Superficial spreading melanomas are the most common subtype representing between 50 and 70% of all cases (14, 23). They occur in relatively younger patients (~50 s) and present on anatomical regions such as the trunk, back, and extremities (22). SSM presents as a flat or a slightly elevated lesion with varying pigmentation (24). Histologically, SSM is marked by atypical melanocytes with nested or single cell upward migration (22). Malignant melanocytes display lateral spreading throughout the epidermis, poor circumscription, and increased melanization in the cytoplasm (14, 22).

Nodular melanomas are a fairly common subtype of melanoma (15–35%) that can present most commonly on the head and neck as a growing nodule that shows ulceration (22–24). Histologically, NMs show similarities to SSMs but differ in that they show distinct circumscription. They do not display radial growth but aggressive vertical growth evidenced by large dermal nests and sheets of atypical melanocytes (14, 22).

Lentigo maligna melanomas present almost exclusively on the sun-exposed upper extremities or head and neck of elderly people (mostly octogenarians) (22). It is relatively uncommon (5–15%), and topically can be seen as patch of discolored skin showing variegated coloring (23, 24). Lentigo maligna (Hutchinson’s freckle) is the term for the in situ melanoma phase, and a small percentage of these patients progress to invasive LMM (23). Histologically, the skin exhibits extensive solar damage resulting in an atrophic epidermis and lentiginous (back-to-back) proliferation of melanocytes, which are hyperchromatic (22). Multinucleated (starburst form) melanocyte cells and solar elastosis are also hallmarks of this type of melanoma (14).

Acral lentiginous melanomas are a fairly uncommon subtype (5–10%) and occur primarily in non-Caucasian populations such as people of African or Japanese descent (23). They present on acral sites such as palms, soles of the feet, or under the nails. On the skin they present as slow growing patches with variegated pigmentation (22). Histologically, this subtype displays single cells or nests of melanocytes along the dermal–epidermal junction, and the association of lymphocyte infiltrates can be used as a diagnostic marker for this subtype of melanomas (14, 22).

Desmoplastic melanoma is a rare form of melanoma comprising 4% of primary melanomas and defined by the histological features observed in its dermal component (22, 25). It occurs primarily on the head and neck region in elderly individuals and is associated with higher probability of recurrence but a lower incidence of metastasis (25). Histologically, it is characterized by spindle-shaped melanocytes and a desmoplastic stroma, i.e., new collagen formation, and usually appears to be amelanotic (22, 25).

Nearly 60% of melanoma cases have mutations in BRAF (v-raf murine sarcoma viral oncogene homolog B) (25, 32). Thus, a brief overview of BRAF signaling is warranted. BRAF codes for a serine/threonine protein kinase constituting part of the RAS–rapidly accelerated fibrosarcoma (RAF)–mitogen-activated protein kinase kinase (MEK)–extracellular signal-regulated kinase (ERK) [mitogen-activated protein kinase (MAPK)] pathway, which is activated by the binding of extracellular growth factors to receptor tyrosine kinases (32). This binding leads to the activation of RAS (named for Rat sarcoma) family of GTPases (proteins that bind and hydrolyze guanosine triphosphate to guanosine diphosphate, i.e., GTP to GDP), which recruit and activate RAF serine/threonine protein kinases, which in turn activate MEK resulting finally in the phosphorylation of ERK (32–35). The activation of ERK leads to downstream signaling and activation of transcription factors that mediate cell differentiation, growth, and inhibit cell death (33, 36).

BRAF is one of three mammalian RAF isoforms, and one that has the highest basal kinase activity and thus is the most common isoform mutated in human cancers that include melanoma but also hairy cell leukemia, papillary thyroid cancer and colorectal cancer (CRC) (33, 36). The missense mutation, V600E, results in a substitution from valine to glutamic acid at the 600th amino acid position and represents the majority (80%) of all BRAF activating mutations in melanoma (25, 28). Other BRAF mutations include V600K (valine–lysine) and V600R (valine–arginine). BRAF-activating mutations result in constitutively active MEK signaling leading to tumor progression. In vitro, the V600E mutation confers 500-fold higher activity in BRAF than normal and promotes the transformation of melanocytes to melanoma (37). BRAF^V600E mutations are also found in benign naevi indicating that alone, these mutations may not be sufficient for tumor progression (38). The presence of these mutations has led to the development and approval of two BRAF inhibitors (BRAFi) for melanoma treatment, namely, vemurafenib (Genentech/Plexxikon) and dabrafenib (GlaxoSmithKline) as well as, a MEK inhibitor trametinib (GSK) (33, 39).

The second most common type of driver mutations in melanomas occur in NRAS (neuroblastoma RAS viral v-ras oncogene) and are found in 15–20% of melanoma patients (28). The most common mutation in NRAS occurs at codon 61 resulting in the replacement of glutamine by lysine or arginine, thereby resulting in a constitutively active RAS (38). This leads to upregulation of both the MAPK and phosphatidylinositol 3′ kinase (PI3K) pathways and results in increased cell proliferation and invasiveness (25). NRAS mutant melanomas have increased thickness and display high rates of mitosis (25). NRAS mutations are also found in benign congenital nevi (28). NRAS and BRAF activations rarely occur in the same melanoma, albeit NRAS mutations being observed in patients with advanced BRAF tumors who had failed BRAFi therapy and which therefore may mechanistically contribute to resistance to BRAFi treatment (28). Efforts to target NRAS have focused on downstream inhibitors for the MAPK pathway and include the MEK inhibitor binimetinib, which is undergoing clinical trials (25).

