Introduction

Neutrophils have long been known to serve as an essential line of resistance against infectious agents in innate immunity and downstream of polarized T helper 17 (TH17) cell-driven adaptive immune responses1,2,3. Moreover, not only do neutrophils represent a hallmark of acute inflammation but they are also an important component of circuits that orchestrate the activation, orientation and regulation of adaptive immune responses and chronic inflammation. By expressing a wide repertoire of cytokines, and immunosuppressive and stimulatory molecules, neutrophils engage in complex bidirectional interactions with lymphoid cells and macrophages4,5.

The tumour microenvironment (TME) has emerged as an essential component of neoplasia6. Inflammatory cells and components of the humoral arm of innate immunity are key players in cancer-related inflammation, and contribute to tumour progression, from initiation to seeding at distant anatomical sites6,7,8,9.

Although attention has long been focused on macrophages as a paradigm of cancer-related inflammation10, several lines of evidence in preclinical and clinical conditions point to a role for neutrophils11,12. Neutrophils infiltrate solid tumours to a variable extent, as assessed by conventional immunohistochemical staining for neutrophil markers (for example, CD66b in humans and lymphocyte antigen 6G (Ly6G) in mice) and expression of a neutrophil transcriptional signature11,12,13,14,15,16,17. In most, but not all, human tumours, high infiltration with tumour-associated neutrophils (TANs) has been associated with poor prognosis12. Accordingly, in tumour cell line transplantation models and genetically engineered mouse models (GEMMs) of cancer, TANs have been reported to be a component of tumour-promoting inflammation3,11. However, neutrophils can engage in pathways of antitumour resistance by killing tumour cells directly and/or by interacting with other components of immunity3,11,12. Thus, neutrophils have the potential to be both pro-tumorigenic and antitumorigenic within the TME, and this dual function is likely a reflection of their unexpected plasticity in response to environmental cues.

Here, we review current evidence for the role of neutrophils in tumour progression and metastasis in light of their diversity and plasticity. Previous reviews on the immunobiology of neutrophils and TANs1,2,3,4,11,12,18,19 provide a framework for the present Review, in which emphasis is placed on describing neutrophil diversity and their prognostic and therapeutic implications.

Homeostasis and recruitment

Development and mobilization

Neutrophils represent 50–70% and 10–25% of circulating leukocytes in humans and mice, respectively20,21. In peripheral blood, neutrophils are short-lived cells and require a constant replenishment from bone marrow precursors dependent on signalling through the granulocyte colony-stimulating factor receptor (G-CSFR)2,22 (Box 1). Indeed, mutations in colony-stimulating factor 3 receptor (CSF3R), which encodes G-CSFR, and in haematopoietic cell-specific Lyn substrate 1-associated protein X1 (HAX1), whose protein product contributes to the G‐CSFR signalling pathway, have been associated with severe neutropenia in humans23,24. Similarly, deficiency in granulocyte colony-stimulating factor (G-CSF) or G-CSFR leads to severe neutropenia in mice25,26. In addition to the essential role played by G-CSF, other mediators, such as granulocyte–macrophage colony-stimulating factor (GM-CSF) and the pro-inflammatory cytokine interleukin-6 (IL-6), are involved in the regulation of the development of neutrophils, in particular during an inflammatory response2,22,27,28.

The trafficking of neutrophils from bone marrow into peripheral blood is tightly regulated, in particular by signalling through the chemokine receptors CXC-chemokine receptor 2 (CXCR2) and CXCR4. Expression of CXC-chemokine ligand 12 (CXCL12) by bone marrow stromal cells mediates the retention of CXCR4+ immature neutrophils (NI); as an operational nomenclature based on the expression of selected molecules, we use NI, NM, NA and NISG to refer to immature neutrophils, mature neutrophils, aged neutrophils and neutrophils with an interferon-stimulated gene signature, respectively (see below) (Box 2). Decreased expression of CXCR4 in bone marrow-localized NM coupled with activation of CXCR2 signalling triggers the entry of NM into the circulation29. Ageing neutrophils upregulate the expression of CXCR4, driving their homing back to the bone marrow and their subsequent elimination by macrophages29,30. The cellular composition and molecular signature of the haematopoietic niche, including the number of CXCL12-abundant reticular mesenchymal stem cells (CAR MSCs), is regulated by the rhythmic clearance of neutrophils by macrophages30.

The process of neutrophil ageing in the circulation is regulated by gut microbiota and is controlled by neutrophils themselves through a cell-autonomous transcriptional programme30,31,32. Whether this applies to other myeloid cells, such as monocytes, remains unknown. Indeed, circadian expression of the transcription factor BMAL1 controls the production of CXCL2 by neutrophils in a cell-intrinsic manner. In turn, CXCL2 signals through CXCR2 to induce neutrophil ageing32. Elimination of apoptotic neutrophils and formation of new cells are interconnected processes functioning as a homeostatic rheostat essential to prevent exacerbated inflammation and tissue damage32,33,34.

Recruitment in cancer

In established neoplasia in mice and humans, altered haematopoiesis is usually observed, a reflection of the production of growth factors (G-CSF and GM-CSF) and inflammatory cytokines (for example, IL-6, IL-1β and IL-17) by tumour cells, tumour-associated stromal cells and tumour-infiltrating leukocytes, including T cells, macrophages and neutrophils35,36,37. Neutrophilia and the appearance in the circulation of immature myeloid cells both occur in humans and mice, and together these cells are potent mediators of immunosuppression as discussed below11,12,38 (Fig. 1).

Fig. 1: Neutrophils in tumour promotion.
figure 1

Neutrophils can sustain tumour growth via different mechanisms, including the suppression of T cell activation, the promotion of genetic instability, tumour cell proliferation, angiogenesis and metastasis. a | The production of growth factors, interleukin-17 (IL-17), complement component C5a, oxysterols and CXC chemokines drives the production, recruitment and survival of neutrophils11,35,43,44. b | Chemokines induce the mobilization of neutrophils in the pre-metastatic niche, which sustain the arrival of the metastatic cells via the production of several mediators. Circulating neutrophils escort circulating tumour cells and promote their survival and extravasation by direct interaction mediated by integrins91,93. c | Molecules present in the tumour microenvironment, including granulocyte colony-stimulating factor (G-CSF) and transforming growth factor-β (TGFβ), induce the expression of arginase 1 (ARG1), reactive oxygen species (ROS) and nitric oxide (NO) by neutrophils, which inhibit the activation of T cells35,38,64. Granulocyte–macrophage colony-stimulating factor (GM-CSF) induces the expression of fatty acid transport protein 2 (FATP2) in neutrophils. FATP2 promotes the uptake of arachidonic acid and the synthesis of prostaglandin E2 (PGE2)115. Neutrophils express the immune checkpoint ligands programmed cell death 1 ligand 1 (PDL1)117,118,119,120 and V-domain immunoglobulin suppressor of T cell activation (VISTA)122. d | Neutrophils induce genetic instability via the production of ROS and the release of microparticles containing microRNAs miR-23A and miR-155, which downregulate the expression of molecules involved in the maintenance of nuclear integrity72,73,74. Neutrophils sustain tumour proliferation via the production of epidermal growth factor (EGF), hepatocyte growth factor (HGF) and platelet-derived growth factor (PDGF)76,136, the release of neutrophil extracellular traps containing high mobility group protein B1 (HMGB1) that activates a Toll-like receptor 9 (TLR9)-dependent pathway in cancer cells, neutrophil elastase (NE) and matrix metalloproteinase 9 (MMP9) that cleaves laminin 11188,90. Cleaved laminin 111 triggers the proliferation of cancer cells through activation of integrin signalling90. Neutrophils sustain tumour angiogenesis through the release of the pro-angiogenic factors BV8, the S100 proteins S100A8 and S100A9, and MMP9 that activate vascular endothelial growth factor A (VEGFA) in the extracellular matrix51,81,82,83. G-CSFR, G-CSF receptor; GM-CSFR, GM-CSF receptor; ICAM1, intercellular adhesion molecule 1; RTK, receptor tyrosine kinase; TAN, tumour-associated neutrophil; TGFβR, TGFβ receptor.

