Neutrophils, or polymorphonuclear granulocytes (PMNs), are the most abundant of the circulating white blood cells and have traditionally been considered a homogenous population of cells that execute stereotypical antimicrobial effector functions. This view has been supported by the short lifespan of neutrophils and their lack of proliferative capacity. These effector functions include phagocytosis, the production of reactive oxygen species (ROS) via the respiratory burst, degranulation to release cytotoxic neutrophil granule enzymes, and release of neutrophil extracellular traps (NETs) that trap microbes to sequester them in locales of high antimicrobial activity and prevent their dissemination [1]. Through these functions, neutrophils enhance local tissue inflammation and microbial death, but, if these processes are dysregulated, exert significant collateral damage on host tissues.

This classical conception of neutrophils has largely been replaced with an understanding that considers neutrophils a heterogenous population comprised of various unique functional states capable of mounting different effector functions in health and disease. The discovery of the following functions of neutrophils has led to their consideration as orchestrators of innate and adaptive immune responses in infectious and sterile disease states. Neutrophils can produce pro- and anti-inflammatory cytokines, chemokines, and immunomodulatory factors that can modulate the recruitment and phenotypes of neutrophils, macrophages, dendritic cells (DCs), T cells, and B cells ^[2][3][2,3]. Although neutrophils are substantially less efficient than macrophages or DCs at producing cytokines and chemokines, their sheer abundance at sites of inflammation—several folds of magnitude greater than macrophages and DCs—suggests that they contribute significantly to shaping the inflammatory milieu and, by extension, the immune landscape at these sites. Secondly, NETs can modulate the activation of DCs, T cells, and B cells in chronic inflammatory states like autoimmune diseases and cancer ^[4][5][4,5]. These roles of NETs are focused on in this entry instead of their well-established contribution to antimicrobial defense and thrombosis ^[6][7][6,7]. Lastly, cytokine-stimulated neutrophils can express class II major histocompatibility complex (MHC), costimulatory molecules, and the antigen-processing machinery required to present antigens to CD4⁺ T cells, thereby functioning as atypical APCs [8].

These functions, the continuous production of neutrophils, their abundance at sites of inflammation, and data now showing that the lifespan of neutrophils may be significantly prolonged by certain cytokines at inflammatory sites like the tumor microenvironment (TME) suggest that neutrophils, alongside macrophages and DCs, are key orchestrators of the immune response. Neutrophils are also key to wound repair after injury and a reversion to homeostasis. There is great interest in characterizing the biological implications of these aspects of neutrophil biology in not only disease but also homeostasis to reveal potential disease biomarkers and targets for therapeutic interventions.

A central mechanism by which SARS-CoV-2 causes severe disease is by delaying the production of type I IFNs, which are key initiators of antiviral immunity ^[9][10][11][188,189,190]. This delay in type I IFN responses and T-cell recruitment is associated with an exaggerated myeloid cell—particularly neutrophil—response ^[12][13][14][191,192,193]. Salient features of severe COVID-19 include an elevated neutrophil-to-lymphocyte ratio ^[15][194], the emergence of immature neutrophil populations in both the blood and lungs ^[16][17][18][195,196,197], and activated pro-inflammatory neutrophil phenotypes in both the circulation and lungs with an enhanced production of TNF-α, IL-6, CXCL8, alarmins like calprotectin (SA100A8/A9), and NETs ^[19][20][21][198,199,200]. Through these mechanisms, neutrophils contribute to local tissue damage, amplify lung inflammation, and activate platelets and the coagulation cascade via NETs to promote thrombosis ^{[22][23][24][25]}[201,202,203,204].

Profound metabolic derangements are observed in neutrophils isolated from severe COVID-19 patients who develop acute respiratory distress syndrome (ARDS) and require intensive care unit (ICU) admission ^[26][205]. These include a depletion of intracellular histidine, higher β-alanine, and elevated oxidative stress, but the impact of these changes on neutrophil function is not known. In addition, the activity of the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is reduced in patients with severe COVID-19, which decreases glycolysis and increases the activity of the HMP shunt. Functionally, the decrease in GAPDH activity is linked to increased NET production, but this is independent of its effects on increasing HMP shunt activity and NADPH production ^[26][205].

