Involvement of eosinophils in the anti-tumor response

Abstract

Eosinophils have long been associated with allergy and parasitic infections. Today, they are considered as multifunctional leukocytes, which participate both in innate and adaptive immune response though the expression of various receptors and mediators. Although the tumor-associated eosinophilia is observed for a long time in many hematological and solid malignancies, with a generally good prognosis value, there is a lack of knowledge on the different mechanisms involved in this phenomenon. Moreover, the recent discovery in human eosinophils of different receptors and mediators, shared with lymphocytes and involved in anti-tumor defense, suggests that eosinophils can play a role in anti-tumoral immunity. We review in the present paper the current knowledge on epidemiology, recruitment, and mechanisms involved in the response of eosinophils toward tumors.

Introduction

Anti-tumor immunity, and particularly immune surveillance, implies both innate and adaptive immune responses from the peri-tumoral tissue microenvironment. These responses involve different cell types that can recognize stress ligands or antigens expressed by transformed cells. Although the majority of publications have focused on lymphocytes, natural killer cells, and dendritic cells, new cell types have emerged that appear to express tumoricidal activity, including eosinophils. While eosinophils are traditionally referred to as effector cells in allergic diseases and parasitic infections, their cytotoxic potential toward tumor cells has been reported in experimental models and in humans. Eosinophils, which express a specific arsenal of cytotoxic molecules, are observed in the peri-tumoral infiltrate of several types of cancers, including hematological malignancies and solid tumors [1]. This infiltration is called TATE for tumor-associated tissue eosinophilia, a term first used in 1981 [2]. This phenomenon has been observed in several types of cancers and is linked with a generally good prognostic value. The present paper summarizes epidemiological, in vivo, and in vitro evidences that suggest the involvement of eosinophils in the anti-tumor response.

Biology of human eosinophils

Eosinophils are produced in the bone marrow. They arise from a CD34-expressing myeloid progenitor under the influence of several transcription factors (GATA-1, PU.1, and CCAAT/enhancer-binding protein) and cytokines (granulocyte macrophage colony-stimulating factor, interleukin [IL]-3, and IL-5). After completing a transitional shift in the blood, eosinophils are mainly found in tissues and particularly in the mucosal tissues where they form an interface with the environment (for example, in respiratory, gastrointestinal, or urogenital tracts). Long regarded as end-effectors in allergic diseases and parasitic infections involving helminths, eosinophils are now considered multifunctional leukocytes. They participate in the initiation and propagation of inflammatory responses, regulation of the immune response, and tissue homeostasis [3] through the production and release of a large panel of cytokines and mediators, as well as the expression of surface receptors involved in innate and adaptive responses [4–8] (Fig. 1).

Fig. 1.

Schematic representation of main receptors, present on eosinophils, implicated in innate and/or adaptive immunity. Eosinophils express receptors of innate immunity such as TLR (−1, −2, −4, −5, −6, −7, −9), PAR (−1, −2), receptors for lipid mediators (leukotrienes, platelet-activating factor, prostaglandin), anaphylatoxin receptors, and chemokine receptors (CCR1, CCR2, CCR3, CXCR3). Receptors of adaptive immunity are Fc receptors (IgE, IgA, and IgG), MHC I and II, costimulatory molecules (CD86, CD40, CD40L, CD28), cytokine receptors as well as the TCRγδ/CD3 complex, a link between innate and adaptive immunity. TLR Toll-like receptors, MHC major histocompatibility complex, PAR protease activated receptors