Full size image

Neutrophils express high levels of the chemokine receptors CXCR1 and CXCR2, which play a major role in their recruitment to the TME where CXC chemokine ligands (for example, CXCL1, CXCL2, CXCL5, CXCL6 and CXCL8 (also known as IL-8)) are expressed by tumour cells, tumour-infiltrating leukocytes, endothelial cells and fibroblasts39,40. In addition to chemokines, inflammatory cytokines (for example, IL-17, IL-1β and tumour necrosis factor (TNF)) have been implicated in neutrophil mobilization and recruitment in cancer. In particular, these cytokines are part of an inflammatory circuit that leads to the production of G-CSF and the subsequent formation and mobilization of neutrophils from the bone marrow11,35,41. Moreover, IL-1β and G-CSF dramatically prolong the survival of neutrophils42. In addition to these molecules, tumour-derived oxysterols and the complement component anaphylatoxin C5a have been shown to contribute to neutrophil recruitment in mouse tumours43,44,45.

The dynamics and lifespan of neutrophils in tumours remain to be fully elucidated. In a transplantation mouse model of head and neck cancer and in primary sarcomagenesis in mice induced by treatment with 3-methycholanthrene (3-MCA), neutrophils were already present at the tumour site or at the carcinogen injection site at early time points (hours and/or days after tumour initiation and carcinogen injection)8,15,46. Specifically, intravital multiphoton imaging revealed that neutrophils infiltrated the tumour within 3 h after tumour cell injection in the transplantation model of head and neck cancer and persisted for up to 3 days in the TME46. Interestingly, the motility of intra-tumoural neutrophils was reduced compared with peri-tumoural neutrophils, suggesting different states of activation or polarization46. Early in cancer development, neutrophils isolated from tumour-bearing mice and patients with cancer show increased spontaneous and chemokine-induced migration in vitro mediated by autocrine ATP signalling through the purinergic receptors P2Y1 and P2Y2, compared with neutrophils isolated from tumour-free individuals or those with late-stage cancer47. Therefore, neutrophils undergo dynamic changes during cancer development and progression as also discussed below.

Neutrophils have also been reported to accumulate in the metastatic niche, where the expression of G-CSF, CXCL1 and CXCL2 by cancer cells and stromal cells promoted their recruitment35,48,49,50,51,52 (Fig. 1b). In an orthotopic transplantation model of breast cancer and a GEMM of oncogene-driven mammary carcinogenesis, the mobilization of neutrophils into the metastatic lung was regulated by the atypical chemokine receptor 2 (ACKR2), a decoy and scavenger receptor for inflammatory CC chemokines, expressed in early haematopoietic precursors49. Genetic deficiency of this molecule in mice increased the expression of inflammatory CC-chemokine receptor 1 (CCR1), CCR2 and CCR5 in haematopoietic progenitors, which accelerated the maturation rate, mobilization and activation of neutrophils, and thus restrained metastasis49. In patients with breast cancer, expression of ACKR2 was found to be inversely correlated with the stage of the disease53. However, ACKR2 was also expressed by non-tumoural and tumoural mammary epithelial cells and the relative importance of haematopoietic versus tumour cell expression in this neoplasm remains to be assessed.

Thus, results obtained in preclinical mouse models and in humans suggest that neutrophil recruitment to and survival in neoplastic tissues involves upstream regulation of myelopoiesis and a complex network of chemokines, cytokines, G-CSF and complement components.

Neutrophil diversity

Neutrophil diversity in health

Under homeostatic conditions, circulating and tissue neutrophils exhibit considerable diversity, with phenotypic and functional heterogeneity driven by maturation and ageing as well as cues from the tissue microenvironment18. Circadian oscillations and ageing affect the neutrophil proteome, including the repertoire of chemokine receptors, pattern recognition receptors and molecules involved in adhesion, the inflammasome and vesicular transport as well as the production of neutrophil extracellular traps (NETs) and the capacity to migrate30,31,32,54.

In the circulation, NM newly released from the bone marrow are characterized by high expression of CD62L and CXCR2 and low expression of CXCR4 (Box 2). NM are released during the night and the early morning, and predominate at zeitgeber time (ZT) 13 (that is, 13 h after initiation of a 12-h light–12-h dark cycle)30. The process of neutrophil ageing is a bona fide circadian process. Over a period of 6–8 h, expression of CD62L is dramatically reduced and NA, characterized by high expression of CXCR4 and CD11b and a hypersegmented nucleus, predominate at ZT5 (that is, 5 h after lights on)30,31,32. These phenotypic variations favour neutrophil clearance and suggest that neutrophil-dependent immune and inflammatory responses are not stable over time and may fluctuate during the circadian cycle. In agreement with this hypothesis, NA display reduced migration into inflamed tissues, compared with NM32.

Divergent results have been reported with respect to the capacity of ageing neutrophils to produce NETs31,54. These apparently conflicting findings may reflect different methods used for enriching NA (that is, prevention of neutrophil extravasation by injection of antibodies to block P-selectin and E-selectin or isolation of neutrophils at ZT5 in untreated mice). Furthermore, in steady-state conditions, circulating neutrophils were shown to undergo homeostatic degranulation and to lose their capacity to form NETs before they penetrate tissues, limiting their tissue-damaging potential54. This process is driven by a cell-intrinsic mechanism controlling the circadian expression of CXCL2 induced by BMAL1, as observed for neutrophil ageing (see above)54. Therefore, neutrophil neutralization and ‘disarming’ and neutrophil ageing share molecular mechanisms and are integrated into a circadian programme, which protects the host from an excessive inflammatory response.

In addition to NM and NA, single-cell RNA sequencing (scRNA-seq) analysis of circulating neutrophils has identified NISG, which are characterized by the expression of a set of interferon-stimulated genes (reported in a preprint55). This neutrophil subset is present in mice and humans and could represent a population of neutrophils primed to fight infections. Interestingly, a similar population has been observed in tumours56 (see below) (Box 2).

Within tissues, neutrophils can undertake important homeostatic functions and acquire specific immunomodulatory properties, as occurs in the lymph nodes and spleen3,34,57. In particular, under homeostatic conditions, neutrophils expressing the major histocompatibility complex (MHC) class II molecule are present in lymph nodes in proximity to T cells, suggestive of a role as an antigen-presenting cell (APC)58. Neutrophils present in the marginal zone of the spleen promote immunoglobulin class switching and production of antibodies by activating B cells through the expression of B cell-activating factor (BAFF), a proliferation-inducing ligand (APRIL) and IL-21 (ref.57).

Thus, in homeostasis, neutrophils exhibit a previously unanticipated heterogeneity and are integrated into regulatory circuits of immunity5,32,34. Among mononuclear phagocytes, those cells originating from embryonic precursors perform mainly homeostatic functions, whereas the main function of macrophages derived from circulating monocytes in postnatal life is to respond to damage and inflammation, with plasticity of these cells being a major driver of diversity59. However, there is no evidence for ontogenetically distinct neutrophils or such strictly defined subsets. It can therefore be surmised that neutrophil diversity is the result of plasticity in response to differentiation and environmental signals.

