Dan Ilkovitch, Department of Microbiology and Immunology, Miller School of Medicine, University of Miami, PO Box 016960 (R-138), Miami, FL 33101, USA, Tel.: 305 243 6632, Fax: 305 243 4409, e-mail: email@example.com
Abstract: Melanomas, while the less common of skin cancers, are highly aggressive and once they metastasize usually indicate a poor prognosis. Melanomas are in many cases immunogenic and thus have been a prime target for immunotherapy, which has resulted in objective responses in some patients. To understand why antitumor immunity fails, and for the purpose of discovering new targets to improve therapy, there has been great interest to analyse the antitumor immune responses which exist in these patients, and uncover mechanisms which block tumor-specific immune responses. It is now evident that immunosuppressive cell networks and factors play a major role in the failure of the antitumor immune responses and therapies to eradicate the tumor. In this review, the factors produced by melanomas which can modulate and enhance these suppressive mechanisms are discussed. The roles of immature dendritic cells, neutrophils, T-regulatory cells, myeloid-derived suppressor cells and M2 macrophages or tumor-associated macrophages are described. Furthermore, taking into consideration of the cross-talk which exists among these different cell types and the cycle of immunosuppression which is evident in melanoma cancer patients and animal models, will be important for future therapeutic approaches.
Skin cancer is one of the most common malignancies and the incidence is rapidly rising. While non-melanoma skin cancers are more common (basal and squamous cell carcinomas), the mortality rate is three times higher for melanoma skin cancers. The immune system plays an important role in preventing tumor development, and it is widely accepted that tumor cells induce and are shaped by host antitumor immune responses. However, the type of immune cell infiltration in tumors coupled with the issue of whether they promote or block tumor progression is controversial, and is largely dependent on tumor type or model. Nevertheless, the specific assessment of immune cell infiltration of a given tumor has been shown to serve as a better predictor of patient survival than histopathological staging (1). Tumor cells have several means of evading this specific biological attack, one of which is by the production of immune-modulating factors such as cytokines and chemokines. Cytokines are produced by various cells for the purpose of communicating or signaling changes in the environment. These signals can either induce or block responses by the immune system, such as induction of inflammation during infection or wound repair, or blocking immune responses in oral tolerance and allergy prevention. Tumors can utilize these same signalling pathways to create a chronic inflammatory environment which exhausts the immune response or block responses altogether by producing suppressive cytokines. Chemokines also serve to signal and communicate, but have the added feature in that they can attract immune cells to a specific site. Tumors use these molecules to recruit cells which promote processes important for tumor progression, including angiogenesis, matrix remodelling and promoting a cycle of immunosuppression. Furthermore, tumors use these molecules to stimulate their own migration to metastasize to other organs.
Many of these factors are also being taken advantage of in immunotherapeutic protocols for the purpose of modifying and influencing antitumor immune responses. Melanoma has been a prime target of immunotherapy as antitumor immune responses already exist within the host, one example being the observation of specific tumor-infiltrating lymphocytes (TIL) (2). The goal is to either induce a more potent response or break suppression associated with the existing unsuccessful antitumor immune response. TILs, either expanded from the tumor or genetically engineered to be specific to the tumor, have had some success in treating patients with advanced melanoma disease when other modalities have failed (2). However, these therapies in patients with metastatic melanoma or animal tumor models have been found to be more successful if they follow pretreatment with lymphodepletion chemotherapy (3–5). This is partly because of removal of immunosuppressive cell networks found in tumor-bearing hosts. Melanoma patients have been shown to have increased levels of circulating immunosuppressive dendritic cells (DC) and regulatory T cells (6). These cells have been shown to prevent tumor immunity (7), and their depletion has proven to be effective for adoptive cell therapy (8). Mature DCs normally present antigens to T cells and serve as a bridge between innate and adaptive immunity. However, immature DCs have been shown to lead to T-cell anergy (9), regulatory T-cell induction (10) and production of immunosuppressive cytokines such as IL-10 (11). T-regulatory (Treg) cells are identified as being CD4+ CD25+ T cells that also express the transcription factor Foxp3. There are several subsets of Treg cells, and they regulate immune responses via various mechanisms such as production of the immunosuppressive cytokines IL-10 and transforming growth factor (TGF)-β (12,13), generation of suppressive DC (14), and induction of T-cell