As summarized in reviews [3, 8–10], eosinophils release an unusual arsenal of cytotoxic molecules depending on the type of stimulus. First, their granules contain highly cytotoxic preformed proteins: major basic protein (MBP), eosinophil cationic protein (ECP), eosinophil peroxidase (EPO), and eosinophil-derived neurotoxin (EDN). These basic proteins, which are selectively released upon activation, exert both cytotoxic properties and immunoregulatory functions. Human MBP is cytotoxic against helminths, some bacteria, tumor cells, and other mammalian cells, as it disrupts the integrity of lipid bilayers [11]. EPO, a member of the peroxidase family, kills a variety of microorganisms, parasites, and tumor cells in the presence of H₂O₂ and halide. ECP and EDN express RNase activity, giving them anti-viral and neurotoxic properties, in addition to their anti-bacterial and anti-helminthic activities. ECP reflects tissue cytotoxicity mediated by eosinophils. In fact, high levels of ECP are often observed during allergic asthma and allergic rhinitis. Their granules also contain Charcot–Leyden crystal protein, or galactin-10, which represents 7–10 % of total protein of the eosinophil. Eosinophils are also able to produce other inflammatory mediators, including newly formed reactive oxygen species (ROS) and lipid mediators (e.g., leukotrienes, prostaglandins, and platelet-activating factor).

Eosinophils also have a role in the regulation of the immune response, through antigen presentation to T cells and the production and release of immunomodulatory molecules. They can internalize, process, and present antigenic peptides within the context of surface-expressed major histocompatibility complex II [12]. In addition, eosinophils have the capacity to provide costimulatory signals to T cells through the surface expression of molecules such as CD80, CD86, and CD40, and the ability to physically interact with CD4⁺ T cells [13]. Cytokines released by eosinophils may influence either tissue cells or immune cells [10]. Indeed, they produce profibrotic and angiogenic factors such as vascular endothelial growth factor, matrix metalloproteinases, transforming growth factors (TGF) alpha and beta, and nerve growth factor (NGF), which are involved in physiological and pathological tissue remodeling, as in asthma or nasal polyps. They also release a range of cytokines that may allow them to favor a Th1 response (IL-2, IL-12, interferon gamma) or, more frequently, a Th2 response (IL-4, IL-5, IL-9, IL-10, IL-13, and IL-25). In addition, they secrete a variety of chemokines, including RANTES (CCL5), eotaxin (CCL11), monocyte chemotactic protein 1 (CCL2), and macrophage inflammatory protein 1 alpha (MIP-1α, also known as CCL3) [8, 9]. EDN located in eosinophil granules has also been described as an immunoregulatory mediator that acts through the activation of Toll-like receptor 2 [14].

The cytotoxic potential of eosinophils may be beneficial against foreign targets or tumor cells, or harmful if directed against normal cells or tissues. Therefore, it would be reasonable to view the eosinophil as a multifaceted leukocyte that contributes to various physiological and pathological processes depending on their location and activation status [15].

Anatomopathological consequences of TATE

Although the exact links between tissue eosinophilia and patients’ outcomes remain to be established, many studies have attempted to assess the prognostic value of TATE in relation to the potential anti-tumor activity of eosinophils. Increased numbers of eosinophils have been associated with either good or poor prognosis, depending on the tumor type and stage of progression. However, conflicting results have been obtained because the criteria used to evaluate TATE vary widely among studies, and there is a lack of powerful statistical studies.

Pretlow et al. [16] studied the influence of eosinophil infiltration on the prognosis and development of metastases in patients with colon carcinoma. Among 67 patients, those associated with strong eosinophil infiltration exhibited a significantly reduced incidence of metastasis (23.5 % vs. 62.0 %), and patients with numerous eosinophils in peri-tumoral stromal tissues have better prognosis. Eighteen months after tumor resection, in patients without metastases, all patients with more eosinophils were still alive, in contrast to 73.7 % of patients with fewer eosinophils [16]. Although these results are interesting, they are preliminary in nature, as revealed by the small number of patients (n = 24) and short follow-up period. Nevertheless, these data were confirmed in another study that included 126 patients. After 5 years of follow-up, patients with high eosinophil counts in tumor tissue exhibited significantly better prognosis than those with low counts [17]. In addition, this beneficial influence of eosinophils appears independent of other usual prognostic factors (e.g., stage, age of the patients, histological grading, vascularization, vascular invasion, and neural invasion). The same conclusions were made in oral squamous cell carcinoma (SCC). Indeed, out of 125 patients with oral SCC, the presence of significant TATE is associated with a favorable prognosis and the role of tissue eosinophilia is independent of other prognostic factors such as age, sex, alcohol or tobacco consumption, tumor site, clinical stage, and vascular embolization [18]. The good prognostic value of tissue eosinophils has also been described in esophageal SCC by Ishibashi et al., who observed a significantly higher number of tissue eosinophils in cases without metastasis or clinical recurrence [19]. In nasopharyngeal carcinoma, the survival rate is better when tissue eosinophil infiltration is important, although the difference in survival rate is not statistically significant [20]. Interestingly, this difference becomes significant in the subgroup of patients with a poor prognosis, defined by the expression of epidermal growth factor receptor on tumor cells [20]. Regarding penile cancer, the presence of TATE has no influence on survival in patients with cancer at stages I and II (TNM classification). However, in advanced stages (III and IV), survival tends to be better in the stromal eosinophil-positive group than in the eosinophil-negative group (60 % vs. 0 % at 5 years) [21]. An increase in tissue eosinophils has also been associated with good prognosis in laryngeal carcinoma, pulmonary adenocarcinoma, and bladder carcinoma [22]. Finally, although eosinophil infiltrates have been detected in tissues from the large majority of tumors, none have been observed in prostate cancer. However, eosinophil crystalloids have been detected at prostate tissue sites, and their presence has been inversely correlated with Gleason grade [23].