Neutrophil diversity in cancer

Cancer has served as a paradigm for the plasticity and diversity of neutrophils, generated by the neutrophil maturation stage, response to tissue cues and cancer progression (Box 2). Neutrophil differentiation and maturation trajectories are profoundly altered in tumour-bearing mice60,61. In mice with advanced neoplasia, immature myeloid cells endowed with immunosuppressive properties appear in the circulation, primary tumours and metastases38,62,63 (Box 2). Similarly, early unipotent neutrophil progenitors (NePs) (Box 1) were found to accumulate in both the bone marrow and periphery in a transplantation mouse model of melanoma and a comparable cell subset (CD66b+;CD117+;CD34+/–) was identified in the blood of patients with melanoma63. NePs represent approximately 1% and 0.02% of all circulating CD45+ cells in healthy humans and non-tumour-bearing mice, respectively, and these frequencies increase to 3–9% in patients with melanoma and 0.2% in tumour-bearing mice63. Although these progenitors do not correspond to a subset of NM, they do contribute to the diversity of neutrophils found in patients with tumours and in tumour-bearing mice.

Transcriptomic analysis of neutrophils from the spleen and blood of mice bearing mammary carcinomas and tumour-free mice revealed a profound alteration of the transcriptional programme in neutrophils from tumour-bearing mice leading to an immunosuppressive phenotype, characterized by production of reactive oxygen species (ROS), nitric oxide (NO) and arginase 2 (ARG2) with potential to inhibit T cell proliferation ex vivo35,60,61 (Fig. 1c). In human and mouse lung cancers, scRNA-seq analysis of tumour-infiltrating myeloid cells revealed that TANs formed a continuum of phenotypic states, which can be resolved into five and six cell subsets, respectively, in humans and mice56. Three modules of gene expression within these cell subsets are conserved between humans and mice, including a module expressing canonical neutrophil markers (for example, matrix metalloproteinase 9 (MMP9), S100A8 and S100A9), a module expressing molecules involved in tumour inflammation and growth (for example, CC-chemokine ligand 3 (CCL3) and macrophage colony-stimulating factor 1 (CSF1)) and a module with a limited number of cells displaying strong expression of type I interferon-response genes (for example, interferon-regulatory factor 7 (IRF7) and interferon-induced protein with tetratricopeptide repeats 1 (IFIT1))56.

Analysis of TANs has revealed how signals present in the TME shape their function. In a GEMM of lung adenocarcinoma induced by oncogenic Kras and transplantation models of lung cancer and mesothelioma, transforming growth factor-β (TGFβ) was found to polarize neutrophil function in a pro-tumour direction characterized by TANs with high expression of ARG1, CCL17 and CXCL14 and low expression of CXCL10, CXCL13, CCL6, TNF and intercellular adhesion molecule 1 (ICAM1)64,65. Mirroring the M1–M2 nomenclature used for polarized macrophages10,59, N1 and N2 have been used to refer to antitumour and pro-tumour neutrophils, respectively64,65,66 (Box 2). However, in 3-MCA-induced primary sarcomas, TANs presented as a hybrid phenotype between N1 and N2 (ref.15).

In contrast to TGFβ, interferon-β (IFNβ) or a combination of IFNγ and GM-CSF drives neutrophils towards an antitumour state67,68,69 (Fig. 2). In early, but not late, non-small-cell lung cancer (NSCLC), IFNγ and GM-CSF have been shown to drive the differentiation of APC-like MHCII+ neutrophils expressing the co-stimulatory molecules OX40 ligand (OX40L), CD86 and 4-1BB ligand (4-1BBL; also known as TNFSF9)68,69 (Fig. 2a). A similar human leukocyte antigen-DR (HLA-DR)+ neutrophil population was observed in human head and neck cancer, spatially associated with activated T cells70.

Fig. 2: Antitumour potential of neutrophils.
figure 2

Neutrophils are involved in different mechanisms of antitumour resistance, including the activation of T cell-dependent antitumour immunity, direct cytotoxic activity against tumour cells or antimicrobial activity. a | Interferon-γ (IFNγ) and granulocyte–macrophage colony-stimulating factor (GM-CSF) present in the tumour microenvironment promote the maturation of immature neutrophils into antigen-presenting cells (APCs) expressing the major histocompatibility complex (MHC) class I and class II molecules and the co-stimulatory molecules CD86, 4-1BB ligand (4-1BBL) and OX40 ligand (OX40L)68,69. b | Different stimuli, such as granulocyte colony-stimulating factor (G-CSF), chemokines, such as CXC-chemokine ligand 8 (CXCL8), CXCL5 and CC-chemokine ligand 2 (CCL2), lipopolysaccharide (LPS) and IFNβ promote an oxidative burst and the production of hydrogen peroxide (H2O2) in neutrophils. Blocking signalling through the transforming growth factor-β receptor (TGFβR) (with the small-molecule inhibitor SM16) enhances the production of H2O2 (ref.64) In turn, H2O2 triggers an intracellular signalling pathway in tumour cells leading to the activation and opening of the non-selective cation channel transient receptor potential cation channel, subfamily M, member 2 (TRPM2), which induces a lethal influx of calcium (Ca2+) into cancer cells134. Hepatocyte growth factor (HGF) acts on the HGF receptor (HGFR; also known as MET) expressed by neutrophils and promotes the expression of inducible nitric oxide synthase (iNOS). In turn, iNOS induces the release of nitric oxide (NO), which results in killing of tumour cells136. c | In colorectal cancer (CRC), signalling through the interleukin-1 (IL-1) receptor type 1 (IL-1R1) in neutrophils enhances their antimicrobial activities, which limits bacterial-driven inflammation and CRC development144,145. d | Neutrophils engage in a tripartite interaction with macrophages and CD4–CD8–TCRαβ+ double-negative unconventional T cells (UTCαβ)15. Neutrophils amplify the production of IL-12 by macrophages, which in turn promotes polarization of the UTCαβ towards a type 1 immune response and IFNγ production. These cells are characterized by the expression of T cell receptor αβ chains (TCRαβ), IL-12R, IL-18R, molecules related to their innate-like phenotype (for example, Ly49 and CD94–NK cell receptor G2 (NKG2)) and IFNγ15. CCR2, CC-chemokine receptor 2; CXCR, CXC-chemokine receptor; G-CSFR, G-CSF receptor; GM-CSFR, GM-CSF receptor; TLR4, Toll-like receptor 4.

Full size image

In general, the evidence indicates that, early in carcinogenesis, TANs are part of cellular networks mediating antitumour resistance, whereas progression is associated with a functional switch setting these cells in an immunosuppressive, pro-tumour state12,47,66,68,69,71 (see below).

Neutrophils in tumour inflammation

Evidence from mouse models and patients, the latter discussed in detail below, strongly suggests that neutrophils are an important component of tumour-promoting inflammation in many types of cancer11,12 (Fig. 1d). In support of this, antibody-mediated neutrophil depletion (see Box 3 for a discussion of neutrophil depletion strategies and their limitations) results in protection from primary carcinogenesis and metastasis in GEMMs and inhibition of tumour growth in transplantation mouse models35,40,64.