anergy (15). Depletion of Treg cells seems to be effective in immunotherapy of both murine and human melanoma (16,17). Another cell population involved in immune suppression is the immature myeloid-derived suppressor cells (MDSC), which accumulate in both cancer models and melanoma patients (18). MDSC, have been phenotypically identified as GR-1+CD11b+, but F4/80 negative which is a marker of macrophage maturity (19). MDSC accumulation has been associated with inflammation in the tumor micro-environment and several tumor-derived cytokines have been shown to influence their phenotypes and suppressive pathways (20–22). These cells have been shown to suppress T cells via several mechanisms such as arginine depletion, nitric oxide production, radical oxide production, CD3ζ down-regulation and suppressive cytokine secretion (23). Besides suppression of T cells, MDSC have been shown to play a role in repressing innate immune cells such as natural killer (NK) cells (24). These immature myeloid cells have also been shown to induce or convert macrophage differentiation from M1 type to M2, normally associated with immunosuppression (25). Tumor-associated macrophages (TAMs) are often described to display the characteristics of M2 suppressive macrophages. TAMs also suppress immunity via several mechanisms, such as cytokine production, poor antigen presentation and production of enzymes which lead to tissue remodelling and angiogenesis, ultimately leading to inflammation and the recruitment of other immune cells (26). In this review, we focus on melanoma-derived factors which may influence and enhance these immunosuppressive networks, the complex interactions of which are illustrated in Fig. 1.
Granulocyte-macrophage colony-stimulating factor (GM-CSF) is released by various immune cells and induces the generation of macrophages and granulocytes from bone marrow (BM) progenitors. This cytokine is important in the differentiation stages of these cells, especially critical for DCs (27). GM-CSF has also been shown to enhance or prolong the activation of neutrophils (28). GM-CSF has been used clinically to mobilize peripheral blood progenitor cells and has been used in tumor cell-based vaccines to enhance recruitment and antigen presentation by antigen-presenting cells (29,30). It is a potent immunostimulatory factor, which can induce production of various cytokines and thus enhance antigen presentation and activation of immunity. In melanoma patients, the local administration of GM-CSF has been shown to increase the number of specific T cells and increase the number and activation state of DCs in sentinel lymph nodes (31,32). While the immune stimulatory capacity of GM-CSF is well documented, there is accumulating evidence that GM-CSF can also mediate immunosuppressive networks. High levels of this factor mediating immunosuppression have been reported to prevent spontaneous abortions in reproduction models (33). Furthermore, many tumors have been shown to secrete GM-CSF at high levels, including both human and mouse melanomas (34–37). However, the correlation of GM-CSF and melanoma disease is controversial with some studies providing evidence that GM-CSF is only expressed in low-grade melanomas (38), while others showing GM-CSF in metastatic melanoma to correlate with a shorter recurrence-free period (39). In our laboratory, GM-CSF produced by murine tumors induces IL-6 by immune cells (40), and others have shown that GM-CSF produced by human melanoma cells induces tumor necrosis factor (TNF) and IL-6 secretion from monocytes (41). These cytokines induce a chronic inflammatory setting, which promotes tumor progression. We have reported that tumor-derived GM-CSF can have deleterious effects on tumor immunity, affecting leucocyte binding, blocking macrophage cytotoxicity of tumors, reducing mitogenic responses and leading to the accumulation of haematopoietic changes seen in tumor-bearing mice (42–44). We have also reported previously that GM-CSF is involved in the induction and expansion of Mac1+ Mac2+ splenic myeloid cells, which exert suppressive functions on the immune system and are now known as MDSC (45,46). Others have also provided evidence that tumor-derived GM-CSF elicits inhibitory MDSC, which suppress antigen-specific T cells (47). We have reported previously that a GM-CSF-secreting murine tumor can affect the function of epidermal Langerhans cells by reducing their ability to present antigens and activate T cells (48). GM-CSF has also been used in a model of autoimmune thyroiditis and shown to suppress autoreactive T cells through the generation of CD4+ Treg cells (49). More important is the finding that high levels of GM-CSF in vaccination leads to MDSC induction and immunosuppression (50). In patients with metastatic melanoma who received GM-CSF-based vaccines, a subset of MDSC (CD14+ HLA-DRlo) was shown to expand and suppressed immune responses by mechanisms involving TGF-β (18). This calls for careful consideration of GM-CSF which is being produced by the tumor and also being used in vaccination protocols. So while GM-CSF is shown to stimulate immune responses when given locally or at low levels (51), certain levels can lead to immunosuppression. This not only raises caution in the use of this cytokine systemically, but also helps clarify controversies in the literature.
