Introduction

The role played by neutrophils in immunity has long been viewed as restricted to the acute phase of inflammation and to resistance against extracellular pathogens [1–3]. This view is consistent with the phagocytic theory of Metchnikoff who proposed more than 100 years ago that polymorphonuclear leukocytes patrol the bloodstream and migrate to the site of infection to phagocytose microbes. Several studies have recently challenged this dogma and placed the neutrophil as a key effector cell in the orchestration of adaptive immunity and in the resolution of inflammatory response [1–3]. Indeed, in addition to their phagocytic activity and the storage of a set of lytic enzymes and antimicrobial components in their intracellular granules, neutrophils are induced to express molecules (e.g. cytokines, chemokines) involved in the regulation of innate and adaptive response. Neutrophils have emerged as a major source of humoral pattern recognition molecules (PRMs) that recognise pathogen-associated molecular patterns and initiate the immune response in coordination with the cellular arm, therefore acting as functional ancestors of antibodies. Neutrophils directly interact with macrophages, dendritic cells, and lymphocyte subsets and modulate their effector functions. For instance, natural killer (NK) cell functions are impaired during neutropenia, and under steady state, neutrophils are crucial for NK cell development both in human and mouse [4]. Consequently, a bidirectional cross talk occurs between neutrophils and NK cells, which stimulates the production of IFN-γ by NK cells and promotes the survival and activation of neutrophils [5, 6]. In addition, polarized T helper (Th) 17 cells and innate IL17-producing cells rapidly activate neutrophilic inflammation through the production of granulopoietic factors and chemokines [7–9]. In this review, we will focus on the most recent findings related to the effector functions and plasticity of neutrophils as well as on their emerging role in regulating the innate and adaptive immune system. Finally, we will describe the relation between neutrophils and diverse pathologies.

Neutrophils in innate immunity

Pathogen recognition by neutrophils

Mammals are constantly in contact with microorganisms, and their ability to mount a protective immune response resides in their competence to identify potential pathogens. The life-threatening condition associated with acquired or congenital abnormalities in neutrophil life cycle or function underlines their essential role in innate immunity and resistance to pathogens [2]. Innate immune molecules involved in pathogen recognition are germline-encoded PRMs which belong to both the cellular and humoral arms of the innate immune system [10]. These receptors represent a class of sensors specialized in the discrimination of self versus non-self and modified self and participate in the initiation and regulation of the inflammatory process [11]. Neutrophils are endowed with a vast repertoire of cellular-associated PRMs which, upon recognition of pathogens or tissue damage, promote neutrophil effector functions (e.g. production of ROS, secretion of antimicrobial peptides) [12, 13]. These include all members of the Toll-like receptor (TLR) family, with the exception of TLR3 and a low or absent expression of TLR7 [14, 15]; the C-type lectin receptors Dectin-1 [16], CLEC-2 [17], Mincle [18] and CLECSF8 [19]; and functional cytoplasmic sensors, such as NOD-1, RIG1, MDA5 and IFI16 [20–22]. Dectin-1 (also known as CLEC7A) is the main β-glucan receptor on neutrophils and promotes phagocytosis and the killing of Candida albicans and Aspergillus fumigatus through the activation of a calcineurin signalling axis [16, 23]. Neutrophils express NOD-1, which primes the innate immune system [20], and the NLRP3/ASC/Caspase-1 inflammasome, which regulates IL-1β processing [24]. In addition, neutrophils express the NOD-like receptor family member NLRP6, a negative regulator of NF-kB and ERK activation after TLR engagement [25]. In contrast to other myeloid cells, the Myd88-independent pathway is not activated in human neutrophils stimulated by lipopolysaccharide (LPS) and neutrophils fail to produce IFN-β and, consequently, CXCL10 and other type I IFN-dependent genes after TLR4 engagement [26]. However, human neutrophils express a set of cytosolic DNA sensors, such as IFI16, MDA5, RIG1, LRRFIP1, DDX41 and STING [21, 22]. As a consequence, the expression of IFN-β and CXCL10 mRNAs was observed in human neutrophils transfected with plasmid DNA or infected by intracellular pathogens (e.g. Bartonella henselae, Listeria monocytogenes, Legionella pneumophila and adenovirus type 5) [21]. Formyl peptides, found in bacteria and mitochondria, activate neutrophils via the seven-transmembrane G-protein-coupled receptors FPR1 (high-affinity receptor for fMLF) and FPR2 (low-affinity receptor for fMLF), which have different effector functions [27, 28]. The activation of FPR1 or FPR2, depending on ligand concentration, promotes p38 or Erk activation, respectively, leading to the activation or inhibition of neutrophil chemotaxis [27]. Mitochondrial-derived formylated peptides induce the recruitment of neutrophils and inflammation and elicit neutrophil-mediated organ injury [29]. Since the production of formylated proteins is limited to bacteria and mitochondria, FPRs can be classified as PRMs recognising microbial moieties and tissue damage [29].

Fluid phase PRMs, including collectins, ficolins and pentraxins, are essential effectors and modulators of the innate resistance in animals and human and act as functional ancestors of antibodies [10]. Neutrophils have emerged as a source of humoral PRMs and notably serve as a ready-made reservoir of a set of PRMs, including long pentraxin PTX3, proteoglycan recognition receptor PGRP-S and M-ficolin (also known as ficolin 1), covering a temporal window preceding gene expression-dependent production (Fig. 1).