Neurofibromin 1 encodes a large protein of more than 2,800 amino acids with multiple functional domains (40). It contains several functional domains with one domain bearing resemblance to the catalytic region of GTPase-activating protein. This is the most well-characterized domain of NF1 and acts as a negative regulator for RAS by converting the active RAS-GTP to the inactive RAS-GDP, thus playing the role of a tumor suppressor gene (40, 41). Germline mutations in NF1 lead to a genetic syndrome called neurofibromatosis type 1 (NF1), a relatively frequent genetic condition with an incidence of 1 in 3,000, resulting in a higher predisposition to multiple tumors arising from various cell types (40). The incidence of melanoma in patients with neurofibromatosis type 1 is very low. However, NF1 somatic mutations are found in a range of cancers, and it is the third common driver mutation in melanoma found in nearly 14% of tumors (25, 41). Mutations in NF1 are more commonly observed on skin with chronic UV exposure and in elderly patients (40). NF1 inactivating mutations were found in 48% of a cohort of wild-type BRAF and NRAS melanomas and are often associated with mutations in other RAS-related genes such as RAS p21 protein activator 2 (RASA2), PTPN11, and SPRED1 (25, 40). Recent studies have also shown that NF1 may be a unique driver mutation in DMs as NF1 loss-of-function in DM is more common than for other histogenic subtypes (25). Due to the crucial role of NF1 upstream of RAS/MAPK and PI3K/mTOR pathways, NF1 mutant tumors have been targeted with tyrosine kinase inhibitors (e.g., imatinib), MEK inhibitors (trametinib), and mTOR inhibitors (sirolimus), but to date, none of these agents have been reported in treatment of NF1 mutant melanomas (40).

Microphthalmia-associated transcription factor is a helix-loop-helix leucine zipper transcription factor required for differentiation, proliferation, and survival of melanocytes and thus, its expression is also necessary for melanoma survival (42, 43). MITF also plays an important antiapoptotic function in melanoma cells by activating the expression of genes such as BLC2A1, BCL2, and BIRC7 (43). MITF is observed to be amplified in 20% of metastatic melanomas and is associated with poor survival (25). MITF is regulated by the MAPK pathway and in particular, BRAF^V600E causes induction of MITF through the transcription factor BRN2 (N-Oct-3) (25). Alternately, increased ERK signaling can also target MITF for degradation (44). Finally, MITF is also purported to contribute to BRAFi resistance through the regulation of the BCL2A1 antiapoptotic gene (44). Although targeting of MITF directly may not be viable, the use of histone deacetylase (HDAC) inhibitors can reduce MITF expression. Hence, the HDAC inhibitor panobinostat in combination with decitabine and chemotherapy is being studied in clinical trials for metastatic melanoma treatment (25).

Phosphatase and tensin homolog (PTEN) is a commonly mutated gene in melanoma and PTEN mutations were found in 14% of all melanoma samples from the TCGA genome classification study mentioned above (25, 30). PTEN codes for a phosphatase which targets phosphatidylinositol (3,4,5)-triphosphate and thus plays a crucial role in the aforementioned PI3K–Akt pathway (45). PTEN silencing therefore results in dysregulated apoptosis, cell cycle progression and migration, contributing to tumorigenesis (25, 45). It has been observed that PTEN mutations are more frequent in metastatic melanomas as opposed to early stage primary tumors (25). The loss of PTEN also interferes with genetic stability, thus sensitizing PTEN-deficient cells to polyadenosine diphosphate ribose polymerase (PARP) inhibitors (46). Currently, there are no PARP inhibitor trials underway for the treatment of metastatic melanoma (46).

As tumors arise from a host’s own tissue, immune recognition of these cells is hindered by the fact that a majority of potentially autoimmune cells are deleted during central (thymic) and peripheral mechanisms of self-tolerance (78). However, as early as 1943, it was observed that mice could immunologically reject chemically induced tumors (79). In the late 1970s, the ability to grow CTL cultures using IL-2 allowed for screening of tumor-derived DNA libraries to characterize tumor antigens (79). In 1988, the gene coding for a murine tumor antigen (P91A) was cloned (80). Shortly afterward, the first human tumor antigen gene was identified in melanoma, namely, MAGEA1 (melanoma antigen family A, 1) and was found to be expressed in various types of tumors (81). Interestingly, the gene was not observed to be expressed in normal tissue except for trophoblastic cells and male germline cells (79). Since then, several tumor antigens have been discovered, and their underlying biology has been the subject of much study (82, 83). There are several types of tumor antigens, but they have been broadly classified into three major categories. The first category includes antigens that are caused by non-synonymous mutations, or are encoded by viral genes in tumors of viral etiology (83). These are labeled tumor-specific antigens (TSA) or “neoantigens” (83, 84). Alternately, tumor-associated antigens (TSA) are usually expressed at low levels in normal tissues but are found to be overexpressed in cancer cells such the surface receptor, human epidermal growth factor 2 (HER2 or ERBB2) in breast cancer, and other malignancies (85). Finally, cancer/testis antigens (CTA) such as the aforementioned MAGE family of proteins are expressed in several tumor types and only in normal germline cells such as trophoblasts, ovaries and the testes (82, 83). The advent of high-throughput next-generation sequencing technology has allowed for relatively low-cost detection of somatic mutations in tumor cells. There are currently several approaches being formulated to tailor individualized immunotherapies for patients on the basis of their expression of tumor neoantigens (83). Although currently personalized approaches are highly expensive, it is posited that with the continuing reduction of sequencing costs and using combinatorial treatments, it may be possible to even target tumors that are non-responsive to immunotherapy (83). Since their discovery, tumor antigens have been used for multiple purposes in cancer treatment. They have been used as diagnostic markers, cancer vaccines, and as targets for adoptive T cell therapy (82, 86, 87). In general, most tumor antigens elicit a weak immune response against cancer and have been tested clinically in combination with adjuvants or additional treatments (87). To date, cancer vaccination or adoptive transfer targeting specific tumor antigens has not shown major survival advantages in melanoma (48, 88). The three major types of tumor antigens that have been described and used in melanoma immunotherapy are discussed below. A majority of described melanoma antigens are restricted to human leukocyte antigen A2 (HLA-A2) (89).

