**About the Editors**

#### **Xianda Zhao**

Xianda Zhao is an M.D. Ph.D. and researcher in the fields of gastrointestinal cancer pathogenesis and novel treatment development at the University of Minnesota. Dr. Zhao has been studying gastric cancer, pancreatic cancer, and colorectal cancer for over ten years. He is a pioneer of advanced gastrointestinal cancer models. His current research focuses on tumor-extracellular vesicles mediated immune responses and extracellular vesicles-based treatment for colorectal cancers. Dr. Zhao has published over 30 full-length original research articles and review articles in top-ranking journals. He has also authored one book chapter focusing on cancer immunotherapy resistance. Dr. Zhao serves as an ad hoc reviewer for multiple journals.

#### **Subbaya Subramanian**

Dr. Subbaya Subramanian is currently an Associate Professor in the Department of Surgery at the University of Minnesota, United States. As a faculty at the University of Minnesota, Dr. Subramanian has established an internationally recognized cancer research program focused on deciphering the molecular mechanisms of antitumor immune evasion in cancer. His current research focuses on understanding the interactions between cancer and the host immune system and how cancer cells manipulate the anti-tumor immune response. Dr. Subramanian has authored over 100 peer-reviewed publications and has written numerous book chapters and review articles. He has delivered over 80 national and international invited lectures/seminars and received the American Cancer Society Research Scholar award in 2013. He serves as the section Editor-in-Chief of the journal *Vaccines* and the Associate section Editor-in-Chief of *Cancers*. Dr. Subramanian also serves on the Editorial Board of several journals.

## *Editorial* **Cancer Immunology and Immunotherapies: Mechanisms That Affect Antitumor Immune Response and Treatment Resistance**

**Xianda Zhao 1,\* and Subbaya Subramanian 1,2,3,\***


The past decade has seen immunotherapy rise to the forefront of cancer treatment. This Special Issue of *Cancers* aims to elaborate on the latest developments, cutting-edge technologies, and prospects in cancer immunology and immunotherapy. Seventeen exceptional studies, including original contributions and review articles, written by international scientists and physicians, primarily concerning the fields of tumor biology, cancer immunology, therapeutics, and drug development comprise the main body of this special issue.

Over the last few years, an increasing understanding has emerged on molecular mechanisms that regulate the anti-tumor immune response and an exponential increase in the use of novel cancer immunotherapies in various cancer types. The field of Cancer immunology and Immunotherapies presents promising therapeutic opportunities for developing novel cancer treatments and improving patient survival outcomes. Chemotherapy is still used as a primary method for treatment, and the standard of care for many cancer types is relatively unselective and presents with the rapid development of treatment resistance. In contrast, cancer immunotherapies stimulate the antitumor immune response via the activation of lymphocytes that can recognize neoantigens, resulting in durable treatment response.

A successful antitumor immune response involves interactions between various cell types that coordinately function to prevent tumor cell proliferation or to effectively eradicate tumor cells. A coordinated functioning of the lymphoid and myeloid lineage cells is critical for killing tumor cells, and is performed by enhancing the activity of cytotoxic cells. Myeloid lineage cells, such as dendritic cells, provide tumor antigens to T cells and secrete cytokines that regulate the activation and function of cytotoxic cells. Despite the demonstrated successes of cancer immunotherapy, most patients do not respond, and the development of resistance has occurred in patients who initially respond to immunotherapies. Recent studies have uncovered novel immune escape mechanisms that affect immune cell infiltration, poor antigen presentation, and tumor intrinsic silencing of the immune response via cytokines and the release of immune suppressive exosomes [1]. Additional mechanisms of antitumor immune escape and immunotherapy resistance are continuously being discovered [2–4].

Based on these factors, significant attention has been directed towards the recent advances in cancer immunology [5–10]. In the past decades, the discovery of Programmed cell death protein 1 (PD-1) and the Cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4) has helped to develop immune checkpoint blockade therapies. The articles by Yuan et al. [5] and Sobhani et al. [6] provide an overview and include recent findings on PD-1 and CTLA-4. The review article by Mehdi et al. [7] focuses on the role of methylation in manipulating cancer immunity. In addition to these general cancer immunology topics, reviews by Krishnamurthy et al. [8], Zheng et al. [9], and Marseglia et al. [10] summarize immune regulation in specific cancer types, such as hepatocellular carcinoma, triple-negative breast cancer, and uveal melanoma.

The second series of articles mainly presents original work deciphering the novel regulatory mechanisms of cancer immunity. For the first time, our group (Wangmo et al. [11])

**Citation:** Zhao, X.; Subramanian, S. Cancer Immunology and Immunotherapies: Mechanisms That Affect Antitumor Immune Response and Treatment Resistance. *Cancers* **2021**, *13*, 5655. https://doi.org/ 10.3390/cancers13225655

Received: 9 November 2021 Accepted: 10 November 2021 Published: 12 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

reported that Atypical Chemokine Receptor 4 (ACKR4) determines the migration of dendritic cells from tumor tissue to the tumor-draining lymph nodes. The loss of ACKR4 expression in tumor cells can affect the migration of dendritic cells and their retention in the tumor microenvironment, impairing T-cell priming in tumor-draining lymph nodes. This finding uncovers a novel mechanism that regulates dendritic cells' migration from the tumor tissue, a critical factor in antigen presentation and in antitumor immune responses. Liang et al. [12] further contribute to the body of research regarding antigen-presenting cells. The authors performed an in-depth analysis of antigen-presenting cells in the human colorectal cancer microenvironment. Interestingly, they observed that antigen-presenting cells within distinct intratumoral and colonic milieus showed different functional statuses but were similarly responsive to induced T-cell activation. The third article in this section focuses on the bystander T-cells in cancers. In a hybrid study of bioinformatics and laboratory analyses, Gokuldass et al. [13] revealed a higher proportion of bystander CD8<sup>+</sup> T cells in non-melanoma cancers than in melanoma cancers. This observation helps to establish a new theory to explain the different immune strengths of various tumors. In the context of innate immunity, Kaur et al. [14] reported on the function of CD16 receptors in both direct cytotoxicity and antibody-dependent cell cytotoxicity, making the use of these receptors as a cancer treatment seem promising.

The overarching objective of studying tumor immunity is to develop the next-generation cancer immunotherapies. In the third series of articles, several novel cancer immunotherapy strategies are proposed. Two original research articles from Jiang's group [15,16] provide modified CAR T Cell therapies to treat malignant B-cell neoplasms and prostate cancer. Their modified CAR T cells are better directed to kill malignant B-cells, while sparing the CD19+HLA-C1<sup>+</sup> healthy B Cells. The next study by Hsu et al. [17] developed a recombinant fusion IL15 protein composed of human IL15 (hIL15) and albumin-binding domain (hIL15-ABD) which has been successfully tested with anti-PD-L1 on CT26 murine colon cancer and B16-F10 murine melanoma models. Horn et al. [18] also reported on the use of IL15 as an agonist adjuvant for other cancer immunotherapies. Utilizing colon and mammary carcinoma models, the study showed that a recombinant adenovirus-based vaccine, targeting tumor-associated antigens with an IL-15 superagonist adjuvant is effective when combined with other immunotherapy regimens. This study also validated the idea that providing tumor-associated antigens as a vaccine helps to overcome immune checkpoint blockade resistance. Another feature in this issue is that we include a report on a new method called the 'chemo-enzymatic conjugation approach' (Bai et al. [19]) to generate bispecific antibodies (BiFab). Using this method, the authors produced BiFabHer2/CD3 and BiFabCD20/CD3 to conjugate both the target and effector cells (T-cells). These BiFabs demonstrated a strong considerable effect for inducing T-cell activation and killing target cancer cells upon conjugation. The BiFabCD20/CD3 also showed anti-tumor activity in vivo.

The findings of Benajiba et al. [20] and Zwart et al. [21] highlight clinical observations relevant to cancer and immunology. Disseminated Kaposi's sarcoma is usually treated by interferons, which is a type of immunotherapy. Benajiba et al. [20] performed a retrospective cohort study to evaluate global disease evolution and to identify the risk factors for systemic treatment initiation, including the use of interferons. They found that 41.2% of classic/endemic Kaposi's sarcoma patients require systemic treatment. They also reported that the mean treatment-free time during the first five years following interferon is similar to that of chemotherapy. Lastly, Zwart et al. [21] contribute through a meta-analysis on immunosuppressive therapy after solid organ transplantation and on the development of cancers. Interestingly, the meta-analysis indicated that patients receiving cyclosporine A and Azathioprine after a solid organ transplant are at a higher risk than patients receiving other immunosuppressive drugs of developing certain types of cancers.

In conclusion, the original research articles and reviews included in this special issue ensure that the key aspects of the next generation of cancer immunology and immunotherapy have been covered. We hope that the novel findings in these articles will inform the readers and provide useful references for developing next-generation cancer immunotherapies.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Review* **Recent Advancements in the Mechanisms Underlying Resistance to PD-1/PD-L1 Blockade Immunotherapy**

**Yu Yuan <sup>1</sup> , Abdalla Adam <sup>1</sup> , Chen Zhao <sup>2</sup> and Honglei Chen 1,\***


**Simple Summary:** Immune checkpoint blockade targeting PD-1/PD-L1 has a promising therapeutic efficacy in different tumors, but a significant percentage of patients cannot benefit from this therapy due to primary and acquired resistance during treatment. This review summarizes the recent findings of PD-L1 role in resistance to therapies through the PD-1/PD-L1 pathway and other correlating signaling pathways. A special focus will be given to the key mechanisms underlying resistance to the PD-1/PD-L1 blockade in cancer immunotherapy. Furthermore, we also discuss the promising combination of therapeutic strategies for patients resistant to the PD-1/PD-L1 blockade in order to enhance the efficacy of immune checkpoint inhibitors.

**Abstract:** Release of immunoreactive negative regulatory factors such as immune checkpoint limits antitumor responses. PD-L1 as a significant immunosuppressive factor has been involved in resistance to therapies such as chemotherapy and target therapy in various cancers. Via interacting with PD-1, PD-L1 can regulate other factors or lead to immune evasion of cancer cells. Besides, immune checkpoint blockade targeting PD-1/PD-L1 has promising therapeutic efficacy in the different tumors, but a significant percentage of patients cannot benefit from this therapy due to primary and acquired resistance during treatment. In this review, we described the utility of PD-L1 expression levels for predicting poor prognosis in some tumors and present evidence for a role of PD-L1 in resistance to therapies through PD-1/PD-L1 pathway and other correlating signaling pathways. Afterwards, we elaborate the key mechanisms underlying resistance to PD-1/PD-L1 blockade in cancer immunotherapy. Furthermore, promising combination of therapeutic strategies for patients resistant to PD-1/PD-L1 blockade therapy or other therapies associated with PD-L1 expression was also summarized.

**Keywords:** PD-L1; resistance; immune checkpoints; immunotherapy

#### **1. Introduction**

T-cell activation and proliferation induced by antigens is regulated by expression of both co-stimulatory and co-inhibitory receptors and their ligands [1]. Inhibitory pathways in the immune system can prevent autoimmunity through maintaining self-tolerance and regulating immunity [2]. While in tumors inhibitory pathways known as "checkpoints" can evade immune surveillance. Programmed cell death -1(PD-1) interacting with its corresponding ligand PD-L1 leads to immune suppression via preventing the T-cell activation in the tumor [3]. PD-1 is expressed on activated CD8<sup>+</sup> T-cells as well as B cells and natural killer cells, and inhibits T-cell receptor (TCR) signaling and CD28 co-stimulation under chronic antigen exposure. As ligands of PD-1, PD-L2 is primarily expressed on antigen-presenting cells (APC) while PD-L1 is expressed on various types of cells including tumor cells and immune cells. Evidence of PD-L1 expression increase and spontaneous

**Citation:** Yuan, Y.; Adam, A.; Zhao, C.; Chen, H. Recent Advancements in the Mechanisms Underlying Resistance to PD-1/PD-L1 Blockade Immunotherapy. *Cancers* **2021**, *13*, 663. https://doi.org/10.3390/ cancers13040663

Academic Editors: Subree Subramanian, Xianda Zhao and Roberta Alfieri Received: 21 January 2021 Accepted: 3 February 2021 Published: 7 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

immune resistance is proved in several types of human cancers [4]. Besides, predictive and prognostic value of PD-L1 immunohistochemical expression has been reported in certain cancers. Moreover, PD-L1 as an inhibitory factor is also involved in other signaling pathways underlying mechanisms in resistance to tyrosine kinase inhibitors (TKIs).

Immunotherapy identified as the most promising approach in cancer treatment compared with chemotherapy and targeted therapy, immune checkpoint inhibitors have reported higher rates of response, remission, and better overall survival rates in a variety of tumors [5]. Immunotherapy has received the US Food and Drug Administration (FDA) approval for 57 indications in 17 solid tumors in less than 10 years, while over 80% are PD-1/PD-L1-targeted antibodies. Beneficial function of the PD-1/PD-L1 axis blockade is confirmed in treating many different types of cancers such as non-small cell lung cancer (NSCLC), melanoma and bladder carcinoma [6,7]. So far, six immune checkpoint inhibitors targeting PD-1/PD-L1 have been approved by the FDA for the first and second line of patients with non-small cell lung cancer including monoclonal antibodies (mAb) pembrolizumab, nivolumab and cemiplimab targeting PD-1 and mAb atezolizumab, avelumab and durvalumab targeting PD-L1. However, limited efficacy has been reported in PD-1/PD-L1 blockade therapy which rarely exceeds 40% in most cancer types and a large number of patients show partial responsiveness [8,9]. Even if there is a consistent rate of initial responses, the majority of patients develop therapeutic resistance and disease progression [10,11]. Focusing on PD-L1, we described all these concepts in this review including its predictive and prognostic value, immune resistance induced by PD-L1 and key mechanisms underlying resistance to PD-1/PD-L1 blockade therapy.

#### **2. The Expression of PD-L1 Levels Predicting Resistance and Poor Prognosis**

PD-L1 expression is increased in many types of human cancers and is regarded as a predictive and prognostic marker in cancer tissues. Prior data have demonstrated that PD-L1 expression is upregulated in cisplatin-resistant lung cancer cells compared with parent cells [12–14]. Resistance to epigenetic therapy is associated with enhanced PD-L1 expression in myeloid malignancies [15]. For example, 7 myelodysplastic syndrome and 6 acute myeloid leukemia patients received treatments with either azacytidine (Aza) or combined Aza and the histone deacetylase inhibitor LBH-589 to investigate the PD-L1 expression levels. Non-responders showed a more than two-fold increase in PD-L1 expression after treatment commenced, and except for two patients, none of the responders demonstrated increased expression of PD-L1.