Fig. 1
figure 1

Neutrophils contribute to the humoral arm of innate immunity. Sensing microbial moieties or tissue damage by cellular receptors induces the release of humoral fluid phase pattern recognition receptors by innate immune cells (e.g. neutrophils, macrophages) and other cell types (e.g. endothelial cells, epithelial cells) with different tempos. Neutrophil granules contain a set of humoral fluid phase pattern recognition receptors (PGRP-S, PTX3, M-ficolin) rapidly released in minutes, covering a temporal window preceding gene expression-dependent production. These humoral sensors share fundamental mechanisms of effector function (e.g. opsonisation, agglutination, complement activation, regulation of inflammation) and participate in the initiation of immune response

Full size image

PTX3 is a member of the long pentraxin family related to classic short pentraxins (e.g. C-reactive protein and serum amyloid P) and has served as a paradigm to study the humoral arm of the innate immune system [10]. PTX3 transcript expression is confined to immature myeloid cells, and mature neutrophils serve as a reservoir of preformed PTX3, ready for rapid release into neutrophil extracellular traps (NETs) [30]. Upon pathogen opsonisation (i.e. A. fumigatus, Pseudomonas aeruginosa), PTX3 interacts with Fcγ receptor IIA (FcγRIIA/CD32) and induces inside-out CR3 (CD11b/CD18) activation and amplification of C3b-opsonized pathogen phagocytosis [31, 32]. Accordingly, neutrophil-associated PTX3 is essential for resistance against A. fumigatus [30]. In addition, PTX3 is translocated from granules to the surface of apoptotic neutrophils and acts as a late “eat me” molecule involved in the recognition and engulfment of apoptotic neutrophils by macrophages [33]. Leukocyte-derived PTX3 also has a regulatory function on neutrophil recruitment and inflammation by interacting with P-selectin [34]. Thus, under conditions of full-blown neutrophilic inflammation, leukocyte-derived PTX3 acts as a negative feedback loop by binding to P-selectin and preventing further neutrophil recruitment [34].

PGRP-S and M-ficolin are stored in secondary and tertiary granules [35–37]. PGRP-S is also localized in NETs, binds to peptidoglycan and exerts bacteriostatic and bactericidal activities against selected microorganism (e.g. Micrococcus luteus, Staphylococcus aureus, Bacillus subtilis) [37, 38]. M-ficolin, which belongs to the lectin family, recognises selected Gram-positive and Gram-negative bacteria, activates the complement lectin pathway and exerts opsonic effects [10]. M-ficolin released from granules is also found associated with the neutrophil surface through a direct interaction with CD43 [36, 39]. This interaction enhances neutrophil aggregation and adhesion and activates complement on the neutrophil surface [39].

Collectively, these observations reveal that neutrophils participate in humoral innate immunity via the expression and release of fluid phase PRMs involved in the recognition and phagocytosis of non-self and modified self, complement activation and regulation of the inflammatory response.

Neutrophil extracellular traps

NETs are an extracellular fibrillary network formed by activated neutrophils [40] and composed of nuclear components (i.e. DNA, histones) [40] decorated by a set of proteins from primary (e.g. myeloperoxidase (MPO) and neutrophil elastase (NE)) [40], secondary (e.g. lactoferrin [40] and PTX3 [30]) and tertiary granules (e.g. MMP-9 [40] and PGRP-S [37]). In addition to genomic DNA, mitochondrial DNA has been reported in NETs [41]. These extracellular structures trap bacteria (e.g. Escherichia coli, Shigella flexneri, S. aureus); fungi (e.g. C. albicans, A. fumigatus); and human immunodeficiency virus-1 (HIV-1), favouring their interaction with effector molecules and their disposal (Fig. 2) [40, 42]. In addition to neutrophil-associated molecules, surfactant protein D present on many mucosal surfaces binds NETs and enhances microbial trapping [43]. However, direct microbicidal activity of NETs has recently been a matter of controversy and could require the presence of H2O2, chloride and the formation of HOCl catalysed by MPO [44, 45]. NET formation is a rapid active process called “NETosis” (Fig. 2) which occurs in vivo in animals and humans and prevents systemic bacterial dissemination [40, 46]. NET formation is induced by the Raf-MEK-ERK pathway, through the activation of the NADPH oxidase complex and upregulation of anti-apoptotic proteins, and by the mammalian target of rapamycin (mTOR), through the induction of hypoxia-inducible factor 1 alpha protein expression [47, 48].

Fig. 2
figure 2

Neutrophil extracellular traps in physiology and pathology. Neutrophil extracellular traps (NETs) are formed by activated neutrophils via an active process called NETosis. This extracellular network, composed of DNA and nuclear components decorated by a set of proteins from granules, traps microorganisms and facilitates their disposal by NET-associated effector molecules. NETs are also involved in a diverse set of pathological conditions

Full size image

After stimulation, neutrophils lose their characteristic nuclear morphology. Chromatin decondensation is an essential event and requires the generation of ROS, the induction of neutrophil autophagy [49, 50], and the citrullination of histones by the peptidyl arginine deiminase 4 (PAD4) [51] and is favoured by NE and MPO [52]. Accordingly, genetic deficiency of these molecules, as observed for instance in chronic granulomatous disease or in MPO-deficient patients, results in defective NET formation, which is likely to contribute to the increased susceptibility to infections observed in these patients [49, 53].

NET formation is delayed in neutrophils isolated from preterm or healthy term neonates compared to neutrophils isolated from healthy adults [54]. Interestingly, this defect seems to be ROS-independent, which suggests additional mechanisms driving NET formation [55]. According to this observation, Leishmania donovani induces NET formation in a ROS-independent mechanism [56].

Microorganisms have acquired a set of mechanisms allowing them to escape NET trapping/killing and, thus, enhancing their virulence. For instance, the cell wall structure of Streptococcus pneumoniae and L. donovani is modified by a d-alanylation of the cytoplasmic membrane-anchored lipoteichoic acid and by the expression of a surface glycolipid lipophosphoglycan, respectively, which protect them from the antimicrobial activity of NETs [56, 57]. Expression of DNase by the M1 serotype strains of the pathogen group A Streptococcus (Streptococcus pyogenes or GAS) and S. pneumoniae inhibits their extracellular killing mediated by neutrophils and enhances their virulence in vivo [58, 59]. In addition, strains of P. aeruginosa isolated from cystic fibrosis patients have developed resistance to NET-mediated killing within the cystic fibrosis airway [60]. Finally, HIV-1 is also able to counteract the antiviral activity of NETs through DC-SIGN (CD209)-dependent IL-10 production, which inhibits NADPH oxidase-dependent NET formation [42]. Thus, as ancient Roman gladiators, neutrophils throw noxious NETs, whilst escape from NETs is an evolutionary strategy adopted by bacteria and HIV-1.

The growing number of neutrophil-derived cytokines

Beyond their classical preformed and rapidly secreted mediators, neutrophils have recently emerged as key regulators in innate and adaptive immunity through cytokine production and secretion. Table 1 summarizes the growing number of neutrophil-derived cytokines. Here, we describe the most recent evidence and refer the reader to previous reviews for background [2, 61].