All of these data show that TATE is rather associated with a beneficial anti-tumor response, particularly in solid tumors. In contrast, TATE appears to be associated with poor prognosis in Hodgkin’s lymphoma (HL), a disease in which inflammatory cells infiltrate the tumor, while the tumor cells, the Reed–Sternberg cells, constitute only a small percentage of the total tumor mass. Although various studies have investigated the relationship between TATE and HL, the most complete approach has been published by von Wasielewski et al. [24] and concerns 1,511 HL biopsies. Tissue eosinophilia was observed in 38 % of cases, which differed among HL histology types: 0 % in lymphocyte predominant, 14 % in lymphocyte-rich classical, 40–55 % in nodular sclerosis, 43 % in mixed cellularity, and 54 % in lymphocyte depleted. In a multivariate analysis, this study showed that TATE is the strongest unfavorable prognostic factor for survival in nodular sclerosis HL. To the contrary, no significant effect of eosinophilia on survival could be demonstrated in the mixed cellularity type [24]. Eosinophils secrete CD30-ligand [25]. The linkage of CD30-ligand to the CD30, present on Reed–Sternberg cells, is known to induce anti-apoptotic and proliferation signals [25]. But, the different clinical significance of eosinophilia between the histopathologic categories is not completely understood.

Recruitment of eosinophils at tumor sites

The mechanisms that control the recruitment of eosinophils (which are predominantly present in tissues) to tumor sites have not been clearly established. Because eosinophils express many chemokine receptors on their surfaces, several chemotactic mediators may be involved in this migration. First, some tumor cells have been identified as a source of IL-5 and/or IL-3, factors that act on the medullary differentiation and migration of eosinophils. This is the case for cancers of the thyroid gland [26], liver [27], and bladder [28]. Another chemokine potentially involved in this recruitment is eotaxin, a potent and selective chemoattractant that acts on eosinophils. In humans, one of the first studies to report a correlation between increased recruitment of eosinophils and tissue expression of eotaxin (CCL11) was performed in HL. By contrast, no significant correlation was identified between TATE and levels of interferon gamma-induced protein 10 (IP-10, also known as CXCL10), RANTES, or MIP-1α [29]. Eotaxin may not be secreted solely by tumor cells. In SCC of the oral cavity, the main source of eotaxin is eosinophils themselves [30], representing an autocrine and/or paracrine pathway for local eosinophil accumulation. It might also be envisaged that this chemokine contributes to the maintenance of tissue eosinophilia in these malignant tumors, rather than the initiation of their migration. Thielen et al. [31] investigated the relationships between different chemokines and TATE in 50 cases of peripheral T cell lymphoma. They demonstrated a significant correlation between intra-tumoral eosinophils and the expression of IL-5 and TARC (CCL17), but not between RANTES or eotaxin and TATE. Investigations of eosinophil localization at tumor sites may also help add to understanding their recruitment. Eosinophils may act as part of the innate response in anti-tumor immunity, particularly by recognizing damage-associated molecular pattern (DAMP) molecules, which are typically released after necrotic tumor death. Neoplastic tissues undergoing necrosis induce eosinophil migration in vitro and in vivo [32, 33].Cormier et al. [32] have demonstrated that the infiltration of tumors by eosinophils is an early, persistent, spatially restricted response. After subcutaneous injection of melanoma cells into mice, significant eosinophilia occurs within necrotic and capsule (fibrous acellular) regions compared with areas of viable tumor cells. Quantitative assessment of eosinophil