CCL2, also termed macrophage chemoattractant protein 1, is a member of the CC chemokine family and is highly chemotactic for monocytes, T cells, basophils and NK cells (52–54). The major producers of CCL2 include endothelial cells, fibroblasts and mononuclear cells, and the main inducers of its production are inflammatory cytokines (55). Impaired tumor growth has been observed in mouse tumor cells transfected with the CCL2 gene, perhaps because of the ability of the transfected cells to recruit inflammatory cells (56). Furthermore, migration of human T cells towards melanoma cells has been shown to be driven by CCL2 (57). However, some controversy exists as to whether CCL2 may also be aiding in the recruitment of TAMs, M2 type macrophages which are pro-tumoral and immunosuppressive. Accumulating evidence suggests that the effect of CCL2 depends on its level of secretion by tumor cells (58,59). This is evidenced in a model of human non-tumorigenic melanoma cells, in which low CCL2 secretion resulted in TAM accumulation promoting tumor formation in immunodeficient SCID mice, while high CCL2 secretion resulted in massive macrophage infiltration, probably M1 type, which destroys the tumor (60). CCL2 effects on tumor growth may be biphasic, and thus may shed light on some of the contradictory literature. Human primary and metastatic melanoma have been shown to produce CCL2, while normal skin and melanocytes do not (60,61). Furthermore, CCL2 produced by human melanoma and resulting in TAM recruitment also led to induction of greater levels of angiogenesis (62), and is correlated with human melanoma stage (63). Our lab has previously reported that GM-CSF produced by murine tumors can induce high levels of CCL2 production by T cells in the spleens and tumors of animals (64). Furthermore, treatment of normal T cells with high levels of CCL2 reduces their production of interferon (IFN)-γ, an important antitumor Th1 cytokine. Some evidence from other types of cancers also points to the negative effects of CCL2 in cervical cancer, lack of CCL2 was associated with increased relapse-free survival and smaller tumor size (65), and in breast cancer high levels of CCL2 was a significant indicator of early relapse (66). Furthermore, tumor cells themselves have been shown to have chemotactic activity to CCL2, leading to migration and metastasis (67).
Interleukin-8 or CXCL-8 is a chemokine that regulates migration of neutrophils, T cells and basophils and is activated by and regulates inflammatory processes. IL-1 and TNF-alpha, both critical inflammatory cytokines have been shown to induce expression of IL-8, which contributes to angiogenesis in tumor development (68,69). High-grade human melanoma tissues have been associated with high levels of IL-8 (38,70) and the chemokine is produced by both primary and metastatic human melanomas (37). Furthermore, elevated IL-8 in the serum is associated with advanced melanoma disease stage and correlates with poor overall survival (71). IL-8 derived from melanoma cells seems to be a very potent stimulus for neutrophils. The chemokine induces MAC-1 (CD11b), and other integrin expression important for migration on neutrophils, and in turn, human neutrophils increase the migration and adhesion of human melanoma cells (72). This sets the stage for a cycle favourable of tumor progression and metastasis. Further increasing the cycle are other tumor-derived factors such as TGF-β, IL-6 and IL-1, which induce the expression of IL-8 by metastatic melanoma cells (72,73). In addition, IL-8 being pro-angiogenic can also stimulate the production of other pro-angiogenic factors such as matrix metalloprotease 9 (MMP-9) and increase melanoma cell motility (74,75). While MMP-9 may aid the tumor in degrading the extracellular matrix, it has also been shown to induce MDSC expansion and promote a tumor stroma supportive for tumor development (76). MMP-9 is produced by both tumor cells and stromal cells and BM progenitor cell-derived MMP-9 plays an integral role in MDSC expansion (76). Thus, the high levels of serum IL-8 seen in melanoma patients may influence the immune system and antitumor immunity on many fronts.
