The Crossroads of Cancer Epigenetics and Immune Checkpoint Therapy

Immunotherapy has revolutionized the treatment of cancer and has become a bona fide “fourth pillar” of anticancer therapy (1). Initially approved for advanced melanoma, antibodies blocking CTLA-4 and PD-1 (2) are now part of standard treatment paradigms for over a dozen human malignancies including non–small cell lung carcinoma, Merkel cell carcinoma, mismatch repair deficient colon cancer, and renal cell carcinoma (3). Novel inhibitors targeting additional immune checkpoint molecules that serve as negative regulators of T-cell activation, such as LAG-3 and TIM-3, have also recently gained FDA approval (4), and more are in development. The estimated percentage of patients with cancer in the United States eligible for immunotherapy has increased from approximately 1.5% in 2011, to over 40% in 2018 (5).

However, despite these tremendous advances, certain tumor types, such as pancreatic adenocarcinoma, exhibit very low response rates to immune checkpoint inhibitors (ICI; ref. 6), partly due to low neoantigen load and insufficient immunogenicity (7). In addition, many patients who initially benefit from ICI lack a sustained T-cell response, due to T-cell exhaustion that occurs with chronic antigen stimulation (3). As a result, out of approximately 40% of cancer patients in the US eligible to receive immunotherapy, only 12% to 20% have durable benefit from ICI (5). These data underscore the need to better understand the cancer-immunity axis and develop new therapeutic strategies to further improve patient outcomes (8).

Epigenetic modifications are heritable changes in gene expression, which do not alter the genetic sequence (9). Epigenetic changes encompass alterations in DNA methylation, posttranslational modifications of histone protein tails, alterations of chromatin structure, and gene regulation by noncoding RNAs. Aberrant epigenetic changes are ubiquitous in cancer, including genome-wide DNA hypomethylation (10), focal hypermethylation (11), loss of histone acetylation (12), histone methylation (13), and others (Table 1).

Epigenetic regulators are recurrently mutated or amplified in human tumors (Fig. 1; ref. 14). For example, chromatin remodeling complexes are estimated to be mutated in one fifth of all human malignancies (15), and epigenetic enzymes are among the most mutated genes in a subset of human tumors, such as DNA methyltransferase 3A (DNMT3A) in T-cell lymphoma and AML (16). Epigenetic regulators are known to play prominent roles during carcinogenesis as well as the antitumor immune response (17, 18). Prominent roles for epigenetic mechanisms are described both in tumor cells and in lymphocytes, including regulating the expression of antigen presenting MHC molecules (19), expression of immune checkpoint molecules (20, 21), lymphocyte activation (22), and differentiation (23), as well as expression of numerous cytokines (Table 2; ref. 24). How genetic alterations of epigenetic regulators impact clinical response to immunotherapy is incompletely understood.

Epigenetic mechanisms play prominent roles in regulating immune evasion (Table 2) and can be predictive of checkpoint therapy response. For instance, the polycomb repressive complex 2 can silence MHC-I expression (19) in lung cancer, and limit T-cell multipotency (25). Epigenetic repression of NKG2D ligands by EZH2 inhibits NK-cell responses and promotes immune evasion in glioblastoma (26). In melanoma, promoter hypermethylation of antigen processing and presentation genes is associated with immune evasion and poor prognosis (27). DNA methyltransferase 1 (DNMT1) together with EZH2 can regulate the expression of CXCL9 and CXCL10, which in turn direct effector T-cell trafficking to the tumor microenvironment (24). IL10 expression and CAR-T survival is partly regulated by DNMT3A, which also plays prominent roles in T-cell fate decisions after priming by methylating the Tcf1 promoter (22, 28). The critical role for epigenetics in immune regulation is also reflected in the ability of certain epigenetic changes to predict response to immune checkpoint inhibitors (29–32). FOXP1 methylation is associated with poor progression free survival or overall survival in some cohorts of non–small cell lung cancer (29). On a broader scale, genomic demethylation is associated with immune evasion and poor response to checkpoint blockade (27).

Recent evidence suggests a previously unappreciated role for epigenetic enzymes in regulating tumor cell-intrinsic immunity, and T-cell exhaustion, which have critical implications on immunotherapy (3). Herein, we discuss the impact of epigenetics on the cancer cell—T-cell interface, specifically focusing on approaches to enhance tumor cell immunogenicity and prevent T-cell exhaustion, two significant challenges limiting the effectiveness of current immunotherapies.