The transcriptomic landscape of circulating neutrophils is significantly dysregulated in severe COVID-19, with the emergence of immature and dysfunctional neutrophils and PD-L1⁺ neutrophils and loss of IFN-active neutrophils ^[20][27][28][199,206,207]. PD-L1⁺ neutrophils decrease in patients recovering from severe disease, whereas IFN-active neutrophils appear in both mild and severe COVID-19 ^[29][30][208,209]; hence, their relevance to the disease process is currently unclear. These findings likely mirror the temporal dynamics of type I IFN responses in varying disease severities. A failure to develop early type I IFN responses is associated with severe COVID-19, but conflicting evidence suggested that type I IFNs and their stimulated genes are elevated in severe COVID and correlate directly with mortality and levels of pro-inflammatory cytokines like TNFα and IL-6 ^{[9][30][31][32][33]}[188,209,210,211,212]. Many of these differences could be partly explained by between-study variations in definitions of severe disease, times of sampling, and methods to measure type I IFNs. A recent study more definitively showed that levels of type I IFNs are indeed decreased in severe COVID-19 ^[34][213]. A model of COVID-19 pathogenesis from an IFN perspective states that their production in early-stage disease results in timely antiviral responses and the recruitment of adaptive immunity, whereas their delayed production in severe COVID-19, as a consequence of a profoundly dysregulated immune system, may enhance innate immune inflammation ^[35][214]. This was demonstrated by an ex vivo stimulation of the whole blood of COVID-19 patients by type I IFN, which induced an inflammatory response in leukocytes of severe COVID-19 patients but not in those with mild/moderate disease ^[34][213]. Dexamethasone treatment to alleviate immunopathology of severe COVID-19 has been shown to induce the depletion of IFN-stimulated neutrophils ^[29][208], while the high level of NET production by neutrophils escapes modulation by dexamethasone treatment ^[36][215]. These findings indicate that combining conventional immunosuppression with NET-targeting therapies may enact a synergistic effect on attenuating COVID-19 hyperinflammation and immunopathology.

Long COVID, also known as post-acute COVID-19 sequelae (PACS), describes the multi-system, non-specific symptoms that survivors of COVID-19 develop after the resolution of the acute phase infection. It has thus far proven difficult to identify universally conserved pathophysiological hallmarks of PACS due to significant biological and clinical disease heterogeneity. Chronic immune dysregulation affecting the innate and adaptive response is a salient yet contentious feature of PACS and may be brought about by SARS-CoV-2 viral persistence/viral reservoirs, reactivation of latent viruses like varicella zoster virus (VZV) and/or Epstein–Barr virus (EBV), autoimmunity, and gut microbiome dysbiosis ^[37][38][216,217]. Neutrophil-derived calprotectin, a marker of neutrophil activation and degranulation, and citrullinated histone H3 (cit-H3), a marker of NET production, are enriched in a subset of patients with so-called ‘inflammatory PACS’ ^[39][218]. In the postmortem lung specimens of patients who died from COVID-19, immature neutrophils were seen co-localizing with CD8⁺ T cells and regenerating alveolar epithelium in areas of diffuse alveolar damage ^[18][197], suggesting that perhaps neutrophils could be playing a role in the persistent stimulation of T cells or impairing the regeneration of alveoli that could be relevant to the chronic pulmonary sequelae of fibrosis that survivors develop ^[40][219].

Autoantibodies are known to emerge in severe acute COVID-19 and also in PACS, although their role in the chronic sequela was cast into doubt by a study applying multidimensional immunophenotyping and unbiased machine learning approaches to study conserved immune trajectories in 275 PACS patients ^[41][220]. Nevertheless, keeping in mind the biological heterogeneity of PACS, severe COVID-19 patients can develop IgG and IgM antibodies against NETs, which can protect these structures against degradation by circulating DNase ^[42][221]; their persistence in survivors of severe COVID-19 who develop PACS may contribute to inflammatory PACS ^[43][222]. A fraction of PACS patients and approximately half of hospitalized COVID-19 patients can develop autoantibodies typical of antiphospholipid syndrome, such as anti-cardiolipin, anti-phosphatidylserine, and anti-β₂ glycoprotein ^[44][223]. These antibodies trigger NETosis when injected into mice ^[44][223], and PACS patients exhibiting positivity for these antibodies demonstrate higher levels of NET markers than healthy controls ^[45][224]. Phenotypically, the persistence of NETs in PACS may contribute to chronic lung injury and aberrant regeneration with fibrosis ^[46][225]. Together, these findings indicate that neutrophil effector functions like NET production can be present in a subset of PACS patients ^[47][226], but further studies are required to substantiate these observations.