Table 1 Neutrophil-derived cytokines
Full size table

A recent report has demonstrated that murine neutrophils constitutively express the NLRP3 inflammasome complex and that LPS-pretreated murine neutrophils are the major source of IL-1β in response to classical inflammasome activators (ATP, silica crystals) [24]. Previous results indicated that murine neutrophils are able to produce IL-1β and process it in a proteinase-3- and elastase-dependent manner [62]. However, Mankan and colleagues [24] showed that neutrophils isolated from proteinase 3/elastase double knockout mice are still able to produce functional IL-1β, whilst neutrophils isolated from NLRP, ASC or caspase-1 knockout mice are not, demonstrating that the NLRP3/ASC/caspase-1 pathway plays a major role in IL-1β production by murine neutrophils. IL-27 expressed by human neutrophils during sepsis suppresses the production of ROS and therefore reduces neutrophil bactericidal activity in vitro, suggesting that IL-27 exerts regulatory effects on neutrophils [63]. Human neutrophils are unable to activate the MyD88-independent/TRIF-dependent pathway upon TLR4 engagement and thereby fail to produce IFN-β upon stimulation by LPS [26]. Tamassia and colleagues [21] recently found that DNA-transfected human neutrophils express IFN-β and related genes via direct activation of IRF3 mediated by intracellular DNA sensors. Moreover, murine neutrophils also express IFN-β and related genes upon infection with encephalo-myocarditis virus via MDA5 engagement [22]. These results highlight the role of neutrophils in recognising intracellular pathogens and modulating innate and adaptive immune response.

Other neutrophil-derived cytokines, such as TNF-related leukocyte-expressed ligand (TRAIL), CCL20, CXCL8, B cell-activating factor BAFF) or IL-1 receptor antagonist, are stored in intracellular pools and rapidly secreted upon stimulation by pro-inflammatory stimuli [64].

Human neutrophils represent an important source of BAFF and A proliferation-inducing ligand (APRIL), two cytokines crucial for the survival, maturation and differentiation of B cells, suggesting a role of neutrophil-derived cytokines in autoimmune and neoplastic B cell-dependent disorders [2]. Under homeostatic conditions, Puga and colleagues [65] described a subpopulation of neutrophils presenting a singular phenotype characterized by high levels of BAFF, APRIL, CD40L and IL-21 production. These neutrophils activate B cells from the marginal zone of the spleen and promote the diversification and production of immunoglobulins, and are therefore called “B-cell helper neutrophils” (NBH) [65].

Neutrophil-derived proteases regulate the biological activity of cytokines in the inflammatory microenvironment. For instance, human and murine neutrophil-derived elastase and cathepsin G cleave full-length IL-33 into mature forms [66]. In a mouse model of P. aeruginosa-induced pneumonia, neutrophil-derived elastase (NE) induces the production of TNF-α, MIP-2/CXCL2 and IL-6 in the lungs, and in particular by macrophages, via a TLR4-dependent mechanism [67].

It has been proposed that both human and murine neutrophils infiltrating psoriatic skin lesions and inflamed synovia of rheumatoid arthritis patients express IL-17A [68, 69]. However, human and murine neutrophils show differences in cytokine expression. In particular, the expression of IFN-γ and IL-10 by human neutrophils remains controversial [2, 61, 70]. Surprisingly, and despite previous negative findings [70], human neutrophils have been reported to produce remarkable amounts of IL-10 upon stimulation with serum amyloid A and LPS [71]. These latter findings have not been reproduced in other laboratories, thus raising the issue of the need for stringent purification and control of monocyte contamination [72]. In addition, the IL-10 genomic locus is in an inactive state in human neutrophils, supporting their incapacity to produce IL-10 [76]. Concerning the murine counterpart, several studies have demonstrated that mouse neutrophils produce IL-10, for instance during disseminated Candida infection [23], methicillin-resistant S. aureus infection [73], pneumonia [74] or Trypanosoma cruzi infection [75].

Immune cell cross talk

Within tissues, neutrophils engage in an intricate cross talk with stromal elements, macrophages, dendritic cells (DC) and lymphocyte subsets. Integration of signals received by neutrophils during the migration process (i.e. cytokines, adhesion, transmigration, microbial products) is a critical step to increase their life span, thereby allowing these cross talks [77]. Activated tissue-resident mesenchymal stem cells and bone marrow-derived mesenchymal cells promote neutrophil recruitment, increase their life span and function in vitro [78, 79].

Activated neutrophils were shown to promote the maturation of human monocyte-derived DC (moDC), through interaction between CD18 and CEACAM1 expressed by neutrophil and DC-SIGN on moDC, and mouse bone marrow-derived DC, through the production of TNF-α [80–82]. However, the cross talk between neutrophils and DC may also inhibit or reduce their maturation and immunostimulatory activity through the secretion of ectosomes and neutrophil elastase [83, 84].

Neutrophils also communicate with T cells, B cells and NK cells. For example, activated neutrophils induce Th1 and Th17 cell chemotaxis through the production of CCL2, and CXCL10 or CCL2 and CCL20, respectively [9]. Accordingly, neutrophils and Th17 cells co-localize in the gut from Crohn’s disease and synovial fluid from rheumatoid arthritis patients [9]. In turn, Th17, γδ T cell or Treg, but not Th1, produce the neutrophil chemoattractant CXCL8 and activated T cells modulate the neutrophil life span and activation through the secretion of selected cytokines (e.g. IFN-γ, GM-CSF and TNF-α) [9, 85–87].

Neutrophils migrate to the lymph node in a CCR7-dependent manner and act as antigen-presenting cells [88–92]. Moreover, neutrophils take up and present exogenous antigens on MHC-I complex in vivo and promote the differentiation of naive CD8+ T cells into cytotoxic T cells, suggesting a direct interaction between neutrophils and CD8+ T cells [91]. In contrast, neutrophils may also interfere with professional antigen-presenting cells via a competition for the antigen and, therefore, reduce the T CD4+ response [93].

A “ménage a trois” composed by neutrophils, dendritic cells and T cells can enhance the development of the immune protective response. DC have the capacity to internalize live and UV-irradiated neutrophils in a CD18-dependent fashion [94]. Therefore, infected neutrophils sustain the maturation, activation and migration to the lymph node of DC through a cell contact-dependent fashion, which in turn cross-present antigens and elicit T cells to produce IFN-γ [94–96]. Mycobacterium tuberculosis inhibits neutrophil apoptosis, leading to a slow down of the bacterial acquisition by DC, a slower migration to the lung-draining lymph node and a delayed activation of CD4+ T cells in vivo [97].