Insufficient tumor cell immunogenicity and low neoantigen load is thought to partly account for lack of response of “cold” tumors to ICI (6). Inducing expression of noncoding sequences, which are ordinarily epigenetically silenced in somatic cells, is emerging as a potential strategy to enhance immunogenicity of “cold” tumors and has important implications on immunotherapy (33, 34). Noncoding sequences make up >90% of the human genome (17). Among the noncoding sequences, repetitive elements make up approximately 50% and include long terminal repeat (LTR) containing endogenous retroviruses (ERV, ∼9%) and non-LTR containing short and long interspersed nuclear elements (SINE, ∼15% and LINEs, ∼20%; ref. 35). ERVs have the potential to be transcribed, but they are normally silenced by DNA and histone methylation in somatic cells, depending on the evolutionary age of their integration into the human genome (36).

Detection of viral nucleic acids is a critical aspect of immunity (37). In mammals, pattern recognition receptors RIG-I/melanoma differentiation antigen 5 (MDA5) and cyclic GMP-AMP synthase (cGAS) recognize viral dsRNA and dsDNA in the cytosol, respectively (37), and activate adaptor proteins MAVS and STING, which results in activation of antiviral IFN signaling (38). ERV expression would ordinarily be considered deleterious, due to accumulation of dsRNA and dsDNA, which trigger cell-intrinsic antiviral pathways. This cell-intrinsic immunogenicity is observed during cellular stress, tissue damage, and is implicated in the pathophysiology of autoimmune disease (39). However, enhancing cell-intrinsic immunogenicity in cancer cells is a desired outcome, and can sensitize previously non-immunogenic cells to ICI.

Two landmark studies linked epigenetic modulation to ERV expression, IFN signaling, and antitumor immunity (33, 34). Chiappinelli and colleagues showed that inhibiting DNA methyltransferases in ovarian cancer cells led to hypomethylation and upregulation of ERV transcripts, activating TLR3 and MAVS and leading to a type I IFN response (33). The response could be abrogated by knocking down TLR3 and MAVS, or blocking IFNβ or its receptor. DNMT inhibitor 5-Azacytidine sensitized nonimmunogenic melanoma cells (B16-F10) to anti-CTLA-4 therapy and suppressed tumor growth in immunocompetent mice (33). Concurrently, Roulois and colleagues reported that low-dose 5-Azacytidine activated ERV sequences, which induced a dsRNA response in colorectal cancer cells, resulting in MDA5 and RIG-I activation and increased type III IFN production (34). These studies provided the initial compelling rationale for combining hypomethylating agents that could induce viral mimicry with ICI.

Histone modifiers play equally important roles in regulating ERV expression and immune evasion. Sheng and colleagues showed that loss of lysine specific demethylase 1 (LSD1) leads to increased H3K4 di-methylation, de-repression of ERV transcription, and accumulation of dsRNA (40, 41). The dsRNA stress leads to MDA5 activation and IFNβ signaling. As a result, “cold” or poorly immunogenic B16 and D4m.3A melanoma cells became immunogenic, and loss of LSD1 in these models led to increased T-cell infiltration and tumor shrinkage in response to anti-PD1 therapy (40, 41). However, LSD1 loss did not cause complete tumor rejection, due to also de-repressing inhibitory TGFβ, suggesting that LSD1 inhibition can be a double-edged sword with both immune-supportive and suppressive effects. In fact, depletion of both LSD1 and TGFβ in combination with PD-1 blockade leads to tumor rejection (42). Similarly, Zhang and colleagues uncovered that loss of another H3K4 demethylase KDM5B also leads to a striking phenotype of increased CD8⁺ infiltration, melanoma tumor rejection, and protection from tumor rechallenge in immunocompetent mice (43). These phenotypes were associated with de-repression of ERV transcripts, and an IFN signaling transcriptional signature, which are mediated by both cytosolic RNA and DNA-sensing pathways. Although KDM5 demethylase inhibitors were found to induce STING expression and activation of robust interferon response in breast cancer cells (44), the roles of KDM5B in immune evasion by mouse melanoma was independent of its catalytic activity and correlated with H3K9 methylation instead. Mechanistically, KDM5B recruits H3K9 methyltransferase SETDB1 to repress ERV transcription and anti-tumor immune responses (43). The histone methyltransferase SETDB1, acts as the catalytic domain of the KAP1 and HUSH silencing complexes (45), silencing retroelements and segmental duplications with transcriptional potential. Amplification of KDM5B or SETDB1 has been shown to correlate with poor overall survival of patients with renal cell carcinoma treated with nivolumab (43, 46).

