Neutrophils represent the most abundant cell type of leukocytes in human blood and the second most in mice
[1]. Neutrophils are named for their ability to be stained with a mixture of alkaline and acidic dyes
[2]. Mature neutrophils are differentiated from hematopoietic stem cells in the bone marrow in a process called granulopoiesis and are produced in high quantities, up to 10
11 per day in healthy individuals
[3]. They are the first line of defense against pathogens, which explains the high susceptibility of people with neutropenia to infections
[4]. Neutrophils were always considered a homogeneous population with specific functions in innate immunity, most likely due to their short life span, which limited the ability to investigate their diverse activities or even expect them. The recent observations of neutrophil heterogeneity in the steady state
[5], in different tissues
[6], and in pathology
[7][8] have dramatically altered the old paradigm of neutrophil homogeneity. The recent reconsideration of neutrophil biology was achieved thanks to advances in biotechnology, which enabled researchers to investigate cells at a single-cell resolution
[9]. In cancer, neutrophil actions are diverse and heterogeneous. Neutrophil blood levels increase during cancer progression
[3]. Neutrophilia is associated with poor prognosis in many cancer types
[10]. In addition to quantitative changes, qualitative changes in neutrophils upon cancer were observed. These changes include alterations in neutrophil morphology and function. The observation of tumor-associated neutrophils (TANs) producing neutrophil extracellular traps (NETs) was a hint of the possible role of neutrophils in the tumor microenvironment
[11]. NETs, first observed by Brinkmann et al. in 2004, are web-like structures of neutrophilic genetic material decorated with the proteins of granules
[12]. Later, NETs were shown to be involved in cancer metastasis
[13]. In addition to NETs, neutrophils, after recruitment to the tumor microenvironment, could gain an anti-tumor (N1) or a pro-tumor (N2) phenotype
[14]. Neutrophil polarization seems to be a complicated process affected by several tumor-derived factors. Besides this classification, a high percent of neutrophils in the circulation of cancer patients were shown to have a lower density (low-density neutrophils, LDNs)
[15] and to exhibit some features of immaturity and immunosuppressive function (granulocytic-myeloid-derived suppressor cells (g-MDSCs)
[16]. The recently described interactions between neutrophils and tumors prompted the scientific community to develop neutrophil-based cancer therapies. Achievements in this field are very promising and have reached the generation of chimeric antigen receptor neutrophils (CAR-neutrophils)
[17].
2. Neutrophil Heterogeneity in Cancer: N1/N2, NDN/LDN, and g-MDSC
2.1. N1 vs. N2
The story of neutrophil heterogeneity in cancer started with Fridlender et al.’s study, suggesting for the first time the N1/N2 functional classification of TANs. The researchers introduced a new classification of neutrophils, analogous to the M1/M2 macrophage classification: N1—neutrophils with pro-inflammatory properties and anti-tumor functions, and N2—neutrophils with anti-inflammatory and pro-tumor functions
[14]. Various factors influence the polarization of the neutrophil phenotype (
Figure 1,
Table 1).
In a pioneer study, using three mouse tumor models: mesothelioma AB12, hybridoma, and Kras-derived lung cancer, the ability of transforming growth factor beta (TGF-β) to play a role in neutrophil polarization was demonstrated
[14]. TGF-β inhibition with the small TGF-β type 1 receptor kinase (ALK5) inhibitor SM16 increased the levels of neutrophil chemoattractants in the tumor microenvironment, resulting in neutrophil recruitment
[14]. In all tumor models, the gene expression profiles of TANs from SM16-treated tumors revealed a significant decrease in arginase levels and a significant increase in tumor necrosis factor alpha (TNF-α) and intercellular adhesion molecule 1 (ICAM1) levels compared with TANs from SM16-untreated mice
[14]. Arginase overexpression could lead to L-arginine depletion in the tumor microenvironment, which impairs T cell function and supports tumor immune escape
[18]. Elevated levels of TNF-α and ICAM1 indicate the pro-inflammatory status of TANs from SM16-treated tumors. Functional analysis revealed enhanced cytotoxicity of TANs isolated from SM16-treated tumors against tumor cells, while TANs from untreated tumors were found to be noncytotoxic. In mesothelioma AB12 tumors of SM16-treated mice, in vivo depletion of CD8+ T cells by mAb injection canceled the reduction in tumor growth, indicating a dependence of TAN anti-tumor effects on CD8+ T cells. In SM16-untreated mice, in vivo TAN depletion with or without CD8+ T cell depletion led to a significant decrease in tumor size, indicating the pro-tumor activities of TANs
[14]. The findings of this research provide a basic understanding of the morphological and functional differences between neutrophil N1 and N2 phenotypes, which are primarily regulated by TGF-β.