The interaction of neutrophils with DC can also result in NK cell activation. A cross talk between human neutrophils and 6-sulfo LacNAc+ myeloid DC (slanDC) increases the release of IL-12p70 by slanDC, which in turn potentiates the production of IFN-γ by NK cells [5]. Moreover, IFN-γ potentiates the interaction between neutrophils and slanDC and the release of IL-12p70, creating a positive amplification loop [5]. This tripartite network is supported by direct interactions between neutrophils and slanDC, via CD18–ICAM1 interaction, and between neutrophils and NK cells, via ICAM3 and probably CD18–CD11d complex expressed by NK cells [5]. Co-localization of neutrophils, slanDC and NK cells has been shown in the colonic mucosa of Crohn’s disease patients as well as in skin lesions of psoriasis patients, suggesting a pathophysiological relevance for this tripartite network [5].

Using a novel form of neutropenia obtained from a point mutation in the transcriptional repressor Gfi1, Jaeger and colleagues have recently demonstrated that neutrophils are crucial for the development of NK cells [4, 98]. These mice, called Genista, have normal viability, but are neutropenic due to a blockage of terminal granulocytic differentiation just after the metamyelocytic stage [98]. Interestingly, neutropenia was associated with poor survival, hyperproliferation, impaired development and hyporeactivity of NK cells [4]. For instance, most of the splenic NK cells are blocked at the double-positive CD27+CD11b+ stage and a low percentage expresses CD107a and IFN-γ upon contact with tumour cells [4]. Moreover, the depletion of neutrophils in wild-type mice is sufficient to induce NK cell hyporeactivity and impair their maturation [4]. Interestingly, NK cells from patients suffering from severe congenital neutropenia display similar maturation and functional defects, suggesting that neutrophils contribute to NK cell development and function also in human [4]. In turn, NK cell-derived products (e.g. IFN-γ, GM-CSF) promote the survival and activation of neutrophils [6, 99, 100], whereas cell–cell contact between NK cells and neutrophils induces neutrophil apoptosis via the natural cytotoxicity receptor NKp46 and the Fas pathway [6, 101].

As discussed above, a population of neutrophils around the marginal zone of the spleen (MZ) has been recently identified and called “B-cell helper neutrophils” (NBH) [65]. These neutrophils colonize the MZ during foetal life and become more prominent after postnatal mucosal colonization by bacteria. Indeed, LPS-activated splenic sinusoidal endothelial cells produce neutrophil chemoattractant molecules (e.g. CXCL8, CXCL1, CXCL2, CXCL3) and IL-10, which reprogram neutrophils towards a NBH phenotype [65]. NBH express higher levels of B cell-stimulating molecules such as BAFF, APRIL, IL-21 and CD40L and B cell chemoattractant molecules such as CXCL12 and CXCL13 compared to circulating neutrophils [65]. Based on various parameters, including their expression levels of CD16 and CD15, NBH were divided into NBH1 (CD15int, CD16int) and NBH2 (CD15low, CD16low). Moreover, higher expression levels of CD27, CD40L, CD86, CD95 and HLA-II associated with a lower expression level of CD24 are found in NBH1 compared to NBH2, suggesting that NBH1 are more activated subsets. NBH, and in particular NBH2, via higher secretion levels of BAFF, APRIL and IL-21 compared to NBH1, activate B cells from the MZ and promote immunoglobulin class switching, somatic hypermutation and antibody production [65]. Interestingly, patients with neutrophil disorders present low levels of IgM, IgG and IgA antibodies against microbial T cell-independent antigens (e.g. LPS, peptidoglycan), whereas immunoglobulin levels against T cell-dependent antigens (e.g. diphtheria toxins, tetanus) are unmodified, suggesting that NBH regulate the immunoglobulin response to T cell-independent antigens in vivo [65]. Though this study suggests that neutrophils may regulate the immunoglobulin response, their implication in human immunoglobulin deficiency, such as in acquired IgA deficiency, has not been demonstrated.

Though invariant NKT (iNKT) cells are known to modulate inflammation and neutrophilic inflammation [102–104], neutrophil modulation of iNKT activity has not been previously reported. Recently, Weingender and colleagues [105] have demonstrated that cell–cell contact between neutrophils and iNKT cells occurs in vitro and impairs iNKT function. Accordingly, the expression levels of GATA3 and T-bet in iNKT as well as the levels of α-galactosyl ceramide-induced cytokines are reduced during neutrophilic inflammation in mice and humans [105].

All these data demonstrate that neutrophils are not isolated players that are quickly substituted by more specialized cells, but they guide, support and regulate the immune response throughout its development.

Neutrophils in resolution of inflammation

Neutrophils have long been viewed as final effector cells of the acute phase of the inflammatory response and as passive cells during the resolution of inflammation. However, the critical role played by neutrophils in the resolution phase of inflammation and to maintain tissue homeostasis is now demonstrated and accepted [2].

The cell death receptor ligand TRAIL is produced by neutrophils [106] and, in turn, accelerates neutrophil apoptosis in vitro and in vivo [107]. Accordingly, TRAIL deficiency has been associated with increased neutrophil number and inflammation during neutrophilic inflammation [107]. This phenotype is reversed by treatment with recombinant TRAIL, suggesting that neutrophil apoptosis driven by TRAIL could be a potential therapeutic target in neutrophilic inflammation [107]. Moreover, neutrophil-associated p40phox, a subunit of the NADPH oxidase complex involved in the generation of ROS, plays a crucial role in the resolution phase of intestinal inflammation [108]. Indeed, NADPH oxidase activity controls the resolution phase of inflammation via the downregulation of CCR1 expression in mouse neutrophils and the upregulation of enzymes involved in glycan modifications (e.g. fucosyl transferases, sialyl transferases), which are regulators of leukocyte trafficking through selectin ligand synthesis [108].