Complement proteins, cytokines, and chemokines like C3a, C5a, GM-CSF, CXCR1/2 agonists, IL-8, IL-17, and IL-1β within the TME (the specific cytokine/chemokine composition varies between tumors) can recruit neutrophils and reprogram them towards pro-tumor (N2 or PMN-derived suppressor cells (PMN-MDSCs)) or antitumor (N1) functionalities ^[48][49][227,228]. N1 TANs display greater cytotoxicity against tumor cells and activate antitumor T cells and natural killer cells, whereas N2 and PMN-MDSCs favor M2 macrophage polarization and T_reg differentiation to suppress antitumor immunity ^[50][229]. It should be mentioned that it is unclear whether N2 TANs and PMN-MDSCs represent distinct neutrophil populations. A recent consensus statement suggested dropping the terminology of PMN-MDSCs and instead considering them both under protumor TANs ^[51][230]. Mounting evidence of neutrophil plasticity/heterogeneity also supports moving past these terminologies.

The notion that TANs can adopt heterogenous phenotypes has been extensively studied in hepatocellular carcinoma (HCC) ^[52][231]. Indeed, 12 transcriptionally distinct TAN subtypes have been described in human HCC ^[53][232]. Functionally, PD-L1⁺ TANs in the HCC TME can suppress antitumor CD8⁺ T-cell responses ^[54][55][233,234]. In addition, HCC-associated TANs can elaborate CCL2, CCL3, and CCL17 to recruit immunosuppressive macrophages and T_regs ^[53][56][232,235]. In contrast, in lung cancer models, TANs can orchestrate potent antitumor immune responses. In this regard, TANs can upregulate class II MHC to stimulate antitumor T-cell responses that can mitigate the tumor progression of early-stage lung cancer ^[57][58][236,237]. The composite of cytokines/chemokines released by N1-like TANs like CCL3, CCL9, TNF-α, and IL-12 can recruit and activate T_H1 and CD8⁺ cytotoxic T cells to exert antitumor effects ^[59][238]. TAN-derived ROS can directly inhibit IL-17 release by pro-tumorigenic γδ T cells ^[60][239].

Clinically, the anti-angiogenic treatment sorafenib—one of the first-line agents to treat HCC—is associated with an increase in TANs that produce CCL2 and CCL17, which can establish immunosuppressive niches within the TME to promote tumor cell survival. Accordingly, inhibiting transcriptional regulators of CCL2/CCL17, like p38 and AKT, significantly improved tumor sensitivity to sorafenib ^[56][235]. Along similar lines, cabozantinib (anti-angiogenic) and anti-PD1 immune checkpoint inhibitor combination therapy showed increased antitumor efficacy compared to monotherapy with either agent ^[61][240]. The combination regimen worked synergistically to increase circulating T-cell numbers, decrease the neutrophil-to-lymphocyte ratio, and decrease intratumoral exhausted PD1⁺ CD8⁺ T cells and T_regs. Interestingly, human HCCs that exhibited a favorable response to treatment showed an increased TAN presence in the TME ^[61][240], suggesting that anti-angiogenic and immune checkpoint therapies can reprogram TAN-driven innate and adaptive responses in the TME towards antitumor immunity. Indeed, anticancer immunotherapies can increase a specific subset of IFN-active TANs that were associated with enhanced antitumor immunity ^[62][241].

Regarding the humoral response, B cells recruited to the TME by TAN-derived TNF-α can become activated and differentiate into IgG-producing plasma cells independent of T-cell interactions ^[63][242]. This was shown to be predicated on cell-to-cell contact with TANs and driven partly by TAN-derived BAFF ^[63][242]. The bone marrow environment in multiple myeloma is characterized by inflammasome-primed neutrophils that secrete BAFF to enhance myeloma cell survival and maintain a pro-inflammatory marrow microenvironment ^[64][65][243,244]. This environment can persist after the treatment of multiple myeloma with hematopoietic stem cell transplant, suggesting its role in disease recurrence.