PD-L1 expression is correlated with poor prognosis in different cancers. In chemotherapy and radiotherapy-treated patients with head and neck squamous cell carcinoma (HNSCC), high PD-L1 mRNA (>125 FPKM) from The Cancer Genome Atlas database had significantly reduced the 5-year survival rate [16]. Other data regarded PD-L1 as a potential biomarker for radiation therapy failure of HNSCC [17]. Following radiotherapy, a panel of radiation-resistant human papilloma virus (HPV)-negative HNSCC cell lines exhibited increased expression of PD-L1, three cohorts of HPV-negative HNSCC tumors with high expression of PD-L1 had much higher failure rates compared to the PD-L1-low expression group. Similar results have been reported in metastatic melanoma patients (MMP) [18]. Forty six and thirty four BRAFi-treated MMP harboring mutant BRAFV600 received vemurafenib and dabrafenib respectively. Patients with PD-L1 expression and an absence of tumor-infiltrating immune cells (TIMC) are related to shorter progression-free-survival compared to those with TIMC and absence of PD-L1. This study also identified PD-L1 overexpression and loss of TIMC as independent prognostic factors for melanoma-specific survival.

Interestingly, an experiment involving 18 patients with epidermal growth factor receptor (EGFR)-mutant NSCLC investigated the change of PD-L1 expression following gefitinib. A proportion of 38.9% (7/18) of NSCLC patients had a significant increase in the median H-score (marked as group A) of PD-L1, while the rest (61.1%) did not vary (group B). Besides, MET positivity by immunohistochemistry in biopsies is significantly correlated with group A. The results described a marked increase in expression of PD-L1 in tumor cells of a subset of patients after gefitinib treatment. Though EGFR-mutated NSCLC is prone to express less PD-L1 than wild type [19]. Similar results in several studies indicated that PD-L1 expression as a biomarker predicts resistance and poor prognosis after gefitinib treatment, rebiopsy should be considered [20]. Nevertheless, combination therapy with Durvalumab targeting PD-L1 and gefitinib has been proved to be more toxic and does not demonstrate a significant augmentation in progression-free survival (PFS) [21]. As a crucial factor predicting resistance and poor prognosis, PD-L1 has absolutely specific mechanisms for leading to resistance.

#### **3. PD-L1-Induced Resistance**

PD-L1 as an inhibitor in the immune system that induces immune resistance through interacting with its ligand PD-1. Besides, it is also involved in other signaling pathways generating resistance to TKIs.

#### *3.1. PD-L1-Mediated Immune Resistance*

In certain cancers, efficacy of antitumor treatment has always been found to be limited, due to the activation of immune checkpoints such as PD-1 and PD-L1. Once recognizing the tumor antigen, T-cells produce an anti-tumor immune response, which eventually leads to PD-1 lymphocyte expression and interferon release. To evade this immune attack, PD-L1 expression is adaptively upregulated by cancer cells and other inflammatory cells in the tumor microenvironment (TME) [22]. IFN-γ is secreted by tumor-infiltrating lymphocytes (TILs) and induces PD-L1 expression in the TME, thus T-cell cytotoxic function is impaired through the interaction of PD-L1 and PD-1. A similar pattern has been observed in other cancers including gastric cancer [23]. Fractionated radiation therapy can lead to increased tumor cell expression of PD-L1 in response to CD8<sup>+</sup> T-cell production of IFN-γ [24]. In HPV-HNSCC, which is highly infiltrated by lymphocytes, IFN-γ-induced PD-L1 on tumor cells and CD68<sup>+</sup> tumor-associated macrophages (TAMs) and highly expressed PD-L1 by CTLs, are found located at the same site [1].

In prior studies, PD-L1 expression is also upregulated followed by drug treatment and mediates an immune resistance. For example, in glioblastoma a compensatory recruitment of tumor-infiltrating myeloid cells elicited by antitumor immune response induced by dendritic cell (DC) vaccination contributed to the majority of PD-L1 expression [25]. Placenta-specific protein 8 (PLAC8) as an oncogene promoting cancer growth and progression is abnormally upregulated in gallbladder carcinoma. Overexpression of PLAC8 conferred resistance to gemcitabine and liplatin (OXA), mainstays of chemotherapy by upregulating PD-L1 expression [26]. 5-Fluorouracil selectively depletes myeloid-derived suppressor cells (MDSCs) and OXA triggers an immunogenic form of tumor cell death. A combined chemotherapy Folfox, 5-Fluorouracil plus OXA, has routinely been regarded as a first line of treatment for advanced colorectal cancer. However, Folfox up-regulates high expression of PD-1 on activated CD8<sup>+</sup> TILs, and induces CD8<sup>+</sup> T-cells to secret IFN-γ which upregulates PD-L1 expression on tumor cells [27]. CD40 stimulation on APC directly activates CTLs without the help of CD4<sup>+</sup> T-cells. Agonistic anti-CD40 antibodies induce antitumor responses and upregulation of PD-L1 on tumor-infiltrating monocytes and macrophages, which are extremely dependent on T-cells and IFN-γ [28]. When co-cultured with human PBMC, trastuzumab, the anti-human epidermal growth factor receptor-2 (HER2) antibody, is shown to upregulate PD-L1 in HER2-overexpressing breast cancer cells via mediating stimulation of IFN-γ secretion on immune cells [29]. Inhibitors of mTOR approved by the Food and Drug Administration to treat advanced metastatic renal cancers and enhance nuclear translocation of transcription factor EB, was bound to PD-L1 promoter and thereby led to increased PD-L1 expression [30].

#### *3.2. Signaling Pathways and Factors Involved in PD-L1-Induced Resistance*

Despite immune resistance, PD-L1 has generated resistance to TKIs in certain cancers. Possible mechanisms by which PD-L1 induced acquired resistance through upregulating Yes-associated protein1 (YAP1), [31] Bcl-2-associated athanogene-1 (BAG-1), [32] and DNA methyltransferase 1 (DNMT1), [33] and generated primary resistance by inducing epithelialto-mesenchymal transition (EMT) have been reported [34] (Figure 1).

α α κ ( β **Figure 1.** Signaling pathways and factors involved in programmed cell death ligand-1(PD-L1)-induced resistance. (1) PD-L1 expression induced by epidermal growth factor receptor (EGFR) mutation activation via extracellular single-regulated kinase (ERK) signaling, indirectly promotes expression of Bcl-2-associated athanogene-1 (BAG-1), the EGFR/ERK/PD-L1/BAG-1 feedback loop reaches the reactivation of ERK signaling which promotes Bcl-2-interacting mediator of cell death (BIM) phosphorylation to help cells escape from apoptosis. (2) PD-L1-induced hypoxia-inducible factor-1α (HIF-1α) expression is stimulated by reactive oxygen species (ROS), hypoxia increases YAP-1 expression which confers resistance via a YAP1/EGFR/ERK/NF-κB loop. (3) PD-L1 regulates DNA methyltransferase 1(DNMT1) via Signal transducer and activator of transcription 3 (STAT3) signaling and thus induces DNMT1-dependent DNA hypomethylation which promotes cancer development. (4) Activation of transforming growth factor-beta (TGF-β)/Smad pathway induced by PD-L1 is crucial in epithelial-to-mesenchymal transition (EMT) expression which leads to resistance to TKIs.

β Activation of MEK/extracellular single-regulated kinase (ERK) signaling furthers phosphorylation and ubiquitination of the Bcl-2-interacting mediator of cell death (BIM), a BH-3-only protein, thereby preventing cells from apoptosis [35]. Resistance to TKI in NSCLC generally occurs through reactivating ERK signaling [36]. EGFR mutation activation induces expression of PD-L1 in NSCLC cells via ERK-signaling [37]. Once triggered by ERK signaling, phosphorylated C/EBPβ induced by PD-L1 can enhance binding to the BAG-1 promoter, thus promoting BAG-1 expression. The PD-L1/BAG-1 axis confers TKI resistance through persistent activated ERK signaling via the EGFR/ERK/PD-L1/BAG-1 feedback loop [32]. Thus combining treatment with TKIs and anti-PD-L1 therapy may provide a promising strategy for tumors with a high expression of PD-L1 and BAG-1, though this has not been researched yet.

YAP1 is another factor known to confer EGFR-TKI resistance in lung cancer cells [38]. Distinct experiments utilizing reactive oxygen species (ROS) scavengers and inducers demonstrated a concomitant change of expression of PD-L1 and hypoxia-inducible factor-1α (HIF-1α), YAP1 [31]. While prior reports described that PD-L1-induced HIF-1α is stimulated by the generation of ROS [39,40], hypoxia promotes formation of YAP1 and HIF-1α complex via regulating SIAH2 ubiquitin E3 ligase and increases YAP1 gene expression [41,42]. TKI resistance may be conferred by PD-L1/ROS/HIF-1α/YAP1 axis and

a YAP1/EGFR/ERK/NF-κB loop [31]. Markedly high expression of YAP and PD-L1 are observed in EGFR-TKI-resistant cells in another study, and they demonstrate a positively related change in expression when given a knockdown of YAP [43]. Thereby, giving an anti-PD-L1 or anti-YAP1 may overcome the EGFR-TKI resistance.

The PD-L1/DNMT1 axis is also a critical mechanism leading to acquired resistance [33]. DNMT1, as a member of the DNA methyltransferase family, maintains the DNA methylation pattern [44]. Signal transducer and activator of transcription 3 (STAT3), a wellcharacterized transcription factor that binds to DNMT1 promoter and positively regulates transcription of DNMT1 [45], since phosphorylated STAT3 induces transcriptional activation via binding with specific DNA elements. PD-L1 regulates DNMT1 through the STAT3-signaling pathway and induces DNMT1-dependent DNA hypomethylation to promote development of cancers [46], thereby resulting in acquired resistance [33]. Currently, a combination therapy with oxaliplatin and decitabine inhibiting DNA demethylation was proved to have a synergistic effect in enhancing anti-PD-L1 therapeutic efficacy in colorectal cancer [47].

The transforming growth factor-beta (TGF-β)/Smad signaling pathway plays a role in PD-L1-induced primary resistance to EGFR-TKIs [34]. EMT can decrease efficacy of drug treatment in NSCLC [48,49]. PD-L1 upregulates phosphorylation of Smad3, which significantly participates in the transcriptional regulation mediated by TGF-β1 [50], and the TGF-β/Smad-signaling pathway has been reported to be crucial in EMT progression [51]. The mechanism of primary resistance to EGFR-TKIs in EGFR-mutant NSCLC may confer through the PD-L1/TGF-β/Smad/EMT axis [34]. In addition, in Kirsten rat sarcoma viral oncogene homolog (KRAS)-mutant NSCLC, KRAS G12 mutation is reported to promote PD-L1 expression via a TGF-β/EMT-signaling pathway [52]. Apparently, PD-L1 expression plays a key role in poor prognosis and resistance after treatment in several types of cancers, thereby adding an anti-PD-1 or anti-PD-L1 therapy may improve the efficacy and become a promising therapeutic strategy.

#### **4. Key Mechanisms Underlying Resistance to PD-1/PD-L1 Blockade**

PD-1/PD-L1 blockade therapy has been approved as a significantly helpful treatment in certain cancers, a problem of its limited efficacy has occurred and the targeting solution is urgently discussed and provided. Focusing on PD-L1, we described key mechanisms underlying resistance to PD-1/PD-L1. Surprisingly, abnormally upregulated PD-L1 expression and a lack of PD-L1 can both lead to inefficacy of PD-1/PD-L1 inhibitors (Figure 2).

#### *4.1. Aberrant PD-L1 Expression*

PD-L1 is generally regulated by tumor cells in two ways: the first is innate immune resistance in which constitutive oncogenic signaling is correlated with PD-L1 expression, the second is an adaptive immune resistance through which IFN-γ produced by TILs induces PD-L1 expression.

K-ras mutation as a common oncogenic driver in the lung adenocarcinoma (LUAD) and upregulates PD-L1 through p-ERK instead of p-AKT signaling [53]. Different subgroups of KRAS-mutant LUAD are dependent on STK11/LKB1 or TP53 mutations, and alterations of the former has been confirmed as a major factor that leads to primary resistance to PD-1 blockade [54]. Besides, EGFR-mutant or ALK-rearranged patients had a PD-L1 tumor proportion score of ≥50% and turned out not to respond to PD-1/PD-L1 inhibitors [55].

The transcription factor Yin Yang 1 (YY1); a major regulator reported participating in various pathways, is involved in cell growth, survival and metastasis. YY1 upregulates PD-L1 expression on tumor cells via signaling pathways, including p53, STAT3, NF-κB and PI3K/AKT/mTOR [56]. PD-L1v242 and PD-L1v229, two secreted PD-L1 C-terminal splicing variants, could capture the aPD-L1 antibody and function as a "decoy" to prevent antibodies from binding to PD-L1 [57].

α γ **Figure 2.** Key mechanisms underlying resistance to PD-L1 (1). The transcription factor Yin Yang 1 (YY1)-induced upregulation of PD-L1 expression triggers NOD-, LRR- and pyrin domain-containing 3 (NLRP3) inflammasome to promote tumor Wnt5α expression via HSP70-TLR4 signaling, and non-canonical WNT ligands activate the YAP pathway to induce chemokine (C-X-C motif) receptor 2 (CXCR2) ligands, while granulocytic subset of myeloid-derived suppressor cells (PMN-MDSCs) relied on CXCR2 to suppress T-cell function. (2) Loss-of-function mutations in JAK1/2 leads to the paucity of PD-L1 expression. (3) Tumor-suppressing microenvironment. Tumor-associated macrophages (TAMs) promote tumor progression, while Indole 2,3-dioxygenase (IDO) generated by tumors enhances Tregs and MDSCs activity, which suppress immunity. (4) Activation of alternative immune checkpoints. T-cell immunoglobulin mucin 3 (TIM-3) and Lymphocyte activation gene-3 (LAG-3) produced by T-cells impair generation of IFN-γ, which activates T-cells. CTLA-4 demonstrates a higher affinity and avidity in conjunction with CD80 and CD86 than CD28 to antagonize costimulation. VISTA is found to be related to MDSC mainly derived CD33 expression and HHLA2 decreases T-cell proliferation.

γ Besides, a tumor-intrinsic signaling pathway involved with NLRP3 inflammasome in response to upregulated expression of PD-L1 was found to drive adaptive resistance to anti-PD-1 antibody immunotherapy [58]. NLRP3 inflammasome triggered by PD-L1 induces tumor Wnt5α expression via HSP70-TLR4 signaling, while non-canonical WNT ligands promote production of CXCR2 ligands through the activated YAP pathway [59,60]. CXCR2 relied migration and recruitment of a granulocytic subset of MDSCs (PMN-MDSCs) play a role in suppressing CD8<sup>+</sup> T-cell infiltration and function, therefore leading to adaptive resistance [61,62].

≥ Previous study showed that tumors can be divided into four categories according to positive/negative tumor PD-L1 expression and presence/absence of TILs. For instance, patients with PD-L1 positive and TILs indicate adaptive immune resistance and those with PD-L1 negative and without TILs show immune ignorance [63]. Among these four types, type I with PD-L1 positive and TILs is the most likely to respond to PD-1/PD-L1 blockade therapy, whilst other types may show unresponsiveness to this monotherapy [64].