During the late phase of the inflammatory response, the biosynthesis of eicosanoids by neutrophils is shifted from leukotriene B4 (LTB4) to the lipoxin A4 (LXA4), reducing tissue neutrophil infiltration through a direct interaction between LXA4 and FPR2, also known as the G-protein-coupled receptor (GPCR) LXA4 receptor [109]. Neutrophils contribute to the biosynthesis of omega-3 essential polyunsaturated fatty acid-derived mediators, resolvins (Rv; i.e. RvE1, RvE2, RvD1, RvD2, RvD5), and protectin D1 (PD1), which inhibit neutrophil infiltration in several in vivo inflammatory models [109–111]. RvD1 and RvE2 have the capacity to interact with GPCRs and RvE1 with leukotriene B4 receptor 1 (BLT1), leading to the inhibition of subsequent migration induced by LTB4 [109, 112]. Accordingly, in a zymosan-induced peritonitis model, the anti-inflammatory effects observed after low-dose administration of RvE1 are lost in BLT1-deficient mice [113]. Moreover, RvE1 and RvD1 dampen leukocyte–endothelial interactions via the modulation of the expression of leukocyte adhesion receptors (e.g. CD11b, CD18) and the regulation of L-selectin (CD62L) shedding [109, 111, 112]. Other mediators, such as the recently described macrophage-derived compound maresin 1 and the new 18S series resolvins, inhibit neutrophil transendothelial migration and tissue infiltration in vivo [114, 115]. In addition to limiting neutrophil infiltration, these pro-resolving mediators enhance phagocytosis and the clearance of microorganisms by neutrophils [110]. For instance, in a model of E. coli-induced peritonitis, RvD5 and RvD1 increased the phagocytosis of E. coli by macrophages and neutrophils, reduced pro-inflammatory cytokine levels and bacterial burden, and enhanced survival, potentiating the effect of antibiotics [110].

Disposal of apoptotic neutrophils regulated by the expression of “eat me” signal on apoptotic cells and receptors on phagocytes is a fundamental step to trigger an anti-inflammatory response and the resolution of inflammation [116]. For instance, recognition and ingestion of apoptotic neutrophils by macrophages induce an IL-10high IL-12low M2-like phenotype that negatively regulates inflammation and stimulates tissue repair [117]. The biosynthesis of pro-resolving mediators is increased in neutrophils during apoptosis and in macrophages after engulfment of apoptotic neutrophils. In turn, these mediators enhance the phagocytosis and clearance of apoptotic cells [109, 118]. RvE1 shortens the life span of neutrophils in the presence of opsonized pathogens, probably through the activation of caspase-8 [119]. Accordingly, apoptotic neutrophils protected mice against LPS-induced septic shock, and in a model of pneumonia, treatment with RvE1 increased the percentage of apoptotic neutrophils and reduced neutrophil infiltration and lung injury [119, 120]. Moreover, the pro-resolving mediators LXA4, RvE1 and PD1 increased the expression of CCR5 on apoptotic neutrophils that assist in vivo in the sequestration and clearance of CCL3 and CCL5 from inflamed sites [121]. Live neutrophils also have the capacity to trap and scavenge chemokines and cytokines. For instance, neutrophils express the IL-1 receptor antagonist, a soluble molecule which binds to IL-1R without inducing any intracellular signal [122] and the type II IL-1 decoy receptor (IL-1RII), both in membrane-bound or released forms, which binds IL-1 and prevents its interaction with its signalling receptor complex IL-1R1 [123].

Neutrophils in pathology

Infection

Neutrophils have long been recognised for their ability to sense and eliminate extracellular pathogens. However, recent evidence demonstrating their recruitment during the IL-17/Th17 response and their involvement in the host response against intracellular pathogens challenges this view [8]. For instance, the granule-associated molecules cathepsin G and neutrophil elastase play a crucial role in pulmonary protective immunity against mycobacterial infection [124]. In a genome-wide transcriptional profile from the blood of patients with active tuberculosis (TB), latent TB and healthy controls, a signature revealed an overexpression of type I IFN-inducible transcripts in blood neutrophils from infected patients, thus supporting a role for these cells in the pathogenesis of TB [125]. Consistently, neutrophils express various cytosolic DNA sensors promoting IFN-β production upon infection by intracellular pathogens (e.g. Legionella pneumophilia, B. henselae, L. monocytogenes) [21]. Neutrophilic inflammation observed during M. tuberculosis infection is detrimental to the host and has been associated with increased infection [126]. IFN-γ controls neutrophilic inflammation via inhibition of the IL-17/Th17 response and may directly act on neutrophils to prevent their accumulation [126, 127].

Upon neutrophil activation via TLR engagement, G-protein-coupled receptor kinase-2 induces CXCR2 desensitization and internalization. Accordingly, TLR9 deficiency enhances neutrophil recruitment to the site of infection, increases bacterial clearance and improves sepsis outcome [128]. IL-33, a member of the IL-1 family, is involved in regulating the activity of neutrophils in infections. For instance, it inhibits the downregulation of CXCR2 expression induced by TLR4 ligation in mouse and human neutrophils. Interestingly, patients who do not recover from sepsis have high serum levels of soluble ST2 (ST2 is also known as IL-1RL1), the decoy receptor of IL-33, and their neutrophils have reduced expression of CXCR2, suggesting a therapeutic potential of IL-33 in sepsis [129]. Moreover, IL-33 increases the expression of CR3 in neutrophils via a mechanism involving both the TLR and dectin-1 signalling pathways [130]. Therefore, IL-33 enhances the phagocytosis and killing of opsonized C. albicans by neutrophils and the administration of IL-33 protects mice against C. albicans infection in a neutrophil-dependent mechanism [130].

Neutrophils in chronic inflammation and autoimmunity

Although neutrophils are generally linked to acute inflammation, recent reports have challenged the dogma and demonstrated their fundamental implication in the development and/or persistence of chronic inflammation. For instance, neutrophil infiltration in response to zymosan-induced peritonitis is enhanced in mice experiencing various chronic inflammatory conditions, probably via IL-17A produced during chronic inflammation [131].