NET release is increasingly being shown to be playing a key role in the TME. The chromatin-based matrix of NETs can act as a physical barrier that prevents contact between tumor cells and antitumor CD8⁺ T cells and natural killer cells ^[66][67][68][245,246,247]. Furthermore, PD-L1 embedded within NETs can induce exhausted phenotypes in CD8⁺ T cells in murine models of liver metastasis ^[69][248]. In a clinical context, a risk score composed of six NETs-related genes could estimate the prognosis of colon cancer patients and their response to immunotherapy ^[70][249]. In addition, NET-derived MPO spatially associates with T_regs and PD-1⁺ T_regs that attenuate responses to immune checkpoint therapies ^[70][249]. PAD4 inhibition to attenuate NET formation increases the sensitivity of pancreatic cancer to immune checkpoint inhibitors ^[67][246]. NETs can modulate tumor responsiveness to the more conventional lines of cancer treatment too. Radiation treatment of bladder cancer increases the production of NETs in the TME, which prevents antitumor CD8⁺ cytotoxic T cells from having access to residual tumor cells, an effect that could be alleviated by degrading NETs via DNase-1 ^[71][250]. Chemotherapy-treated gastric cancer cells elaborate IL-1β to stimulate NET production that confers chemoresistance by increasing local TGF-β concentration ^[72][251]. Moreover, NETs-mediated immunosuppression may be one of the direct facilitators of HCC development by TLR4-mediated metabolic reprogramming of naïve T cells to T_regs ^[73][252]. Blocking NETs in vivo using PAD4-knockout mice or DNAse-1 reduces the numbers of T_regs, significantly attenuating the progression of non-alcoholic steatohepatitis into HCC ^[73][252].

Multiple lines of evidence show that NETs can augment antitumor immunity. For example, intravesical BCG therapy for bladder cancer recruits TANs and stimulates NET release, which, in turn, enhances the recruitment of antitumor monocytes and T cells to the TME ^[74][253]. In melanoma, NETs adhere to tumor cells via integrins to prevent their migration and exert direct cytotoxicity ^[75][254]. Melanoma-specific T-cell immunotherapies and immune checkpoint inhibitors significantly increase antitumor immunity against melanoma cells in the early phase of treatment. However, complete tumor eradication, including the killing of antigen loss variants, during late-phase treatment is dependent on neutrophil recruitment and NET production ^[76][255].

These phenotypes of TANs are partially due to metabolic reprogramming. The depletion of extracellular glucose in the TME, as well as factors derived from tumor cells like GM-CSF in select tumor models, can cause TANs to upregulate fatty acid uptake, reprogramming them towards a phenotype of protumor TANs that suppress antigen-specific antitumor CD8⁺ T cells ^{[77][78][79][80][81][82]}[256,257,258,259,260,261]. The TME is also highly heterogeneous in terms of its blood supply, oxygen and glucose availability, cytokine/chemokine/growth factor milieu, and immune and stromal cell composition and phenotype. Further layers of complexity can be added when considering the specific organ affected. It is likely that neutrophils in spatially distinct compartments of the TME are driven to diverging phenotypes that co-exist and partake in homotypic and heterotypic cellular interactions to shape the immune response to tumors, either towards antitumor immunity or immunosuppression. A spatial understanding of neutrophil identity in the TME combined with single-cell approaches would provide more insights into understanding their divergent individual phenotypes in the TME.

Another aspect of tumor biology that influences tumor response to immunotherapy is the genomic makeup of tumors. Performing immunogenomic analyses of 67 spatially distinct regions of an anti-PD-1-resistant melanoma sample, Mitra et al. reported that tumor regions harboring a subclonal gain of chromosome 7 exhibit an accumulation of immunosuppressive TANs in the TME, the transcriptomic signature of which was associated with resistance to immune checkpoint inhibitors like anti-CTLA4 and anti-PD-1 ^[83][262]. Therefore, both tumor-extrinsic factors, like blood flow and nutrient and oxygen availability; and intrinsic factors like genomic abnormalities may influence the abundance and activation phenotypes of TANs.