Figure 1. Neutrophil heterogeneity during tumor development. In the peripheral blood of cancer patients, three distinct populations of circulating neutrophils can be found: NDNs, LDNs, and g-MDSCs. Tumors recruit neutrophils via various mediators. These mediators include G-CSF
[19], CXCL1
[20], CXCL2
[21], CXCL5
[22], CXCL8
[23], CXCL12
[24], IL-10
[19], IL-17
[25], and TGF-β
[26]. After infiltration into the tumor microenvironment, neutrophils gain an N1 or N2 phenotype under the action of IFN-β
[27] or TGF-β
[14], respectively. Neutrophils in their turn reshape the tumor microenvironment: N1 TANs secrete pro-inflammatory anti-tumor mediators
[14][28], while N2 TANs support tumor progression and angiogenesis and enhance the immunosuppressive tumor microenvironment
[24][28]. NDNs—normal-density neutrophils, LDNs—low-density neutrophils, g-MDSCs—granulocytic-myeloid-derived suppressor cells, G-CSF—granulocyte colony-stimulating factor, CXCL—C-X-C motif chemokine ligand, CCL—C-C motif chemokine ligand, IL—interleukin, TGF-β—transforming growth factor beta, IFN-β—interferon beta, TNF-α—tumor necrosis factor alpha, ROS—reactive oxygen species, VEGF—vascular endothelial growth factor, MMP9—matrix metallopeptidase 9, TME—tumor microenvironment.
Later, interferon beta (IFN-β) was identified as the orchestrator of neutrophil polarization toward the N1 phenotype in cancer patients and tumor-bearing mice
[24][27][29]. In
Ifnb1−/− mice after B16F10 melanoma implantation, enhanced tumor growth, angiogenesis, and metastasis were observed and accompanied by higher levels of TANs compared with tumors developed in
Ifnb1+/+ mice. TANs isolated from
Ifnb1−/− mice (
Ifnb1−/−-TANs) highly expressed C-X-C motif chemokine receptor 4 (CXCR4) and its regulators c-Myc and signal transducer and activator of transcription 3 (STAT3), vascular endothelial growth factor (VEGF), and matrix metallopeptidase (MMP9)
[24]. CXCR4 traffics neutrophils via a gradient of CXCL12, which was overexpressed in the tumors of
Ifnb1−/− mice compared to controls
[24][30]. MMP9 is a proteolytic enzyme that degrades the ECM and paves the way for new vessels
[31]. VEGF plays a well-known key role in angiogenesis and is an important suppressor of anti-tumor immunity in the tumor microenvironment
[32][33][34]. Altogether, high expression of CXCR4, VEGF, and MMP9 could serve as an ideal triad for successful neutrophil-induced angiogenesis. Interestingly, in vitro treatment of
Ifnb1−/−-TANs with exogenous IFN-β decreased the expression of the abovementioned genes
[24]. This study sheds light on the regulatory role of IFN-β in the acquisition of pro-angiogenic properties by neutrophils.
The absence of IFN-β was also associated with a prolonged life span of blood neutrophils and TANs
[29]. Pro-angiogenic TANs from
Ifnb1
−/− mice were shown to have a prolonged life span in tumor-bearing mice, which could be explained by lower expression of FAS, active caspase 3 and 9, and an imbalance in the expression profiles of pro-apoptotic and anti-apoptotic genes
[29]. Moreover, TANs from IFN-β–deficient mice showed a reduction in reactive oxygen species (ROS) production
[29].
Table 1. Diverse neutrophil subpopulations in cancer in comparison with mature neutrophils in healthy individuals.
Andzinski et al. clearly showed the ability of IFN-β to polarize neutrophils in anti-tumor phenotype
[27]. In tumor-bearing mice, upon IFN-β deficiency, neutrophil turnover and mobilization were faster and were combined with a higher percentage of immature neutrophils with ring-shaped nuclei in the blood
[27]. In a co-culture with tumor cells, TANs from IFN-β-deficient mice showed significantly lower cytotoxicity and TNF-α expression in comparison with TANs from wild-type mice. However, the anti-tumor cytotoxicity of TANs was recovered after adding exogenous IFN-β to the co-culture
[27]. Thus, the phenotypic switch of neutrophils could be regulated by TGF-β and type 1 IFN antagonistic signaling pathways
[60][61].