Three independent CRISPR-screen studies identified the roles of SETDB1 complexes in immune evasion. Griffin and colleagues conducted a chromatin regulator screen to identify genes whose knockout increases ICI sensitivity of B16 melanoma cells and Lewis lung carcinoma (46). Most of the chromatin targets were tumor-specific, however the few shared hits were Setdb1, Ep300, and Kdm1a. Melanoma specific hits included other notable epigenetic modifiers Dnmt1, Ezh2, and Trim28 (HUSH complex component), whereas Smarca4 and Pbrm1 were specific to lung cancer (46). Setdb1 had the strongest signal in both cell types. They found that Setdb1 loss activated transposable elements containing viral open reading frames and upregulated MHC class I binding peptides. Compared with previous reports of retroelement re-activation (33), nucleic acid sensing pathways and IFN signaling were not affected (46), likely due to the extent of retroelement activation or existing expression levels of retroelements in their model systems. Consistently, Hu and colleagues performed an epigenome-wide CRISPR-screen in a lung adenocarcinoma model and uncovered that loss of ATF7IP, a transcription factor interacting with SETDB1, suppressed tumor cell growth in immunocompetent mice (47). Mechanistically, they report that either Atf7ip or Setdb1 loss leads to decreased H3K9 trimethylation and increased expression of ERV-derived antigens (including gp70 and p15E), triggering IFN signaling, increased T-cell infiltration, and potentiating antitumor responses (47). Using a CRISPR-screen, Lin and colleagues identified TRIM28 as one of the top hits regulating PD-L1 expression in ovarian cancer (48). Setdb1 and Trim28 loss led to increased Granzyme B⁺ CD8⁺ T-cell infiltration, activation of cGAS/STING signaling, and greater decrease of ovarian cancer tumor growth with ICI compared with anti-PD-L1 therapy alone (48).

SETDB1 was also found to be a critical regulator of ERV by Guo and colleagues in a study of WEE1, a tyrosine kinase that serves as a mitotic gatekeeper by negatively regulating cyclin-dependent kinases (49). Interestingly, it can also phosphorylate a tyrosine residue on histone H2B and suppress histone transcription, a critical step during S-phase of mitosis. WEE1 inhibition also increased ERV expression through downregulating SETDB1 expression and H3K9 tri-methylation via FOXM1 (49). ERV re-expression triggered a dsRNA stress response and RIG-I activation. Using a WEE1 inhibitor, they observed upregulated IFN signaling, which ultimately resulted in PD-L1 elevation and enhanced sensitivity to ICI in ovarian cancer cell lines. They also observed enhanced tumor clearance of the MC38 colon cancer, and B16 melanoma cell lines in immunocompetent mice in a CD8⁺ T-cell-dependent manner. WEE1 inhibitors have not been trialed in combination with immunotherapies but have shown some benefit against refractory ovarian and pancreatic cancer, in combination with gemcitabine (49–51).

In addition to SETDB1, two other histone H3K9 methyltransferases SUV39H1 and G9A were shown to repress ERV expression. FBXO44, a member of the F-box protein family, was identified in an RNAi screen of genes that affect retroviral element expression (52). FBXO44 knockdown led to decreased H3K9 tri-methylation by abrogating the targeting of the histone methyltransferase SUV39H1 to repetitive element loci. FBXO44 plays an important role in recruiting the epigenetic machinery to the replication fork and coordinating silencing. Loss of FBXO44 activated both RIG-I/MDA5-MAVS and cGAS-STING pathways to trigger IFN signaling, resulting in enhanced tumor cell immunogenicity and sensitivity to checkpoint blockade (52). Combining the effects of inhibiting DNA methylation and the histone methyltransferase G9A, Liu and colleagues reported a synergistic effect of the two silencing mechanisms on ERV re-expression (53). Their findings suggest that not all ERVs are equally repressed, and certain loci are dually regulated by H3K9 methylation and DNA methylation. Interestingly, a distinct cell line specific pattern of ERV upregulation was observed upon epigenetic therapy. This cell-type specificity is incompletely understood but has also been described by other studies (46) and is an important potential caveat to using viral mimicry to stimulate immunogenicity. In general, DNMT inhibition triggered more ERV expression than G9A inhibition, and dual inhibition de-repressed up to 10-fold higher number of ERV transcripts in ovarian cancer cells (53).