Within the TME, TANs have been shown to affect epithelial genetic instability, tumour cell proliferation, angiogenesis, tissue remodelling and suppression of innate and adaptive lymphoid cell-mediated immunity (immunosuppression is discussed in detail in the next section). Production of a high quantity of ROS is a fundamental property of neutrophils3. In cancer, neutrophil-derived ROS have been associated with DNA damage and genetic instability in epithelial cells72,73,74. However, ROS-independent mechanisms for inducing DNA damage also exist and include neutrophil-derived microparticles, which deliver specific pro-inflammatory microRNAs (that is, miR-23A and miR-155) into intestinal epithelial cells, promoting the accumulation of DNA double-strand breaks via downregulation of the nuclear envelope protein lamin B1 and RAD51, a regulator of the homologous recombination pathway72.

Neutrophils can express a host of cytokines and growth factors relevant to tumour growth and progression, including epidermal growth factor (EGF), hepatocyte growth factor (HGF) and platelet-derived growth factor (PDGF)75,76. In a GEMM of lung adenocarcinoma induced by oncogenic Kras, the tumour burden was dramatically reduced with genetic deficiency of neutrophil elastase (NE) and this was associated with a reduction in tumour cell proliferation77. In vitro, NE activated the proliferation of human and mouse lung cancer cells by entering into an endosomal compartment and degrading the insulin receptor substrate 1 (IRS1), which promotes an interaction between PI3K and the PDGF receptor (PDGFR)77. This in vitro mitogenic activity of NE was also observed with epithelial and tumour cells of different origin, including human oesophageal cell lines and mammary epithelial cells through the transactivation of the EGF receptor (EGFR) and Toll-like receptor 4 (TLR4), and human prostate cancer cells through activation of the MAPK signalling pathway78,79,80.

Neutrophils play an important role in promoting tumour angiogenesis through the production of pro-angiogenic factors, including BV8, MMP9 and vascular endothelial growth factor A (VEGFA)51,81,82,83,84 (Fig. 1d). In the extracellular matrix, TAN-derived MMP9 induced the liberation and activation of VEGFA and consequent angiogenesis, whereas BV8 induced myeloid cell mobilization and acted as a mitogen for endothelial cells81,83. Neutrophil-derived BV8 has been implicated in resistance to anti-VEGF therapy and inhibition of G-CSF or IL-17 increased the therapeutic efficacy of anti-VEGF85,86,87. Collectively, these studies support a role for neutrophils in the initial angiogenic switch during tumorigenesis81,83,85,86,87.

NETs have been observed in different tumour types (that is, liver, breast, intestinal and gastric cancers) and their production has been shown to be driven by hypoxia, complement or fatty acids88,89,90,91,92. NET-associated molecules such as high mobility group protein B1 (HMGB1), NE and MMP9 can induce the proliferation of cancer cells88,90 (Fig. 1d). In a mouse model of lung metastasis, sustained lung inflammation promoted the formation of NETs, which in turn induced the proliferation of dormant cancer cells90. Indeed, the proteolytic remodelling of the extracellular matrix component laminin 111 by NE and MMP9, contained within NETs, induced the generation of a new epitope that triggered the proliferation of dormant cancer cells through α3β1 integrin activation90. In addition, entrapment of circulating tumour cells (CTCs) within NETs promoted the formation of metastases, and in vivo administration of DNase to eliminate NETs could reduce this effect88,91,93,94. In mouse models of liver metastases induced by intrasplenic injection of lung and colon cancer cells, intravital microscopy revealed decreased adhesion of CTCs to liver sinusoids in mice treated with DNase94. In vitro studies indicated that HMGB1 within NETs induced a TLR9-dependent activation of cancer cells, promoting their proliferation, migration and invasion capacity88. In addition, myeloperoxidase (MPO) contained in NETs promoted a hydrogen peroxide (H2O2)-induced TLR4-dependent pro-angiogenic response in endothelial cells, characterized by proliferation and motility95. Therefore, NETs can participate in tumour-promoting inflammation by driving angiogenesis, extracellular matrix remodelling and proliferation of tumour cells.

The tumour-promoting function of neutrophils occurs throughout the multistep process of dissemination and seeding at distant anatomical sites. Neutrophils have been reported to prepare the metastatic niche in organs as diverse as the lung and the liver50,96,97. Moreover, neutrophils have been reported to engage with CTCs in the bloodstream and to favour extravasation and subsequent metastatic growth98,99,100,101 (Fig. 1b).

GEMMs of cancer, including mammary tumours induced in keratin 14 (K14)–Cre;E-cadherin (Cdh1)fl/fl;Trp53fl/fl (KEP) mice35,36 and mouse mammary tumour virus (MMTV)-polyoma middle T antigen (PyMT) mice36,50, and colorectal cancer (CRC) induced in villinCreER;KrasG12D/+;Trp53fl/fl;Rosa26N1icd/+ (KPN) mice97, have provided insights into molecular mechanisms underlying neutrophil-mediated promotion of metastasis. In KEP mice, systemic accumulation of neutrophils with immunosuppressive activity was associated with increased formation of spontaneous lung metastases35. Mechanistically, the loss of p53 in cancer cells promoted the secretion of WNT ligands that stimulated the production of IL-1β by macrophages36. In turn, IL-1β activated the production of IL-17 by γδ T cells that drives neutrophil accumulation in the circulation and in the lung and promoted formation of metastases36. As IL-17 lies upstream of G-CSF in the signalling pathway, it increases the formation of neutrophils and their polarization into cells with immunosuppressive activity11,35,37. Indeed, G-CSF can promote the generation of immunosuppressive neutrophils and in vivo neutralization of G-CSF in KEP mice reversed the immunosuppressive phenotype of neutrophils35,61. Therefore, IL-17 derived from TH17 cells or γδ T cells participates in the neutrophilia observed in tumour-bearing individuals and drives the neutrophil-derived pro-tumour activities11,35. In the pre-metastatic niche of the lung, neutrophils produced factors facilitating the extravasation and growth of metastasis-initiating cells, including the pro-angiogenic molecules BV8 and MMP9 (observed in MMTV-PyMT mice)51, the chemoattractants S100A8 and S100A9 (observed in KEP mice)35, the proteases NE and cathepsin G that mediate the cleavage of thrombospondin 1 (TSP1; the resultant outcome of which is degradation of the extracellular matrix and abrogation of the TSP1-mediated suppression of tumour cell growth)102,103 (observed in MMTV-PyMT mice)104 and the pro-inflammatory cytokine IL-1β and the neutrophil chemoattractant leukotriene B4 (LTB4) (observed in MMTV-PyMT mice)50 (Fig. 1d).

Immunosuppression

NI and NM can express a host of mediators capable of suppressing innate and adaptive lymphoid cell function. These include ROS, reactive nitrogen intermediates (RNI), ARG1, prostaglandins and ligands of immune checkpoints.

Neutrophil-derived ROS have long been associated with suppression of T cell activation in cancer (Fig. 1c), in particular in advanced tumours11,47,64,105,106. In a transplantation mouse model of breast cancer, glucose deprivation in the TME triggered a metabolic switch in neutrophils that resulted in enhanced mitochondrial fatty acid oxidation, increased ROS production and consequent T cell suppression107. In addition to ROS, neutrophils can inhibit T cell activation through the inducible NO synthase (iNOS)-dependent production of NO, as observed in neutrophils from tumour-bearing KEP mice35,38 (Fig. 1c).

The production of ARG1 by TANs reduced the availability of l-arginine in the TME, which controls T cell metabolism and promotes T cell survival. In turn, this resulted in T cell dysfunction and alteration of T cell-mediated antitumour immunity64,66,108. The expression of ARG1 by TANs can be driven by TGFβ64,66 (Fig. 1c). Importantly, production of ARG1 by neutrophils has been shown to hamper the T cell response in human cancer, including in renal cell carcinoma and advanced-stage NSCLC109,110.