However, the fate of neutrophils to be friend or foe is probably decided by multiple factors, and not only in the tumor microenvironment but outside it. For example, Yan et al. showed that interleukin 6 (IL-6) along with granulocyte colony-stimulating factor (G-CSF) induces the neutrophil N2 phenotype in the bone marrow, a process most likely regulated by the immune suppressor cytokine IL-35
[62][63]. Moreover, it has also been suggested that neutrophils act differently depending on the stage of tumor development
[64][65]. TANs isolated from early tumors produced higher levels of NO, H
2O
2, and TNF-α and demonstrated greater cytotoxicity against tumor cells in comparison with TANs isolated from late-stage tumors
[64]. Interestingly, tumor growth was unaffected by neutrophil depletion during the early stages of tumor development. In contrast, after tumor establishment, neutrophil depletion led to a significant reduction in tumor growth, indicating a pro-tumorigenic effect of neutrophils at the late stage of tumor development
[64]. Neutrophil migratory properties also vary in different stages of tumor development
[65]. At early stages, neutrophils show enhanced migratory and metabolic potential with no immunosuppressive function. However, in later stages, neutrophils lose their elevated migratory and metabolic properties and gain an immunosuppressive phenotype
[65].
Shaul et al. deeply analyzed the N1 and N2 phenotypes of neutrophils using microarray analysis and identified different transcriptomic signatures of N1 versus N2 neutrophils
[28]. In the N1 profile, 136 genes were overexpressed and 2 genes were downregulated with a fold change of ≥10
[28]. N2 TANs showed a relative downregulation of genes associated with cytoskeletal organization and actin polymerization compared with bone marrow neutrophils and N1 TANs, suggesting that after neutrophil infiltration into the tumor, N2-polarized TANs lose the ability to organize the cytoskeleton and to leave the tumor microenvironment
[28]. N1 TANs showed an upregulation of many genes associated with antigen presentation, especially major histocompatibility complex type 1 (MHC-I)-related loci. Moreover, many integrins and membrane receptors associated with neutrophil immune responses are overexpressed in N1 compared with N2 TANs. For example, IFN-γ receptor 1 is expressed in bone marrow naive neutrophils and N1 TANs but is significantly downregulated in N2 TANs, which may result in a loss of communication between neutrophils and IFN-γ-releasing cytotoxic T cells
[28]. N1 TANs have pro-inflammatory properties with higher expression levels of the pro-inflammatory cytokines IL-12 and TNF-α as well as various chemokines that attract T cells and macrophages—C-X-C motif chemokine ligand 10 (CXCL10) and C-C motif chemokine ligands 2, 3, and 7 (CCL2, CCL3, and CCL7)
[28]. CCL17, which recruits Tregs, is downregulated in N1 TANs compared to N2 TANs, another mechanism of the immunosuppressive function of N2 TANs
[28].
Ohms et al. first polarized human neutrophils in vitro
[45]. A cocktail containing lipopolysaccharide (LPS), IFN-γ, and IFN-β was used to polarize neutrophils toward an N1-like phenotype, while L-lactate, adenosine, TGF-β, IL10, prostaglandin E2 (PGE2), and G-CSF together were used to polarize neutrophils toward an N2-like phenotype
[45]. Since neutrophils have a short life span and spontaneously undergo apoptosis, pan-caspase inhibitor was added during the polarization process
[45]. The cytokine profile and functional features of in vitro-polarized neutrophils correspond to those of in vivo-polarized ones, allowing the investigation of deeply different phenotypes of neutrophils in vitro
[45]. Lovászi et al. applied the protocol provided by Ohms et al.
[45] to investigate the role of the neutrophilic A2A adenosine receptor (A2AAR) in neutrophil polarization
[66]. A2AAR-specific agonist CGS21680 was added to the N1 polarization cocktail, and A2AAR-selective antagonist ZM241385 was added to neutrophils before adding the N2 polarization cocktail. The activation of A2AAR skewed N1 neutrophils to the N2 phenotype, while blocking A2AAR suppressed N2 polarization, which indicates the crucial role of the adenosine–A2AAR axis in N2 neutrophil polarization
[66]. The discovery of the pro- and anti-inflammatory profiles of N1 and N2 neutrophils, respectively, has led to a wide investigation of these two phenotypes in several physiological and pathological conditions, including inflammatory diseases
[67][68], bone regeneration
[69], ischemia
[70], myocardial infraction
[71], and Alzheimer’s disease
[72]. Of note, N1/N2 neutrophil classification in terms of infection could differ from N1/N2 TANs described in terms of tumor, which should be considered when moving from one research field to another. However, LPS-stimulated neutrophils showed a phenotype similar to that of anti-tumor N1 neutrophils, which may indicate a relationship between the pro-inflammatory and anti-tumor functions of neutrophils
[73].