Autoimmune diseases are characterized by a loss of tolerance, leading to the failure of the adaptive immune system to distinguish self from non-self-antigens and consequently causing it to attack host tissues. Neutrophils in the blood of patients with systemic autoimmune diseases exhibit an activated phenotype and are present at inflamed sites like the vascular wall in vasculitis, the kidney in SLE, and the synovium in rheumatoid arthritis ^[84][263]. At these sites, neutrophils enact their classical effector functions of phagocytosis, degranulation, and NET release, as well as their immunoregulatory roles of cytokine/chemokine production and antigen-presentation. NETs constitute a rich source of intracellular alarmins, like HMGB1, that are normally shielded from the immune system. The release of these alarmins results in antigen sequestration and increased autoimmune activity ^[85][134]. Additionally, immune complexes (ICs), the presence of which is a cornerstone of many autoimmune diseases, can stimulate the production of lytic NETs or non-lytic mitochondrial DNA-containing NETs ^[86][87][70,264]. Lastly, autoimmune diseases like SLE are associated with an increase in T_H17 responses ^[88][265], and this may be related to NET production as cit-H3 within NETs can directly stimulate T_H17 differentiation via TLR2 signaling ^[89][140].

Neutrophils are abundant in the synovial fluid of inflamed joints in rheumatoid arthritis, juvenile idiopathic arthritis, and monosodium urate crystal-induced arthritis ^[90][91][92][266,267,268]. Functionally, chemokines upregulated by synovium-infiltrating neutrophils are associated with the chemoattraction of more neutrophils (CXCL1, CXCL2, and CXCL8), DCs (CCL2, CCL4, and CXCL16), T_H1 cells (CCL2 and CXCL10), and T_H17 cells (CCL2 and CCL20) ^[93][94][96,269]. Synovial-fluid neutrophils in rheumatoid arthritis express and secrete BAFF and APRIL that could promote autoantibody production ^[95][96][270,271]. Rheumatoid arthritis patients typically exhibit positivity for anti-citrullinated peptide antibodies (ACPAs), and this positivity correlates with disease progression ^[97][272]. Inoculating control neutrophils with synovial fluid from rheumatoid arthritis patients elicits NET production with exposure to cit-H3-positive DNA ^[94][269], indicating that NETs can constitute a major source of citrullinated peptides against which antibodies are specifically generated in rheumatoid arthritis ^[98][99][273,274]. In support of this, serum from RA patients who are positive for ACPA cross-reacts with the cit-H4 present within NETs ^[100][275]. Mechanistically, citrullinated proteins within NETs can be endocytosed by MHC II-expressing synoviocytes and presented to antigen-specific CD4⁺ T cells to initiate T cell-dependent B-cell maturation and autoantibody production ^[101][276]. Alternatively, NET-derived neutrophil elastase can damage the intra-articular cartilage of affected joints, liberating its proteins for citrullination and subsequent presentation by synoviocytes to CD4⁺ T cells ^[102][277]. Considering these findings with the observations that (1) neutrophils are one of the most abundant cells in the synovium of joints affected by rheumatoid arthritis, (2) they can directly present antigens to CD4⁺ T cells, and (3) they stimulate B-cell differentiation and antibody production through BAFF/APRIL implicates neutrophils as being key to the loss of immune tolerance and production of pathogenic ACPA.

Neutrophils in SLE appear to be skewed towards LDGs, which are primed population with augmented pro-inflammatory activity and NETosis ^{[103][104][105]}[278,279,280]. Ribonucleoprotein (RNP)-antibody ICs in SLE induce the secretion of BAFF by neutrophils that promote B-cell survival, proliferation, and plasmablast differentiation, indicating that perhaps neutrophils may sustain autoantibody production in lupus ^[106][281]. Neutrophils isolated from SLE patients display reduced NADPH oxidase-dependent ROS production ^[107][282]. These findings indicate an impairment of respiratory burst-dependent processes in SLE that play a key role in the phagocytic clearance of apoptotic cellular debris, thereby leading to the persistence of intracellular debris that can act as neoantigens or substrates for autoantibodies to form ICs ^[108][283]. Indeed, NADPH oxidase-knockout mice have a higher risk of developing SLE and progression of existing disease ^[109][284]. Similarly, pristane-induced lupus (PIL) was exacerbated in mice deficient in NADPH oxidase or PAD4, whereas treating PIL mice with NADPH oxidase activators induces lytic NET formation and ameliorated disease severity ^[110][285].