Transforming growth factor-β is a critical player in immune system homeostasis and can have varying effects on differentiation or activation of monocytes, DCs and T cells (77). While TGF-β has been shown to be growth inhibitory for normal epithelial cells and melanocytes, resistance to growth inhibition is characteristic of melanoma cells (78). Human melanomas have been shown to produce TGF-β (37), and cell lines established from metastatic lesions are less sensitive to its antiproliferative effects (79,80). This resistance was correlated with high levels of the factor being produced by the melanoma cells themselves and may be critical for invasion and metastasis (81). Microvesicles released by melanoma and isolated from the plasma of advanced melanoma patients have been shown to skew monocyte differentiation into suppressive TGF-β-secreting cells that suppress T cell function (82). This monocyte subset was expanded in the peripheral blood of melanoma patients compared to healthy donors. These cells are similar to what others describe as MDSC, which have also been shown to induce Treg development via production of IL-10 and TGF-β (83). Others have also reported that immature DC subsets may secrete TGF-β leading to the generation of IL-10 secreting CD4+ T cells (84). In cancer patients, it has been shown that Treg mediate suppression of the immune system by production of IL-10 and TGF-β (85), in part, by inhibiting T-cell function through blocking of the secretory machinery in T helper cells (86). TGF-β has also been shown to induce IL-8 expression by metastatic melanoma cells, immune effects of which are described above (73). Thus TGF-β in the tumor micro-environment can affect every arm of the immune system.
Interleukin-1 is a pleiotropic cytokine critical in inflammatory processes and plays an important role in organizing the secretion of various cytokines, chemokines and proinflammatory molecules which amplify its actions. IL-1 consists of two members, IL-1α and IL-1β, which bind to the same receptor and have overlapping activities. However, they may have different physiological roles, as they are expressed in an organ-specific manner (87). With chronic inflammation gaining ground as an important factor promoting tumor progression (88), the understood role of IL-1 in this process is quickly expanding. In one study, IL-1 mRNA expression was found in more than half of metastatic human non-small cell lung cancers, colorectal adenocarcinomas and melanomas (89). Other studies on human melanoma cells have also confirmed high expression of the cytokine (36,37). Furthermore, IL-1 expression has been found up-regulated in metastatic lesions relative to primary tumor lesions, suggesting a possible selection of high expressers in the process of metastasis (90). It has been implicated in promoting the various stages of tumor progression, including inducing production of vascular endothelial growth factor (VEGF) and TNF for angiogenesis (91) and MMPs for invasion and metastasis (92). IL-1 receptor antagonist (IL-1Ra) is a biological antagonist of IL-1 function. Treatment of animals with IL-1Ra or expression of the protein by human melanoma cells, while not altering their in vitro growth, does down-regulate their production of pro-tumorigenic factors such as IL-8 and VEGF (89), reduces their in vivo growth in a xenograft model (93) and blocks metastasis in a murine melanoma model (94). IL-1β expression by tumor cells has been shown to cause great haematological alterations, namely, the accumulation of MDSC in the bone marrow, spleen and blood of tumor-bearing mice, which results in vast immunosuppression (95). Furthermore, IL-1Ra has been shown to attenuate the accumulation of MDSC in animals harbouring such tumors (95). Other studies have shown that inflammation induced by IL-1β expression leads to a rapid accumulation of MDSC, which are phenotypically different from MDSC accumulated in tumor-bearing animals not expressing the cytokine (96). The same group also showed that IL-1 receptor-deficient mice have delayed accumulation of MDSC and reduced tumor growth, probably because of reduced immunosuppression (20). Furthermore, IL-6 partially restored the accumulation of MDSC in that model, thus is a possible downstream mediator of IL-1. While IL-1-deficient mice display lower inflammation in the presence of a tumor, IL-1Ra-deficient mice display excessive inflammation, with greater accumulation of MDSC, and enhanced immunosuppressive activity (20). Thus, IL-1 normally stimulates an inflammatory response that recruits immune cells and enhances antigen presentation, but in the context of a tumor producing various immunosuppressive factors, IL-1 further promotes tumor development and recruits immunosuppressive cells.