κ

#### *4.2. Paucity of PD-L1 Expression*

The interaction between PD-L1 and its receptor PD-1 leads to immune escape and inhibits T-cell function and blockade of PD-L1 and PD-1 enhances the antitumor immunity in several cancers. However, the expression of PD-L1 or PD-1 is a prerequisite for the therapeutic efficacy. Evidence of the relation of rare PD-L1 expression and poorer responses to PD-1 blockade has been proved in prostate cancer [65]. DNA hypomethylating agent upregulate PD-L1 gene expression [66]. Anti-PD-1 therapy curbs the expression of PD-L1 through either eliminating the tumor cells that overexpress PD-L1 and possess a hypomethylated PD-L1 promoter or switching off the PD-L1 expression through epigenetic modulation, therefore leading to resistance [67]. Loss-of-function mutations in JAK1/2 can lead to primary resistance to anti-PD-1 therapy due to the inability to respond to IFN-γ for a lack of PD-L1 expressions [68]. Despite the effect of aberrant PD-L1 expression, an abnormal process from antigen expression to T-cell activation can result in resistance to PD-1/PD-L1 inhibitors. Moreover, a recent study demonstrated that PD-L1 expression is enhanced via nicotinamide adenine dinucleotide (NAD<sup>+</sup> ) metabolism, in which nicotinamide phosphoribosyltransferase (NAMPT) functions as the rate-limiting enzyme [69]. NAMPT increases PD-L1 expression induced by IFN-γ and leads to immune escape in tumors with the help of CD8<sup>+</sup> T-cells. Thus NAD<sup>+</sup> metabolism is a promising strategy for resistance to anti-PD-L1 therapy [69].

#### *4.3. Aberrant Antigen Expression, Presentation and Recognition*

Tumors with a higher tumor mutation burden (TMB) are likely to have more neoantigens, which can be recognized by the immune system as "non-self" in response to checkpoint inhibition. In Naiyer's study, the result of the treatment of PD-1 targeting antibody pembrolizumab in NSCLC described that a higher burden of nonsynonymous tumors is correlated with a better response and PFS [70]. Besides, strong immunogenicity and extensive expression of immune checkpoint ligands make the microsatellite instability subtype more susceptible to immunotherapeutic methods, for example, with anti-PD-L1 and anti-cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) antibodies [71]. Tumors with defective mismatch repair possess more DNA mutations and show an improved responsiveness to anti-PD-1 therapy [72]. In short, a low mutational burden, microsatellite stability and efficient DNA repair mechanisms are involved in innate resistance to immune-checkpoint blockade therapy. Moreover, evolution of neoantigen loss can produce an acquired resistance [73]. A study also demonstrates that deficiency of heterogeneity in HLA genes is observed in cancer development, a high level of HLA loss results in acquired resistance during immunotherapy [74].

Resistance to immune checkpoint blockades also involves impaired DC maturation, which is an essential process in T-cell activation, through it is displayed in various costimulatory factors expression including MHC class I/II, CD80, CD86 and CD40 [75]. IL37b decreases CD80 and CD86 expression through the ERK/S6K/NF-κB axis and suppresses DC maturation [76]. A transcription factor STAT3 that facilitates tumor growth and metastasis leads to the induction of other immunosuppressive factors that possess a suppressive function on DC maturation, including IL10, Tregs and TGF-β [77–80].

Despite inducing PD-L1, IFNs have been reported to (re-)activate T-cells to control the tumor development via advancing DC cross-priming [81–83]. It is well-known that CTLs recognize MHC class I-presented peptide antigens on the surface of tumor cells. Heterozygous mutations, deletions or deficiency in β-2-microglobulin (β2M); a crucial factor in MHC class I antigen presentation, generally reduces antigen recognition by antitumor CD8<sup>+</sup> T-cells and mutation of β2M gene leads to resistance to anti-PD-1 therapy [84,85]. IFN-γ can induce tumor cells to express MHC class I molecules, significantly promoting CTL differentiation and enhancing apoptosis. Mutations or loss of IFN-γ pathway-related proteins on tumor cells (such as STATs, IFN-γ receptor chain JAK1 and JAK2) can cause escape from immune recognition and resistance to immune checkpoint inhibitions [68,86].

#### *4.4. Aberrant Immunity of T-Cells*

Despite normal antigen expression, presentation, recognition and successfully activated T-cells, resistance to the PD-1/PD-L1 blockade inhibitors may occur owing to the T-cell itself. The aberrant immunity of T-cells include insufficient T lymphocytes infiltration, dysfunction of T-cell and exhausted T-cells.

#### 4.4.1. Insufficient T Lymphocytes Infiltration

Despite the expression of PD-L1, a lack of T lymphocyte infiltration can cause unresponsiveness to anti-PD-L1 therapy. A crucial prerequisite for the therapeutic efficacy is the existing and tumor-infiltrated anti-tumor CTLs [87]. LIGHT, a member of the tumor necrosis factor superfamily, may activate lymphotoxin β-receptor signaling, resulting in the generation of chemokines that recruit a huge number of T-cells [88].

The PI3K-AKT-mTOR pathway, a crucial oncogenic signaling pathway, is involved in a multitude of cellular processes including cell survival, proliferation, and differentiation. PTEN, a lipid phosphatase, inhibits the PI3K signaling activity which activates the pathway. Loss of PTEN has been reported to reduce CD8<sup>+</sup> T-cells infiltrating into tumors and lead to resistance to PD-1 blockade therapy. A selective PI3Kβ inhibitor treatment enhanced the efficacy of anti-PD-1 antibodies [89]. The MAPK pathway also plays a major role in cell proliferation, inhibits T-cell recruitment and functions by inducing VEGF and IL-8 [90]. An inhibited MAPK pathway promotes CD8<sup>+</sup> T-cell activation and infiltration in melanoma [91,92]. Furthermore, studies showed that the combination of PD-1 blockade, BRAFi and MEKi enhances tumor immune infiltration and improves treatment outcomes [93].

A crucial oncogenic signaling pathway Wnt/β-catenin has been highly related to immune escape [94,95]. An activated Wnt/β-catenin pathway is correlated to loss of T-cell gene expression in metastatic melanoma [96]. Another study reported that the activation of the Wnt/β-catenin pathway in tumors brings about a non-inflammatory environment via numerous mechanisms. For instance, it acts on CD103<sup>+</sup> DCs of the Batf3 lineage and induces the transcription inhibitor ATF3 (activating transcription factor 3) expression to decrease production of Chemokine (C-C motif) ligand 4 (CCL4), thereby reducing initiated and infiltrated CTLs. Moreover, the Treg survival rate is enhanced by β-catenin [97].

Recently immune tumors are divided into three phenotypes: immune-desert, excluded and inflamed. Among these, the first and second phenotypes, which are non-inflamed tumors, show a low density of CTLs in the tumors and poor prognosis in immune checkpoint blockade therapy [98].

#### 4.4.2. Dysfunction of T-Cells

Accumulation of extracellular adenosine is exploited by tumors to escape immunosurveillance through the activation of purinergic receptors [99]. CD38 expression expressed on Tregs and MDSCs is infiltrated in the tumor microenvironment and stimulated adenosine production via the CD38–CD203a-CD73 axis, and therefore inhibits CTL function [100,101].

#### 4.4.3. Exhausted T-Cells

In vitro studies have reported that the PD-1 signal intensity determines the severity of T-cell exhaustion, which in turn affects the efficacy of anti-PD-1 treatment. In Nigow's animal model, high expression of PD-1 and extremely unresponsive T-cells showed relevance with resistance of anti-PD-1 therapy [102]. PD-1 treatment helps patients with low or moderate PD-1 expression to re-invigorate exhausted CD8<sup>+</sup> T cells and exert their immune effects. However, the cellular, transcriptional, and epigenetic changes following the PD-1 pathway blockade suggested limited storage potential after TEX re-invigoration, which means re-exhaustion following PD-L1 blockade [103].

#### **5. Tumor-Suppressing Microenvironment**

Apart from abnormal T-cells and PD-L1 expression, there are some other types of cells and cytokines that benefit tumor development inside the tumor microenvironment, they form the tumor-suppressing microenvironment to play a key role in resistance to the PD-1/PD-L1 blockade.

#### *5.1. Tregs*

Tregs are involved in maintaining self-tolerance, and inhibit autoimmunity through secreting cytokines, including TGF-β1. The ratio of CD8<sup>+</sup> Teff cells/Tregs is strongly associated with the prognosis of immunotherapy [104,105]. The administration of low-dose TLR-7 agonist resiquimod could transform Treg accumulation-caused resistance to the PD-L1 blockade [106]. Combination of radiation therapy and dual immune checkpoint blockade restores antitumor immunity of consumed Tregs [107]. Currently, anti-CD25 therapy is believed to take effect through Treg depletion when combined with PD-1 blockade therapy [108].

#### *5.2. MDSCs*

MDSCs suppress immunity mainly through preventing T-cell activation and function, Arg1 and ROS are the common molecules used. Besides, they downregulate macrophage production of the type I cytokine IL-12 to polarize macrophages toward a tumor-promoting phenotype [109,110], suppress tumor cell lysis mediated by NK cells and induce and recruit Tregs [111–114]. In the presence of MDSCs, the levels of PD-1 expression show a decrease, while PD-L1 expression shows an increase [115]. MDSC-targeted therapy, which decreases MDSC frequency and transforms its function, is studied to overcome the resistance to immune checkpoint inhibitors, thus combining MDSC-targeted therapy and immune checkpoint blockades is considered a promising strategy for the future [116].

#### *5.3. TAMs*

Protumor macrophages are differentiated through interaction with tumor cells and turn to polarize into M2-like TAM, which play a significant role in immunosuppression, invasion and metastasis. For the sake of overcoming the latent resistance of macrophages, CSF-1R blockade reduces the frequency of TAMs, therefore increasing production of interferon and tumor regression [117], and synergizing with immune checkpoint blockades [118].

#### *5.4. IDO*

Indole 2,3-dioxygenase is generated by tumors and immune cells to enhance Tregs and MDSCs production and activity. IDO, an enzyme catalyzing the degradation of tryptophan along the kynurenine pathway, is induced in response to inflammatory stimuli and its activity is known to have an inhibition of effector on T-cell immunity [119]. A report conducted on B16 melanoma demonstrated that following PD-1 blockade treatment, a subset of mice with IDO knockout had an obviously slower tumor development and better overall survival rates compared with wild type [120]. Thus, a combination therapy of IDO inhibitors and PD-1/PD-L1 antibodies may demonstrate a better efficacy than single agent [121].

#### *5.5. VEGFA*

TMB with hypoxia and hyper-angiogenesis is obviously crucial for tumor growth and progression, and vascular endothelial growth factor A (VEGFA) plays a significant role in it. High expression of VEGFA is reported to impair infiltration of effective anti-tumor T-cells, thus leads to innate resistance in PD-1/PD-L1 blockade [122]. Unfortunately, combining treatment with inhibiting the VEGFA and PD-1/PD-L1 blockade demonstrates more toxic and harbors more adverse effects than monotherapy.

#### *5.6. Immunosuppressive Cytokines*

TGF-β inhibits the expansion and function of many components of the immune system, either by stimulating or inhibiting their differentiation and function, therefore it maintains immune homeostasis and tolerance. Specific chemokines are capable of recruiting cells into tumors. CXCL9, CXCL10, CXCL11, CCL3, CCL4 and other chemokines and their receptors are recruited to cause antitumor response via recruiting CTL and NK cells while CCL2 CCL22, CCL5, CCL7 and CXCL8 recruit immunosuppressive cells to suppress the immune response. Research reveals that epigenetic silencing of CXCL9 and CXCL10 can suppresses T-cell homing [123].

#### **6. Activation of Alternative Immune Checkpoints**

As one of the most prospective approaches in cancer treatment, immunotherapy has reached notable achievements, especially with the PD-L1 blockade. However, the efficacy of PD-L1 inhibitor therapy has been found to be limited due to activation of other immune checkpoints including TIM-3 and VISTA. So far, some studies have reported that the combination therapy targeting distinct types of immune checkpoints has been proved effective in several cancers.

#### *6.1. TIM-3*

T-cell immunoglobulin mucin 3 (TIM-3) has been identified as a critical regulator of CTL exhaustion with co-expression of PD-1 [124]. TILs with co-expression of TIM-3 and PD-1 do not produce IL-2 and IFN-γ, and they are prone to exhaust. In response to radiotherapy and PD-L1 inhibition, TIM-3 is upregulated and subsequently caused acquired resistance in HNSCC [107]. Combination therapy targeting TIM-3 and PD-1 signaling pathways simultaneously is proved to be effective against cancer [124].

#### *6.2. HHLA2*

HHLA2, a member of the B7 family, can predict poor overall survival in several cancers, including human clear cell renal cell carcinoma and colorectal carcinoma [125]. HHLA2 can suppress T-cell activation and proliferation in the presence of TCR and CD28 signaling [126], and can do this more robustly than PD-L1 [127].

#### *6.3. VISTA*

V-domain Ig suppressor of T-cell activation (VISTA) expression induced by IL-10 and IFN-γ is observed to be higher in immature DCs, MDSCs and Tregs compared with peripheral tissues [128,129]. The synergistic effect of the combining VISTA and PD-L1 monoclonal therapy in colon cancer can be taken as an example, reduction of tumor growth and better OS are observed compared with monotherapy [130].

#### *6.4. LAG-3*

Lymphocyte activation gene-3 (LAG-3) is responsible for maintaining immune homeostasis through repressing activation of T-cells and cytokines secretion [131]. Interaction between LAG-3 and Galectin-3, a soluble lectin regulating antigen-specific T-cell activation, expands the immunomodulatory effect of LAG-3 on tumor-infiltrating CD8<sup>+</sup> T-cells in the TME [132]. Sinusoidal endothelial cell lectin binds to LAG-3 to reduce IFN-γ expression produced by activated T-cells [133]. An amazing synergistic effect in suppressing immune responses is found in LAG-3 with PD-1 under distinct conditions [134].

#### *6.5. CTLA-4*

CD28 interacting with the CD80 dimer and the CD86 monomer mediates T-cell costimulation along with TCR signals, while CTLA-4 demonstrates a higher affinity and avidity in conjunction with the two ligands than with CD28, which in turn antagonizes CD28-mediated co-stimulation [135]. A combination of PD-1-targeted mAb nivolumab

and CTLA-4-targeted mAb ipilimumab has been approved as the first-line treatment for renal clear cell cancer patients with moderate or poor prognosis [136].

#### *6.6. Siglec-15*

As a member of the sialic acid-binding immunoglobulin-like lectin (Siglec) gene family, Siglec-15 is found to impair anti-tumor immunity through suppressing T-cell functions. Siglec-15 is expressed only on some myeloid cells normally, while it is upregulated on TAMs and tumor cells [137]. Interestingly, an antagonistic relationship between Siglec-15 and PD-L1 has been reported, mainly due to regulation of IFN-γ [138]. M-CSF induces expression of Siglec-15 on macrophages and IFN-γ, identified as a crucial factor promoting PD-L1 expression, inversely decreases it [137].

#### *6.7. TIGHT*

T-cell immunoglobulin and ITIM domain (TIGIT), expressed mainly on Tregs, is a co-inhibitory checkpoint receptor which has a significantly higher affinity in binding to CD155 than the co-stimulatory receptor CD226 [139]. TIGIT/CD155 signaling causes T-cell exhaustion to impair anti-tumor immunity in several types of cancer, including melanoma and HNSCC [140,141]. Furthermore, the phenomenon that TIGIT expression often accompanies PD-1 has been observed in both normal tissues and tumors [142].