In contrast, LDGs stimulated by ribonucleoprotein (RNP)-ICs in SLE upregulate mitochondrial ROS production and the oxidation of mitochondrial DNA that is then extruded in oxidized mitochondrial DNA-containing NETs in an NADPH oxidase-independent manner ^[111][112][286,287]. The mechanism of NET formation in SLE was recently clarified. Circulating ICs activate the Fcγ receptor (FcγR) on the surface of neutrophils, leading to the inactivation of serpinb1 and consequent activation of intracellular caspase-1/caspase-11 that cleave and thereby activate gasdermin D ^[113][23]. Importantly, gasdermin D can mediate both lytic NETosis (with nuclear DNA) and non-lytic mitochondrial NET formation (decorated with oxidized mitochondrial DNA), whereas PAD4 is only involved in NADPH oxidase-dependent lytic NETosis (which may be defective in SLE—see above). Accordingly, the RNP-IC-mediated pathway of NET production induced mitochondrial ROS production and required gasdermin-D activation but not PAD4. This was confirmed pharmacologically in vitro, where PAD4 inhibitors (GSK484) had no effect on the release of RNP-IC-related NET release from control human neutrophils and LDGs derived from lupus patients, whereas gasdermin-D inhibition (disulfiram) or mitochondrial ROS attenuation (Mito-TEMPO) significantly attenuated oxidized mitochondrial DNA release within NETs ^[113][23].

Downstream, NETs containing oxidized mitochondrial DNA can trigger the TLR9-dependent activation of pDC, leading to the production type I IFNs that stimulate autoreactive T cells and lower the activation threshold of autoreactive B-cells ^[114][288]. NETs decorated with IL-33 also directly activate TLR7/9 signaling in pDCs and the production of type I IFNs ^[115][289]. Alternatively, the uptake of the oxidized mitochondrial DNA of NETs (promoted by NET proteins LL-37 and HMGB1) by pDCs activates the cGAS/STING pathway of type I IFN production in SLE ^[112][116][287,290]. Independently, LL-37-DNA complexes within NETs can directly get endocytosed by polyclonal B cells, activating TLR9 signaling, clonal expansion, and anti-LL-37 antibody production, which are autoantibodies seen in SLE patients ^[105][280]. The degradation of NETs can be impaired in SLE, as NET-directed autoantibodies like anti-dsDNA antibodies and circulating factors like C1q protect NETs against degradation in SLE, and genetic polymorphisms that negatively affect the function of DNase enzymes can increase the risk of the development of lupus ^{[117][118][119][120][121][122]}[291,292,293,294,295,296]. This persistence of NETs may promote exposure to autoantigens and augment autoimmunity.

Together, these findings indicate that SLE immunopathogenesis may be initiated or exacerbated by an increase in the circulating proportions of LDGs prone to mitochondrial DNA-enriched NETs, which stimulate IFN responses in pDCs, as well as activate autoreactive B cells. What genetic, epigenetic, bone marrow, circulating, lifestyle, and/or environmental factors skew neutrophils towards LDGs to kickstart this whole process remain to be elucidated. Importantly, the link between mitochondrial DNA-containing NETs and pDCs with regards to type I IFN production may need to be revisited given recent data describing that pDCs may not be the major producers of type I IFNs in SLE, as evidenced by their defective production of type I IFNs in SLE ^[123][297]. Instead, nonhematopoietic cells like keratinocytes in cutaneous lupus and tubular epithelial cells of the kidney in lupus nephritis can be major producers of type I IFNs ^{[123][124][125]}[297,298,299]. Furthermore, neutrophils can produce type I IFNs like IFNα in response to extracellular chromatin by upregulating cytosolic DNA sensors like the cGAS/STING pathway, which is associated with further NET production ^[126][300]. This suggests the presence of a positive feedback loop where NET production drives IFN production by neutrophils, thereby driving further autoimmune inflammation and NET production.