Histone K27 methyltransferase EZH2 has also been implicated in regulating ERV expression (54). Using a model of small cell lung cancer, Canadas and colleagues identified a population of fibroblastic mesenchymal cells with high PD-L1 expression (54). Investigating the mechanism behind PD-L1 upregulation, they uncovered that EZH2 was downregulated, and this was associated with de-repression of ERVs, increased cytosolic dsDNA, STING upregulation, and IRF3 phosphorylation. The effect could be reverted by MAVS or MAVS/STING deletion. Pharmacologic inhibition of EZH2 also de-repressed ERV expression and enhanced IFNγ induced cytokine production, bolstering the proposed mechanism (54).

In summary, several recent independent studies converged on epigenetic modifiers (DNMTs, LSD1, KDM5B, SETDB1, SUV39H1, G9A, EZH2) as powerful regulators of the antitumor immune response. Notably, several KAP1 and HUSH complex members were identified as key regulators of immunogenicity in different tumor types, making SETDB1 a promising target to improve immunotherapy. Taken together, these studies suggest that targeting epigenetic enzymes to re-express endogenous antigens and trigger IFN signaling and MHC-I antigen presentation is a compelling strategy to sensitize nonimmunogenic tumor cells to immune checkpoint inhibitors (Fig. 2).

Epigenetic changes have long been known to play important roles during T-cell development, differentiation, activation, cytokine production, and memory differentiation (55, 56), summarized in Table 2. More recently, epigenetic changes were shown to regulate the expression of immune checkpoint molecules in several types of cancer (Table 2). PD-1 is regulated by DNA methylation (57, 58), and its expression is re-enforced by H3K4me1 and H3K27 acetylation at conserved upstream regulatory elements (59). Conversely, H3K9me3, H3K27me3, and H4K20me3 repress its expression following resolution of its stimulation (60). Similarly, CTLA-4 and PD-L1 expression (21) are regulated by DNA methylation, H3K9 and H3K27 trimethylation (61). Epigenetic changes of checkpoint molecules have been shown to predict response to immunotherapy in melanoma (62), prostate cancer (63), colorectal (64), and other malignancies in individual studies.

A critical role for epigenetic changes is emerging in regulating T-cell exhaustion, with critical implications for immuno-oncology. Originally described in experimental models of chronic viral infection, T-cell exhaustion is an phenotypic adaptation of T cells thought to arise as a consequence of chronic antigen exposure and T-cell receptor signaling (65). The functional hallmark of exhaustion is decreased ability to produce IL2, IFNγ, TNFα, and overall decreased capacity to clear antigen (66). In addition to terminally exhausted CD8⁺ T cells, a self-renewing population of progenitor exhausted CD8⁺ T cells has been described, which has a preserved effector function, proliferative capacity, and is critical for response to immunotherapy. Multiple recent studies using single-cell technologies on human tumor samples identified tumor-infiltrating lymphocyte, exhibiting properties of exhausted and progenitor exhausted CD8⁺ T cells in several types of cancer (67, 68). Importantly, the ratio of progenitor exhausted to terminally exhausted T cells directly correlates with response to immunotherapy. It was demonstrated that the progenitor exhausted population, marked by transcription factor TCF1, is the primary source of T-cell expansion after immune checkpoint therapy (69), whereas the terminally exhausted T cells are largely inert and do not respond to ICI. Transcriptional programs and cell surface markers of the progenitor and exhausted T-cell phenotypes have been relatively well described (70–78). For instance, the progenitor population hallmark markers are cell surface expression of SLAMF6 and intermediate expression of PD-1 and a T-bet^high and Eomes^low transcriptional program. Conversely, the terminally exhausted T cells express TIM3, LAG3, and high levels of PD-1, whereas their transcriptional profile is the opposite—T-bet^low and Eomes^high.