recruitment in solid tumors has shown that tissue infiltration by eosinophils is mediated by factors (e.g., DAMPs) released directly from necrotic tumor tissues [34]. All of these data suggest that DAMPs or alarmins act in anti-tumor immunity by recruiting and activating eosinophils into the tumor. One such molecule is high-mobility group box 1 (HMGB1). This possible role of this factor is likely for several reasons. First, HMGB1 is released from necrotic but not from apoptotic tumor cells [35]. Next, Lofti et al. showed that eosinophils express the receptor for advanced glycation end products (RAGE), one of the first defined receptors for HMGB1, and that this alarmin acts as a chemoattractant for these cells and induces their degranulation [36]. Notably, IL-33 (another alarmin) may also act in this context. Indeed, serum IL-33 is increased in gastric cancer [37], and IL-33 has been shown to recruit eosinophils in vivo [38]. Taken together, chemoattractant factors influencing the migration of eosinophils in areas of tumor development appear to depend on the cancer type and stage.

Tumoricidal properties of eosinophils

Although TATE is often associated with favorable prognostic value, little is known about the exact role of eosinophils in anti-tumor responses. Eosinophils are multifunctional leukocytes involved in cytotoxicity, inflammatory processes, tissue remodeling, and the modulation of innate and adaptive immunity. Their frequent state of degranulation in close proximity to the tumor [39] suggests that their cytotoxic potential might be linked to the reduction in tumor growth. Reports of immunotherapeutic approaches and recent in vitro and in vivo studies suggest that eosinophils are involved in tumoricidal activity.

Immunotherapy

Eosinophilia is frequently observed during immunotherapy protocols, particularly with IL-2 [40, 41] or IL-4 [42, 43]. The impact of eosinophil infiltration on therapeutic efficiency remains unclear.

Interleukin-2 immunotherapy is used to treat certain types of cancers, such as melanoma or renal carcinoma. The association of the anti-tumor efficacy of systemic IL-2 administration with the presence of degranulated eosinophils within in the tumor suggests that eosinophils may play an effective role in the anti-tumor response [40, 41]. Several hypotheses have been advanced to explain this association. First, eosinophils may induce direct tumor lysis in an innate manner. Huland et al. [40] showed that eosinophils release their toxic granules on bladder cancer cells after treatment with IL-2. Alternatively, antibody-dependent mechanisms of eosinophil activation may also be involved [44], as well as the ability of eosinophils to modulate the tumor microenvironment through their immuno-regulatory properties. However, despite the anti-tumor properties of eosinophils, the prognosis of intra-tumoral eosinophilia observed in IL-2-treated patients has not been sufficiently explored thus far.

Concerning immunotherapy with IL-4, studies in mouse models have suggested a link between eosinophils and scores of therapeutic anti-tumor responses [43]. Phase 1 clinical trials have demonstrated that administering IL-4 to patients with cancer induced eosinophil degranulation in a dose-dependent manner, based on increased levels of MBP in serum and urine [42].

More recently, it has been demonstrated that IL-25 (or IL-17E) has anti-tumor potential in vivo. In fact, administration of IL-25 showed some efficacy in the treatment of human melanoma, pancreatic, lung, colon, and breast cancer xenograft models [45]. The results demonstrated that IL-17E treatment leads to eosinophilia, which is correlated with tumor inhibition [45]. However, the link between immunotherapeutic anti-tumor efficacy and eosinophilia is mainly based on correlation analysis, and no conclusions can be drawn regarding the mechanism of action of eosinophils in the modulation of tumor growth.