Endoplasmic reticulum (ER) stress has been associated with altered lipid metabolism, pathological activation and immunosuppressive activity of myeloid cells in cancer111,112,113, including neutrophils from patients with NSCLC and head and neck cancer114,115,116. In patients with NSCLC, a GEMM of pancreatic cancer induced by oncogenic Kras coupled with a Trp53 mutation and a transplantable mouse model of lymphoma, immunosuppressive neutrophils were characterized by their low density and increased expression of genes associated with the ER stress response (for example, C/EBP-homologous protein (CHOP; also known as DDIT3), X-box-binding protein 1 (XBP1), binding-immunoglobulin protein (BIP; also known as HSPA5) and AMP-dependent transcription factor (ATF4))116. Induction of ER stress in neutrophils upregulated the expression of lectin-like oxidized LDL receptor 1 (LOX1), a scavenger receptor involved in lipid metabolism, together with the onset of potent immunosuppressive activity114. In patients with NSCLC and head and neck cancer, LOX1+ neutrophils showed higher expression of ROS and ARG1 compared with LOX1– neutrophils and defined the neutrophil population as having immunosuppressive activity114. In addition to LOX1, immunosuppressive neutrophils present in tumour-bearing mice and patients with head and neck, breast and lung tumours exhibited an upregulation of other proteins involved in trafficking of lipids, such as CD36 and fatty acid transport protein 2 (FATP2)115. Although the role of LOX1 in the immunosuppressive activity of neutrophils remains to be defined, increased uptake of arachidonic acid by FATP2-expressing neutrophils drives the biosynthesis of prostaglandin E2 (PGE2) and subsequent immunosuppression115 (Fig. 1c). Therefore, administration of a FATP2 inhibitor in tumour-bearing mice reduced both the immunosuppressive activity of neutrophils and tumour growth115. These results may pave the way for new strategies targeting neutrophil lipid metabolism.

Neutrophils can express ligands that activate immune checkpoints on T cells (Fig. 3a). Programmed cell death 1 ligand 1 (PDL1) was shown to be induced by the hypoxia-inducible factor 1α (HIF1α) pathway in the mouse and by inflammatory cytokines (for example, IL-6, IFNγ and GM-CSF) in humans117,118,119,120. PDL1-expressing neutrophils have been identified in human hepatocellular carcinoma and gastric carcinoma and shown to have prognostic significance120,121. Therefore, neutrophils are part of the myeloid and stromal cell network expressing PDL1 that drives immune checkpoint engagement and T cell exhaustion. However, further studies are needed to evaluate the expression and function of PDL1 in neutrophils in different cancer types.

Fig. 3: Therapeutic targeting of neutrophils.
figure 3

a | Neutrophils express a set of myeloid checkpoints, including signal regulatory protein-α (SIRPα), CD200 receptor (CD200R), leukocyte immunoglobulin-like receptor B2 (LILRB2), paired immunoglobulin-like type 2 receptor-α (PILRα) and, expressed on neutrophil precursors, programmed cell death 1 (PD1) and atypical chemokine receptor 2 (ACKR2)49,166,167,188,191,241,242. The significance of targeting SIRPα, LILRB2 and ACKR2 on neutrophils has been established in preclinical models (see main text). However, the potential antitumour effects of blocking CD200R243,244, PILRα241 and PD1 (ref.242) have not been similarly demonstrated. Neutrophils also express a set of ligands for lymphocyte checkpoints (that is, V-domain immunoglobulin suppressor of T cell activation (VISTA), PD1 ligand 1 (PDL1), CD86, 4-1BB ligand (4-1BBL) and OX40 ligand (OX40L)), representing potential targets to limit the process of neutrophil-mediated immunosuppression in cancer68,69,120,121,122. The interaction with their cognate receptors expressed by T cells (P-selectin glycoprotein ligand 1 (PSGL1), PD1, cytotoxic T lymphocyte-associated antigen 4 (CTLA4), 4-1BB and OX40, respectively), delivers a positive (+) or negative (–) signal to T cells, dependent on the specific receptor–ligand pairing. b | Inhibition of CXC-chemokine receptor 1 (CXCR1) or CXCR2 dampens the recruitment of immunosuppressive neutrophils in cancer39,40. c | Blocking the transforming growth factor-β receptor (TGFβR), interferon-β (IFNβ) signalling or antagonism of angiotensin II type 1 receptor (AGTR1) can increase the cytotoxic activity of neutrophils towards cancer cells. CXCR4 blockade increases the production of interleukin-18 (IL-18) by neutrophils and the activation of natural killer (NK) cells64,161,162,163. Lastly, the immunosuppressive effect of neutrophils can be impaired by blocking the fatty acid transport protein 2 (FATP2), which, in response to arachidonic acid, induces the synthesis of prostaglandin E2 (PGE2)115. d | Neutrophils express the immunoglobulin G (IgG) Fc receptors (Fcγ receptor (FcγR)) and the IgA Fc receptor (FcαRI) and are involved in the elimination of antibody-opsonized cancer cells through the process of antibody-dependent cellular cytotoxicity (ADCC)183,184. HLA-G, human leukocyte antigen-G; IFNAR, IFNα/β receptor 1.

Full size image

In addition to PDL1, V-domain immunoglobulin suppressor of T-cell activation (VISTA) is expressed on tumour-associated myeloid cells, including monocytes, macrophages, dendritic cells and neutrophils122,123. In a transplantation mouse model of melanoma, blockade of VISTA generated a myeloid differentiation primary response 88 (MYD88)-mediated pro-inflammatory response that resulted in the development of antitumour immunity122. VISTA inhibition enhanced the production of IL-12 by tumour-associated dendritic cells and monocytes and reversed their immunosuppressive activity on T cells122. By contrast, the immunosuppressive activity of neutrophils was not altered following VISTA inhibition, indicating that further investigation is needed to determine the role of VISTA in neutrophils and its impact on tumour immunity122. Furthermore, these data suggest that tumours with elevated levels of immunosuppressive TANs may be resistant to treatment with antibodies targeting VISTA, and approaches combining VISTA inhibitors and neutrophil depletion or reprogramming should be assessed in preclinical models.

Although the existence of important crosstalk between neutrophils and innate lymphoid cells (ILCs), in particular natural killer (NK) cells, in an inflammatory context is well established124,125, only limited findings have been reported on the bidirectional interaction between these two innate cells in the TME. Neutrophils have been shown to promote metastatic dissemination by preventing NK cell-mediated clearance of tumour cells from initial sites of dissemination126. Evidence from in vitro experiments and in vivo transplantation models of tumour cell lines has shown that ligands of CXCR1 and CXCR2 produced by tumour cells can induce the formation of NETs in the TME, which in turn coat and protect tumour cells from the cytotoxic activity of NK cells and T cells by impairing their contact with tumour cells127. In humans, G-CSF-mobilized neutrophils inhibited the activation of NK cells128. Conversely, NK cells can control the tumour-promoting and angiogenic function of neutrophils in an IFNγ-dependent manner by inhibiting VEGFA expression129. However, significant antitumour NK cell-mediated activity attributable to enhanced NK cell activation and survival has also been reported following NK cell interactions with neutrophils in haematopoietic stem cell transplantation recipient mice transplanted with a syngeneic colon cancer cell line130. Thus, the interaction of neutrophils with NK cells in the TME can be context-dependent.