In chronic obstructive pulmonary disease (COPD), immunoglobulin-free light chains (IgLC), found in serum and lung tissue of patients, bind to neutrophils, activating and inducing them to secrete CXCL8 [132]. Interestingly, in a murine model of lung emphysema induced by cigarette smoke, the administration of IgLC antagonist reduces neutrophil influx in the broncho-alveolar space and activation [132]. Moreover, selective neutrophil chemoattraction promoted by the tripeptide proline–glycine–proline (PGP) has been implicated in the persistence of COPD [133, 134]. The leukotriene A4 hydrolase, which promotes the synthesis of the chemotactic factor, LTB4, also has the capacity to degrade PGP via its aminopeptidase activity [133]. However, the aminopeptidase but not the hydrolase activity of LTAH4 is inhibited by cigarette smoke, thereby leading to sustained neutrophil recruitment and chronic lung inflammation by the combined action of LTB4 and PGP [133, 134]. Since the extracellular levels of chloride ions selectively activate the peptidase activity of LTAH4, a similar mechanism may be at play in cystic fibrosis, characterized by chronic neutrophilic inflammation [133]. Indeed, due to mutations in the cystic fibrosis transmembrane conductance regulator protein, the extracellular levels of chloride ions are reduced in cystic fibrosis and are likely to be responsible for the measurable levels of PGP in the sputum of these patients [135]. In addition, gamma-glutamyl transferase, observed in high levels in cystic fibrosis sputum and involved in the catabolism of the antioxidant and mucolytic glutathione, are found associated with the neutrophil secretory vesicles and released after stimulation [136]. Thus, neutrophils can directly contribute to the low concentration levels of glutathione found in cystic fibrosis airways and in worsening respiratory function [136].

The nature of the contribution of neutrophils in autoimmune disorders is not well defined, despite their recognised role in pathogenesis. Recent observations demonstrated the involvement of NETs in autoimmunity. Indeed, the degradation of NETs by DNase I, normally observed in healthy human serum, is compromised in a subset (36.1 %) of patients with systemic lupus erythematosus (SLE), a multi-organ autoimmune disease characterized by an interferon and granulopoiesis signature [137, 138]. NETs activate the classical complement pathway leading to C1q deposition, which inhibits DNase I and, thus, their degradation [139]. A positive correlation has been reported between undegraded NETs and the levels of antinuclear and anti-NET antibodies and with the frequency of lupus nephritis development [138]. Accordingly, netting neutrophils are found in the skin and kidney of lupus patients, and serum from lupus patients contains immune complexes composed of autoantibodies, notably anti-ribonucleoproteins (anti-RNP IgG), self-DNA and antimicrobial peptides, such as LL-37 or HNP [140–142]. Therefore, defective NET clearance leads to the expression of a set of autoantigens and danger-associated molecular patterns, known to trigger and promote inflammation [116]. In Felty’s syndrome, a variant of rheumatoid arthritis defined by an enlarged spleen and an abnormally low white blood cell count, circulating autoantibodies, in particular directed against PAD4-deiminated histones, are found associated with NETs [143]. These immune complexes are transported across pDC membranes via binding to CD32 (FcγRIIB), where self-DNA induces the production of IFN-α. In turn, IFN-α primes neutrophils for further NET formation [140–142]. Interestingly, a distinct subset of neutrophils (low-density granulocytes) found in SLE patients display an activated phenotype; secrete high levels of type I IFNs, TNF-α and IFN-γ; overexpress a set of immunostimulatory proteins and alarmins; have a high capacity to produce NETs; and induce significant endothelial cell cytotoxicity [140, 144].

Neutrophils are also linked to vascular diseases where the presence of anti-neutrophil cytoplasmic antibodies (ANCAs) is a hallmark of ANCA-associated vasculitis (e.g. Wegener’s granulomatosis, Churg–Strauss syndrome and microscopic polyangiitis) [145]. NETs promote injury of the endothelium and the surrounding tissue via the presence of extracellular histones, which have been involved in cytotoxic activity, organ failure and death [146, 147]. Accordingly, pre-incubation of NETs with antibodies against histones reduced NET-induced cytotoxicity in vitro [148]. NETs are produced by ANCA-stimulated neutrophils and found in kidney biopsies from patients with small-vessel vasculitis and SLE, where they contribute to endothelial damage [140, 149].

In a mouse model of transfusion-related acute lung injury (TRALI), neutrophils capture circulating platelets, which results in the production of ROS by neutrophils and subsequent vascular injury [150]. Moreover, activated platelets, found during vessel injury and sepsis, induce NET formation, causing endothelial and tissue damage [146, 151–153]. Disruption of NETs by intranasal DNase I treatment in TRALI improves blood arterial oxygenation and reduces lung oedema, lung vascular permeability and mortality [151, 152]. In humans, NETs were detected in plasma and lungs of patients with TRALI, suggesting that clinical trials to target platelet activation and platelet–neutrophil interactions should be considered in this condition [151].

Neutrophils, and in particular NET formation, are involved in the coagulation cascade and thrombus formation by supporting platelet adhesion and aggregation, thrombin-dependent fibrin formation and blood clot formation [153–155]. Accordingly, in an experimental model of deep vein thrombosis, NETs are found in the thrombus and the administration of DNaseI protects mice from thrombosis [154, 156]. Moreover, NET formation has been involved in cancer-associated thrombosis, one of the major causes of death in cancer patients [155].

Neutrophil recruitment into joints, induced by LTB4 and chemokines produced by stimulated synovial cells, is a hallmark of arthritis [157]. Their recruitment into joints requires the GPI-anchored protein Ly6G, which is closely associated with β2-integrins at the neutrophil surface [158]. Moreover, Syk-dependent signalling in neutrophils, required for FcγR-induced signalling, is essential to establish an immune complex-mediated arthritis [159].

Finally, CXCR2-dependent neutrophil activation and consequent inflammatory disease has been involved in susceptibility to murine multiple sclerosis, an inflammatory demyelinating disorder of the central nervous system [160, 161]. For instance, transfer of CXCR2-positive neutrophils into CXCR2-deficient resistant mice restores susceptibility to autoimmune encephalomyelitis [161]. Moreover, in a model of cuprizone-induced demyelination, CXCR2 knockout mice are resistant due to deficient neutrophil effector responses [160]. Thus, CXCR2 not only contributes to neutrophil migration into tissues but also promotes their effector functions.

Neutrophils and cancer

Genetic instability and inflammation, such as inflammatory cell infiltration, chemokine and cytokine expression in the microenvironment of most neoplastic tissues, have been proposed to represent hallmarks of cancer [162]. Among myeloid cell subsets infiltrating the tumour stroma, tumour-associated macrophages (TAM) were the most prominent and best-characterized cells implicated in tumour progression, stroma deposition and remodelling, angiogenesis and antitumour T cell-dependent immunity [163]. However, tumour-associated neutrophils (TAN) have recently emerged as key mediators in malignant transformation, tumour progression and in the regulation of antitumour immunity [2, 127].