A cardinal feature of cancer is the immunosuppressive TME (106, 107). As the disease progresses, T cells in the TME exhibit a phenotype analogous to that seen in chronic viral infection known as T cell exhaustion (108). T cell exhaustion denotes a state of hyporesponsiveness to antigen with reduced cytokine secretion and cytotoxic function (108, 109). Nevertheless, the overwhelming majority of studies in human patients have demonstrated a correlation between TIL and better disease outcomes in cancers (110, 111). An exception to this observation is that FOXP3 expression, a marker of Treg that has been shown to correlate to poor prognosis in various types of human cancer (112, 113). The term TIL was first described by Wallace Clark, who was instrumental in developing the first histological classifications for melanoma as mentioned above (114, 115). TIL have been described in primary tumors, tumor-bearing LN, and in metastases of melanoma and various other cancer types (114). The range of immune cells that infiltrate a tumor, i.e., the “immune contexture” of a tumor is heterogeneous and consists of various types of T lymphocytes, B cells, NK cells, macrophages, and DC (111, 114). In 1989, Clark published a classification of the three major patterns of lymphocyte infiltration that are commonly used today (115). The brisk pattern is indicated by interposed lymphocytes between tumor cells that may be diffusely present throughout the tumor nodule or along the advancing (basal) periphery of the nodule (114, 115). The non-brisk pattern delineates a scattered multifocal presence of lymphocytes throughout the vertical growth phase of the nodule. Finally, an absent pattern is associated with a lack of lymphocytes in the tumor, or if they are present, their lack of interaction with melanoma cells (115). In recent years, various groups have attempted to further classify TIL or propose novel grading schemes, but the Clark model remains widely accepted and highly reproducible (114). In a recently published report, it was shown that melanoma tumors with brisk TIL patterns in primary melanoma H&E tissue, even in the absence of immunohistochemistry for specific markers, was associated with increased OS in patients versus tumors with non-brisk and absent patterns (116). The importance of TIL has been used to establish a novel classification system for cancer based on an “Immunoscore,” which relies upon the quantitation of CD3 and CD8 lymphocytes with the additional marker CD45RO used to mark memory T cells. The “Immunoscore” was found to be superior to the conventional AJCC TNM system for prognosis of stage I–III colorectal cancer (CRC) (117). Similar approaches are now being tested for immunoscoring of melanoma but have not been tested in large patient cohorts (118).

An additional feature observed in cancer, and other situations of chronic inflammation is the formation of tertiary lymphoid structures (TLS—also called tertiary lymphoid organs) (119, 120). These TLS can range from loose aggregates of various immune cells to complex structures that resemble secondary lymphoid organs such as LN. They consist of T cell-rich regions containing mature DC expressing DC-LAMP (lysosomal associated membrane protein), B cells, and high endothelial venules, which play a role in immune cell extravasation and production of key chemokines (120). In 2012, Messina et al. reported that a gene expression profile consisting of 12 chemokines could accurately predict the histological presence of LN-like TLS in stage IV melanoma (primary tumors and metastases), and the TLS correlated strongly with improved overall patient survival (121). Other studies have shown that the presence of TLS is a positive prognostic indicator in melanoma and a range of other cancer types including breast carcinoma, CRC, and pancreatic cancer (120). Thus, these results suggest that lymphocyte infiltration mediates a protective immune response to cancer.