Neutrophils in antitumour resistance

The results discussed above and clinical correlative evidence suggest that neutrophils are an important component of tumour-promoting inflammation and immunosuppression in numerous mouse and human tumours. In apparent contrast to these observations, neutrophils have also been shown to mediate antitumour resistance in vitro and in vivo, suggesting a dual potential for these cells.

It has long been known that substantial recruitment and activation of neutrophils can result in antitumour activity131. Neutrophils can kill tumour cells through direct contact and via the generation of ROS64,132,133 (Fig. 2b). ROS-mediated killing involves the transient receptor potential cation channel, subfamily M, member 2 (TRPM2), an H2O2-dependent channel that induces a lethal influx of calcium (Ca2+) into target cells134 (Fig. 2b). Expression of TRPM2 was upregulated in cancer cells undergoing an epithelial-to-mesenchymal transition (EMT) and this was also associated with an increase in the secretion of CXCL2, suggesting that, in addition to triggering an apoptotic cascade, TRPM2 sustains the recruitment of neutrophils134,135.

The neutrophil killing armamentarium includes expression of NO, TNF-related apoptosis inducing ligand (TRAIL) and TNF136,137. The latter induced the expression of the hepatocyte growth factor receptor (HGFR; also known as MET) on neutrophils136. Studies performed in different tumour cell line transplantation mouse models (for example, Lewis lung carcinoma and fibrosarcoma) showed that HGF present in the TME induced neutrophil recruitment and production of NO, which resulted in killing of tumour cells136. However, in a transplantation mouse model of melanoma, HGF–MET signalling in neutrophils led to an immunosuppressive phenotype associated with a limited expansion of antitumour T cells and a reduced response to adoptive T cell transfer and immune checkpoint blockade therapies138. Thus, the impact of the expression of MET on neutrophils remains to be fully elucidated in different tumour contexts and therapeutic conditions136,138.

Accumulation and activation of neutrophils in the metastatic niche can reduce the formation of metastases through the elimination of cancer cells48,49,139. In an orthotopic transplantation mouse model of breast cancer, expression of G-CSF and CCL2 by the primary tumour induced the mobilization and activation of neutrophils in the pre-metastatic lung, and consequent ROS-dependent killing of tumour cells48. Antibody-mediated neutrophil depletion did not affect the growth rate of the primary tumour or the number of CTCs but increased the metastatic burden in the lung, suggesting that neutrophils act in the metastatic niche (see also above)48. This result suggests interplay between primary tumours and neutrophils to activate their antitumour activity and control metastatic progression. In immunodeficient mice, the transplantation of human breast cancer cells with low spontaneous metastatic potential into the mammary fat pad resulted in the reprogramming of neutrophils in the pre-metastatic lung, with high cytotoxic activity associated with expression of transmembrane protein 173 (TMEM173; also known as STING)139. In this case, breast cancer cells with low spontaneous metastatic efficiency showed increased expression of CCL2 compared with breast cancer cells with high metastatic potential. Tumour-derived CCL2 induced the recruitment of IFNγ-producing CCR2+ monocytes. In turn, IFNγ upregulated TMEM173 and enhanced the cytotoxic activity of neutrophils139. These observations highlight the capacity of neutrophils to act as effector cells.

In addition to mediating direct killing, neutrophils engage in networks of T cell-dependent antitumour immunity. TANs have been shown to produce chemokines including CXCL10, CCL2, CCL3, CXCL1 and CXCL2, which recruit T cells as well as other leukocytes3,15,140. Neutrophils can acquire an APC phenotype, and in early-stage human lung cancer a population of immature CD11b+CD15hiCD10–CD16int/low neutrophils stimulated the proliferation of both CD4+ and CD8+ T cells68,69. In vitro experiments showed that in response to GM-CSF and IFNγ present in the TME, these NI acquired APC features, characterized by the expression of HLA-DR and CD86 and the capacity to amplify the antitumour T cell response68 (Fig. 2a). In addition, neutrophils isolated from human CRC biopsy specimens amplified the activation of CD8+ T cells in response to T cell receptor (TCR) triggering in vitro141.

The intestinal microbiota play a role in inflammation and colorectal carcinogenesis142,143. Neutrophils were reported to have a tumour-suppressive effect in CRC via the response to IL-1 produced in the TME by monocytes and tumour cells, which enhanced the expression of antimicrobial peptides by neutrophils and their subsequent antibacterial activities144,145 (Fig. 2c). In addition to CRC, lung carcinogenesis in a GEMM induced by Kras mutation coupled with Trp53 loss has also been associated with dysregulation of the airway microbiota, which stimulates IL-17 production by resident γδ T cells resulting in neutrophilia and tumour growth37. Therefore, neutrophils can play a role in the control of microbiota-induced tumour-promoting inflammation144,145.

In 3-MCA-induced primary sarcomagenesis, a tripartite interaction between neutrophils, macrophages and a subset of unconventional T cells (UTCs), known as CD4–CD8–TCRαβ+ double-negative UTCs (UTCαβ), was found to be essential for the establishment of effective antitumour immunity15 (Fig. 2d). During the early phase of sarcoma development, neutrophils amplified the production of IL-12 by macrophages, which in turn promoted polarization of the UTCαβ towards a type 1 immune response and IFNγ production15 (Fig. 2d). However, further investigation is needed to determine the mechanism(s) by which UTCαβ, as a subset of T cells, can act as antitumour effector cells, as well as their presence, significance and role in human tumours. Interestingly, in silico analyses suggest that this neutrophil-dependent antitumour axis is relevant in select human tumours15.

Thus, neutrophils can exert dual, seemingly opposite, functions in tumour immunity. The disease stage as well as the tumour type and tissue context are key determinants of the specific role of these cells in promoting or restraining cancer. The levels and nature of inflammatory mediators found in the context of different tumour types and at different tumour stages may dictate the phenotype of neutrophils47,66,71. The complexity of the regulatory pathways involved is underlined by the fact that the same growth factor, G-CSF, can drive the differentiation and activation of both antitumour and pro-tumour neutrophils, by stimulating their cytotoxic activity or the acquisition of immunosuppressive activity, depending on the conditions61,131,146. In an attempt to synthesize available data on the dual role of neutrophils in tumours, we surmise that at early stages of tumour development, myeloid cells are set in an antitumour state68,147,148, whereas progression to invasion and metastasis is associated with and driven by the acquisition of a pro-tumour, immunosuppressive phenotype47,66,71.

Neutrophils in human cancer

Occurrence and significance

As discussed above, increased myelopoiesis is a common feature of advanced neoplasia and neutrophil diversity has been also observed in patients with cancer, including lung cancer, head and neck cancer and melanoma11,12,38,56,63,114,149. Using mass cytometry by time-of-flight (CyTOF) on blood samples, a study identified distinct phenotypes of CD66b+ neutrophils at different stages of melanoma progression (reported in a preprint149). Notably, the abundance of the terminally differentiated NM subset, characterized by expression of CD66b+, CD10+, CD101+ and CD16+, gradually decreased during tumour progression, whereas NI, characterized by expression of CD66b+, CD117+, CD49d+ and CD79b+, increased149. Nevertheless, it is important to note that the association between neutrophil immaturity and immunosuppressive activity remains a matter of debate. Indeed, immature human CD10–CD66b+ neutrophils have been described to promote T cell activation whereas an opposite effect has been reported for the CD10+ NM150.