#### *6.8. BTLA*

B and T lymphocyte attenuator (BTLA), expressed mostly on B-cells, is upregulated on CD19<sup>+</sup> high B-cells through AKT and STAT3 pathways once triggered by IL-6 and IL-10 [143]. BTLA is regarded as one of the factors leading to resistance to anti-PD-1 therapy, though they do not suppress T-cell signaling through an identical mechanism related with src-homology-2 domain-containing phosphatase (SHP)1 and SHP2 [143,144].

#### **7. Current Combination Therapies with PD-1/PD-L1 Inhibitors**

With regard to clinical the limitations of anti-PD-1/PD-L1 monotherapy, it exists more and more in combination therapies based on mechanisms underlying resistance to the PD-1/PD-L1 blockade. Among all of them, chemotherapy, VEGF/VEGFR-targeted therapy and anti-CTLA-4 rank in the top three. Other treatments that are considered to combine with PD-1/PD-L1 blockade include radiotherapy, vaccines, cytokine therapy and chemokine inhibition. Radiotherapy is identified to alter differentiation and function of T-cells and promote the expression of PD-L1, which means adding radiotherapy may enhance the effects of anti-PD-L1 treatment [145]. A triple therapy with anti-PD-1 antagonist antibody, anti-CD137 agonist antibody and vaccine therapy has been reported to significantly enhance T-cell activation in pancreatic ductal adenocarcinoma in a preclinical study [146]. Recently, another immune checkpoint inhibitor tiragolumab targeting TIGIT has been granted breakthrough therapy designation by the FDA and combining anti-PD-L1 and anti-TIGIT has been reported as highly effective in clinic with metastatic NSCLC patients [147]. Combining TNF-α-loaded liposomes and anti-PD-1/PD-L1 further enhances the anti-tumor immunity [148]. Even utilizing newly emerged neoantigens may improve the therapeutic efficacy of immune checkpoint blockade treatment [148].

#### **8. Conclusions**

As an inhibitor in the immune system, PD-L1 plays multiple roles in tumors. PD-L1 has been confirmed as a prospective and prognostic biomarker in certain cancers, while rebiopsy should be considered when PD-L1 expression is increased due to treatment (such as gefitinib treatment). Immune resistance induced by PD-L1 following various therapies inspired a combination therapy of PD-L1 blockade and these therapies. To date, immunotherapy, especially PD-1/PD-L1 blockade, which is at forefront of clinical therapy, has benefited many patients. However, primary and acquired resistance to this blockade therapy still exists and limits its efficacy. So far, key mechanisms suggest complement

approaches for patients who cannot respond well to PD-1/PD-L1 antibodies. For example, modulating the immunosuppressive tumor microenvironment, such as depletion of Tregs, IDO, or MDSCs, interfering suppressive cytokines and inhibiting alternative immune checkpoints, may enhance the therapeutic efficacy of the PD-1/PD-L1 blockade. Other mechanisms underlying resistance to this blockade therapy and individual treatments for more patients requires further investigation.

**Author Contributions:** Conceptualization: Y.Y., C.Z., H.C.; Writing—Original: Draft Y.Y., A.A.; Writing—Review and Editing: A.A., C.Z., H.C.; Supervision: H.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partly funded by the National Natural Science Foundation of China [No. 81872443].

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Review* **CTLA-4 in Regulatory T Cells for Cancer Immunotherapy**

**Navid Sobhani 1,\* , Dana Rae Tardiel-Cyril <sup>1</sup> , Aram Davtyan <sup>2</sup> , Daniele Generali <sup>3</sup> , Raheleh Roudi <sup>4</sup> and Yong Li 1,\***


**Simple Summary:** In the fight against cancer, immunotherapies have given great hope after encouraging results in clinical investigations showing complete remission in some patients with melanoma. In fact, directing the immune system against cancer has been a very innovative strategy fostered during the past three decades. Despite this fact, the disease is serious, the mortality is still very high, and only a minority of patients are responsive to immunotherapies. Therefore, there is a need for a better understanding of the molecular mechanisms of resistance to immune checkpoint inhibitors such as antibodies against cytotoxic T-lymphocyte-associated protein 4 (CTLA-4). In this article, we discuss the molecular mechanism of CTLA-4 in T regulatory cell inhibition, while highlighting the knowledge gap.

**Abstract:** Immune checkpoint inhibitors (ICIs) have obtained durable responses in many cancers, making it possible to foresee their potential in improving the health of cancer patients. However, immunotherapies are currently limited to a minority of patients and there is a need to develop a better understanding of the basic molecular mechanisms and functions of pivotal immune regulatory molecules. Immune checkpoint cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and regulatory T (Treg) cells play pivotal roles in hindering the anticancer immunity. Treg cells suppress antigenpresenting cells (APCs) by depleting immune stimulating cytokines, producing immunosuppressive cytokines and constitutively expressing CTLA-4. CTLA-4 molecules bind to CD80 and CD86 with a higher affinity than CD28 and act as competitive inhibitors of CD28 in APCs. The purpose of this review is to summarize state-of-the-art understanding of the molecular mechanisms underlining CTLA-4 immune regulation and the correlation of the ICI response with CTLA-4 expression in Treg cells from preclinical and clinical studies for possibly improving CTLA-4-based immunotherapies, while highlighting the knowledge gap.

**Keywords:** CTLA-4; Treg cells: immune checkpoint inhibitors; CD28; antigen-presenting cells

#### **1. Introduction**

Globally, cancer remains the leading cause of mortality and morbidity, with nearly 9 million deaths every year [1]. Early diagnosis and advances in cancer treatment have improved the survival of cancer patients, but there were more than 1.7 million new cases of cancer in the United States in 2019 [1]. A considerable percentage of these patients manifested drug resistance, metastasis, and recurrence [2].

A promising paradigm in the dilemma and challenge of cancer therapy is immunotherapy, and the T cell population has generated considerable enthusiasm among scientists due to its ability to kill malignant tumor cells directly [3].

**Citation:** Sobhani, N.; Tardiel-Cyril, D.R.; Davtyan, A.; Generali, D.; Roudi, R.; Li, Y. CTLA-4 in Regulatory T Cells for Cancer Immunotherapy. *Cancers* **2021**, *13*, 1440. https:// doi.org/10.3390/cancers13061440

Academic Editor: Anne-Marie Schmitt-Verhulst

Received: 5 February 2021 Accepted: 18 March 2021 Published: 22 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

There are two major types of T cell: Conventional adaptive T cells (including helper CD4+ T cells [Th1, Th2, Th17, Th9, and Tfh], cytotoxic CD8+ T cells, memory T cells, and regulatory CD4+ T cells [Treg]) and innate-like T cells (including natural killer T cells, mucosal associated invariant T cells, and gamma delta T cells (γδ T cells)) [4]. The CD4+ T cells subset can target malignant tumor cells using different approaches, either by directly killing tumor cells or indirectly modulating tumor microenvironments (TME) [5,6]. These cells can increase the response of cytotoxic T cells (CTL) and quality of B cells [7]. The major killers of tumor cells are cytotoxic CD8+ T cells [8].

Innate-like T cells, representing one of the major groups of T cells, can be grouped into natural killer T cells (NKT cells), mucosal associated invariant cells (MAIT), and gamma delta T cells (γδ T cells) [9–11]. During development, innate-like T cells, called innate lymphoid cells (ILCs)-Natural Killer (NK) cells, acquire an effector function, whereas conventional T cells remain in a naive state [12]. The first group, NKT cells, express T-cell receptors (TCRs) and cell surface markers of NK cell lineages [13]. They are involved in the recognition of glycolipid antigens and present them to antigen-presenting cells (APCs) in the context of major histocompatibility complex (MHC) class I-associated protein CD1d [14]. T cells with γδ expression, representing the first layer of defense, constitute nearly 2% of the T cell population in peripheral blood and secondary lymphoid organs, while they are mainly found in the epithelia of the skin, gut, lung, and other organs [15,16]. Another group of innate-like T cells, called MAIT cells, constitute approximately 5% of all T cells and have considerable similarities to NKT cells [17,18].

Treg cells are one of the most fascinating immunosuppressive subsets of CD4+ (CD25+) T cells, mainly represented by master transcription factor 3 (FOXP3), and they account for nearly 5% of the total CD4+ T cell population under normal conditions [19]. Treg cells increase dramatically in response to the early stages of malignant tumor initiation and growth [20]. In the tumor microenvironment, Treg cells can suppress the immune system activity of cytotoxic T lymphocytes (CTLs) [21]. A panel of immune-modulatory receptors expressed on the Treg cell population includes cytotoxic T lymphocyte antigen 4 (CTLA-4), the vascular endothelial growth factor receptor (VEGFR), and programmed cell death protein 1 (PD1) [22]. CTLA-4 is expressed on activated T and Treg cells [23,24] https://paperpile.com/c/d61gxv/defR (accessed on 5 February 2021). Atkins et al. showed that an immune checkpoint blockade of CTLA-4 improved the survival rate of renal cell carcinoma, melanoma, non-small cell lung cancer (NSCLC), and head and neck squamous cell cancer [25]. This protein was the second receptor of the T-cell costimulatory ligand CD80/86 and, therefore, an immune checkpoint whose function is critical for downmodulating the immune response. In contrast to the first receptor (CD28), which is antigen-dependent, CTLA-4 is antigen-independent [26]. In 2011, ipilimumab was the first immunotherapy drug targeting CTLA-4 to receive FDA approval to treat late-stage melanoma [27]. This approval came after encouraging results of a large randomized phase III clinical trial improving patients' survival compared to standard therapy. Since then, several immunotherapies targeting the PD-1/PD-L1 axis have received FDA approval to treat multiple types of cancer [27].

This review will describe the mechanisms of CTLA-4 immune checkpoint inhibition, the role of Treg cells in tumorigenesis, and how anti-CTLA-4 antibodies can provoke an alteration in the expression of CTLA-4 on Treg cells while exerting anti-cancer therapeutic activity.

#### **2. Mechanism of CTLA-4 Immune System Inhibition**

A better understanding of the biological mechanisms and functions of negative and positive co-stimulatory molecules has been shown to be essential for improving current and potentially new CTLA-4 or Programmed Cell Death 1 (PD-1) inhibitors for anti-cancer immunotherapies.

Once bound to B7-1 (CD80) or B7-2 (CD86), CTLA-4 switches-off antigen-presenting cells [28]. CTLA-4 was immediately increased after T-cell receptor (TCR) engagement,

reaching its highest level of expression as a homodimer at 2–3 days after the activation of conventional CD4+ and CD8+ T cells [29,30]. CTLA-4 competes with costimulatory molecule CD28 for the CD80/86 ligands CD80 and CD86, for which it has a higher affinity and avidity [31,32]. It is necessary to inhibit interactions with both CD80 and CD86 with antibodies to optimally block the CD28-dependent proliferation of T cells in an allogenic mixed lymphocyte reaction stimulated with B lymphoblastoid cell lines. Since both CD80 and CD86 exert a positive costimulatory signal through CD28, the role played by CTLA-4 in the competitive inhibition of CD28 is important for attenuating T-cell activation, thereby fine-tuning the immune response [33]. Rapid binding kinetics with a very fast dissociation rate constant (koff) of both CTLA-4 and CD28 to CD80 has been observed (koff ≥ 1.6 and ≥0.43 s−<sup>1</sup> ) [34], which permits their instant competition. The function of T cells can be suppressed by Treg cells through multiple mechanisms [35]. Treg cells constitutively express CTLA-4 on their suppressive functions. CTLA-4-expressing T cells (Treg or activated conventional T cells) have been shown to lower levels of CD80/86 costimulatory molecules available on APCs by CTLA-4-dependent sequestration via transendocytosis [36]. This event can negatively regulate the proliferation of non-Treg T cells, as well as the production of cytokines.

RAG2-deficient mice reconstituted with CTLA-4-deficient bone marrow developed lethal inflammation of multiple organs and died around 10 weeks after reconstitution, whereas control mice (reconstituted with normal bone marrow) were healthy. Intriguingly, the mouse chimeras reconstituted with a mixture of normal and CTLA-4-deficient bone marrow remained healthy, without developing any disease [37]. The authors concluded that the disease observed in CTLA-4−/<sup>−</sup> mice is not due to a T cell autonomous defect and that CTLA-4 triggering on normal T cells produces factors inhibiting the disease induced by CTLA-4-deficient T cells. It has been shown that mice selectively deficient in CTLA-4 in Treg cells (Foxp3+) develop systemic lymphoproliferation and fatal T cell-mediated autoimmune disease, indicating that Treg cells critically require CTLA-4 to suppress immune responses and maintain immunological self-tolerance [38,39].

Additionally, after T-cell activation by TCR, CTLA-4 within intracellular compartments is immediately transported to the immunologic synapse [40]. The stronger the TCR signaling, the more CTLA-4 transported to the immunological synapse [40]. After reaching the synapse, CTLA-4 becomes stable through its binding to the CD80 and CD86 ligands, leading to its accumulation and effective out-competition against CD28 [28]. Differences in both the affinity and avidity in ligand-binding cause selective CD28 or CTLA-4 recruitment to the immunological synapse. The major ligand leading to CTLA-4 localization in the synapse is CD80, while for CD28, it is CD86 [28]. In this way, CTLA-4 attenuates the positive co-stimulation of CD28, thereby limiting the downstream signaling of CD28, which is primarily achieved through PI3K and AKT [41,42]. This mechanism allows a fine-tuning of TCR signaling and therefore T-cell activity. The negative co-stimulation of CTLA-4 is intrinsically linked to CD80/86 and CD28 positive co-stimulations. CTLA-4 mainly regulates T cells at priming sites (e.g., gut or lymphoid organs such as spleen and lymph nodes). Since CTLA-4 plays a crucial function in the activation of T cells, its negative co-stimulation plays a critical role in tolerance. As a matter of fact, the biallelic genetic *Ctla-4* deletion in mice leads to their death at 3–4 weeks of age because of pronounced lymphoproliferation with multi-organ lymphocytic infiltration and tissue destruction, particularly with pancreatitis and myocarditis [43–45]. Mice lethality can therefore be prevented by normal T cell factors. Several groups foster the idea that extrinsic cell suppressive functions of CTLA-4 are mainly mediated through Treg cells [38,46]. Others support the idea that CTLA-4's ability to inhibit T cells is Treg cell-independent [47,48]. An argument for the first line of thought is that a particular loss of CTLA-4 in Treg cells was enough to induce abnormal T-cell activation and autoimmunity [38,49]. In fact, Wing et al. showed that the loss of CTLA-4 in Treg cells was capable of hyper producing immunoglobulin E, systemic lymphoproliferation, fatal T cell-mediated autoimmune disease, and powerful tumor immunity [38]. After losing the CTLA-4-expressing subpopulation, the Treg cells were not capable of exerting their T cell

suppressive functions; in particular, they were not able to down-modulate the dendritic cell expressions of CD80 and CD86 [38]. It must be noted that the lack of CTLA-4 in Treg cells also leads to an aberrant expression and expansion of Tconv cells, which can cause the latter cells to infiltrate and fatally damage nonlymphoid tissues and cells [49]. Therefore, CTLA-4 in Treg cells is also needed to prevent the accumulation of T cells that may harm vital organs.