Interleukin-6 is a pro-inflammatory cytokine, which can aid in tumor progression by manipulating immune cells such as monocytes in the tumor microenvironment, and act as a growth or survival factor for tumor cells (97). IL-6 expression is detected in human melanoma tumors (36,37) and is elevated in the serum of high-risk melanoma patients along with other inflammatory factors such as IL-8, GM-CSF, CCL2, TNF and VEGF (98). It is a predictive factor for survival in metastatic malignant melanoma patients (99,100). In other tumors chronic IL-6 exposure in vitro led to a change from the factor being a growth inhibitor to one that is a growth stimulator, with induction of endogenous IL-6 production by the tumor cells themselves (101). Furthermore, high levels of IL-6 are correlated with tumor burden and resistance to chemotherapy in metastatic malignant melanoma patients (102). In one study, the crossing of IL-6-deficient mice with MT-ret transgenic mice, which are predisposed to develop melanomas, significantly reduced the frequency and size of melanomas that developed (103). Cyclooxygenase (COX) inhibitors which block production of prostaglandins that mediate inflammation were partially successful at pushing advanced melanoma patients into remission, and implicated a lower IL-6 production by tumor cells in the process (104). COX inhibitors lower prostaglandins E, which leads to lower intracellular 3′,5′-cyclic adenosine monophosphate necessary for IL-6 synthesis. Our lab has also found that tumor derived factors such as GM-CSF can induce immune cells such as B cells to produce IL-6 in tumor-bearing mice (40). Furthermore, IL-6 induces neutrophil mobilization and promotes IL-8 synthesis (72), maintaining a cycle of chronic inflammation. Others have also shown that tumor exosomes released into circulation mediate the induction of IL-6 by BM cells causing a block in BM DC differentiation (105). Immature progenitor cells are poor at antigen presentation, and thus are poor at activating the immune system. Additionally, immature cells produce immune-suppressive factors. Tumor IL-6 has also been shown to play a role in skewing monocyte differentiation into TAMs, which also are poor at presenting antigens, block T cell proliferation, and suppress DC function (106). Macrophage-derived IL-6 has also been shown to induce IL-10 expression in some cancer cells promoting the cycle of immunosuppression (107).
Interleukin-10 is thought to be a cytokine which plays a role in putting the brakes on inflammation. Its actions on immune cells serve to inhibit production of pro-inflammatory cytokines and in the context of cancer leads to heightened immunosuppression. IL-10 has been detected in metastatic melanoma-derived cells, but not found in skin biopsies from healthy volunteers or healthy skin of melanoma patients (108,109). It is correlated with a significantly shorter survival in advanced melanoma patients (110,111). IL-10 is thought to have autocrine functions in stimulating tumor cell proliferation, but also blocks immune cell recognition of human melanoma cells by reducing their expression of major histocompatibility complex molecules (108,109). Tumor IL-10 has been found to induce the expression of the negative co-stimulatory molecule B7-H4 on TAMs, which upon contact with T cells blocks T-cell proliferation and cytokine production (112). TAMs also produce IL-10 and TGF-β which blocks differentiation of myeloid cells into DC, forcing them to differentiate into TAMs (97). Tregs have also been shown to induce macrophages to make IL-10 and IL-6, which leads to the induction of B7-H4 on macrophages, and such macrophages can then recruit more Tregs via production of chemokines such as CCL22 (113). IL-10 produced by human melanoma cells has been shown to convert DC into tolerogenic DC, which force melanoma-specific T cells into anergy (114). These anergic T cells can then also suppress other antigen-specific T cells (115). Human Tregs block the maturation of DC and their antigen presentation, rendering them incapable of stimulating an immune response, but also inducing them to secrete IL-10 (116). IL-10 can also mediate cross-talk between macrophages and MDSC, causing MDSC to up-regulate their production of IL-10 and down-regulating macrophage production of IL-12, a pro-inflammatory cytokine (25). Thus IL-10 can mediate immunosuppression and cross-talk between nearly every arm of the immune system, rendering it completely ineffective at mounting an antitumor immune response.