Neutrophils and prognostic significance

The relationship between TAN infiltration and prognosis in human cancer has not been systematically investigated, as has been done for macrophages [163]. However, evidence based on epidemiological studies and animal models is consistent with the view that neutrophil infiltration and accumulation within neoplastic tissues may be associated with a poor clinical outcome, as observed in human patients with bronchoalveolar carcinoma, hepatocellular carcinoma, colorectal carcinomas, aggressive types of pancreatic tumours, or head and neck squamous cell carcinoma [164–169]. In contrast, neutrophil infiltration has been associated with a favourable prognosis in patients with gastric carcinoma [170]. Collectively, these results suggest that depending on the localisation, the prognostic significance of infiltrating neutrophils may differ, as observed for other leukocyte populations [163].

Neutrophil recruitment into the tumour and tumour promotion

The tumour-associated inflammation, a hallmark of cancer, triggers the production of CXC chemokines (CXCL8, CXCL1, CXCL2, CXCL3, CXCL5) by cells infiltrating or surrounding the tumour (e.g. TAM) and by tumour cells themselves [2, 171, 172]. This family of chemokines was linked to cancer progression and in particular to tumour angiogenesis and metastasis [173]. In animal models, a correlation was found between the expression levels of CXC chemokines and lung tumour progression, and CXCR2 inhibition reduced pancreatic ductal adenocarcinoma progression [174, 175]. Most recently, Jamieson and colleagues [176] reported that in models of inflammation-induced skin papillomas, colitis-associated tumour or spontaneous malignancy, CXCR2 deficiency or neutrophil depletion suppressed the inflammation-associated tumourigenesis and the spontaneous development of tumours.

More than 30 years ago, Clark and Klebanoff [177] suggested a neutrophil-dependent tumour cell cytotoxicity mediated by the peroxidase system. In contrast, in 1995, Pekarek et al. [178] found that granulocytes are required for the rapid growth of tumour cells and that their depletion inhibits tumour development. Evidence suggesting the involvement of neutrophils in the promotion and progression of cancer has followed these original observations [2, 127].

Neutrophil-associated proteins stored within granules and directed to defence towards pathogens are also involved in tumour promotion. Neutrophil elastase shows a dual role in tumour initiation. For instance, NE is taken up into a specific endosomal compartment of adjacent epithelial tumour cells and hydrolyses insulin receptor substrate-1, which normally interacts with a subunit of the PI3K and blocks its activity [179]. Therefore, upon NE activity, PI3K enhances the signalling of the platelet-derived growth factor receptor, thereby leading to tumour cell proliferation [179]. In contrast, NE taken up by breast cancer cells cleaves cyclin E into a truncated isoform, which is subsequently presented in the context of HLA-I molecule and promotes a T lymphocyte-mediated lysis of cancer cells [180].

Neutrophil-derived cytokines contribute also to tumour progression. The expression of GM-CSF by breast tumour cells promotes the production of oncostatin M by neutrophils, which in turn induces the production of vascular endothelial growth factor (VEGF) by breast cancer cells, reduces cancer cell adhesiveness and increases their invasive capacity [181]. Hepatocyte growth factor (HGF), a cytokine involved in cell proliferation and motility, is stored in secretory vesicles and granules of neutrophils [182]. Interestingly, neutrophils enhance the invasiveness of human cholangiocellular carcinoma and hepatocellular carcinoma cells in vitro via a HGF secretion-dependent mechanism [183]. Moreover, in patients with bronchoalveolar carcinoma, tumour cells express the HGF receptor and neutrophil-derived HGF promotes their migration [166]. Accordingly, the levels of HGF found in the bronchoalveolar lavage fluids of patients are correlated with the neutrophil counts and associated with poor prognosis [166].

A mechanism facilitating the occurrence of metastasis has been directly linked with the recruitment and activation of neutrophils. Indeed, melanoma cell-derived CXCL8 increases the expression of β2-integrin on neutrophils and promotes a neutrophil–melanoma cell interaction via the expression of ICAM-1 by melanoma cells [171]. In turn, this cross talk favours the transmigration of melanoma cells across the endothelium and facilitates the development of lung metastasis [171]. In contrast, in mice orthotopically implanted with breast cancer cells, TAN accumulate in the premetastatic lung and their depletion increases the metastatic burden. The authors suggested that tumour-educated neutrophils are recruited to the premetastatic lung and inhibit tumour cell seeding through an H2O2-dependent cytotoxic activity [184]. Neutrophils acquire this cytotoxic phenotype upon accumulation and sequestration in the lung mediated by granulocyte colony-stimulating factor (G-CSF) and a subsequent activation induced by CCL2 [184].

Angiogenic switch mediated by neutrophils

TAM and TAN are involved in tumour angiogenesis through the production of growth and matrix remodelling factors such as VEGF, basic fibroblast growth factor, platelet-derived growth factor, urokinase-type plasminogen activator and metalloproteinases (MMPs). For instance, neutrophil-derived VEGF has been identified as the major factor responsible for the in vivo angiogenesis activity induced by CXCL1 [185]. In addition, neutrophil-derived MMP-9 induces VEGF expression in the neoplastic tissue and therefore catalyses tumour angiogenesis [186]. In patients with hepatocellular carcinoma and head and neck cancer, neutrophils are the major source of MMP-9 in the peritumoural stroma and within the tumour, respectively [164, 187]. Interestingly, a microvascular architecture promoting blood-borne metastasis is observed in hepatoma samples presenting an elevated number of neutrophils [164]. Moreover, depletion of granulocytes reduces tumour angiogenesis and growth in murine hepatoma, demonstrating that neutrophils control the progression of tumour angiogenesis in vivo [164].

In a model of subcutaneous injection of melanoma or fibrosarcoma cells, neutrophils were identified as the major cells with angiogenic activity negatively controlled by IFN-β, a potential adjuvant in cancer vaccines. High expression levels of VEGF, CXCR4 and MMP-9, associated with better developed blood vessels, are found in neutrophils from IFN-β mice [188]. Interestingly, neutrophil depletion in IFN-β-deficient mice reduces tumour growth [188], suggesting the relevance of the interplay between INF-β and neutrophils in the early stages of cancer development.