However, many tumors are not T cell inflamed, and the mechanisms underlying T cell infiltration into the tumor are poorly understood (89, 122). In the context of melanoma, a recent study compared all major classes of melanoma tumor antigens between T cell inflamed and non-T cell inflamed tumors and found that there were no differences between both groups in terms of antigen load (123). Rather it was shown that non-T cell inflamed melanomas displayed reduced gene expression associated with Batf3-dependent, CD141⁺ DC (123). Furthermore, studies have pointed to the ability of tumors to interfere with chemokines that recruit leukocytes to tumors. Finally, the abnormal tumor vasculature may express reduced adhesion molecules required for homing and directly or indirectly suppress T cells by expression of molecules such as PD-L1, PD-L2, VEGF, and TGFβ (122). Once T cells infiltrate the TME, they are acted upon by a range of immunoregulatory mechanisms that prevent complete eradication of the tumor (72). These can be tumor-specific escape mechanisms or the recruitment of suppressive immune cells. For instance, mutations in BRAF or PTEN loss are associated with increased T cell inhibition by production of IL-1 and VEGF (72). Furthermore, conserved immunoregulatory mechanisms are also at play within the TME the production of immunosuppressive mediators [TGFβ and indoleamine 2,3 dioxygenase (IDO)], and the recruitment of regulatory myeloid and lymphoid cell populations (72). Another important consideration is that although, CD8⁺ T cells are canonically considered the primary cytotoxic cells involved in tumor eradication, CD4⁺ T cells can also kill tumor cells (89). However, the precise mechanisms of CD4⁺ antitumor immunity are not well described, and the role of CD4⁺ T cell infiltration in the TME has not been explored significantly with the exception of FOXP3⁺CD4⁺ Treg (72, 89). A recently concluded meta-analysis demonstrated that FOXP3⁺ Treg infiltrates were predominantly associated with worse OS in a review of over 17 types of cancer (124). In most tumors, such as cervical, renal, breast cancers, and melanoma, FOXP3⁺ Treg infiltrates correlated with shorter OS whereas they were associated with improved survival in patients with colorectal, head and neck, and esophageal cancers (124). In recent years, several studies have described the heterogeneity in FOXP3-expressing cell populations (125). In 2016, Saito et al. showed that human CRCs could be distinguished by the extent of infiltration of two distinct FOXP3⁺CD4⁺ T cell populations (126). Type A CRCs had low frequencies (<9.8%) while Type B had comparatively higher frequencies (>9.8%) of infiltrating non-suppressive FOXP3^loCD45⁻ T cells. Infiltration by these non-suppressive T cells was correlated with the presence of intestinal bacteria, in particular Fusobacterium nucleatum within the tumor (126). Furthermore, Type B CRCs were marked by high mRNA expression of IL12A and TGFB1 compared with Type A and tumors with high expression of these mRNAs exhibited significantly longer disease-free survival versus low expressing tumors. Thus, FOXP3⁺ T cell infiltration must be considered in combination with other immune signatures while determining the immune status of a tumor. In addition to T cells, the roles of B cells in the TME are being currently explored as they have both APC and effector lymphocyte functions (127). Studies in melanoma have demonstrated that CD20⁺ infiltrating B cells are found in most tumors and higher levels of these infiltrates correlated with improved patient survival (127). Furthermore, B cells are known to produce IgG antibodies that can recognize tumor cells and within a murine model of organ transplantation have been observed to promote chronic allograft rejection through antigen presentation rather than their antibody secreting functions (127, 128). Finally, recent studies have also focused on the roles of putative regulatory B cells in the context of transplantation and autoimmunity, as these cells can produce potent immunosuppressive mediators such as IL-10 and TGFβ (129). The multiple immunoregulatory mechanisms that effect TIL are the targets of a majority of current immunotherapies. However, as the aforementioned observations indicate, there are several functionally redundant pathways that allow for immunological escape of tumors in immunocompetent individuals. Thus, to be successful, the field of immunotherapy must move toward combinatorial and multipronged approaches for tumor treatment. This involves investigation of the mechanisms of innate immune cells such as NK cells, TAM, and TIDC within the TME.

Despite their discovery over 40 years ago, the exact mechanisms underlying DC dysfunction in cancer remain poorly understood (107). In both mice and humans, DC are classified into two major subsets comprised of conventional or cDC, and plasmacytoid DC (pDC) (130). In non-steady state conditions such as cancer or autoimmune disease, inflammatory DC derived from monocytes have also been described in humans and in mice (130, 131). Despite the fact that nearly all DC subsets express the surface marker CD11c, there are unique transcription factors and surface proteins that characterize the major DC subsets in human and mice. These markers have been extensively reviewed in the literature, but further study is needed to accurately profile each subset (130, 132, 133). DC canonically present extracellular antigens on MHC class II while intracellular or self-antigens are presented on MHC class I (134). However, murine and human DC also possess the capacity to cross-present antigens of extracellular origin on MHC class I to activate CD8⁺ CTL (135, 136). In humans, the primary cross-presenting DC subset is characterized by CD141 (BDCA-3) while in mice this subset is marked by surface expression of CD8α or CD103 (137). The mechanistic roles played by various DC subsets in both tumor progression and the response to treatment are a key area of research for cancer immunotherapy with little consensus as to their frequencies and functions (102, 107). In 2008, it was reported that knocking out Batf3 in mice eliminated CD8α⁺ DC, and consequently it was demonstrated that these mice were incapable of cross-presenting antigen or rejecting highly immunogenic fibrosarcomas (138). Although pDC are purportedly not efficient at cross-presentation, studies have shown their capacity to mediate direct tumor killing and to activate NK cells via the production of type I IFN (139). Despite the key roles played by TIDC in promoting antitumor responses, generally TIDC are skewed in both phenotype and function toward an immunosuppressive role in the microenvironment (107). These alterations in TIDC have been mechanistically studied in murine models (107, 140). The TME has been reported to induce a “paralyzed” state in TIDC resembling an immature phenotype with reduced expression of costimulatory CD80 and CD86 molecules and a diminished capacity to present antigens (107). This induction is a result of various immunosuppressive factors such as VEGF, TGFβ, IDO produced by tumor cells as well as by other cells in the TME (72, 107). Furthermore, DC paralysis in mouse models has been observed to be associated with upregulation of immune checkpoint receptors such as PD-1 and T cell immunoglobulin and mucin-domain containing-3 (TIM-3), which was reported to interact with the alarmin protein high mobility group box 1 (HMGB1) resulting in reduced DC sensing of tumor-derived nucleic acids (107). TIDC with immature and paralyzed phenotypes themselves suppress immune cells in the TME through various mechanisms such as but not limited to, expression of inhibitory molecules (PD-L1), production of regulatory cytokines such as IDO and induction of Tregs (107, 141).