In vivo data

Several in vivo studies suggest a link between tumor eradication and eosinophil recruitment. In 1992, Tepper et al. [43] showed that mouse tumor cells that were engineered to produce IL-4 by transfection exhibited reduced or absent tumorigenicity when reintroduced into animals. The injection of malignant cells induced a tumor infiltrate composed predominantly of eosinophils and macrophages, and the role of eosinophils in IL-4-mediated tumor cytotoxicity was clearly evidenced in this model. Later, in a cytotoxic T lymphocyte-resistant mouse melanoma model, the clearance of lung metastases by Th2 lymphocytes was under the control of eotaxin and signal transducer and activator of transcription 6 (STAT6). The eradication of these metastases is associated with an influx of degranulated eosinophils into the tumors [46]. Although the incubation of eosinophils with B16 melanoma cells revealed no lysis of these tumor cells, the cytotoxicity of eosinophil lysates was demonstrated. Furthermore, immunohistochemical staining detected eosinophil-derived MBP in sections containing lung metastases [46]. Therefore, it appears that the tumor microenvironment may provide additional signals for eosinophil degranulation and tumor destruction. Nevertheless, one limitation of these models is the use of cytokine- or OVA-expressing malignant cells to facilitate the development of a Th2 response. Using a different model with unmanipulated tumor cells in wild-type mice, Cormier et al. [32] investigated eosinophil function in a more natural setting and showed that the infiltration of tumors by eosinophils is an early and persistent response. In another study, Simson et al. investigated the role of eosinophils in tumor immune surveillance in several genetically modified mouse models of fibrosarcoma. When they used IL-5 transgenic mice, which expressed high eosinophil counts, a significant reduction in tumor establishment and growth was demonstrated. This result was significantly correlated with a high level of eosinophil recruitment to the tumor and surrounding connective tissue. By contrast, elevated tumor incidence and reduced influx of eosinophils into tumors was observed when they used mice with lower eosinophils counts (CCL11 ⁻/⁻ mice), as well as in eosinophil-deficient IL-5/CCL11 ⁻/⁻ and ∆dbl GATA mice [47]. Taken together, these results indicate that a Th2-type response involving eosinophils is associated with tumor eradication in several animal models.

In vitro data

Such in vivo experiments indicated a role for eosinophils in anti-tumor immunity; however, the mechanisms involved remain misunderstood. Although limited, in vitro data provide some answers. Recently, the demonstration that eosinophils express receptors and mediators shared with cytotoxic T cells (which are known to be involved in the anti-tumor response) provides additional arguments favoring the potential tumoricidal role of eosinophils. Munitz et al. [7] showed that eosinophils express functional 2B4, a receptor that belongs to the CD2 subfamily of the immunoglobulin superfamily and is also expressed by T cells. The activation of eosinophils through 2B4 leads to cytotoxicity against two malignant cell lines, the mouse mastocytoma P815 and EBV-infected 721.221 B cell lines [7]. More recently, our own studies revealed the expression of another receptor shared with T cells, the γδTCR/CD3 complex [6]. The γδ T lymphocytes are centrally involved in the defense against cancer, and phosphoantigens (ligands of the γδTCR) are expressed by many tumor cells. The γδTCR/CD3-mediated activation of human eosinophils induced ROS production and the release of cytotoxic granule proteins, confirming that this receptor was functional. Human eosinophils were able to induce apoptosis and necrosis in the Colo-205 colon carcinoma cell line [5, 6]. These data, which suggest that eosinophils have tumoricidal properties, led us to further investigate the interactions of eosinophils with these tumor cells. Close contact between eosinophils and tumor cells is essential to induce cytotoxicity, as demonstrated by the role of adhesion molecules CD11a/CD18 in this cytolytic process [5] as well as by electron microscopy studies [39]. These studies provide evidence that cross talk exists between eosinophils and viable tumor cells. Mediators involved in the anti-tumor response include ECP, EDN, and tumor necrosis factor (which are produced by eosinophils), as well as granzyme A. This mediator, which is classically associated with cytotoxic T cells, was recently identified in eosinophils and acts synergistically with the cytotoxic granule ECP [5]. These in vitro data suggest that through receptors and mediators involved in tumor cell cytotoxicity, eosinophils participate in the anti-tumor response.

