*5.2. Resistance to Immunotherapy*

The development of immunotherapy, which has transformed the management of metastatic tumors, has undoubtedly been fostered by a comprehensive understanding of the tumor microenvironment (TME) and its immunophenotype. Tumors can be reduced to two main compartments that are closely intertwined: the malignant cells and TME. TME is composed of a variety of stromal cells embedded in an extracellular matrix irrigated by a complex network of blood and lymphatic vessels [174]. The cells within the stromal compartment can include immune cells (macrophages, B lymphocytes and cytotoxic T lymphocytes (CTLs), natural killer cells (NKCs), neutrophils, and dendritic cells), mesenchymal cells (fibroblasts, myofibroblasts, cancer stem cells- CSCs, mesenchymal stem cells- MSCs, adipocytes, and endothelial cells), and MDSCs [175,176]. Moreover, in melanoma's TME, TAMs are abundant and due to their pro-tumoral M2 phenotype these "tumor hijacked" cells sustain therapy resistance [15]. Stromal cells are in close communication with tumor cells and help them adapt to a changing microenvironment, survive, and replicate. As melanomas have a clear immune fate since immunosurveillance favors efficient tumor elimination and immunotolerance promotes tumor survival [175], therapy resistance has a clear immunological background. Additionally, it is well known that melanomas have an increased mutational rate and express a plethora of antigens, for example CTAs, which attracts immune cells that can eradicate the tumor or can be diverted towards pro-tumoral activity [177].

Natural immunosuppression has emerged as a physiological mechanism, but in TME immunosuppression usually interferes with CTL activity and functions [178,179]. Tumor cells mediate immunosuppression taking over inhibitory checkpoint proteins such as PD-1, T cell immunoglobulin and mucin domain 3 (TIM-3), lymphocyte activation gene 3 (LAG-3), and CTLA-4 expressed on the surface of T lymphocytes [180,181]. PD-1 and CTLA-4 bind to specific ligands such as PD-L1 and PD-L2, or CD80 and CD86, respectively, to negatively regulate T cell activity, leading to immune cell escape [181]. Therefore, the task of immunotherapy is much more challenging than it seems, as its goal is not to induce apoptosis but to modulate TME to induce a state of immunosurveillance that destroys cancer cells [109]. Anti-CTLA-4 antibody (Ipilimumab/Tramelimumab), approved for clinical management of CM more than five years ago [182] and followed closely by the approval of anti-PD1 (Nivolumab) [183], is the main immune therapeutical player that changed the fate of melanoma patients. Recently published data regarding the long-term therapy with individual and/or combined therapies in melanoma patients has shown that complete response is witnessed in 28% of patients and that there is still a great percentage of patients with incomplete response, patients that gain resistance and/or patients that due to immune-related adverse effect have to cease their immune therapy [184].

To date, the proposed mechanisms for CM resistance to immunotherapy are downregulation of MHC molecules, loss of antigenic expression, T-cell exhaustion, aberrant

expression of PD-L1 in response to IFN-γ production by T cells, along with the altered expression of chemokines such as CCL3, CXCL1, and CCL4 [9]. Although it is well known that epigenetic regulation has critical roles in shaping the identity and differentiation of immune cells, the administration of epigenetic therapy should be done with caution in immunotherapy-resistant CMs, as these agents can affect other cells within TME in addition to tumor cells [59].

#### 5.2.1. DNA Methylation and Resistance to Immunotherapy

The discovery that the immune system can be harnessed to fight cancer and improve clinical outcomes in CM was recognized with a Nobel prize in 2018 [185]. Nevertheless, further studies drove research toward elucidating how DNA methylation can impact the function and activity of immune system components [57]. Earlier studies have shown that DNA methylation appears to be involved in regulating T cell differentiation and exhaustion [186,187], but also in modulating immune checkpoint genes [188], which are the main biomarkers for the response to immunotherapy.

Several authors have reported a mechanistic link between DNA methylation status and immune checkpoint gene expression that may have important predictive and monitoring implications for immunotherapy-treated CM patients. For instance, methylation of immune checkpoint CTLA4 has recently been associated with worse response and progression-free survival (PFS) in stage IV CM patients treated with ipilimumab [189]. The same study also highlighted an inverse correlation between CTLA4 promoter methylation and the presence of tumor-infiltrating lymphocytes (TILs), which play a critical role in tumor control and response to immunotherapy. Therefore, melanoma samples with a low level of methylation are likely to have an increased immune cell infiltration, and an increased number of TILs is an indicator of a good clinical response. However, no significant correlation was reported between CTLA4 promoter methylation and CTLA-4 protein expression, suggesting that the level of protein expression of CTLA4 cannot be used as a predictive biomarker in CM [189]. Other authors have shown that the pattern of PD-L1 methylation may also be suggestive of the response to immunotherapy. Briefly, it has been shown that the degree of DNA methylation of melanoma cells facilitates the stratification of CM patients into four subgroups based on the expression of PD-L1 and TILs and that this information may provide clues about the therapeutic response and survival rates of these CM patients [190,191]. Earlier studies have shown that highly responsive patients showed elevated levels of TILs and PD-L1, while the nonresponsive group displayed low levels of TILs and PD-L1 [190]. In the meantime, while clinical information has been gathered about this clear-cut PD-L1 high expression, efficient immune therapy has been shaken due to various newly discovered molecular mechanisms [192]. DNA methylome analysis of 52 stage III patients from TCGA revealed that low/absent PD-L1 expression is associated with high DNA methylation, differential expression of immune-related genes and worse survival [193]. In parallel, Micevic at al. confirmed that PD-L1 methylation regulates PD-L1 expression and identified for the first time the existence of methylated CpG loci at the PD-L1 promoter [194]. The authors further stratified melanomas into 2 groups based on PD-L1 status and observed that PD-L1 hypomethylation is associated with increased PD-L1 expression and superior OS in CM patients regardless of the diagnosed stages [194]. Moreover, studies on melanoma cells showed that treatment with the hypomethylating agent 5-azacytidine can orchestrate transcriptional derepression in hypermethylated PD-L1 tumors, leading to amplification of PD-L1 expression [194]. Finally, some other studies confirmed the epigenetic regulation of immune checkpoint gene LAG3 via DNA methylation in CM [195]. LAG3 is a molecule involved in blocking tumor cell proliferation and regulating the production of IFN-γ and TNFα cytokines. Interestingly, it has been shown that LAG3 promoter hypomethylation positively associates with increased levels of tumor-infiltrating immune cells and better PFS in CM patients [195].

