The frequent occurrence of partial and sometimes even complete spontaneous regression of primary melanomas and squamous cell carcinomas in the skin has puzzled me ever since I started training as a dermatologist. It is generally believed that incipient cancer cells are destroyed by cells of the immune system because dense lymphocytic infiltrations can frequently be found histopathologically in these tumors. These observations fit well with the cancer immunosurveillance theory originally proposed by Burnet and Thomas in the 1950s and recently refined by Schreiber and coworkers. According to this concept, the immune system protects the host against malignant tumors by specifically detecting and eliminating incipient cancer cells. The importance of this process has been convincingly demonstrated for virally induced cancer both in mouse models and in man. Over the last 20 years, we have learned that antigen-specific lymphocytes with the ability to destroy tumor cells are not only directed against viral oncogenes but can also recognize tumor-specific or lineage-specific germline-encoded antigens. Whether the immune system can actually control tumors of non-viral origin is a controversial issue. Indeed, in the case of melanoma, it has been reported by a number of investigators that regressive changes at the primary tumor site are associated with tumor neoangiogenesis and subsequent metastatic spread. These observations seem to contradict the immunosurveillance hypothesis and touch upon fundamental questions in the field: When and how are immune cells alerted to neoplastic transformation of melanocytes? Is there a (positive or negative) correlation between regression of primary melanomas and the development of metastatic disease?
To understand the pathogenetic role of the immune system in cancer, it is helpful to remember what has recently been learned regarding the regulation of immune responses in the context of infection. Invading microorganisms are first detected by an array of germline-encoded receptors that recognize common ‘non-self’ microbial structures, which are foreign to the host (Takeuchi and Akira, 2010). The discovery of toll-like receptor 4 (TLR4) as the first of these pattern recognition receptors (PRR) that bind bacterial lipopolysaccharides as a pathogen-associated molecular pattern (PAMP) by Medzhitov and Janeway in 1997 represents a milestone in immunological research. PRRs of the membrane-bound TLR family are complemented by the cytosolic RIG-I-like receptors (RLR) and NOD-like receptors (NLR), which recognize intracellular replication of viruses and bacteria, respectively. Triggering of PRRs stimulates evolutionarily highly conserved signaling pathways resulting in broad activation of IRF- and NF-κB-driven transcriptional responses. This leads to the following: (i) the production of proinflammatory chemokines, interferons and other cytokines, which upregulate endothelial adhesion molecules and recruit immune cells to attack pathogens; (ii) the activation of dendritic cells, which carry microbial antigens to draining lymph nodes where they initiate an adaptive immune response; (iii) the clearance of infection and the removal of cellular debris; and (iv) the induction of a regenerative response that restores tissue integrity. Tissue destruction, remodeling and fibrosis may ensue if pathogens and inflammatory responses persist.
How does the immune system recognize incipient cancer cells that derive from normal host cells? If transformation occurs because of infection with an oncogenic virus, cytosolic PRRs of the innate immune system are able to detect foreign viral nucleic acid structures and initiate the recruitment and activation of dendritic cells. They take up viral antigen, migrate to lymph nodes and stimulate antigen-specific T cells, which can specifically destroy virally infected cells. If transformation occurs because of alterations in the genome, the picture is much less clear. In this case, PRRs could be stimulated by inappropriate ‘immunogenic’ cell death or by inadequate invasive growth, which are associated with the production or release of endogenous molecules representing danger-associated molecular patterns (DAMP). The DNA-binding nuclear protein HMGB1 as well as fragments of extracellular matrix proteins such as hyaluronan, fibronectin and versican are examples of DAMPs, which have been reported to activate TLRs. As an alternative to this tumor cell-extrinsic triggering of PRRs, it has been described that genetic changes that drive oncogenic transformation can also activate innate immune signalling pathways in a tumor cell-intrinsic manner. Oncogene-induced DNA damage can lead to the production of proinflammatory chemokines and cytokines, which reinforce the senescence response. The early activation of IFN-driven inflammatory responses can favor cytotoxic immune effector mechanisms (similar to those observed in viral infections) that are capable of inducing tumor cell death. In progressively growing tumors, NF-kB-driven inflammatory responses ensure tumor cell survival, proliferation and invasive growth associated with loss of immune cell control (similar to those observed in tissue repair, chronic infection or fibrosis). Thus, tumor-associated inflammatory responses turn from swords that initially promote immune cell-mediated tumor regression into ploughshares, which support regenerative proliferation, tumor growth and immune escape (Grivennikov et al., 2010).