As previously noted, there has been significant research on TIL in melanoma. On the other hand, the mechanistic roles of TIDC in melanoma are not well studied. Melanoma is of particular interest due to the fact that skin contains multiple DC subsets. The five major DC subsets found in human skin are Langerhans cells, CD14⁺ DC, CD1c⁺ DC, CD1a⁺ DC, and CD141⁺ DC (133). The correlations between various TIDC subsets and disease outcome, their association with other cells and specific functions have not yet been fully elucidated (102). However, recently it was demonstrated that intratumoral CD103⁺ DC in mice were crucial for trafficking of melanoma tumor antigen to LN and were dependent on surface expression of CCR7 (142). Enhanced CCR7 mRNA expression in human melanoma samples was also correlated to increased T cell infiltrates and improved patient outcomes (142). In general, it is observed that there are higher frequencies of TIDC in the peritumoral region than within the tumor (102). These peritumoral DC include arguably the most mature population of DC-LAMP⁺CD83⁺fascin⁺ cells (102). In fact, DC-LAMP expression is associated with positive prognosis in not only melanoma but also lung, breast, and metastatic CRC (120). On the other hand, CD123⁺ pDC that do in principle possess the capacity to promote antitumor responses are found to be associated with early relapse and poor prognosis in human melanoma (102, 143). It was shown in both ex vivo patient samples and in that a humanized melanoma mouse model that pDC in melanoma are directed toward a T_H2 promoting phenotype by induction of the molecules OX-40L (TNFSF4) and ICOSL (inducible T cell costimulator ligand), which then drive tumor progression (143). To comprehensively characterize TIDC in melanoma, it is crucial to obtain genomic data to appropriately distinguish and profile TIDC subsets. Pyfferoen et al. performed transcriptomic profiling of DC in a murine model of lung carcinoma and demonstrated that TIDC had significantly increased expression of PD-L1, acquisition of TAM surface markers and a pro-metastatic microRNA signature (144). To date, similar studies have not been performed in human melanoma. There have been several studies in murine models that have demonstrated the therapeutic reprogramming of TIDC (107). Thus, manipulation of TIDC represents a hitherto unexplored target for future melanoma immunotherapies. Many of the same agents that have been shown to induce DC activation and maturation in vitro have been tested for direct targeting of DC in vivo (133, 145). For instance, direct administration of BCG has been utilized for the treatment of bladder cancer for over 30 years although its precise mechanisms of action in vivo are still under study (146). Direct modulation of DC in vivo using DC maturation agents and mAbs is a highly desirable goal in tumor immunotherapy. This is due to the excessive costs, safety considerations, and practical limitations of using cellular products (147). As such, the identification of both targetable DC receptors and maturation stimuli continues to be an active area of research interest. In particular, targeting antigen-coupled antibodies to DC C-type lectin receptors (CLRs) such as DEC205 (CD205), Clec9A, and DC-SIGN in murine and in vitro studies resulted in effective CD4⁺ and CD8⁺ T cell responses (145, 148). Additional receptors such as XCR1 (expressed entirely on CD141⁺ DC) are also being studied for their effects on DC function (133). Clinical trials for multiple cancer types are presently underway to investigate the efficacy of anti-DEC205 conjugated to the cancer–testis antigen NY-ESO-1, which is also used for melanoma immunotherapy (133, 149). Recently, a series of seminal papers have shown the importance of the cytosolic DNA sensor cyclic GMP-AMP (cGAMP) synthase (cGAS) in promoting antitumor immunity (150–152). DNA introduced to the cytosol as a result of viral infections or cellular damage is a potent immune activator that leads to the production of type I IFN (153). Upon detection of DNA by cGAS, it catalyzes the production of cGAMP that binds to the adaptor protein stimulator of interferon genes (STING) ultimately resulting in the production of type I IFN (153). In 2014, Woo et al. demonstrated in a mouse model that tumor-derived DNA was responsible for inducing IFNβ production and the consequent activation of APC and CD8⁺ T cells versus melanoma in vivo (150). Alternately, mice deficient in STING failed to reject these tumors highlighting the crucial role played by this pathway in the immune response to cancer (150, 151). In a more recent paper, Wang et al. showed the role of cGAMP in mediating the effects of ICB (152). It was reported that in mice lacking either cGAS or STING, PD-L1 blockade did not result in significant shrinkage of tumor volume or increase in survival compared with WT mice. Moreover, intramuscular injection of cGAMP in combination with PD-L1 significantly enhanced survival, compared with PD-L1 or cGAMP alone (152). Finally, it was also shown that cGAMP treatment of BMDC enhanced expression of DC activation markers and increased DC antigen cross-presentation. Another molecule that has recently gained interest for its effects on DC is IL-32. In 2012, Schenk et al., identified an IL-32-dependent mechanism for DC differentiation in response to nucleotide-binding oligomerization domain containing protein (NOD2) activation through its ligand muramyl dipeptide (154). DC obtained from IL-32 differentiation were found to express higher levels of MHC class I and CD86, as well as, present antigen to CD8⁺ T cells more effectively than GM-CSF differentiated DC (154). These studies highlight the multiple pathways that may be targeted to generate effective DC in vivo, which is essential for antitumor immunity.

Natural killer cells were characterized over 40 years and are the first population of ILC to be described and studied (155, 156). NK cell defects lead to enhanced susceptibility to viruses and many forms of cancer in humans and in mouse models (156). NK cell functions are modulated by a number of surface receptors that provide either NK activating or inhibitory signals (156, 157). NK cells are broadly defined as CD3⁻CD56⁺ in humans and CD3⁻NK1.1⁺ in mice while both murine and human NK cells express the surface receptor NKp46 (CD335) (156). In humans, NK cells are further divided into CD16⁺CD56^dim which predominate in blood, and CD16⁻CD56^bright populations (156). Canonically, NK cells can recognize tumor cells that have downregulated MHC class I molecules or upregulated induced stress molecules (155, 156). NK cells can also bind to antibodies bound to tumor antigens and mediate antibody-dependent cellular cytotoxicity (156). As with CD8⁺ CTL, NK cells mediate their cytotoxic functions through perforin and granzymes, as well as, by expressing death mediating ligands such as FasL (CD95L) and TRAIL (TNF-related apoptosis inducing ligand) (156). Activated NK cells also produce IFNγ, among other cytokines, which leads to recruitment of other immune cell populations (156).