In peripheral blood, high neutrophil counts and a high neutrophil-to-lymphocyte ratio (NLR) are associated with bad prognosis for patients with a wide spectrum of solid tumours (for example, CRC, melanoma and breast, prostate and lung cancers)12. The prognostic significance of NLR was validated in a meta-analysis involving 100 studies with 40,559 patients and 22 solid tumours151. However, the relevance of NLR in the clinic remains to be proven152,153. For instance, in patients with metastatic breast cancer, NLR was found to be associated with the stage of the disease, involvement of the central nervous system and the presence of visceral metastases but its prognostic significance was lost in multivariate analysis153.

Neutrophils are present in variable numbers in human solid tumours, as assessed by conventional immunohistochemistry (for example, by staining for CD66b) and neutrophil transcriptional signatures13,14,15,16,17,154. In general, high TAN infiltration is associated with worse prognosis12. For instance, in a large study using the CIBERSORT (cell type identification by estimating relative subsets of known RNA transcripts) method to quantify 22 leukocyte populations in approximately 18,000 patients with 39 different tumour types, a neutrophil signature emerged as the most significant negative prognostic factor14. In early NSCLC, TANs were the most represented leukocyte population and were negatively correlated with T cell infiltration16. Thus, suppression of T cell-mediated immunity is likely one of the mechanisms underlying their adverse clinical significance. In addition, correlative analysis on hepatocellular carcinoma biopsy specimens revealed an association between the occurrence of neutrophils and angiogenesis155.

In apparent contrast to the above results, in select human tumours high levels of TANs as assessed by immunohistochemistry or neutrophil transcriptional signatures were associated with better prognosis. These include CRC13,141,156,157,158, endometrial cancer132, invasive ductal breast carcinoma132, low-grade glioma132 and undifferentiated pleomorphic sarcoma (UPS)15. In CRC, TANs co-localized with CD8+ T cells and combined infiltration of TANs and CD8+ T cells was associated with a better prognostic value compared with CD8+ T cells alone141. In UPS, but not in other sarcomas, neutrophil signatures were associated with a type I immune response and better clinical outcome15. As UPS is likely to be the human counterpart of 3-MCA-induced primary sarcomagenesis in mice, here neutrophils may engage in antitumour resistance mediated by UTCαβ15 (Fig. 2d). Thus, in select human tumours it would appear that TANs can mediate antitumour resistance by direct killing of tumours cells132,133 or by engaging in cooperative networks with innate and adaptive lymphoid cells15,141. Collectively, current data suggest that the significance of neutrophils and their functions, in the circulation and in the neoplastic setting, may be strongly influenced by the tissue and tumour context.

In response to chemotherapy, radiotherapy and immunotherapy

TAN infiltration affects response to different anticancer treatment modalities (Table 1). High neutrophil infiltration was generally reported to be associated with a worse response to chemotherapy and radiotherapy (Table 1). Notable exceptions were CRC, gastric cancer and high-grade ovarian cancer, where higher levels of TANs were associated with a better response to chemotherapy13,159,160. These discordant observations regarding the predictive value of TANs in response to cytoreductive regimens is likely a reflection of fundamental differences in the immunobiology of these cancers.

Table 1 Predictive value of neutrophils in response to therapy
Full size table

Peripheral blood neutrophilia and high NLR have been associated with a poor response to immune checkpoint inhibitors (ICIs) (Table 1). As mentioned above, neutrophils express ligands of immune checkpoints including PDL1 and VISTA120,121,122. High levels of PDL1-expressing neutrophils in the TME have been associated with a poor survival rate in patients with hepatocellular carcinoma and gastric cancer120,121. Therefore, neutrophils represent both a target for and a mechanism of resistance to ICIs.

In summary, neutrophils are an important determinant of the antitumour efficacy of established treatment modalities, ranging from chemotherapy to ICIs and monoclonal antibodies (mAbs) mediating antibody-dependent cellular cytotoxicity (ADCC) (see below). Moreover, experimental therapies including myeloid checkpoint targeting strategies, new immune checkpoint blockade immunotherapies and TGFβ inhibitors have neutrophils as one of their therapeutic targets49,64,161,162,163,164,165,166,167.

Neutrophil targeting and reprogramming

A better mechanistic understanding of the complex role of neutrophils in tumour progression will provide a basis to design therapeutic approaches. In particular, dissecting the diversity and plasticity of neutrophils in different cancer types and sites (that is, in the peripheral blood or the TME) represents a significant challenge for specifically targeting these cells and setting them in an antitumour state. As described above, chemokine receptors CXCR2 and CXCR1 expressed by neutrophils are important for their recruitment to the TME, modulation of their activation state and their circadian oscillations32,39,54. Based on results in preclinical models, inhibition of neutrophil recruitment by blocking CXCL8, CXCR1 and/or CXCR2 (refs39,40,52,168) has now entered clinical evaluation (Fig. 3b). Recent studies showed that higher levels of CXCL8 in serum and tumours of patients with advanced solid cancers (that is, 1,344 patients with melanoma, NSCLC and renal cell carcinoma169, and 1,445 patients with urothelial carcinoma and renal cell carcinoma170) were associated with increased tumour infiltration by neutrophils, shorter survival and decreased clinical response to ICIs (for example, the programmed cell death 1 (PD1) antibody nivolumab169, nivolumab plus the cytotoxic T lymphocyte-associated antigen 4 (CTLA4) antibody ipilimumab169 or the PDL1 antibody atezolizumab170). The clinical benefit of a fully human CXCL8 mAb (BMS-986253) is under evaluation in patients with advanced solid tumours (NCT03400332 (ref.171); BMS-986253 in combination with nivolumab) or with hormone-sensitive prostate cancer (NCT03689699 (ref.172); BMS-986253 in combination with nivolumab and with intermittent androgen deprivation). An initial phase I clinical trial of BMS-986253 in 15 patients with metastatic or unresectable advanced solid tumours showed that the treatment was safe and well tolerated173. CXCR2 inhibitors (that is, AZD5069, reparixin174, SX-682 and navarixin) are undergoing clinical evaluation in patients with metastatic castration-resistant prostate cancer (NCT03177187 (ref.175); AZD5069 in combination with the non-steroidal anti-androgen therapy enzalutamide), with early breast cancer (NCT01861054 (ref.176); reparixin), with metastatic breast cancer (NCT02001974 (refs177,178) and NCT02370238 (ref.179); reparixin in combination with the chemotherapy paclitaxel), with metastatic melanoma (NCT03161431 (ref.180); SX-682 in combination with the PD1 antibody pembrolizumab) and with NSCLC and CRC (NCT03473925 (ref.181); navarixin in combination with pembrolizumab). It will be important to assess whether these agents indeed affect TAN infiltration and/or the activation state given the disappointing results so far obtained with chemokine inhibitors in inflammatory conditions39,182.

Reprogramming neutrophil function in the TME presents a challenge for which different approaches have been proposed, including blocking TGFβ64 (Fig. 3c). Similarly, targeting angiotensin-converting enzyme (ACE) and the angiotensin II type 1 receptor (AGTR1), nicotinamide phosphoribosyltransferase (NAMPT) or CXCR4 in mouse models has been reported to switch neutrophils to an antitumour state161,162,163. Consistent with the role of HIF1α in setting neutrophils in a pro-tumour state, hyperoxygenation and reversion of hypoxia activated the antitumour potential of neutrophils in a GEMM of uterine cancer induced by Pten loss133. Below, we focus on two other approaches, namely ADCC and myeloid checkpoints, because of their significance to neutrophil diversity and reprogramming.