Given that anti-PD-L1 and anti-PD-1 antibodies are the most extensively used immunotherapies in the clinical setting, and PD-L1 hypermethylation renders CM resistant to

ICIs, applying DNMTi treatments appears a tempting strategy to reverse CM immunotherapy resistance. Interestingly, it has been noted that DNMTi hs the ability to activate endogenous retroviruses (ERVs) and virus defense-related pathways in melanoma cells [196]. In the TCGA, the expression levels of viral defense genes may help in stratifying primary samples from multiple tumors, including CM, into two risk groups where a high defense signature positively associates with improved OS and more durable clinical response. Moreover, combining anti-CTLA-4 with low doses of DNMTi proved to be an effective strategy in augmenting the immunotherapy efficiency in a mouse melanoma model [196]. Therefore, the use of immune checkpoint inhibitors in conjunction with DNMTi may be a promising strategy for maximizing the therapeutic benefit in CM patients. Certainly, we will soon find out more about the efficacy and safety of combining DNMTi with immunotherapy, given that ongoing clinical trials are studying the oral use of azacitidine with pembrolizumab in patients with metastatic CM (NCT02816021).

#### 5.2.2. Histone-Modifying Enzymes and PTMs Involved in Immunotherapy Resistance

To date, information on histone enzymes and PTMs involved in immunotherapy resistance are even vaguer than in the case of epigenetic regulation by DNA methylation. Several studies have suggested a mechanistic link between EZH2 activity and resistance to immunotherapy. For example, Zingg at al. observed that anti-CTLA-4 or IL-2 immunotherapy leads to TNFα amplification and T cell infiltration, resulting in EZH2 overexpression and loss of tumor control in melanoma mouse models [197]. Mechanistically, EZH2 catalyzes the deposition of H3K27me3 marks and the suppression of a plethora of immune-related genes. Notably, EZH2 inactivation reversed the drug-resistant phenotype and amplified the effects of anti-CTLA-4 and IL-2 immunotherapy in melanoma mouse models, thus blocking CM growth and dissemination [197]. Therefore, in this study, Zingg at al. have demonstrated not only that EZH2 expression is a valuable biomarker for monitoring the response to immunotherapy but can also be exploited as a therapeutic target to restore and enhance the effects of immunotherapy.

Additional studies have reinforced that the involvement of EZH2 in resistance to CM immunotherapy could be much broader. Tiffen at al. analyzed 471 cases of CM in the TCGA and found that 20% of patients displayed copy number amplifications and mRNA upregulation, along with activating mutations in EZH2 [198]. RNAseq analysis further showed that these alterations correlated with DNA hypermethylation and downregulation of certain genes involved in tumor suppression, antigen processing, and presentation pathways. Treatment with the EZH2 inhibitor GSK126 reversed the transcriptional silencing driven by EZH2 alterations in CM cells, suggesting that EZH2 inhibition is a promising pharmacological strategy for improving the therapeutic response in CM [198]. From these observations, it appears that there may be a close link between EZH2 and the activity of DNMTs in regulating tumor biological properties, including the response to immunotherapy in CM. It is well documented that the ATRX\_DNMT3\_DNMT3L (ADD) domain of DNMT3A may interact with several epigenetic players, such as SUV39H1 methyltransferases, HDAC1, and EZH2, among others [199]. Moreover, it was postulated that the activity of DNMTs is supported by EZH2, a well-known target of PI3K/Akt signaling, and that they cooperate in cancer pathogenesis [200]. In support of this idea is the observation that EZH2 and DNMTs are regulated by similar upstream signaling cascades, such as MEK/ERK and PI3K/Akt, and transcription factors such as NF-kB2 [57]. Last but not least, it appears that EZH2 can pre-mark genes for DNMTs methylation [201]. Therefore, although still in its infancy, the study of epigenetic mechanisms concerning the response to immunotherapy is expected to not only guide and revolutionize the treatment of refractory patients, but also to stratify them into risk groups to provide personalized therapeutic solutions.