The roles of NK cells in the TME are currently not fully described (155, 157). Several studies have indicated that NK cell infiltration is generally a positive prognostic factor in various types of cancer (155). In the context of melanoma, the roles of NK cells are an important venue of research. Analysis of several melanoma cell lines indicated that a high percentage of melanoma cells possess ligands for a NK activating receptors such as NKG2D and DNAM1, while ligands have also been identified for NK-bound NCR (natural cytotoxicity receptors) such as NKp30 (157). Melanoma cells are also known to have decreased MHC class I expression as a mechanism to escape CD8⁺ T cells, thus making them targets for NK cells (157). Despite these observations, melanoma immunoediting leads to tumor escape from NK cells via multiple mechanisms (157). Melanoma immunoediting by NK cells increases expression of MHC I, or downregulates NK ligands supported by the decreased expression of MICA reported in metastatic versus primary melanoma (157). IDO and prostaglandin E2 (PGE2) produced by melanoma cells act directly to inhibit NK cells while increased expression of ligands to regulatory receptors such as TIGIT modulate NK cell activity (157). In light of these observations, it will be important to identify NK populations that have persistent antitumor activity and characterize their phenotypes to better understand the mechanism involved in effective NK immunity. Recently, it was reported that tumor-bearing/infiltrated LN in melanoma patients contained twice as many NK cells as ipsilateral tumor-free LN (158). These tumor-infiltrated LN also contained a population of highly cytotoxic CD56^dimKIR⁺CCR7⁺ NK cells that may have prognostic potential for melanoma (158). Conversely, melanoma, breast, and colon cancers were found to be infiltrated by CD56^bright NK subsets, which are similar to decidual NK cells during pregnancy thus implying a potentially regulatory role for this subset (159). NK cells remain an important target for immunotherapy. Along with T cells, NK cells were used early on for adoptive cell transfer therapy of melanoma in the 1980s and both autologous and allogeneic NK cell adoptive transfers are being studied in clinical trials (156, 157). Currently, two antibodies for the blockade of NK checkpoints are under clinical development, namely, lirilumab (anti-KIR-studied in combination with ipilimumab) and IPH2201 (anti-NKG2A) for various types of cancers including melanoma (157). However, further study of NK cells in the melanoma TME is required to understand the several mechanisms of immune escape from NK cells and CD8⁺ CTL and thus devise, rational combinatorial immunotherapies.

Cytotoxic T lymphocyte antigen-4 (also termed cytotoxic T-lymphocyte-associated protein 4), is a crucial regulator of T cell activation and ipilimumab, a human IgG1 mAb targeted to this molecule was the first ICB drug to show clinical efficacy in advanced melanoma and a number of other cancer types (48, 195). CTLA-4 plays a key role in T cell immunity and its molecular biology has been recently reviewed elsewhere (167, 196). However, to understand the clinical role of CTLA-4 blockade, a brief summary of its mechanism of action is warranted. Naive T cells are modulated by APC through the interaction of multiple surface receptors in a region referred to as the “immunological synapse” (197). Canonically, naive T cells require 3 signals for complete activation (Figure 2 inset panel) (198). The engagement of the TCR by peptide antigen presented in the context of MHC, provides the first signal of T cell activation (signal 1) (198, 199). T cells require further signaling from the binding of costimulatory molecules on T cells such as CD28, to its respective ligands CD80/86 on APC (signal 2). Finally, the complete activation requires cytokines (IL-2) binding to their cognate receptors on T cells (Signal 3) (199). As an evolutionary checkpoint to autoimmunity, activated T cells induce surface CTLA-4 expression, which binds with greater affinity to CD80/86 and mediates T cell inhibition and cell cycle arrest (195, 200). CTLA-4 is also expressed constitutively on Treg (167). The crucial role of CTLA-4 in maintaining tolerance is demonstrated by the severe multiorgan autoimmune pathologies and early mortality (3–4 weeks) observed in CTLA-4^−/− mice (201). Humans with heterozygous germline mutations in CTLA-4 also exhibit autoantibodies, increased intra-organ lymphocyte infiltration and other symptoms of immune dysregulation (167).

In 2010, Hodi et al. demonstrated the clinical efficacy of ipilimumab in patients with stage III and IV unresectable and metastatic melanoma whose tumors were refractory to prior treatments (187). The treatment subjects received ipilimumab alone, ipilimumab plus the peptide gp100 or gp100 alone. Patients receiving ipilimumab alone or ipilimumab plus gp100 had significantly increased median OS compared with those receiving gp100 alone (roughly 10 versus 6 months) (187). Currently, ipilimumab has only received FDA approval for melanoma. However, a number of studies have shown modest responses to ipilimumab in other tumor types such as metastatic RCC and NSCLC, and it continues to be studied in clinical trials as combination therapy with PD-1/PD-L1 (discussed below) (160, 167). As mentioned previously, a number of immunological toxicities (IRAEs) are commonly observed to occur in patients treated with ipilimumab primarily in the skin, GI tract, and the endocrine system and in some rare cases result in deaths (193). The frequency of severe toxicities (grade 3 or 4) in the preliminary phase III trials of ipilimumab was demonstrated to be 20%, but this value was not significantly higher than the toxicities associated with many chemotherapy or targeted therapy drugs (163, 195). Most IRAEs can be resolved within 6–12 weeks of steroid therapy but for steroid-resistant adverse events, patients can also be treated with immunosuppressive antimetabolite drugs such as azathioprine and mycophenolate mofetil (193). Novel CTLA-4 blockade agents including modified versions of ipilimumab are also currently under study for a number of advanced solid tumors with the aim of improving safety profiles and tumor-specific delivery (202).