The factors described in this review play an integral role in melanoma development and set the stage for a cycle of immunosuppression. Not all the cytokines and chemokines covered are elevated in every study and produced by every melanoma (37) and may depend on various factors including staging of tumors, culture conditions and immune status of patients. Furthermore, melanomas make some factors constitutively, while others are inducible, suggesting a heterogeneous expression (117). We do know that these factors are not only expressed in culture, but are elevated within tumors as well as systemically in patients. Some studies have identified correlations between staging and the presence of several factors by in situ staining of melanomas, such as IL-6 and IL-8 with early malignancy and TGF-beta, GM-CSF and IL-1 with later more aggressive malignancy (118). While many of the factors described are pro-inflammatory, and in many cases a lymphocytic infiltrate in the tumors can be detected, some cytokines such as IL-10 are anti-inflammatory and immunosuppressive. Furthermore, the actions of the factors described can be far reaching. Systemically elevated levels can result in the described accumulation of such cells as MDSC, which affect other arms of the immune system at immune compartments and sites distant from the tumor. In addition, some of the factors play an autocrine role in promoting melanoma cell proliferation, invasion and migration (119). While many of these cytokines can induce inflammation and immune responses which target and clear tumor cells initially, in a scenario where inflammation is not resolved they exacerbate the chronic inflammatory environment and promote the activation of key transcription factors such as nuclear factor kappa B (NF-κB) through cytokine stimuli and matrix degradation products (120). Activation of NF-κB in turn leads to an amplification loop for the release of the same factors, promoting DNA damage in cancer cells, promoting up-regulation of anti-apoptotic genes, activation of various genes involved in metastasis and angiogenesis and leading to accumulation of immunosuppressive cells [A review on NF-κB and its role in cancer and inflammation by Li et al. and Naugler and Karin (121,122).].
It may thus be necessary in future therapeutic protocols to target these tumor-derived factors and it may be more effective to target master regulators than going after each factor individually. For example, targeting NF-κB may be one option, however, would be difficult as this important factor is also important for initiation of immune responses, and thus important for the induction of antitumor responses. Another such master regulator is a pro-tumorigenic transcription factor known as signal transducer and activator of transcription 3 (STAT3). STAT3 can induce the production of several of the above described factors in tumor cells, which in turn, leads to induction of STAT3 in immune cells, ultimately resulting in expression of the same suppressive molecules [A review on STAT3 and cancer by Yu et al. 123)]. STAT3 is found constitutively activated in mouse and human melanomas and contributes to regulation of all arms of the immune system (124,125). This pro-tumorigenic transcription factor plays an opposing role to the antitumor and IFN responsive element STAT1, and up-regulation of one or the other may tip the scale in the respective direction (126). Thus targeting STAT3 up-regulation or finding mechanisms to up-regulate STAT1 may be attractive strategies. In one study, targeting solely, the CCL2 factor in combination with chemotherapy showed a reduction in tumor burden compared with chemotherapy alone (127). While some studies target specific factors, others have shown that DC cell based vaccine efficacy can be enhanced by rendering DC insensitive to tumor-derived factors such as TGF-beta (128). Furthermore, treating a DC-based vaccine with a non-steroidal anti-inflammatory molecule, a COX-2 inhibitor, blocks IL-6 and IL-10 production induced by tumor-derived factors, and significantly reduces the progression of a murine melanoma (129). In summary, attempting immunotherapeutic protocols for melanoma [some of which are reviewed in (130–132)] without taking into account these melanoma-derived factors which help shape the context of therapy may prove to be discouraging. Furthermore, an analysis of melanoma tissue from different sites may be necessary to determine whether different factors are produced based on organ site or primary versus metastatic status (133–135). This may help explain why some current strategies have failed as a global approach to melanoma treatment.