As a hypothetical molecular biology explanation, it is possible that Treg cells with CTLA-4 may limit the availability of CD80/86 ligands for the positive co-stimulation of CD28 in effector T cells. Through such a mechanism, the CTLA-4 would indirectly and cell-extrinsically dampen T-cell activation. It is also known that CTLA-4 on effector T cells can trans-compete for CD80/86 ligands [50]. Another mechanism by which CTLA-4 can lower the total availability of CD80/86 ligands is through APC-mediated trans-endocytosis of CD80/86 ligands [36]. The last two mechanisms explain how CTLA-4 could prevent anti-cancer immune reactions without the need for Treg cells. Overall, it is noteworthy that these mechanisms are not yet fully understood and each contribution remains elusive in the context of cancer immunity and drug design.

Furthermore, unexpectedly, the depletion of CTLA-4 from a Treg cell population of adult mice conferred resistance to autoimmune encephalomyelitis (EAE) and did not enhance anti-tumor immunity [51]. This was accompanied by an expansion of functional CTLA-4-deficient Treg cells expressing immunosuppressive molecules (IL-10, LAG-3, and PD-1) capable of protecting them from EAE, demonstrating that CTLA-4, in addition to previously described mechanisms of action, has a Treg-intrinsic effect in limiting Treg expansion.

Additionally, since CTLA-4 expression has been correlated with the TCR signal strength, high Treg cell and CTLA-4 expressions are concomitant [52,53]. The inhibition efficacy of any cell by CTLA-4 depends on the affinity between the major histocompatibility complex (pMHC) ligand and its TCR. The higher the affinity of TCRs, the more those cells can be inhibited through CTLA-4 [54,55]. Additionally, the induction of CTLA-4 also restricts CD4+ T-helper clonal expansion. Ultimately, through such a mechanism of action of CTLA-4, the TCR signal is fine-tuned in response to specific immunological threats.

Furthermore, a number of structures of the extracellular domain of human CTLA-4 are available in Protein Data Bank (PDB), including apo structures and various complexes. The very first structure of CTLA-4 was determined using solution NMR spectroscopy (PDB ID: 1AH1), revealing an Ig-like V (variable)-type domain, where two beta-sheets of the V-fold are connected by two disulfide bonds (21 to 94 and 48 to 68) [56]. Another apo structure of CTLA-4 was later published in the physiological dimeric state (PDB ID: 3OSK) [57]. CTLA-4 binds its native ligands CD80 and CD86 at the A 'GFCC' face, which contains a number of charged residues that are highly conserved between CTLA-4 and CD28 (and across species). A key role in these interactions is also played by the <sup>99</sup>MYPPPY<sup>104</sup> loop connecting F and G strands [56]. The structures of CTLA-4 with CD80 and CD86 (PDB IDs: 1I8L and 1I85) manifested a mostly convex binding surface at CTLA-4, free of any notable cavities that could have been targeted with traditional small-molecule campaigns [58,59]. It is also interesting to note that while the CD80-bound conformation of CTLA-4 is very similar to the apo form, CD86 binding requires some structural rearrangement, most significantly, in the FG loop [57–59]. Finally, several structures of CTLA-4 bound to monoclonal antibodies have also recently been reported (PDB IDs: 5GGV, 5TRU, 5XJ3, and 6RP8) [60–62]. These structures reveal that ipilimumab and tremelimumab directly compete with CD80 and CD86 at their binding surface, sterically displacing and preventing their interactions with CTLA-4. Moreover, subtle differences in the CTLA-4 structure, such as a slightly larger distance between G and F stands, and extended interactions of antibodies with non-conserved residues on the opposite side of the FG loop, enable selectivity between CTLA-4 and CD28 [61]. Interestingly, the amino acid sequence of the intracellular tail of CTLA-4 is conserved in 100% of all mammalian species, meaning that its intracellular domain must have an important role in the inhibition of T-cell activation [63,64]. In fact, the inhibitory functions of CTLA-4, by competing with CD28 for CD80 and CD86 or

through its transmission of negative signals, can be accomplished because of its intracellular domain, but such a downstream mechanism of CTLA-4 signal transduction deserves further investigations [64,65]. Based on the primary amino acid sequence of the CTLA-4 cytoplasmic region, there are two potential binding sites for Src homology domain 2 (SH2) and an SH3 potential binding motif [66]. CTLA-4 was found to be capable of becoming associated with SH2-containing tyrosine phosphatase-2 (SHP-2) through the SH2 domain of SHP-2. Such an association resulted in phosphatase activity against Ras regulatory protein p52SHC [67]. Therefore, CTLA-4 might be able to start a signal transduction cascade leading to the dephosphorylation of TCR-associated kinases or substrates.

While the antitumor activity and clinical benefits of antibodies such as ipilimumab that block CTLA-4 interactions with ligands have been demonstrated [61], it is always desirable to have bioavailable and cheaper options in the form of small molecules or peptides. In cases of traditionally undruggable targets, such as CTLA-4, where no suitable smallmolecule binding pockets can be immediately identified at the ligand-binding interface, peptide drugs can present a viable alternative. Like antibodies, peptides can achieve a high affinity and specificity by capturing a larger interaction area with the target. At the same time, they are easier to synthesize and have greater tissue penetration due to their smaller size compared to the antibodies. Moreover, peptides have recommended themselves in a variety of therapeutic areas, including cancer [68,69]. In addition, targets similar to CTLA-4 can be amenable to less-standard small molecule campaigns. One such approach is allosteric modulation. In this case, a small molecule bound to a distant site can activate or inhibit the protein function or its interactions with other molecules as a result of structural changes that it induces at a distance [70]. However, for CTLA-4, such sites still have to be determined through either experimental or computational techniques [71,72].

#### **3. Regulatory T Cells and Anticancer Immunity**

#### *3.1. First Insights into Treg Cells*

Treg cells are a population of CD4 T cells constitutively expressing CTLA-4. They are crucial for both immune-oncology and autoimmunity, as we will describe in this review. The focus of this article is on the CTLA-4-positive population of Treg cells in cancer. After Treg cells were discovered for the first time in the CD4+ CD25+ T cell subpopulation in 1995 [73], mutations of *FOXP3* recapitulated the impaired formation or improper function of Treg cells, causing an immune dysregulation syndrome in mice, termed polyendocrinopathy enteropathy X-linked syndrome, which ultimately leads to multiple autoimmune disorders [74]. Corroborating the importance of Treg cells for a functional immune response, mice carrying spontaneous alterations of *Foxp3*—that ultimately lacked Treg cells—died due to systemic autoimmunity [75,76]. As expected, the external expression of FOXP3 bestowed naïve CD4+ T cells (Tconv, without Treg cells) with the same immune-suppressive capacity typical of Treg cells. Therefore, FOXP3 is a master transcription factor that regulates Treg cell phenotypes and their function as immunosuppressants. The role of Treg cells in cancer is mainly observed at inflammatory sites, where they migrate and inactivate different types of effector T cells, such as CD4<sup>+</sup> T helper (TH) cells and CD8<sup>+</sup> cytotoxic T cells (CTLs) [77–80]. As a consequence, intervening in this activity of Treg cells could induce the immune system in the fight against cancer.

#### *3.2. Inhibitory Effects of Treg Cells on APC*

Treg cells represent a crucial component of the immune system, being essential for controlling self-tolerance, and thereby play essential roles in various medical conditions. Treg cells have a crucial role in the suppression of the immune response in cancer [73,75,81–85]. Treg cells inhibit APC by three main mechanisms: (1) Depleting immune-stimulating cytokines [86–89]; (2) producing immunosuppressive cytokines (like TGF-β, IL-10, and IL-35); and (3) constitutively expressing CTLA-4. Treg cells express Interleukin 2 (IL2) receptors that bind to IL2, thereby limiting the amount of this cytokine available for Tconv

cells [90,91]. As a consequence, the constitutive expression of CTLA-4 blocks the priming and activation of Tconv cells to APCs [38,92].

Figure 1 summarizes the role of CTLA-4 in Treg cells modulating Tconv activation.

**Figure 1.** Regulatory T (Treg) cells inhibit antigen-presenting cells (APC) by three main mechanisms: (1) Depleting immunestimulating cytokines; (2) producing immunosuppressive cytokines (e.g., TGF-β, IL-10, and IL-35); and (3) constitutively expressing CTLA-4, which blocks the priming and activation of naïve CD4+ T (Tconv) cells to APCs.

Treg cells block normal protective immune-surveillance and inhibit the antitumor immune response in cancer patients. Thereby, if Tconv cells are like tumor suppressors, Treg cells could be considered as oncogenes because they are suppressing antitumor immunity [81,82,93,94], although the definitions of oncogenes and tumor suppressors refer to genes in tumors that, when expressed, cause or prevent cancer, respectively [95]. Likewise, CTLA-4 and PD-1 immune checkpoints, since they block the immune system's recognition of cancer cells, could also be comparable as tumor suppressors.

#### *3.3. Conflicting Roles of Treg Cells in Malignant Tumors*

The role of Treg cells in immunoncology was discovered by two Japanese groups in 1999 [93,94]. The two groups independently reported that anti-CD25 antibodies, capable of depleting CD4+CD25+ Treg cells, led to higher tumor rejection and retarded tumor growth in normal and T cell reconstituted nude mice [93,94]. CD25 is the α chain of the interleukin-2 receptor. Onizuka et al. showed that a single dose (less than 0.125 mg) of anti-CD25 was capable of causing the regression of multiple tumors derived from four different inbred mouse strains (five leukemias, myeloma, and two sarcomas) [93]. Similarly, Shimizu et al. showed that the elimination of CD25-expressing T cells caused a powerful immune response in syngeneic tumors in mice, leading to tumor regression within 1 month, thereby allowing the host to survive > 80 days [94]. Among CD4+ T cells, the percentage of Treg cells is higher in the blood of cancer patients compared to that of healthy individuals [83,96,97]. Expectedly, the relatively higher Treg cell levels in the tumor microenvironment correlated with a poor prognosis in various cancer types, such as melanoma and non-small cell lung, ovarian, and gastric cancers [82,83]. The Treg cell population is not large in

the periphery blood of cancer patients compared with the TME, implying that T cells' interaction with tumor cells is important [97]. On the contrary, in certain tumors, such as colorectal cancer (CRC), a high level of FOXP3+ T cells is correlated with a better prognosis [98]. This is because the accumulation of FOXP3+ occurs together with inflammatory cytokines, possibly implying that Treg cells play a role in repressing tumor inflammation. It was brought to light that two populations of FOXP3 (+) CD4 (+) T cells had distinct roles in controlling the prognosis of CRCs, contributing in opposing ways. FOXP3 (hi) Treg cells are correlated with worse survival, whereas FOXP3 (lo) non-Treg T cells are correlated with better survival. This is possibly because the FOXP3+ (lo) non-Treg T cell population leads to an inflammatory TME against the tumor. In fact, it was observed that FOXP3+ non-Treg T cells in CRCs are correlated with high levels of tumor necrosis factor (TNF), IL2, and TGFβ [96]. Depleting FOXP3 (hi) Treg cells from tumor tissues could be deployed to increase the antitumor immunity to treat CRC or other cancers, whereas other strategies enhancing the levels of FOXP3(lo) non-Treg T cells could also be used to suppress or prevent tumorigenesis [96].

There are conflicting reports regarding the prognostic value of tumor-infiltrating Treg cells. Shang et al. demonstrated that FOXP3+ Treg cells are correlated with shorter overall survival in breast, hepatocellular, gastric, melanoma, renal, and cervical cancers, and longer overall survival in head and neck, colorectal, and esophageal cancers, whilst they display no correlation for pancreatic and ovarian cancers [99].

In conclusion, Treg cells inhibit anti-cancer immunity, blocking the immune surveillance of tumors, which ultimately leads to cancer spreading [81–83,93,94]. Immunosuppressive Treg cells, producing cytokines, are observed in both human chronic inflammatory disease and cancers, where they promote tumorigenesis through a mechanism similar to that of chronic inflammation [48,100,101]. The depletion of Treg cells in mice is capable of promoting lymphocyte recruitment and as a consequence, a decrease in the tumor growth rate and the presence of high endothelial venules, indicating destruction of the tumor tissues [102,103].

#### *3.4. Treg Cells and the Tumor Microenvironment*

The TME is mainly comprised of a subpopulation of Treg cells called bona fide Treg cells that enhance the expression of immunosuppressant molecules such as CTLA-4 and T-cell immunoreceptors with Ig and ITIM domains (also called TIGIT), whose expression is very low in naïve Treg cells [83,96,104]. A transcriptome analysis of 15 human lung cancer samples and 14 colorectal cancer samples demonstrated that tumor-infiltrating Treg cells have very high levels of different Treg activation markers, such as T cell immunoglobulin mucin receptor 3 (HAVCR2), glucocorticoid-induced TNFR-related protein (GIRT), lymphocyte-activation gene 3 protein (LAG3), and inducible T cell co-stimulator (ICOS). Interestingly, this phenotype was not observed in peripheral blood samples from the same patients, whose expression levels in the blood remained the same. This could indicate that Treg cells become activated in TME, where they exert their immune-suppressive functions [105].

#### *3.5. Cross-Talk between Treg Cells and the Tumor Microenvironment*

It has recently been shown that adenosine produced by apoptotic Treg cells present within the TME exerts higher immunosuppressive effects compared to live Treg cells [21,106]. A weak NRF2-associated antioxidant pathway leads to a vulnerable system against reactive oxygen species in TME, possibly causing apoptosis in Treg cells, which is a process that has been shown to convert high ATP levels into adenosine through Treg cell-expressed ectoenzymes CD39 and CD73. In turn, the resulting abundance in adenosine engages purinergic adenosine A2A receptors (also known as ARORA2A), which is a family of G protein-coupled receptors with seven transmembrane alpha helices whose function is to regulate the oxygen demand and increase vasodilatation, as well as suppress immune cells. Apoptotic Treg cells use the A2A pathway to suppress immune cells [21,106]. The

mechanism postulated to explain the activation of Treg cells in TME is that proliferating and dying tumor cells have loads of self-antigens, which are best recognized through Treg cells and thereby become activated in TME [107]. Another explanation comes from results from mice experiments of two research groups showing that immune dendritic cells expressed in mice tumors activate Treg cells in a TGFβ-dependent manner [107,108]. Treg cells recognize specific self-antigens and can become clonally expanded in TME [109,110]. Treg cells typically have higher affinity TCRs for self-antigens than Tconv cells and therefore, should be predominantly activated, even when in competition with Tconv cells. It must be stated, however, that these data come from animal studies and Treg cells induced by TFGβ have not yet been fully demonstrated in humans. As for the epigenetic profile of tumor-infiltrating Treg cells, very little is understood [111–113]. Epigenetic studies of Treg cells are limited and future studies could shed more light on the subject, in order to better understand the origin and mechanisms of activation of Treg cells. Treg cells move to the TME by chemotaxis via chemokines and their receptors, such as CXCL12-CXCR4, CCL5-CCR5, CCL22-CCR4, and CCL1-CCR8 [83,105,114–118]. Blocking such chemotactic signals can reduce the accumulation of Treg cells inside tumors [119]. Such chemokines are produced in the TME by the tumor and/or macrophages [83,105,114–116]. Additionally, some chemokines, such as CCL1 and CCL22, can be produced within tumors by exhausted or dysfunctional CD8+ T cells [119,120]. Therapies targeting chemokines could be considered to lower the Treg:Tconv ratio in the tumor microenvironment, in order to produce more Tconv and less Treg cells. Cancers engage various immune escape mechanisms that can be dependent on specific tumor intrinsic factors. In fact, alterations in tumor suppressor PTEN; Liver Kinase B1 (LKB1); or oncogenes WNT/β-catenin, KRAS, or basic leucine zipper transcriptional factor ATF-like 3 (BATF3), affect effector T-cell recruitment to the tumors [121–125]. On the contrary, tumor hyper-activation of FAK leads to a recruitment of Treg cells, together with chemokine-driven CD8+ T cell exhaustion or poor infiltration within the tumor [126,127]. In fact, Jiang et al., using tissues from pancreatic ductal adenocarcinoma (PDAC) patients, observed that FAK was elevated and correlated with high levels of fibrosis and poor CD8+ cytotoxic T cell infiltration, which are signs of an immune-suppressive TME. The use of a FAK inhibitor (VS-4718) substantially limited tumor progression and doubled the survival of a humanized mice model of PDAC [126]. In squamous cell carcinoma (SCC) cells, FAK was shown to drive the exhaustion of CD8+ T cells and recruitment of Treg cells in TME through the regulation of chemokines/cytokines and ligand-receptor networks (such as Ccl5/Ccr5), ultimately permitting tumor growth. FAK kinase inhibitor VS-4718 drove Treg cell depletion and promoted the anti-tumor response of CD8+ T cells [127].