The role of chronic inflammation in tumor development has been intensively studied experimentally using the well-characterized mouse model of multistage chemical carcinogenesis in the skin. The molecular mechanisms that link malignant transformation and inflammatory responses in this system are incompletely understood. Using knockout mice, it has been found that MyD88, an adaptor protein which connects signalling of all TLRs except TLR3 to the activation of NF-κB, plays an important role in tumorigenesis. Mittal et al. (2010) now investigated the role of different TLRs and TLR ligands in the inflammatory response, which is required for the development of skin tumors. They found that the number of skin tumors arising after epicutaneous application of the carcinogen dimethylbenzanthrene (DMBA) and the tumor promoter croton oil was significantly reduced in TLR4-deficient but not in TLR2- or TLR9-deficient mice. The expression of proinflammatory cytokines (IL1-β, TNF-α and IL-6) and the recruitment of inflammatory cells following application of croton oil were strongly decreased in the skin of TLR4-deficient mice when compared to wild-type mice. Experiments with bone marrow chimeric mice revealed that TLR4 expression on both radio-resistant and bone marrow-derived cells was important for tumor development. Furthermore, the authors provide evidence that HMGB1 released by dying cells but not bacterial lipopolysaccharides derived from skin bacteria was important for croton oil-induced inflammation. These results establish a critical role for endogenous DAMPs but not microbial PAMPs as TLR4 ligands, which activate NF-κB-driven inflammatory responses through the adaptor protein MyD88 and support tumor growth in this experimental setting.
To emphasize the general importance of their findings, Mittal et al. show that TLR4 as well as MyD88 is expressed in human colon carcinomas and melanomas. These observations are in line with reports showing that melanoma cell lines express TLR4 and respond to bacterial lipopolysaccharide or hyaluronan fragments with upregulation of cytokines and metalloproteases (Voelcker et al., 2008). The pathogenetic relevance of TLR4 expression by melanoma cells remains to be determined in future studies. It is tempting to speculate that triggering of TLR4 on melanoma cells supports proliferation and tumor growth through the activation of NF-κB-driven transcriptional responses, which are physiologically required for regenerative proliferation. This has already been demonstrated in different tumors of epithelial origin (Grivennikov et al., 2010).
Firm evidence that activation of NF-κB signalling in melanoma cells not only supports proliferation and apoptosis resistance in vitro but is indeed required for the development and progression of primary tumors in vivo has recently been obtained in a novel genetically engineered mouse melanoma model by the group of Ann Richmond (Yang et al., 2010). They generated mice that allowed the inducible genetic ablation of Ikkβ (a kinase which phosphorylates IκB leading to NF-κB nuclear translocation and transcriptional activity) specifically in melanocytes expressing the oncogenic HRasG12V mutation on the Ink4a/Arf-deficient background. Deletion of Ikkβ significantly inhibits the development of melanomas in these mice in vivo and promotes p53-dependent apoptosis and cell cycle arrest in cultured melanocytes in vitro. These results support a role for tumor cell-intrinsic activation of NF-κB signalling in the pathogenesis of melanoma in agreement with reports in genetically engineered mouse models for other tumor types (Grivennikov et al., 2010). NF-κB-driven inflammatory responses may also participate in the development of melanoma following burning doses of UV irradiation. This interesting question can now be experimentally addressed using UV-sensitive mouse models.
What is the impact of the NF-κB-driven tumor-associated inflammatory respon-ses on T cell-mediated immunosurveillance? This question was addressed by Soudja et al. in another genetically engineered mouse melanoma model, in which expression of oncogenic HRasG12V and simultaneous deletion of Ink4a/Arf in melanocytes can be induced by tamoxifen treatment (Soudja et al., 2010). Two types of melanoma can be induced in these ‘TiRP’ mice: slowly growing, pigmented melanomas and more rapidly growing, non-pigmented melanomas that are infiltrated by Gr1+/CD11b+ inflammatory immune cells. TiRP mice express the T cell-defined tumor antigen P1A in addition to oncogenic HRasG12V in melanocytes. This enables targeting of melanoma cells with adoptively transferred T cells that specifically recognize a P1A-derived peptide epitope bound by H2-Ld molecules on the cell surface. Using this experimental system, the authors show that an initially protective adaptive immune response is subsequently suppressed by chronic tumor-induced inflammation associated with infiltrating bone marrow-derived immature myeloid cells and a systemic Th2/Th17-oriented cytokine production profile. Experiments using ‘TiRP’ mice lacking T and B lymphocytes ruled out adaptive immunity as a basis for the tumor-associated inflammation in this model.
Can chronic NF-κB-driven inflammatory responses in progressively growing tumors be therapeutically converted back into early IFN-driven inflammatory responses to promote tumor regression? In other words, can the ploughshares of regenerative inflammation be converted into the swords of cytotoxic immunity? Pharmacological inhibitors of NF-κB signalling may represent one possible approach (Yang et al., 2007). NF-κB inhibition could promote apoptosis in tumor cells, inhibit angiogenesis and also functionally modulate immune cells in the tumor microenvironment. This could be combined with triggering of cytosolic PRRs directly in tumor cells, which imitates an antiviral immune defence program leading to proliferation arrest, autophagy and apoptosis (Tormo et al., 2009). Further investigations in clinically relevant genetically engineered mouse models are ideally suited to transfer pathogenetic insights generated in basic research into novel and innovative therapeutic concepts for the (immuno-) therapy of patients with melanoma in the future.