Tumour-associated G-CSF enhances the expression of Bv8 (also known as prokineticin-2) by neutrophils, which promotes their mobilization and tumour angiogenesis [189, 190]. Interestingly, treatment of tumours refractory to anti-VEGF therapy with anti-G-CSF or anti-Bv8 reduces tumour growth and angiogenesis [191].

Finally, NET formation has been recently linked to cancer-associated thrombosis, a highly common cause of death in cancer patients [155]. Neutrophils isolated from leukaemic mice or tumour-bearing mice are primed to release NETs, most likely via the increased levels of G-CSF found in cancer, which thus promotes a prothrombotic milieu [155].

Neutrophilic inflammation is linked to genetic instability

More than 25 years ago, neutrophilic inflammation and, in particular, neutrophil-derived ROS were linked to genetic instability [192]. Today, several lines of evidence are consistent with the view that neutrophils are linked with the process of carcinogenesis through ROS-dependent and ROS-independent mechanisms. Accordingly, the expression of neutrophil-derived inducible nitric oxide synthase and ROS, notably the MPO-mediated formation of HOCl, have been linked with point mutations or DNA damages, and levels of TAN significantly correlated with DNA mutations [193–195]. In ulcerative colitis, neutrophils located in crypt abscesses are activated and induce DNA damage, as suggested by G2/M cell cycle checkpoint arrest in colon epithelial cells [196].

Plasticity of TAN

In 1989, the plasticity of neutrophils was suggested in the context of tumours in which neutrophils isolated from tumour-bearing animals significantly increased the metastatic potential of adenocarcinoma cells compared to neutrophils isolated from healthy animals [197]. More recently, reports have confirmed that neutrophils are endowed with unexpected plasticity [70, 172]. Therefore, mirroring the M1–M2 and Th1–Th2 paradigms, neutrophils can be polarized towards a pro-inflammatory phenotype with antitumour activity (N1) or towards a protumoural N2 phenotype [172]. TGF-β, which has a regulatory role on neutrophil functions, including chemotaxis or cytotoxicity, and plays a critical role in tumour initiation, progression and metastasis (e.g. suppressor or promoter depending on the context and stage of the tumour), is a crucial effector in this polarization [172]. In models of lung adenocarcinoma and mesothelioma, TGF-β drives neutrophils towards a N2 phenotype, whereas TGF-β blockade enhances the infiltration of N1 TAN, which are characterized by a cytotoxic activity against tumour and a pro-inflammatory phenotype (i.e. TNF-αhigh, CCL3high, ICAM-1high, arginaselow) [172]. The authors have demonstrated that TGF-β blockade is associated with the activation of a CD8+ T cell-dependent arm that involves neutrophils as effectors [172].

Since studies have indicated that neutrophils can also exert an antitumour immunity [184], in analogy with their “big brothers”, macrophages, and depending on environmental signals, neutrophils can exert a dual influence on tumour growth. Moreover, in parallel to the differences found between mouse and human macrophage polarization, the existence and functional properties of N1 and N2 in human have to be carefully investigated.

Relation between neutrophils, TAN and MDSC

Cancer has provided a paradigm for myeloid-derived suppressor cells (MDSC), which represent a heterogeneous group of myeloid cells composed of monocytic (M-MDSC) and granulocytic (G-MDSC) subsets, recognised for their immune-suppressive activity [198]. Neutrophils and TAN share phenotypic (e.g. cell markers and morphology) and functional properties (e.g. recruitment within tumours, production of arginase, promotion of tumour angiogenesis) with G-MDSC [198]. Moreover, G-MDSC and neutrophils were usually defined and isolated using the same phenotype markers (i.e. CD11b+ Ly6G+ Ly6Clow), creating confusion to discriminate these cells. Therefore, whether TAN activity can be attributed to MDSC remains a matter of debate. Reports have suggested that activated neutrophils present an immunosuppressive activity in cancer patients and could account for the arginase I-mediated suppression of lymphocytes in renal cell carcinoma [199, 200]. Recently, in a genetic model of lung adenocarcinoma, but also in patients with invasive cancer, splenic granulocyte/macrophage progenitors have been proposed as TAN progenitors [201]. Consistent with the hypothesis that a part of the MDSC activity can be attributed to TAN, a subset of MDSC accumulated within the spleen of tumour-bearing animals are indeed immature myeloid cells [201–203]. However, other studies have reported that MDSC and neutrophils are functionally and phenotypically different [202, 204]. For instance, transcriptomic analysis of mouse TAN, naive neutrophils and G-MDSC has suggested that TAN and G-MDSC are distinct populations of cells and that naive neutrophils and G-MDSC are more closely related to each other than to TAN [205]. Interestingly, upon stimulation by GM-CSF, G-MDSC acquire the same characteristics as neutrophils, suggesting that G-MDSC are immature neutrophils [204]. These immature neutrophilic MDSC have also been reported in the peripheral blood of patients with cancer, and high levels are associated with poor prognosis [202, 203]. Collectively, MDSC are a heterogeneous population of myeloid cells which includes immature granulocytes. Therefore, further investigations are required to identify new biomarkers and discriminate the distinct subpopulation of cells within MDSC.

Concluding remarks

Long viewed as ultimate short-lived effectors, several lines of evidence have demonstrated in vitro, but also in vivo, that neutrophils are endowed with unsuspected diversity and plasticity. Conforming to Metchnikoff’s model, neutrophils patrol the bloodstream to detect and phagocyte pathogens. However, this classical mechanism has been complemented by the discovery of NETs, which trap and kill extracellular pathogens. Their importance in innate immunity was notably put into evidence by evolutionary strategies adopted by pathogens to escape these poisonous NETs.

Neutrophils use a set of membrane and intracellular molecules to sense their local environment signals and switch their phenotype towards a pro-inflammatory and antitumour (N1) or anti-inflammatory and protumoural (N2) programs. In turn, neutrophils are involved in a bidirectional cross talk with most other types of leukocytes and can directly or indirectly modify their maturation, activation or effector functions, depending on the context. These new vistas shed new light on neutrophil function, but also raise new questions. For instance, the existence of neutrophil plasticity and diversity in humans together with their role in human pathologies require further studies.