The most clinically successful agents for ICB to date target the inhibitory PD-1/PD-L1 axis (169, 195). The transmembrane receptor PD-1 (CD279) plays a crucial role in regulating antigen-specific T cell responses (169, 203). PD-1 is not only expressed on activated effector T cells but also on NK cells, B cells, macrophages, and Tregs (167, 203). Similar to the activating co-receptor CD28, PD-1 is acted upon by two distinct ligands PD-L1 (B7-H1, CD274) and PD-L2 (B7-DC, CD273) (203). Whereas PD-L2 expression has hitherto been observed only on professional APC (including B cells), PD-L1 is expressed on various tissue types such as epithelial tissue, vascular endothelium, stromal cells as well as tumor cells and virus-infected cells (167, 203). The induction of PD-L1 expression is generally in response to pro-inflammatory cytokines such as interferons, TNF-α, and VEGF (167, 169). PD-1 does not, as its name implies, directly induce cell death. The binding of PD-1 to its ligands instead serves to attenuate T cell activation by recruiting the tyrosine phosphatase SHP-2, which interferes with signaling downstream of the TCR and leading to decreased T cell growth and reduced cytokine production (203). However, PD-1 signaling can also reduce the expression of antiapoptotic genes while upregulating proapoptotic gene expression thus impairing T cell survival (167).

PD-1-deficient mice do not display as severe a phenotype as CTLA-4^−/− mice, developing glomerulonephritis and arthritis in a C57BL/6 background and autoantibody induced dilated cardiomyopathy in BALB/c mice as they age (204, 205). This is arguably due to the more direct inhibitory and Treg-related functions of CTLA-4, whereas PD-1 serves to limit T cell activation indirectly and prevent peripheral autoimmunity (169). As noted previously, in certain conditions of persistent antigen exposure such as in chronic viral infections or in cancer, T cells are observed to develop a dysfunctional or “exhausted” phenotype (72, 167). Such T cells are also marked by elevated expression of PD-1 and other inhibitory receptors such as TIM-3 and LAG3 (72). Furthermore, PD-L1 and/or PD-L2 are both observed to be expressed on a number of tumor-infiltrating APC and tumor cells themselves, not only as a result of cytokines but also due to alternative factors such as gain of chromosomes carrying PD-L1 and PD-L2 or the signaling of the epidermal growth factor pathway (167). Recent studies have shown that APC and tumor cells bearing PD-L1 play additive non-redundant roles in the suppression of antitumor immunity (206). Thus, blockade of the PD-1/PD-L1 axis remains a critical area of interest in tumor immunotherapy with studies on its efficacy in nearly 20 types of solid tumors and hematological cancers (169).

In the context of melanoma, nivolumab, and pembrolizumab, both of which target PD-1 have been shown to have significant clinical efficacies (160, 169, 195). In 2012, results from a phase I study comparing various doses of nivolumab in NSCLC, prostate cancer, CRC, renal cell carcinoma, and melanoma patients were reported (188). The highest activity was demonstrated in melanoma patients where the cumulative response rate (for all doses) was 28% compared with 27% for renal carcinoma and 18% for NSCLC (188). In the same year, an αPD-L1 antibody (BMS-963559) was tested in advanced cancers ranging from melanoma to RCC and was shown to have comparatively low response rates (6–17%) (191). A number of recently concluded trials have also demonstrated the potency of pembrolizumab. The large multicenter phase II trial KEYNOTE-002 examined the efficacy and safety of pembrolizumab in patients who had progressed on ipilimumab therapy, and in patients with BRAF mutations, those who had received either BRAF or MEK inhibitor treatment (207). Patients received either two separate doses of pembrolizumab (2 or 10 mg/kg) or chemotherapy of the investigators choice (carboplatin, dacarbazine, paclitaxel, and temozolomide). The results were highly encouraging as the 6-month PFS was shown to be 38% (10 mg/kg) and 34% (2 mg/kg) in the pembrolizumab group compared with only 16% in the chemotherapy group (207). Similar efficacy over investigator choice chemotherapy (32 versus 11%) has also been reported from an open-label phase III trial of nivolumab in patients who had progressed on ipilimumab (195). Furthermore, pembrolizumab was shown to have significantly higher activity than ipilimumab in patients with advanced melanoma. Robert et al. compared two dosing schedules (every 2 or 3 weeks) of pembrolizumab to ipilimumab and reported 6-month PFS in the range of 46–47% (response rates of roughly 33%) for the pembrolizumab group versus 26.5% (RR of 11.9%) for the ipilimumab-treated patients (208). Finally, in a phase III trial of nivolumab in previously untreated advanced melanoma patients (without BRAF mutations), ICB therapy was demonstrated to have significantly higher efficacy compared with dacarbazine with a 1 year survival rate of 72.9% in the nivolumab treated group versus 42% in the dacarbazine group (189). The successes of αPD-1 in melanoma treatment have also been observed (albeit at lower rates) in a range of other cancer types (167, 169). Furthermore, the rate of grade 3 or 4 treatment related adverse events is lower in patients receiving PD-1 blockade therapy versus ipilimumab which is similar to the decreased severity of autoimmune pathologies observed in PD-1 versus CTLA-4 knockout mice (169, 193). In contrast to PD-1 blockade antibodies, the αPD-L1 agent atezolizumab (MPDL3280A) has thus far received FDA approval only for urothelial bladder cancer and lung cancer (169, 209). Recently, studies have further complicated the role of PD-L1 by demonstrating that it binds to CD80 on T cells and provides another inhibitory signal (210). Thus, further studies are warranted to determine the role of PD-L1 in T cell inhibition in tumors and investigate which tumor types may benefit most from PD-L1 versus PD-1 blockade. A large number of clinical trials are currently underway targeting PD-1/PD-L1 as well as novel combination approaches (169). As previously mentioned, further study will be required to determine biomarkers of response to ICB and further mechanistic knowledge will be necessary to design effective combinatorial immunotherapies. Four clinical biomarker profiles for ICB treatment have already been proposed based on the presence of PD-L1 and TIL (211). The tumor are characterized as type I (PD-L1⁺TIL⁺), type II (PD-L1⁻TIL⁻), type III (PD-L1⁺TIL⁻), and type IV (PD-L1⁻TIL⁺) (211). In melanoma, where the data are most complete, the majority of patients are either type I (~38%) or type II (~41%). Type I patients are deemed to be the best responders to PD-1 blockade whereas type II tumors are estimated to have very poor prognosis due to their lack of immune cell infiltrates (211). Currently, the mechanisms that regulate the immune composition of a tumor are not well understood and there is a significant interest in treatments that can convert T cell non-inflamed (non-infiltrated) tumors to T cell inflamed (infiltrated) tumors (212).