#### *3.6. Treg Cells and Nonself Antigens*

At the location of tumor cells, there are two types of antigens recognized by Treg cells: Shared antigens and neoantigens. The first ones arise from highly or aberrantly expressed endogenous proteins encoded by the germ line. The second ones derive from either abnormal self-proteins formed from somatic genetic alterations or from oncogenic viral proteins. Experiments with animals have shown that Treg cells primed to nonself antigens increased the affinity of CD8+ T cells, most likely by the inhibition of T cells carrying TCRs with low-avidity to antigens [128]. APCs can render CD8+ T cells targeting self-antigens self-tolerant through the control of Treg cells [129]. In fact, using human T cells in vitro, the authors showed that Treg cells were able to make the self-reactive human CD8+ T cells anergic upon antigen stimulation. In addition, they observed the proliferative activity of self-antigen-specific T cells in CTLA-4+ and CTLA-4- fractions. The CTLA-4+ fraction was highly proliferative, had a low expression level of BCL2, and was prone to death upon self-antigen stimulation. On the contrary, Treg cells did not suppress non-self-specific CD8+ T cells [129]. Therefore, Treg cell-mediated immunosuppression could be more effective in shared antigen-expressing tumors compared to those expressing neoantigens. This could be a reason why tumors expressing neoantigens respond better to immune checkpoint blocking than tumors with a low mutational burden [130,131]. One of the major aims

of immunotherapy research is to understand why some cancer patients respond very well to immune checkpoint inhibitions while others do not, as well as discovering new biomarkers useful for just-in-time determination of treatment-responsive patients, before administrating immunotherapies.

#### **4. Correlation between Anti-CTLA-4 Treatment and Its Effect on Treg Cells**

The anti-CTLA-4 monoclonal antibody ipilimumab (Yervoy, Bristol-Meyers Squibb) gained FDA approval in March 2011 for the treatment of advanced melanoma, which is the most dangerous type of skin cancer, after a large randomized phase III clinical trial consisting of 676 patients demonstrated that ipilimumab improved the overall survival (OS) of melanoma patients who did not respond to standard therapy. In fact, the median OS in 403 patients randomly assigned to receive 3 mg/kg ipilimumab with an investigational vaccine made of HLA-A\*01201-restricted glycoprotein 100 with incomplete Freund' adjuvant was 10.0 months (gp100, 95% Confidence Interval [CI], 8.5–11.5) vs. 6.4 months observed for 136 patients treated with gp100 only (Hazard Ration [HR] for death = 0.68; *p* = 0.001). In total, 137 patients were treated with ipilimumab alone and had an OS of 10.1 months vs. 6.4 months for the gp100 alone group (95% CI, 9.0–13.8; HR for death = 0.66, *p* = 0.003) [132]. After its approval, the drug was added as a category 1 recommendation in the National Comprehensive Cancer Network (NCCN) guidelines for the systemic treatment of advanced or metastatic melanoma.

This clinical evidence shows that the antibody enhanced the ability of the immune system to attack cancer through CTLA-4 inhibition. It must be mentioned that adverse events occurred in 10–15% of patients treated with ipilimumab alone compared to patients treated with gp100 only [132].

In 2014, another pivotal phase III clinical trial (CA184-024) including 502 metastatic melanoma patients tested ipilimumab. The current standard of care treatment for the disease is chemotherapy (decarbazine), which has not been shown to increase OS. Interestingly, the treatment of patients with 850 mg/m<sup>2</sup> decarbazine with 10 mg/kg ipilimumab improved OS compared to an arm with only chemotherapy with the placebo. The OS of patients treated with ipilimumab plus decarbazine vs. decarbazine plus placebo was 47.3% vs. 36.3% at the first year, 28.5% vs. 17.9% at the second year, and 20.8% vs. 12.2% at the third year (HR for death with ipilimumab/decarbazine, 0.72; *p* < 0.001). The risk of progressing through the disease decreased by 24% when using ipilimumab/decarbazine vs. decarbazine/placebo (HR for progression, 0.76; *p* = 0.006). The ratios of the disease to control were similar for the two groups (33.2% for ipilimumab/decarbazine and 30.2% for decarbazome/placebo; *p* = 0.41). This study was important because it showed how ipilimuamb could be used as the first line treatment for metastatic melanoma [133]. The study tested a higher concentration (10 mg/kg) of ipilimuab than the approved 3 mg/kg [134]. Consequently, more adverse events were observed using higher doses of anti-CTLA-4, possibly because of CTLA-4 molecular degradation. In fact, CTLA-4 is needed to prevent immune-related adverse reactions and its degradation can be deleterious.

Interestingly, a recent report demonstrates that the immune-related Adverse Events (irAEs) of ipilimumab and alike result from the lysosomal degradation of CTLA-4 in Treg cells. The study used the CTLA-4 mutant (Y201V), which is incapable of being recycled because it lacks interaction with the lipopolysaccharide (LPS)-responsive and beige-like anchor protein (LRBA). This indicates that the specific region of CTLA-4 is an essential mediator of CTLA-4 recycling. The investigators made antibodies targeting CTLA-4 (HL12 and HL32) that were not able to degrade the CTLA-4 of Treg cells. In fact, in contrast to ipilimumab or TremeIgG1, the use of novel anti-CTLA-4 antibodies had no effect on the CTLA-4 level of Treg cells in the same model. Additionally, HL12 and HL32 could more effectively lead to tumor rejection, with fewer irAEs in mice [135]. Such knowledge is useful for the generation of novel antibodies or molecules that could inhibit CLTA-4 without eliciting its degradation and could therefore be used in combination with other PD-1 or PD-L1 inhibitors with less toxicity.

Various studies show that consolidated or novel types of CTLA-4 therapies correlate with different expression levels of Treg cells. Ji et al. showed that the treatment of mice with 0.25 mg anti-CTLA-4 monoclonal antibody correlated with a lower level of the CD25+Foxp3+ Treg cell population (*p* < 0.05) [136]. Qu et al. observed that anti-CTLA-4 monoclonal antibodies enhanced IL36-stimulated antitumor activity by depleting Tregs in the tumor [137]. Mihic-Probst et al. showed that anti-CTLA-4 antibody ipilimumab, anti PD-1 antibody nivolumab, or pembrolizumab decreased the number of CD25+ Treg cells [138]. Sun et al. observed that the number of Treg cells decreased after treating mice with anti-CTLA-4 or anti PD-1 antibodies in an HPV16 E6/E7<sup>+</sup> syngeneic mouse tumor model [139]. Kvarnhammar et al. showed that new IgG1 bispecific anti-CTLA-4 and anti-OX40 induced the activation of T cells and Treg cell depletion in vitro and in vivo in the tumor [140]. Sharma et al., using samples from 19 melanoma, 17 prostate, and 9 bladder cancer patients treated with ipilimumab and 18 samples from melanoma cancer patients treated with tremelimumab, observed that the monoclonal antibodies depleted intratumoral FOXP3 Treg cells in tumors [48]. Pai et al. devised a dual variable domain immunoglobulin of the anti-CTLA-4 antibody (anti-CTLA-4 DVD) possessing an outer tumor-specific antigen-binding site engineered to shield the inner anti-CTLA-4-binding domain. The latter only became available upon reaching the tumor after cleavage of the construct by proteases present in the tumor. In a preclinical tumor model, treatment with the anti-CTLA-4 DVD led to the depletion of tumor-resident Treg cells, while preserving tissue-resident Treg cells, resulting in an efficient antitumor response with a reduced multi-organ immune toxicity [141]. Morris et al. observed that anti-CTLA-4 antibodies IgG2a and IgG2b isotypes of 9D9 clone decreased the number of Treg cells in syngeneic murine tumors of B78 melanoma and/or Panc02 pancreatic cancer [142]. Duperret et al. observed that, upon treatment with anti CTLA-4 in combination with a TERT DNA vaccine administered once a week for four rounds of immunization in C57BL/6 mice, the level of Treg cells decreased within the tumors, while it remained unchanged within the peripheral blood [143]. Tang et al. observed, through IHC and quantitative real-time PCR, that the anti-CTLA-4 monoclonal antibody decreased the presence of Treg cells in the mice tumor microenvironment, but not in peripheral lymphoid organs [144]. Son et al. showed that the anti-CTLA-4 antibody and radiotherapy suppressed CD25 Treg cells in C57BL mice injected with lung cancer [145]. Schwarz et al. investigated the effect of using different doses of anti-CTLA-4 in the presence of Treg cells in mice. They used a low dose of 0.25 mg CTLA-Ig antibody (LD, 10 mg/kg body weight), high dose of 1.25 mg CTLA-Ig antibody (HD, 50 mg/kg body weight), and very high dose of 6.25 mg CTLA-Ig antibody (VHD, 250 mg/kg body weight). Treg cell levels decreased, independently of the doses [146]. Marabelle et al., using a combination of anti-CTLA-4 and anti-OX40 with CpG therapy, observed a reduction of Treg cells in tumors [147].

Interestingly, Du et al. observed that anti-CTLA-4 antibodies are capable of efficiently inducing Treg cell depletion and tumor regression in mice [148].

In contrast, several other groups reported an increase of Treg cells in cancers after anti-CTLA-4 treatment. In fact, Sandi et al. observed that high dose treatment of anti-CTLA-4 increased the accumulation of Treg cells in secondary lymphoid organs [149]. Kavanagh et al. observed that the anti-CTLA-4 antibody ipilimumab in four cohorts of patients increased Treg cell levels in a dose-dependent manner. The drug was administered every 28 days [150]. Quezada et al. observed that a CTLA-4 blockade with GM-CSF combination immunotherapy in an in vivo B16/BL6 mouse model of melanoma led to a self-expansion of Treg cells in tumors [47]. The reason for such discrepancies between the last four studies and the majority of studies described in the previous paragraphs remains unknown. A possible explanation could be that different subpopulations of Treg cells were detected by the groups, such as bona fide and naïve Treg cells, or that the organisms' TMEs of either animals or humans were different across the different experimental settings.

Of note, CTLA-4 has two opposing and crucial properties in cancer and autoimmunity. For self-tolerance it is important to have functional CTLA-4. Current antibodies developed

against CTLA-4 have the property of reducing the levels of CTLA-4 by 50% by lysosomal degradation, which is directly responsible for their toxicity [135]. Therefore, since CTLA-4 is crucial for preventing autoimmunity, which is the major cause of irAE triggered by monoclonal antibodies such as ipilimumab and tremelimumab [135], new drugs should be developed considering such a gap. Encouraging results have already been produced by Zhang et al. HL12 and HL32 anti-CTLA-4 antibodies did not change the CTLA-4 level total or that in the Treg cell fraction, while exerting powerful anti-CTLA-4-induced tumor inhibition [135]. Table 1 summarizes all the studies investigating anti-CTLA-4 therapies' effect on Treg cell levels.


#### **Table 1.** Effects of anti-CTLA-4 therapy on Treg cells.


**Table 1.** *Cont.*

Moreover, in clinical routines, it should also be considered that T cells are made of multiple subpopulations with their own peculiar effects. The modulation of Treg cells and/or Teff cells and pro-inflammatory responses is critical for cancer. An immunosuppressive state (increased Treg and/or decreased Teff) may facilitate the growth and spread of abnormal cancer cells. Therefore, the Treg:Teff ratio could be used in a clinical setting. The new checkpoint inhibitors attempt to pharmacologically modulate the Treg:Teff ratio in the treatment of cancer therapy. However, in cancer progression, the expression of co-inhibitory molecules by tumors favors an imbalance in the tumor microenvironment toward an immune suppression status by increasing Treg infiltration and decreasing Teff activity [151]. On the contrary, the ratio of Treg:Teff should be in favor of Treg depletion and an increase of activated effector T cells, in order to potentiate an anti-tumor response [152]. Tremelimumab was shown to improve the proliferative response of Teff and to abrogate the Treg suppressive ability, suggesting that monitoring these populations may allow for the proper selection of responsive patients from those who would not obtain a benefit from immunotherapy [153]. With regards to the patients' management, it seems to be crucial to understand and monitor the "ping-pong" effect produced by treatment of the Treg:Teff ratio in the regulation of autoimmunity and anti-tumor immunity. Clinicians should pay attention to monitoring this effect in order to maintain an effective anti-tumor response and immune homeostasis preventing the onset of IRAEs [154].

#### **5. Conclusive Remarks and Future Directions**

In conclusion, most studies have shown that anti-CTLA-4 antibodies mainly depleted Treg cells in cancers, whereas very few have observed that the number of Treg cells increased or remained the same because of different experimental settings or in some cases, the design of their therapeutic agents. It is generally known that Treg cells inhibit anti-cancer immunity, blocking the immune surveillance of tumors, ultimately leading to cancer growth. In our opinion, antibodies or small molecules that inhibit CTLA-4, but do not alter CTLA-4 levels in Treg cells, could be innovative and ultimately more effective in eradicating cancer cells. In fact, such drugs would not cause the degradation of CTLA-4 and consequently, do not interfere with Treg cells' function in preventing autoimmunity. Consequently, the inhibition of CTLA-4 could be achieved without the degradation of CTLA-4 and adverse related events caused by toxicity. Testing their efficiency, together with other checkpoint inhibitors, such as anti-PD1 and anti-PD-L1, could further improve the therapy efficacy.