In contrast, epigenetic regulation of T-cell exhaustion is incompletely understood (79, 80), but has critical implications for response to immunotherapy. Namely, ICI can transiently reprogram or “rejuvenate” exhausted T cells toward a more effector-like transcriptional and functional phenotype. However, rejuvenated cells quickly become re-exhausted, due to the largely stable chromatin state of exhausted T cells, which is unperturbed by PD-1 blockade (81). This phenomenon of epigenetic reinforcement of an exhausted state has been termed epigenetic “scarring” (55, 56). The exact epigenetic mechanisms that establish and maintain an exhausted T-cell state are also incompletely understood, but both DNA methylation and histone modifications are likely to play important roles. De novo DNA methylation may be required for development of fully exhausted T cells, which exhibit greater than 1,000 de novo methylation events (80). In a model of T-cell exhaustion during chronic infection with the lymphocytic choriomeningitis virus (LCMV), inhibition of DNMT3A synergized with PD-L1 blockade to enhance T-cell infiltration and antiviral immunity. Conversely, depletion of DNMT3A in CAR-T cells was shown to potentiate the effectiveness of adoptive cell transfer therapy and stimulate T-cell effector function, through IL10 (28). Taken together with the previously discussed ability of DNMTi to induce cancer cell intrinsic IFN signaling, there is potential for DNA methylation inhibitors to act on both sides of the cancer cell-immunity synapse. However, pan-DNMT inhibition has also been reported to impair T-cell development, and may pose cytotoxicity challenges (13). Of note, a new generation of DNMT inhibitors, including compounds with higher specificity and lower toxicity have been developed (17).

In addition to regulating the innate immunogenicity of cancer cells as discussed previously, the histone demethylase LSD1 may also play important roles in the epigenetic enforcement of T-cell exhaustion (41, 82). Liu and colleagues recently reported that the LSD1/CoREST complex interacts with TCF1 and acts to suppress its transcription (82). LSD1 loss led to a shift in the T-cell milieu toward a progenitor exhausted TCF1+ profile with an enhanced response to anti-PD1 treatment in the MC38 tumor model (82). Many irreversible and few reversible LSD1 inhibitors have been studied in phase I trials of refractory malignancies and showed some promise in refractory prostate cancer (83). A potential caveat are demethylase-independent activities of LSD1, which have also been implicated in AML and prostate cancer (84, 85). Similarly, histone methyltransferase SUV39H1 also limit cytolytic effects of CD8⁺ T cells (86). Depletion or inhibition of SUV39H1 induces chromatin accessibility of cytolytic effector loci, delays melanoma growth, and enhances antitumor responses to anti-PD-1 treatment.

An important role for transcriptional repressor BACH2 was recently identified in enforcing the progenitor exhausted CD8⁺ T-cell population state (71). The BACH2 genomic locus exhibits enrichment for H3K27 acetylation and is highly expressed in progenitor exhausted antigen-specific CD8⁺ T cells compared with terminally exhausted cells (71). Ectopic BACH2 expression polarizes the transcriptional phenotype of CD8⁺ T cells toward a progenitor program, including heightened expression of stem-like state markers such as Tcf7, Id3, and Slamf6. BACH2 may be required for establishing the epigenetic landscape of progenitor exhausted T cells. Chromatin accessibility analysis comparing BACH2 deficient and sufficient antigen-specific CD8⁺ T cells suggests that BACH2 is necessary for maintaining an open chromatin state at the loci associated with stem-like CD8⁺ T cells by silencing the repressive program of Prdm1, in a chronic viral infection model. It is unclear which histone deacetylase (HDAC) is responsible for regulating BACH2 expression during exhaustion, and whether pharmacologic HDAC inhibitors can induce a progenitor CD8⁺ T-cell state via BACH2 re-expression. H3K27 tri-methylation antagonizes H3K27 acetylation, raising the possibility of using pharmacologic depletion of H3K27 tri-methylation to promote H3K27 acetylation. Related to this possibility, Weber and colleagues used a model of tonic CAR-T signaling to investigate the epigenetic changes associated with exhaustion and whether they can be reversed by intermittent blockade of CAR signaling (“rest”; ref. 79). They found that “rest” was associated with functional rejuvenation/IL2 production and significant changes in the epigenetic landscape at exhaustion-associated genes encoding T-bet, NFATC1, AP-1, DNMT3A, and NR4A family transcription factors. The epigenetic “rest”-associated rejuvenation was dependent on EZH2 activity, and could be abolished by the selective EZH2 inhibitor, tazemetostat (79).