AllergoOncology

The ongoing debate regarding the relationship between allergy and cancer is not new and remains valid. Two general hypotheses can be considered: allergic inflammation enhances tumor immunosurveillance, making it more efficient, or alters the functioning of the immune system, promoting tumor development.

First, epidemiological studies have attempted to demonstrate an association between a history of IgE-mediated allergy and cancer. These studies have been summarized in recent reviews [48, 49]. Although the results are not entirely clear, there is some evidence to suggest the existence of possible inverse relationships. For instance, experiments conducted in IL-5 transgenic mice demonstrated that TATE is associated with tumor suppression, and not with increased vascularity or tumor progression [47]. This finding suggests that eosinophils may exert anti-tumor functions in an IL-5-rich environment (e.g., allergies). Moreover, our own studies showed that eosinophils from allergic donors induced significantly increased apoptosis of tumor cells compared with eosinophils purified from normal donors (Fig. 2) [5], suggesting efficient in vivo priming of eosinophils in allergic patients. Such heterogeneity of eosinophil-mediated tumor cytotoxicity led to the suggestion that the response to tumor development may be more efficient in allergic patients, with a potential tumor sensing role for IgE [50]. Eosinophils express both low-affinity and high-affinity receptors for IgE (FcεRI), and surface IgE is detected on eosinophils from allergic patients [51]. Therefore, an antibody-dependent cell-mediated toxicity mechanism toward tumor cells via surface IgE might be involved, as has been shown regarding parasitic targets [52]. The possible participation of IgE together with eosinophils in mechanisms of cytotoxicity directed against tumor cells requires investigation.

Fig. 2.

Cytotoxic potential of human eosinophils purified from normal donors (n = 6) or allergic patients (n = 4) toward the colon carcinoma cells Colo-205. Colo-205 cells were stained with PKH-26. Eosinophils-induced tumor cell death was performed by Annexin V staining after 0.5 and 6 h coincubation with eosinophils. Eosinophils from allergic donors displayed exacerbated cytotoxicity compared with those from normal donors. (*P < 0.05)

In addition to IgE- or IL-5-related processes, it must be noted that the existence of a gene polymorphism in ECP leads to the production of ECPArg97. The presence of this mutated ECP is associated with the occurrence of allergic symptoms [53] and HL nodular sclerosing [54].A correlation between the ECPArg97 and allergic symptoms [53] and the development of fibrosis in lymphoma [54] support an altered biological function for ECP depending on genotype. Although limited, these epidemiological, experimental, and genetic approaches suggest a relationship between allergy and cancer, which needs to be further investigated.

Summary and perspectives

Altogether, these studies indicate that eosinophils, whose function has long been restricted to an effector activity in parasitic infections or allergic diseases, likely play a role in innate defense and anti-tumor surveillance, acting synergistically with other effectors. Eosinophils are capable of integrating danger signals and responding quickly and selectively. Although the in vivo relevance of these new features attributable to eosinophils remains to be demonstrated, their tissue location, possible recruitment within many tumors, and their well-known cytotoxic potential are all arguments that support a potential role for eosinophils as an effector in the anti-tumor response. Further studies are needed especially to decipher the molecular interactions between eosinophils and tumors, in different preclinical and clinical situations. These results could lead to consider new therapeutic perspectives, including the activation of eosinophils via the stimulation of some of eosinophils receptors by their agonists.

Abbreviation

DAMP

Damage-associated molecular pattern molecule

ECP

Eosinophil cationic protein

EDN

Eosinophil-derived neurotoxin

EPO

Eosinophil peroxidase

HL

Hodgkin’s Lymphoma

HMGB1

High-mobility group box 1

IL

Interleukin

MBP

Major basic protein

RAGE

Receptor for advanced glycation end products

ROS

Reactive oxygen species

SCC

Squamous cell carcinoma

TATE

Tumor-associated tissue eosinophilia