**Author Contributions:** N.S. conceptualized and drafted the initial version of the manuscript, researched the literature, and edited the manuscript; D.R.T.-C. contributed to the draft and revised the manuscript; A.D. contributed as a chemist, providing insights on the mechanisms of CTLA-4 inhibition in immunotherapies; D.G. contributed as an oncologist to the relevance of immunotherapies for cancer and revised the manuscript; R.R. revised the literature and edited the manuscript; Y.L. improved the idea, and revised and finalized the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was financed by the seed fund provided to Y.L. from Baylor College of Medicine and a Recruitment of Established Investigator Award from Cancer Prevention & Research Institute of Texas (grant number RR190043). The funds had no influence on the content of this review.

**Institutional Review Board Statement:** "Not applicable" for studies not involving humans or animals.

**Informed Consent Statement:** "Not applicable" for studies not involving humans.

**Data Availability Statement:** Data sharing not applicable. No new data were created or analyzed in this study. Data sharing is not applicable to this article.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Role of Methylation in Pro- and Anti-Cancer Immunity**

**Ali Mehdi 1,2 and Shafaat A. Rabbani 1,2,\***


**Simple Summary:** Epigenetic mechanisms including methylation play an essential role in regulating gene expression not only in cancer cells but also in immune cells. Although role of DNA methylation has been extensively studied in tumor cells in tumor microenvironment (TME), the understanding of transcriptional regulation of pro- and anti-cancer immune cells in TME is beginning to unfold. This review focuses on the role of DNA and RNA methylation in regulating immune responses in innate and adaptive immune cells during their activation, differentiation, and function phase in cancer and in non-cancer pathologies. Uncovering these crucial regulatory mechanisms can trigger discovery of novel therapeutic targets which could enhance immunity against cancer to decrease cancer associated morbidity and mortality.

**Abstract:** DNA and RNA methylation play a vital role in the transcriptional regulation of various cell types including the differentiation and function of immune cells involved in pro- and anti-cancer immunity. Interactions of tumor and immune cells in the tumor microenvironment (TME) are complex. TME shapes the fate of tumors by modulating the dynamic DNA (and RNA) methylation patterns of these immune cells to alter their differentiation into pro-cancer (e.g., regulatory T cells) or anti-cancer (e.g., CD8+ T cells) cell types. This review considers the role of DNA and RNA methylation in myeloid and lymphoid cells in the activation, differentiation, and function that control the innate and adaptive immune responses in cancer and non-cancer contexts. Understanding the complex transcriptional regulation modulating differentiation and function of immune cells can help identify and validate therapeutic targets aimed at targeting DNA and RNA methylation to reduce cancer-associated morbidity and mortality.

**Keywords:** DNA methylation; RNA methylation; S-adenosylmethionine (SAM); cancer; tumor microenvironment; innate immunity; adaptive immunity; T cells; m6A

#### **1. Introduction**

Epigenetic modifications are heritable changes regulating the cellular gene expression patterns required for the normal development and maintenance of various tissue functions [1–3]. Whereas genetic mutations result in the activation/inactivation of certain genes playing a pivotal role in carcinogenesis, abnormalities in the epigenetic landscape can lead to altered gene expression and function, genomic instability, and malignant cellular transformation (Figure 1) [3,4]. The three most studied epigenetic mechanisms that result in cancer are alterations in DNA methylation, histone modification, and non-coding RNA (ncRNA) expression.

**Citation:** Mehdi, A.; Rabbani, S.A. Role of Methylation in Pro- and Anti-Cancer Immunity. *Cancers* **2021**, *13*, 545. https://doi.org/10.3390/ cancers13030545

Academic Editors: Günther H.S. Richter and Subree Subramanian Received: 17 December 2020 Accepted: 27 January 2021 Published: 1 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** A balance between carcinogenesis and cancer immunosurveillance system. Abnormal genetic modifications such as gene mutations, deletions, amplifications, copy-number variations (CNVs), chromosomal abnormalities, or instability and gene fusions can all result in abnormal expression of genes and proteins leading to transformation of a normal cell into a pre-cancer state and/or cancer stage. Similarly, abnormal epigenetics, such as aberrant DNA methylation patterns, histone modifications, and ncRNA expression (e.g., miRNA) levels, also cause tumorigenesis. Recently, abnormal RNA methylation patterns, such as m6A RNA post-transcriptional modifications (epi-transcriptomics), have been shown to result in the initiation and progression of cancer. Although these abnormalities in malignancy promote tumorigenesis, the cancer immunosurveillance system acts as a tumor suppressor working against the formation of pre-malignant and cancer cells. The cancer immunosurveillance system comprises the innate and adaptive immune systems that have various components that help to regress or eliminate tumor cells. However, some immune cells can be pro-tumor, which paradoxically help tumor progression in the tumor microenvironment. Cancer can evolve and escape the immune system by developing immunosuppressive escape mechanisms (such as high expression of PD-L1) that allow it to progress. This state can be reversed with immunotherapy, such as immune checkpoint inhibitors (ICPi).

#### *1.1. DNA Methylation: Writers, Readers, Erasers, and Co-Factors*

DNA methylation is the most well-characterized epigenetic mechanism, and was linked to cancer as early as the 1980s [5]. Specific DNA methylation patterns are crucial for parental imprinting, genomic stability, and importantly, regulation of gene expression [6,7]. DNA methylation is the covalent addition of a methyl (-CH3) group at the cytosine (C) base adjacent to 5′ of a guanosine (G) [8,9]. The methyl donor for this methylation reaction is s-adenosylmethionine (SAM). In the human genome, more than 28 million CpG dinucleotides exist, and 60–80% show methylation in any given cell [10]. In contrast, there are specific regions where CpG dinucleotides are enriched, called CpG islands, which are primarily located near gene promoters [10]. Increased methylation at CpG islands is typically associated with gene silencing. However, varying levels of DNA methylation at other regions, including gene bodies, enhancers, 5′ and 3′ UTRs, and partially methylated domains (PMDs), can also differentially affect gene expression to regulate dynamic biological processes [11–14].

In mammals, the addition of methyl groups to DNA is carried out by "writers", DNA methyltransferase (DNMT) 1, DNMT3A, and DNMT3B, converting unmodified C into 5-methyl-cytosine (5mC) [15]. DNMT3A and DNMT3B add methyl groups to DNA without

template DNA and hence, undertake de novo methylation, whereas DNMT1, maintenance DNMT, adds methyl groups to hemi-methylated DNA by copying DNA methylation patterns from the parental strand to the daughter strand during cell division. DNMTs utilize methyl groups from SAM, which is a universal methyl donor and acts as a co-factor in this reaction [16].

DNA methylation can be recognized by readers including methyl-CpG-binding domain (MBD) proteins, certain transcription factors, and zinc finger (ZNF) proteins [17]. Generally, methylation of the CpG can directly affect gene transcription by interference with the binding of the transcription factors at a regulatory site leading to transcriptional silencing. In addition, DNMTs and MBD proteins can recruit histone modifiers to the methylated promoter region, and stimulate chromatin condensation and gene silencing [15,18–21].

Methyl groups from DNA can be removed either passively or actively. Active DNA demethylation is performed by "erasers", called ten-eleven translocation (TET), which remove methyl groups from DNA by oxidizing 5mC into 5hmC (5-hydroxymethyl-cytosine), 5fC (5-formylcytosine), and 5caC (5-carboxylcytosine) [22]. The 5fC and 5caC marks are later identified by thymine DNA glycosylase (TDG), and repaired and replaced by unmodified C. Passive DNA demethylation occurs when DNA methylation maintenance proteins are altered or the DNMT1/UHRF1 complex is unable to read 5hmC, 5fC, or 5caC, leaving C on a newly formed strand unmethylated and, due to multiple rounds of cell division, the original DNA methylation patterns are lost [22].

#### *1.2. m6A RNA Methylation: Writers, Readers, and Erasers*

An emerging crucial layer of post-transcriptional gene regulation, N6-methyladenosine (m6A) RNA methylation, plays an essential role in gene expression regulation and development, and human diseases [23–30]. m6A is the most common and characterized modification in RNA amongst 150 other post-transcriptional modifications in eukaryotes [23–30]. Alterations in m6A RNA methylation and its regulators target different genes in various cancers, including melanoma, acute myeloid leukemia (AML), liver cancer, glioblastoma, and breast and pancreatic cancer (Figure 1) [24,26–30]. m6A RNA regulators include writers/methyltransferases, erasers/demethylases, and readers that can add/methylate, remove/demethylate, and read/recognize m6A modified sites on RNA, respectively [23,25,26,28]. The major methyltransferases of m6A are methyltransferase-like (METTL) 3 and METTL14 complexes that add a methyl group donated from SAM to the 6th Adenosine of the RNAs [23,25,26,28]. In contrast, active demethylation of m6A is performed by demethylases AlkB homolog 5 (ALKBH5) or fat mass and obesity-associated (FTO), which remove the methyl groups from the RNA [23,25,26,28]. Readers recognize the m6A modification either directly using the YTH domain (e.g., YTH-domain containing reader; YTHDF1/2/3; or YTHDC1/2) or indirectly, which leads to either RNA degradation or enhanced translation of the mRNA [23,25,26].

#### *1.3. Immune System: Pro- and Anti-Cancer Immunity*

Humans have evolved their immune system, including the innate and adaptive immune systems, to combat a broad range of diseases, including cancer (Figure 1) [31–33]. The innate immune system consists of immune cells including natural killer (NK) cells, dendritic cells (DC), macrophages, and neutrophils. The innate immune system is typically the first line of defense, has a nonspecific and immediate response against pathogens, and exhibits germline inheritance [31–33]. Innate immune cells use pattern recognition receptors (PRRs), such as toll-like receptors (TLRs), and identify pathogens based on nonspecific molecular patterns including single-stranded RNAs or lipopolysaccharide. The adaptive immune system, by comparison, is highly specific and forms the immunological memory. Adaptive immunity comprises lymphocytes, and T and B cells, which produce cytokines and antibodies to counter pathogens [31–33]. A large number of extremely diverse but highly specific receptors on T cells—T cell receptors (TCRs)—and B cells—B cell receptors (BCR)—which recognize and differentiate self from non-self antigens are

extremely useful in response to foreign pathogens. Long-lasting memory cells generated after pathogen clearance provide a rapid and robust pathogen control upon re-exposure to the same pathogen.

After a century of controversy, it has now been established that a functional cancer immunosurveillance system indeed exists, and acts as a tumor suppressor or killer (Figure 1) [31–35]. Interestingly, both innate and adaptive immune systems can recognize and eliminate malignant cells. Components of the immune system in the tumor microenvironment (TME) can be either anti-tumor, regressing or killing tumor cells; or pro-tumor, helping tumor progression. TME is a complex interaction of tumor cells, immune cells, and stromal cells, and is influenced by various factors including cytokines, chemokines, the extracellular matrix, tissue-specific factors, and inflammation [31,36]. Tumor inhibition or progression depends on TME factors, which can be anti- or pro-tumorigenic. Tumor progression is suppressed or eliminated by the cancer immunosurveillance system; however, tumor cells can evolve and develop mechanisms that allow them to evade or escape the immune system (Figures 1 and 2) [31,36,37]. There are three main immune escape mechanisms: (1) loss of antigenicity—tumor cells increase defects in antigen processing and presentation machinery resulting in lower presentation of antigens to immune cells; (2) loss of immunogenicity—tumor cells produce low levels of immunogenic tumor antigens and high levels of immunosuppressive ligands (e.g., PD-L1); and (3) creating an immunosuppressive TME—tumor cells transform to cause alterations in oncogenes and tumor suppressor genes to increase inflammation and recruitment of pro-tumor immune cells in TME.

Solid tumors typically have immune cells that can be anti-tumor or pro-tumor as a result of factors including differentiation (Figure 2). In summary, pro-tumor factors include high type II M2 macrophages; high CD4<sup>+</sup> regulatory T cells (Tregs); high type II CD4<sup>+</sup> Th2 cells; typically low or exhausted tumor infiltrating lymphocytes (TILs) (cold tumor); and low antigenicity and immunogenicity of the tumor cells. In contrast, anti-tumor factors include high NK cells; high type I M1 macrophages; high type I CD4<sup>+</sup> Th1 cells; low Tregs; high tumor infiltrating CD8<sup>+</sup> T cells (memory, cytotoxic); high type I cellular immune response (e.g., IFN-g, IL-2, granzyme B); more and functional TILs (hot tumor); and high antigenicity and immunogenicity of the tumor cells (Figure 2) [31–33,36–44].

Epigenetic mechanisms including miRNAs and histone modifications are crucial for the regulation of the immune system in the TME and has been extensively reviewed [45–49]. DNA methylation also plays an essential role in the differentiation and function of immune cells into various subtypes, and the manner in which these immune cells influence each other in the TME, which ultimately results in tumor progression or suppression. Schuyler et al. [50] carried out analysis of large whole-genome bisulfite sequencing datasets (112 datasets from the BLUEPRINT Epigenome Project) to delineate trends of changes in DNA methylation in different lineages of immune cells, including myeloid and lymphoid cells in TME of various cancer models. Global methylation, in general, increases during macrophage differentiation and activation, whereas it reduces during lymphocyte differentiation (T and B). Numerous studies have also shown methylation changes in the differentiation and activation of proor anti-cancer myeloid and lymphoid cells [22,51,52].

The role of methylation in hematopoiesis and in immune disorders is now well established [22,51,52]. The focus of this review is to discuss the role of DNA and RNA methylation (m6A) and its regulators in key pro- or anti-cancer immune cells of innate and adaptive immune systems. Examples from other non-cancer immune triggering pathologies are also included. Additionally, the translational potential of targeting methylation with DNA methyltransferase inhibitors (DNMTi), methylating agents such as SAM, and m6A RNA demethylase inhibitors in the treatment of liquid and solid cancers is also discussed.

γ α **Figure 2.** An imbalance between pro-tumor and anti-tumor immune cells and factors in the tumor microenvironment (TME) can lead to tumor growth and progression or tumor suppression and elimination. Pro-tumor immune cells can promote tumor progression, including type II M2 or TAMs (tumor-associated macrophages), regulatory T cells (Tregs), and type II Th2 cells. Moreover, factors that influence tumor progression are low tumor infiltrating lymphocytes (TILs) in the TME, low antigenicity and immunogenicity of tumor cells, and inflammation. Anti-tumor immune cells can reduce tumor growth and suppress tumor progression in the TME. These include CD8+ T cells, type I Th1 cells, NK cells, and type I M1 cells and their type I cytokines such as IFNγ, TNFα, IL-2, and granzyme B. Furthermore, anti-tumor immune factors can also influence tumor suppression, including high infiltration of functional TILs, and greater antigenicity and immunogenicity of the tumor cells, such as high MHC-I expression and tumor-associated antigen expression.

#### **2. Role of DNA Methylation in Innate and Adaptive Immunity**

#### *2.1. Innate Immunity*

#### 2.1.1. Dendritic Cells (DCs)

DCs and macrophages are the first innate immunity cell types which are triggered for defense against pathogen invasion. DCs are professional antigen presenting cells (APCs) that are essential for triggering adaptive T cell responses in an antigen-specific manner. DCs can undergo marked changes in their phenotype and function under various stimuli and inflammatory conditions [53]. For instance, DCs can be polarized towards producing specific type of cytokines (e.g., IL-12, IL-23) and Notch ligands (e.g., DLL1/4) to induce different effector CD4 (Th1, Th2, Th17) and CD8 (cytotoxic) T cells [53].

