burden of which has been indicated by numerous studies (246249). On a gross scale, a UK BioBank study analyzed brain MRI scans pre- and post-infection and observed a greater clinical burden of cognitive decline and radiological changes of reduction in brain size and gray matter atrophy affecting the hippocampus and orbitofrontal cortex in the post-infection group (250).

Mechanistically, direct viral invasion of the brain does not appear to be prominent except in very severe acute COVID-19 cases (7). Instead, neuroinflammation, micro-clots, and the activation of CNS-resident glial cells are hypothesized to be crucial mediators of LC-associated neurological sequelae (251). Experiments in ACE2-transgenic mice have indicated that even a mild SARS-CoV-2 respiratory infection raises systemic cytokine levels, such as CCL11, which can be neurotoxic by inducing reactive states in microglia (252). Indicators of glial cell reactivity are elevated amongst LC patients with persistent depressive symptoms (253). A study of 76 LC patients experiencing “brain fog” (encompassing headache, fatigue, malaise, and altered level of consciousness) was the first to objectively demonstrate COVID-19-associated BBB disruption by utilizing neurological biomarkers and dynamic contrast-enhanced magnetic resonance imaging (254). Sustained elevations in S100β, IL-8, TGF-β, and GFAP were observed in brain fog LC patients, indicating persistent inflammation. Importantly, the adhesion of peripheral blood mononuclear cells to brain microvascular endothelial cells was enhanced in patients with brain fog, and exposing endothelial cells to the serum of these patients triggered endothelial cell activation (254).

A distinctive feature of LC is the higher propensity for developing ischemic strokes (251). Lee et al. documented the presence of microvascular injury indicated by scattered microthrombi, endothelial activation associated with the adhesion of autoantibodies and complement proteins, and BBB disruption indicated by the perivascular presence of fibrinogen in brain autopsies of individuals who died suddenly with or after COVID-19 (255, 256). Co-localizing with fibrinogen were microglia/macrophages, CD8+ T-cells, and reactive astrocytes, with areas of neuronal loss observed from microglia phagocytosis (255, 256). Fibrin clots are also generally elevated in the blood of LC patients (216). A recent study associated elevated serum fibrinogen and D-dimer levels relative to C-reactive protein during acute admission with neurocognitive deficits 6 and 12 months after acute COVID-19, supporting the notion of an acute inflammatory response being responsible for the long-lasting neurological effects of LC (257). Together, these findings suggest that BBB disruption during acute COVID-19 leads to the spillover of fibrinogen which, along with propagating a hypercoagulable state and micro-clots (258), foster neurotoxic resident glial cell phenotypes that damage neurons in a subtle but debilitating manner with long-lasting neurocognitive consequences (see (259) for a detailed review on the neurological effects of fibrinogen).

The brain is typically devoid of neutrophils and NETs due to the integrity of the BBB (260). However, in scenarios where the BBB is compromised, neutrophils infiltrate and NETs are often visualized in brain tissue where they contribute towards ongoing neuroinflammation (260, 261). Recent studies proposed that COVID-19 facilitates BBB disruption by increasing the expression of matrix metalloproteinase-9 (MMP-9) which leads to basement membrane degradation (262, 263). The levels of NETs in the brain have been directly related to the degree of neuroinflammation in model organisms of Alzheimer’s disease, meningitis, ischemic stroke, and traumatic brain injury (264). That being said, direct evidence of the involvement of NETs in cerebral micro-clots and neuroinflammation in acute and LC is lacking. The fact remains that many of the autoantibodies (82), complement proteins (121), fibrinogen (265), Von-Willebrand factor (266), and platelets (267) that are hypothesized to be key to the micro-clot formation in LC and driving its neurological manifestations are known to be intertwined with neutrophil biology and NET production. Hence, directing research efforts toward identifying and tackling NETs may improve our understanding of the neurological sequelae of LC.

5.6. Cellular senescence and NETosis

Cellular senescence refers to a state of cell cycle arrest accompanied by the release of inflammatory molecules collectively termed the senescence-associated secretory phenotype (SASP) (268). Cellular senescence is associated with a chronic low-grade inflammatory state and is causally implicated in aging and various chronic diseases, including pulmonary fibrosis, neurodegeneration, and cardiovascular disease (269277). Viral infections, including SARS-CoV-2, trigger a cellular stress response, culminating in the induction of senescence, termed virus-induced senescence (VIS) (278).

The intracellular signaling pathways mediating SARS-CoV-2—induced VIS have been extensively reviewed elsewhere (278). Suffice it to say that multiple studies have shown elevated levels of senescence markers in the upper and lower respiratory tract of COVID-19 patients, indicative of SARS-CoV-2 infection-related VIS, which results in the elaboration of a pro-inflammatory SASP that recruits and induces pro-inflammatory M1 phenotypes in macrophages (279282). Notably, some samples with a high burden of senescent cells did not show detectable viral infection, suggesting that senescence can persist even after clearance of SARS-CoV-2 (278, 280). Linking senescence induction to COVID-19 pathology, Lee et al. demonstrated that supernatant from SARS-CoV-2-induced VIS cells can induce endothelial cell senescence or apoptosis, promote M1 macrophage polarization, activate platelets, trigger NET production, and accelerate thrombosis (280). Treating with senolytic drugs, which eliminate senescent cells by causing their selective apoptosis, reduced the levels of SASP-reminiscent pro-inflammatory cytokines in hamster models of SARS-CoV-2 infection. However, the histopathological impact on inflammation and thrombosis did not significantly improve with senolytic treatment (280).

In the context of LC, it is important to note that senescent cells are typically cleared by the immune system because of their chemoattractive SASP (283). However, in certain circumstances, such as tumorigenesis, senescent cells can persist and contribute to long-term disease recurrence by evading immune surveillance (284286). Although data on whether senescent cells persist and drive LC phenotypes are currently lacking, ongoing studies have demonstrated SARS-CoV-2 infection-induced VIS of human brain organoids in corticothalamic neurons and GABAergic ganglionic eminence neurons, which are responsible for modulating neuronal circuitry and processing of sensory information (287). SARS-CoV-2 was shown to induce the loss of dopaminergic neurons in the brainstem responsible for coordination and consciousness, potentially explaining abnormalities in these processes in LC (287). However, multiple studies have suggested that direct SARS-CoV-2 infection of the brain likely plays an insignificant role in acute and long neuro-COVID (251).

Additionally, the downstream consequences of senescence induction and pro-inflammatory SASP elaboration—in terms of changes in neuronal function, synaptic plasticity, astrocyte or microglial activation, or the recruitment of circulating innate and adaptive immune cells—are currently unclear. The BBB disruption in LC does support the role of neutrophils and NETs, as is indicated by their role in other CNS pathologies that involve BBB disruption (261). Other than the data above, no studies directly observing senescent cell persistence and their roles in LC in other organ systems such as the lungs have not yet been conducted, but several hypotheses have been put forward (288, 289). Exploring the link between senescence and NETosis is particularly important given the rapid evolution of senolytic therapies into clinical trials and the feasibility of targeting senescent cells; such research may uncover therapeutic targets with great potential for translation into clinical applications.