An important role for noncoding RNAs has also been demonstrated in regulating T-cell exhaustion. Ectopic expression of miR-29a was recently shown to potentiate antiviral responses in a LCMV chronic infection model of T-cell exhaustion (87). Stelekati and colleagues showed that forced miR-29 expression affected the expression of Eomes, a known master transcriptional regulator of T-cell states (88) and induced a TCF1⁺ “memory-like” CD8⁺ transcriptional profile rather than an exhausted state. Interestingly, miR-29 is a known regulator of DNMT expression (89), however the described effect appears to occur through a different mechanism.

In summary, ICI are unable to epigenetically reprogram terminally exhausted CD8⁺ T cells, which have a distinct epigenetic landscape. Targeting specific DNA methyltransferases, histone demethylases and deacetylases has shown promise in polarizing CD8⁺ T cells toward an immunotherapy-responsive phenotype in experimental models of T-cell exhaustion (Fig. 3). Given the genome-wide effects of epigenetic drugs, it will be challenging to specifically target T cells without causing systemic effects. This may be achievable by engineering high-affinity small molecules or with more judicious approaches, such as targeting epigenetic enzymes in lymphocytes ex vivo using genetic or pharmacologic approaches. Further research is needed to pinpoint the epigenetic drivers of T-cell exhaustion and how to optimally target them.

Low immunogenicity (“cold tumors”) and resistance to immune checkpoint therapy are two of the main factors limiting the benefit of cancer immunotherapy. Epigenetic mechanisms play important roles in regulating the antitumor response on both sides of the cancer cell: T-cell synapse. On the tumor cell intrinsic side, epigenetic enzymes regulate antigen presentation and processing, immune checkpoint expression, tumor cell proliferation. In lymphocytes, epigenetic mechanisms regulate T-cell development, differentiation, cytolytic activity, and cytokine production. Recent evidence, reviewed here, uncovered critical roles for epigenetic modulation of cancer cell immunogenicity and T-cell exhaustion, key components limiting immune checkpoint efficacy. Considering these new findings, leveraging epigenetic therapies is an attractive approach to improve current immunotherapies by acting both on tumor cells and T cells to boost immunogenicity and eschew T-cell exhaustion, respectively.

Beyond critical roles in response to immune checkpoint inhibition, epigenetic modifiers have important implications on other immunotherapies, including CAR-T therapy. In particular, the landscape of DNA methylation is associated with clinical response and may be used as a biomarker to predict CAR-T response in some malignancies (90). It was reported recently that DNMT3A deletion from CAR-T cells upregulates IL10 and enhances antitumor responses in pre-clinical models, thus DNMT3A may be a target to enhance CAR-T therapy (28).

Currently, epigenetic therapies are FDA approved for treatment of acute myeloid leukemia, myelodysplastic syndrome, multiple myeloma, epithelioid sarcoma, cutaneous and peripheral T-cell lymphoma (17). A summary of current clinical trials combining epigenetic and immune checkpoint therapies in solid and hematologic malignancies is shown in Table 3. Many epigenetic agents evaluated in past clinical trials were noncompetitive inhibitors, which nonspecifically target entire classes of epigenetic enzymes, such as pan-DNMT inhibitors (14). Trials were frequently conducted in combination with traditional chemotherapy, resulting in significant toxicity and modest tumor responses (13). With the emergence of: (i) new classes of competitive, high affinity, and enzyme-specific epigenetic inhibitors; (ii) possibility of using epigenetic therapies in cell-based approaches such as adoptive cell transfer; (iii) increased safety of CRISPR-based genetic editing capabilities; and (iv) technologies to target specific proteins for degradation, there is tremendous potential for epigenetic therapies to improve anticancer immune responses and herald a next generation of responses in immuno-oncology.

M.W. Bosenberg reports a patent for WO 2008/066878 A3 issued. Q. Yan reports grants from NIH, Department of Defense, and Melanoma Research Alliance during the conduct of the study as well as personal fees from AstraZeneca outside the submitted work. No disclosures were reported by the other author.

G. Micevic was supported by an NIAID-funded fellowship T32AR007016–47 to Yale Department of Dermatology. G. Micevic has been supported by the Dermatology Foundation and American Skin Association. M.W. Bosenberg was supported by NIH grants P50CA121974, U01CA233096, U01CA238728, P30CA016359, and a Melanoma Research Alliance Team Science Award and was issued a relevant patent WO2008066878A3. Q. Yan was supported by NIH grants P50CA121974, R01CA237586, P30CA016359, a Department of Defense Breast Cancer Research Program Breakthrough Award W81XWH-21–1-0411, and a Melanoma Research Alliance Team Science Award.