2.2. NDN vs. LDN
In differential density centrifugation, the main proportion of neutrophils is purified in a high-density layer and called high-density neutrophils (HDNs). However, a significant proportion of neutrophils were found to co-purify with the low-density mononuclear cell layer and are called low-density neutrophils (LDNs)
[15] (
Figure 1,
Table 1). This heterogeneity in neutrophil density was described in 1983
[74]. To avoid confusion, since the term HDN does not refer to a specific neutrophil subpopulation except neutrophils with unaltered normal density, normal-density neutrophils (NDNs) seems to be a more suitable term
[75]. It should be noted that TANs can come from both NDNs and LDNs
[42], but because LDNs are more likely to have a pro-tumor phenotype
[76], researchers hypothesized that N1 TANs come from the NDN fraction and N2 TANs come from the LDN fraction after entering the tumor microenvironment from the bloodstream.
The elevated levels of LDNs in the blood of cancer patients and tumor-bearing mice resulted in the study of their functions and the molecular pathways involved in their elevation during cancer development
[15][47][77][78]. Interestingly, TGF-β was also involved in NDN to LDN switching
[15]. Guglietta et al. showed that NETosis-induced blood clots could also switch NDN to LDN and suggested, based on gene expression profiling, that LDNs have an intermediate profile between an NDN and N2
[79]. In comparison to NDNs, LDNs from cancer patients overexpress CD66b, CD11b, and CD15
[15][80]. Shaul et al. performed cytometry by time-of-flight (CyTOF) analysis of NDNs and LDNs from healthy individuals and patients with lung cancer. Their data showed significant differences in the expression of CD10, CXCR4, CD94, and programmed death-ligand 1 (PD-L1) between NDNs and LDNs. In both healthy individuals and cancer patients, two populations of NDNs were identified: CD66b
high/CD10
high/CXCR4
med/PDL1
low and CD66b
high/CD10
med/CXCR4
med/low/PDL1
low neutrophils. Heterogeneous subsets in the LDN fraction from cancer patients were demonstrated and a unique subset defined by CD66
high/CD10
low/CXCR4
high/PDL-1
high/med was identified
[78].
In patients with pancreatic ductal adenocarcinoma (PDAC), increased levels of circulating LDNs, which included cycling and non-cycling precursors, immature as well as mature neutrophils were observed
[5]. The LDN fraction, isolated from the peripheral blood of stem cell donors receiving recombinant G-CSF, is composed of both immature (CD10−) and mature (CD10+) neutrophils
[81]. Valadez-Cosmes et al. performed a high-dimensional screening of human cell surface markers and identified various markers that are overexpressed in LDNs which allowed them to discriminate between LDN and NDN subpopulations in cancer patients
[47]. In the LDN subpopulation, the highest fold change was found in the CD36, CD41, CD61, and CD226 markers
[47]. Functional analysis revealed impaired phagocytic activity, impaired ROS production, and the absence of anti-tumor activity in the LDN mature fraction, which corresponds to the results published by Marini et al. where mature (CD10+) LDNs inhibited T cell functions whereas immature (CD10−) LDNs enhanced them
[15][81]. Furthermore, compared with NDNs, LDNs express higher levels of PD-L1 and can inhibit cytotoxic T cells and natural killer (NK) cells
[49][82]. In a recent study, Arasanz et al. showed a possible role of circulating LDNs in the development of resistance to PD-1/PDL1 immunotherapy in non-small-cell lung cancer (NSCLC) patients
[83]. In breast cancer patients, LDN levels were associated with a worse prognosis and were significantly higher in the case of metastatic cancer than in non-metastatic cases
[77]. Similar results were observed in breast-cancer-bearing mice, where LDNs were involved in promoting liver metastasis
[48]. In addition to studying the role of LDNs in cancer development, LDNs are actively investigated in inflammatory diseases
[84], infections
[85][86], and autoimmune diseases
[87].
2.3. g-MDSCs
In the field of neutrophil heterogeneity, researchers should mention myeloid-derived suppressor cells (MDSCs) (
Figure 1,
Table 1). MDSCs are a population of immature myeloid cells derived from the granulocytic (g-MDSCs) or monocytic (m-MDSCs) lineages with a remarkable ability to suppress T cells
[57]. MDSCs have been shown to accumulate in cancer patients and tumor-bearing mice and have also been observed under different conditions, including infection, chemotherapy, experimentally induced autoimmunity, and stress
[88]. The similarity in the morphology and phenotype of g-MDSCs and mature neutrophils makes it difficult to distinguish between these populations
[89]. Que et al. summarized the studies in which g-MDSCs were believed to be a neutrophil subset or a distinct population
[90]. The researchers described a TAN as a “similar entity” to a g-MDSC, which is a suitable description in this context
[90].