Despite the tremendous successes of ICB, to date, only a subset of patients achieve durable clinical responses (160, 167). However, the potency of immune checkpoint therapies has ushered in a new era of cancer treatment by offering the possibility of combining these drugs with conventional cancer treatments such as radiation, chemotherapy, and targeted molecular therapy (e.g., BRAF/MEK inhibitors). The prospects for such combination treatments in melanoma and other cancer types, as well as the clinical findings to date using such approaches have been expertly reviewed this year (213–215). The primary focus of this section will be to discuss the approaches involving combination checkpoint blockade therapies for melanoma that have demonstrated efficacy thus far. Nevertheless, it is pertinent to note that currently there are no clinical data to distinguish between ICB or BRAFi/MEKi targeted therapy as first line treatment for melanoma and a clinical trial (NCT02224781) is being conducted to provide direct comparisons between clinical outcomes in patients receiving checkpoint blockade drugs following targeted therapies and vice versa (215).

The success of combined ipilimumab and nivolumab has also been recently reported in a number of clinical trials. In 2015, Postow et al. reported the results of a study where previously untreated patients with metastatic melanoma received either ipilimumab in combination with nivolumab or with placebo preceding a subsequent treatment with nivolumab or placebo (216). The ORR was 61% in the combination treatment group versus 11% in the ipilimumab plus placebo group. Moreover, nearly 22% of patients treated with combination therapy achieved CR compared with none of the patients given ipilimumab and placebo (216). In the same year, results were published from a phase III trial in 945 patients with unresectable stage III or IV melanoma treated with nivolumab alone, nivolumab plus ipilimumab, or ipilimumab alone. The median PFS was 11.5 months for the combination group, 6.9 months for the nivolumab group, and 2.9 months for the ipilimumab group (217). However, serious (grade 3 or 5) treatment related adverse events in the combination treatment group were significantly higher reaching 55% compared with 27% for the ipilimumab group (217). These studies also indicate the superiority of combinatorial checkpoint blockade over monotherapy leading to the approval of ipilimumab and nivolumab dual therapy for melanoma in the USA, while its efficacy in other tumor types continues to be investigated (218). The successful use of combined checkpoint blockade has also sparked clinical interest in additional immune checkpoints some of which are undergoing preclinical or clinical investigation (167, 169, 218). A target of particular interest is the CD4 homolog lymphocyte activation gene-3 (LAG-3), which is expressed on Treg, effector CD4⁺ and CD8⁺ T cells, NK cells, B cells, and pDC and which also binds to MHC class II (167, 219). LAG-3 is an important negative regulator of CD4⁺ and CD8⁺ T cells and is required for Treg activity (219). The αLAG-3 antibody BMS986016 is currently being examined in a clinical trial (NCT01968109) for several advanced tumors both as a monotherapy and in combination with nivolumab (167). Another immune checkpoint that has exciting potential for tumor immunotherapy is TIGIT (T cell immunoreceptor with immunoglobulin and ITIM domain) (167). TIGIT is expressed by activated T cells, NK cells and is also expressed on highly functional subsets of Treg (219, 220). TIGIT has two ligands, namely, CD155 (poliovirus receptor, PVR) and CD112 (PVRL2) that are expressed on APC as well as on tumor cells (167). Likewise, TIGIT is reportedly expressed on TIL (219). The immunoregulatory functions of TIGIT are only recently beginning to be described (221). TIGIT can bind to CD155 on DC resulting in increased IL-10 and decreased IL-12 secretion (167). Ligation of TIGIT on Treg results in the expression of fibrinogen-like protein 2 (Fgl2), a Treg effector molecule that has broad immunosuppressive effects such as mediating Th1 and Th17 phenotype suppression in favor of Th2 (167, 222). In human melanoma, tumor-specific CD8⁺ T cells in peripheral circulation and CD8⁺ TIL were found to express both TIGIT and PD-1 and furthermore, TIGIT was upregulated in response to PD-1 blockade (223). Thus, the described functions of TIGIT further complicate our understanding of the immune response to αPD-1 treatment and provides further proof of the need of combinatorial approaches to overcome current barriers to ICB treatment. The positive results associated with ICB treatment have also renewed interest in a parallel treatment approach involving the development of agonistic antibodies for T cell costimulatory molecules such as CD137 (4-1BB), GITR (glucocorticoid-induced TNFR family related gene), and OX40 (CD134) many of which are currently undergoing clinical trials in combination with nivolumab (167, 169, 218). In 2016, early results were showcased for the antibody urelumab (αCD137) in combination with nivolumab (202). In melanoma, the ORR was observed to be 50% in patients who had not previously received checkpoint blockade therapy and was found to be independent of tumor PD-L1 status (202). Thus immune agonistic antibodies have revealed a plethora of novel possibilities for cancer treatment. Future studies will involve analyses of various combinations aimed at developing immunotherapies tailored to the specific tumor immune microenvironment (224).