Antibody-dependent cellular cytotoxicity

Neutrophils share with monocytes, macrophages and NK cells the expression of Fcγ receptors (FcγRs) and mediate tumour cell elimination via ADCC183 (Fig. 3d). Highlighting the requirement for neutrophils in this process, depletion of neutrophils reduced the efficacy of treatment with mAbs directed against CD52 (alemtuzumab) and CD20 (rituximab) in mouse lymphomas183.

Human neutrophils also express the high-affinity receptor for immunoglobulin A (IgA), FcαRI (also known as CD89), a potent inducer of ADCC; expression of this receptor leads to increased killing of IgA-opsonized cancer cells compared with neutrophil-elicited FcγR-mediated ADCC of IgG-opsonized target cells183,184. Furthermore, chimeric IgA anti-CD20 mAb containing the human constant region domain was found to be more efficient than chimeric IgG anti-CD20 mAb in transgenic mice expressing functional human FcαRI and protected against lymphoma development185. Interestingly, IgA-elicited neutrophil-mediated ADCC can be enhanced by concomitant blocking of the CD47–signal regulatory protein-α (SIRPα) myeloid checkpoint186 (see below). In a three-dimensional collagen culture model of human breast cancer cells, endothelial cells and neutrophils, triggering of neutrophil FcαRI through treatment with an FcαRI antibody promoted the release of LTB4, a potent chemoattractant for neutrophils, IL-1β and TNF, which in turn amplify the recruitment of neutrophils via the production of CXCL8 by endothelial cells187. In effect, antitumour IgA treatment sustains the activation of neutrophils and creates an amplification loop for neutrophil recruitment187. Thus, FcαRI-mediated ADCC may represent a valuable neutrophil-centred therapeutic strategy.

Myeloid checkpoints

The function of myeloid cells is under the control of numerous negative regulators (known as checkpoints), which are expressed by neutrophils, monocytes and macrophages. These include SIRPα, CD200 receptor (CD200R), leukocyte immunoglobulin-like receptor B2 (LILRB2), paired immunoglobulin-like type 2 receptor α (PILRα) and, expressed on neutrophil precursors, PD1 and ACKR2 (Fig. 3a). Preclinical evidence suggests that neutrophils contribute to the antitumour activity of agents that block the CD47–SIRPα signalling axis and LILRB2, whereas for the other molecules presented in Fig. 3a the significance of their expression on neutrophils is currently unknown.

SIRPα is highly expressed by neutrophils, monocytes and macrophages, and acts as a phagocytosis checkpoint via its interaction with the ‘do not eat me’ signal CD47 presented on target cells188. CD47 is overexpressed on cancer cells, rendering them resistant to myeloid cells188,189. Interestingly, CD47–SIRPα checkpoint blockade increased the elimination of cancer cells, including non-Hodgkin lymphoma cells, melanoma cells and breast cancer cells, during an antibody-based treatment and potentiated the cytotoxic activity of neutrophils in vitro against breast cancer cells opsonized by trastuzumab, an anti-human epidermal receptor 2 (HER2) mAb, through a process of trogoptosis164,165. Furthermore, in transgenic mice expressing human SIRPα, the administration of a SIRPα mAb increased the elimination of tumour cells by macrophages and neutrophils when combined with antitumour mAbs (for example, anti-CD20, anti-HER2 or anti-EGFR mAbs)166. Importantly, in this case, complete antitumour activity was neutrophil-dependent166. Anti-CD47, combined with anti-CD20, was reported to have remarkable antitumour activity in patients with non-Hodgkin lymphoma190. However, the significance of neutrophils in this context and more generally in the activation and orientation of adaptive immunity downstream of blocking CD47–SIRPα remains to be defined.

LILRB2 is expressed by myeloid cells, including neutrophils, and acts as a negative regulator of cell activation191. LILRB2 binds to classical and non-classical HLA class I molecules and contains immunoreceptor tyrosine-based inhibitory receptor motifs (ITIMs) in its cytoplasmic tail. Activation of LILRB2 on neutrophils by one of its ligands, HLA-G, inhibited their phagocytic activity and production of ROS191. In a mouse model of lung cancer, LILRB2 blockade suppressed infiltration of immunosuppressive neutrophils and significantly promoted antitumour immunity when combined with anti-PDL1 (ref.167).

ACKR2 is expressed on haematopoietic precursors and is virtually absent on NM (ref.49). Genetic deletion of ACKR2 resulted in an increase in the mobilization of neutrophils endowed with antitumour properties, characterized by ROS-mediated cytotoxic activity towards cancer cells49. Thus, although targeting ACKR2 may unleash CC-chemokine-mediated lymphocyte and monocyte mobilization in the periphery, it might also release neutrophil effector function.

Evidence obtained from dissecting the function of the CD47–SIRPα signalling axis has suggested that blocking myeloid checkpoints unleashes adaptive immune responses, by eliciting T cell-dependent immunity including CD8+ cytotoxic T cells192,193,194,195,196. This indicates that targeting neutrophil checkpoints may represent a new frontier in cancer immunotherapy. However, not all myeloid checkpoints are expressed by neutrophils and, indeed, some negative regulators (for example, common lymphatic endothelial and vascular endothelial receptor 1 (CLEVER1)) are present only on macrophages197,198,199. Therefore, as these molecules enter the clinical arena as therapeutic targets, it will be important to carefully assess neutrophil numbers, diversity and function as candidate correlates of antitumour activity.

Conclusions and perspectives

The presence and significance of neutrophils in cancer have long been overlooked. More rigorous approaches to quantify their infiltration into the TME and analyse their diversity and plasticity have revealed new insights into TAN immunobiology. As a consequence, TANs have emerged as an important component of the niche of many mouse and human tumours. Current views on the opposing roles of neutrophils in cancer are based on neutrophil depletion studies using antibodies in mice or correlative analysis between neutrophil numbers found in the TME or peripheral blood and the survival of patients with cancer. However, a more systematic effort using gene-targeting approaches for neutrophil depletion and abrogation of select neutrophil functions in carcinogen-induced and GEMMs of cancer, rather than in tumour cell line transplantation mouse models, is needed to truly determine the specific roles of neutrophils in different neoplasias. Deconvoluting the diversity of TANs at the single-cell level and relating this complex information to neutrophil function as well as patient prognosis and response to therapy represent important challenges in the field.

The current nomenclature for the diversity of neutrophils and related myeloid-derived suppressor cell (MDSC) populations can be confusing for researchers both outside and within the field (Box 2). However, even imperfect nomenclatures can have some value as communication tools and hence are a heuristic process200. Therefore, we call for a consensus effort to develop a provisional nomenclature for neutrophil plasticity and diversity, along the lines of previous exercises conducted for ILCs, macrophages, IL-1 and other cell types and molecules201,202,203.

Myeloid cells at different stages of differentiation and activation represent a major pathway to immune suppression both at a systemic level and at the local TME level. Dissecting the relative importance and diversity of the monocytic versus the neutrophil differentiation pathway in different tumour contexts and integrating it with the general immunological landscape may pave the way for personalized therapeutic approaches.

TANs can be part of antitumour resistance pathways. Neutrophils express myeloid checkpoints and there is now proof of principle that targeting the negative regulator CD47 and unleashing myeloid cell function can result in clinical therapeutic benefit190. We surmise that harnessing the antitumour potential of neutrophils in those tumour types for which there is evidence of their protective role (for example, sarcomas and CRC) and in patients currently resistant to immunotherapy may represent a strategy worth pursuing to complement the established T cell-centred therapeutic armamentarium.