The role of DNA methylation is crucial for regulating differentiation and activation of DCs; however, this has not been fully elucidated, particularly in the TME. Nevertheless, DNA methylation changes have been reported during differentiation of monocytes into DCs and immature DCs (iDCs) into mature DCs (mDCs) [54–57]. Bullwinkel et al. investigated epigenetic changes occurring at *CD14* and *CD209* gene loci, which are essential for the function of monocytes and DCs, respectively, and found CD14 expression was lost, whereas CD209 expression was elevated, upon differentiation from monocytes to DCs [54]. The reciprocal expression changes in CD14 and CD209 were associated with histone modifications at the *CD14* locus leading to *CD14* silencing, whereas loss of "repressive" histone marks and DNA demethylation at the *CD209* locus resulted in CD209

transcriptional activation. Zhang et al. carried out a comprehensive study of DNA methylation changes at single nucleotide-resolution for human monocytes and monocyte-derived iDCs and mDCs [56]. Several known genes and pathways regulating DC differentiation and maturation were identified. A total of 1608 differentially methylated positions (DMPs) from monocytes to iDCs and 156 DMPs from iDC to mDCs were identified. Major DNA demethylation occurred at the binding sites of the transcription factors of genes involved in DC differentiation and function that ultimately increased transcription of these genes. Moreover, the demethylation was locus-specific, and is associated with changes in DNA methylation regulators, including DNMT1, DNMT3A, DNMT3B, and TET2 [56]. Interestingly, DNA methylation reader, MBD2, in DCs was previously shown to have a dominant role in inducing CD4+ T cells differentiation into the Th2 cell type. Specifically, loss of Mbd2, resulted in reduced phenotypic activation of DCs and capability to initiate Th2 immunity against helminths or allergens [58]. In addition, during IL-4-mediated differentiation from human monocytes to DCs and macrophages, TET2 was identified as the main regulator of DNA demethylation of dendritic cell-specific or macrophage-specific gene sets mostly in intergenic regions and gene bodies [57]. Essentially, the IL-4-JAK3-STAT6 pathway is required for dendritic cell-specific demethylation and expression signature, and STAT6 also prevents demethylation of macrophage-specific genes required for monocyte to macrophage differentiation. Pacis et al. performed a comprehensive epigenome and transcriptome analysis of DCs infected with a live pathogenic bacterium (*Mycobacterium tuberculosis*) [59]. A rapid and active DNA demethylation at distal enhancers was identified that activates master immune transcription factors such as NF-κB and IFN regulatory families [59]. Although the above studies provide strong evidence of DNA methylation regulating monocyte to DC differentiation, and activation of DCs, the role of DNA methylation in the TME needs further characterization.

#### 2.1.2. Macrophages

Macrophages are myeloid cells that have a spectrum of phenotypes in which M1 or M2 subtypes are the extreme ends. M1 cells are "classically activated" by IFNγ, and destroy tumor cells through their production of nitric oxide and type 1 cytokines and chemokines [31,60]. Moreover, M1 act as APCs to activate cytotoxic CD8<sup>+</sup> T cells in an antigen (Ag)-specific manner. M2 cells are activated by "alternative" pathways via IL-4, IL-13, and/or TGFβ [31,60]. M2 secrete type II chemokines and cytokines, thereby promoting tumor growth and progression. Stromal and tumor-associated factors in the TME can shift macrophages to M2 types, specifically the tumor-associated macrophages (TAMs) type that promotes angiogenesis, tumor progression, and metastasis [60–62]. The differentiation from monocyte into macrophages and between the M1 or M2 (or TAMS) phenotype is regulated by DNA methylation at lineage-specific promoter and enhancer regions.

Upon examining global DNA methylation between human monocytes, naïve macrophages, and activated macrophages, Dekkers et al. reported major DNA methylation changes during monocyte to macrophage differentiation [63]. Differential methylation was generally fixed to short regions or single CpGs, and was prevalent at lineage-specific enhancers. The differential methylation was either gain (e.g., *IRF8*, *CEBPB*) or loss (e.g., *PPARG*) of methylation at specific transcription factor binding sites involved in monocyte to macrophage transition. Authors also analyzed different types of activated macrophages and found some genes for lipopolysaccharide (LPS)/IFNγ macrophage-specific activation (e.g., *CCL5*). In another study, the transcriptome and epigenome of human monocytes differentiated into macrophages with colony-stimulating factor 1 (CSF1) identifying several RNAs (mRNA and miRNAs) that are differentially expressed [64]. In addition, 100 differentially methylated regions (DMRs) between monocytes and macrophages were identified in enhancer regions that were uniquely demethylated in macrophages and repressed in monocytes, and were linked to actin cytoskeleton, phagocytosis, and innate immune response [64]. Evidence has shown that both methyltransferases DNMT1 and DNMT3A/B play a vital role in differentiation and macrophage polarization [51]. For instance, knock-down (KD) of

DNMT3B in RAW264.7 cells showed a higher polarization towards the M2 macrophage phenotype compared to M1, and leads to suppressed inflammation; the opposite pattern was observed for overexpression of DNMT3B [65]. During chronic inflammation, DNMT1 expression is elevated and has been associated with DNA hypermethylation. A study examined the role of TAMS in DNA methylation of a tumor suppressor gene gelsolin (*GSN*) during gastric cancer progression. Firstly, DNMT1 overexpression was shown to methylate and silence the *GSN* gene, and secondly, DNMT1 overexpression was associated with higher TAMs infiltration in the TME of gastric cancer [66]. Further analysis revealed that TAMs secreted CCL5 that triggered DNMT1 overexpression by activating the JAK2/STAT3 pathway in gastric cells, resulting in GSN silencing and tumorigenesis. In another study, DNMT1 was associated with M1 polarization by silencing the *SOCS1* gene and a subsequent increase in tumor necrosis factor (TNF) and IL-6 production [67]. Furthermore, DNMT1 overexpression was shown to promote M1 activation induced by LPS and IFNγ [67].

In contrast, TET proteins appear to have a role in the downregulation of inflammatory gene expression in normal myeloid cells [22]. In a model of TET2-deficient macrophages and DC, a higher expression of IL-6 was observed upon stimulation [68]. TET2 was shown to reduce IL-6 expression by interacting with Iκbζ (a member of the nuclear IκB family) and binding to the IL-6 promoter region in addition to recruitment of histone deacetylase 2 (HDAC2) [69]. Furthermore, Tet2-deficient mice are more susceptible to septic shock and colitis induced by endotoxin and dextran sulfate sodium (DSS), respectively, both due to elevated IL-6 expression [69]. TET2 expression is elevated in tumor infiltrating myeloid cells of both melanoma patients and mouse models via the IL-1R-MyD88 pathway. Moreover, TET2 acts as an oncogene in melanoma tumorigenesis by suppressing anti-cancer immune cells [70]. This is consistent with the TET protein acting as anti-inflammatory to myeloid cells [22]. Overall, these studies show the role of DNA methylation in regulating monocyte to macrophage differentiation and macrophage polarization.

#### 2.1.3. Natural Killer (NK) Cells

NK cells can directly lyse MHC class I-deficient tumor cells [31,35]. NK cells have activating receptors that identify malignant cells expressing stress-induced ligands (e.g., MICA) [31,35]. NK cells kill the tumor cells by making them undergo apoptosis through either expressing death ligands (e.g., Fas ligand) or by releasing granzymes and perforin [31,35].

The role of DNA methylation in NK cells' activation or differentiation has not been fully elucidated. However, it was reported that the MHC-I cytotoxicity of NK cells, which is mediated by the KIR (killer cell Ig-like receptor) family, is regulated via methylation. In progenitor cells, KIR genes are silenced via hypermethylation and histone modifications, whereas in KIR-expressing cells, such as NK cells, KIR genes are demethylated and expressed [71]. Furthermore, work with human cytomegalovirus (HCMV) viral infection has shown that, upon infection, subjects have elevated levels of a "memory-like" subtype of NK cells which survive long term and have increased response upon re-exposure of the same pathogen. These memory-like NK cells are characterized by activation of NKG2C, which is in turn epigenetically regulated. In addition, in some HCMV-infected patients, memorylike NK cells were reported to lack B-cell and myeloid signaling proteins such as tyrosine kinase SYK. Further analysis showed that the gene promoter of *SYK* was hypermethylated and SYK expression was downregulated [72]. HCMV-associated NK cells also have low expression of signaling adaptors, including EAT-2, FCER1G, and transcription factor PLZF due to hypermethylation at their DNA [73]. Wiencke et al. examined human naïve vs. activated NK cells' DNA methylome and found reproducible genome-wide DNA methylation changes [74]. Methylation analysis showed primarily CpG hypomethylation (81% of significant loci) during activation of NK cells. Several previously reported and novel genes or pathways associated with activation of NK cells were identified. The high priority gene *BHLHE40* had high demethylation in activated NK cells, whereas it had low demethylation in naïve NK cells and was shown to be a potential biomarker for NK activation in

peripheral blood. Interestingly, increased NK cells and CD8+ T cells tumor infiltration was reported using the DNA methyltransferase inhibitor (DNMTi), AzaC, through type I IFN signaling while reducing the tumor burden of the murine epithelial ovarian cancer model [75]. Histone deacetylase inhibitors (HDACi) lead to further activation of these anti-tumor immune cells and reduction in pro-tumor macrophages in the TME. Furthermore, ligands (such as ULBPs and MICA) of NK cells activating receptor NKG2D, which are essential for NK cell lytic activity, are downregulated in gliomas and hepatocellular carcinoma (HCC) cells via DNA methylation and histone methylation, respectively [76,77]. Indeed, treatment with DNMTi and Enhancer of zeste homolog 2 (EZH2) inhibitor was shown to upregulate NKG2D ligand expression, resulting in the lysis of glioma and HCC cells by NK cells, respectively. These studies show that DNA methylation not only controls the critical gene expression in NK cells that regulates differentiation and activation of NK cells but also genes in cancer cells that regulate NK cell tumor lytic activity.

#### *2.2. Adaptive Immunity*

Binding of the T cell receptor (TCR) present on T cells to the antigen/MHC complex (signal 1) expressed on APCs is essential for the activation of naive T cells [78]. Additional binding of positive co-stimulatory molecules present on activated APCs, called signal 2 (e.g., CD80/86 and B7RP1 on APCs onto CD28 and ICOS on T cells, respectively), helps in further activation. TCR activation is a multistep process that leads to an intracellular signaling cascade that results in activation, differentiation, and proliferation (clonal expansion) of T cells, and transforms them into effector cells producing cytokines [78]. DNA methylation has a key role in regulating these processes. For instance, upon TCR stimulation of T cells, IL-2 is highly expressed and is required for T cell activation and clonal expansion in mouse [79]. The increase in IL-2 cytokine results from active demethylation at a promoter-enhancer region of the *IL-2* locus upon T cell activation and remains demethylated afterwards [79]. In addition to IL-2 cytokine, DNA methylation also plays an important role in the activation, proliferation, and effector functions of CD4 and CD8 T cells as discussed below.

#### 2.2.1. CD4<sup>+</sup> T Cells

CD4<sup>+</sup> T cells are unique T cells that can, depending on the nature of the Ag signal and type of cytokine stimulation, differentiate into various subtypes including helper T cell 1, 2, and 17 (Th1, Th2, and Th17) and Tregs (Figure 3). Th1 produce type I cytokines, including IL-2 and IFNγ, facilitating optimal expansion, trafficking, and effector functions of CD8<sup>+</sup> T cells, thereby reducing tumor growth and progression [31,36,37]. In contrast, Th2 produce type II cytokines (IL-4, IL-5, and IL-13) and polarize immunity towards tumor progression [31,36,37]. This differentiation of CD4<sup>+</sup> T cells into various subtypes is regulated by DNA methylation (Figure 3) [31,36,37]. The differentiated CD4 T cells then regulate downstream immune functions, such as enhancement of CD8 T cells, macrophages, and B cell effector functions, and immunological memory.

Numerous studies have analyzed the methylation status of immune genes and correlated it with immune responses in the TME (Figure 3). Upon antigenic stimulation, naïve CD4<sup>+</sup> T cells differentiate into Th1 and Th2 by epigenetically activating or silencing a certain set of genes, usually by DNA demethylation and hypermethylation, respectively [80–82]. By analyzing the methylation status of a key gene, *IFNG* or *IFNγ*, essential for anti-tumor activity, Janson et al. reported demethylation of the *IFNγ* gene promoter and enhancer, and upregulation of IFNγ in Th1 cells [83]. In contrast, Th2 cells had hypermethylation at the *IFNγ* gene promoter and had low IFNγ expression. Studies show that naïve T cells that develop in the thymus have hypermethylated DNA at enhancer regions of the *IFNγ* and *IL-4* cluster (IL-4, IL-5, IL-13), and methylated H3K27me3 marks [80,81]. These marks limit chromatin accessibility and inhibit transcription of these genes and hence, naïve T cells minimally transcribe these genes. Interestingly, these regions become demethylated in T cell lineages that require expression of these cytokines—for instance, the demethylated

promoter of the *IFNγ* gene in Th1 and CD8<sup>+</sup> T cells [81]. These CpGs are maintained by Dmnt1 as deletion of *Dnmt1* results in global hypomethylation in naïve precursors, including DNA regions which are normally hypermethylated at these cytokine regulatory regions [84]. For instance, in Dnmt1-deficient mice, naïve T cells produce effector cytokines such as IFNγ immediately after activation. This shows that Dmnt1 is required to maintain these hypermethylated regions during T cell development to suppress and induce cytokine gene expression in naïve and active T cells, respectively [84,85]. Indeed, Th1 cells produce 100-times more IFNγ transcripts than naïve T cells but the *IL-4* gene loci are silenced [81]. γ

γ *γ* γ *γ* **Figure 3.** Role of DNA methylation in regulating differentiation and activation of naïve CD4<sup>+</sup> T cells into effector cells including Th1, Th2, Th17, and Tregs subtypes. DNA methylation changes during differentiation can lead to formation of subtypes of CD4<sup>+</sup> T cells. The black boxes are cytokines that help in the differentiation and activation process for each subtype. For instance, Th1 are formed when naïve CD4<sup>+</sup> T cells are stimulated by IL-12 and IFNγ cytokines and the *IFNγ* gene promoter remains hypomethylated and IFNγ is highly expressed. For the Th2 subtype, the *IL-4* gene is demethylated and is highly expressed, whereas *IFNγ* is methylated and repressed. For Th17 cells, the *IL-17* gene is demethylated and highly expressed. For Tregs, *FOXP3* is demethylated at various regions, including promoter and enhancer, thereby markedly increasing FOXP3 expression. These methylation levels are maintained by DNMT1, DNMT3A, and TET2. The green boxes indicate the cytokines released from differentiated cells. These immune cells and released cytokines can further lead to tumor progression or suppression.

*γ γ* γ *γ* γ In contrast, some genes have the opposite pattern, i.e., they have hypomethylation in naïve cells but hypermethylation in differentiated T cells. For example, the *IFNγ* promoter region is unmethylated in naïve CD4<sup>+</sup> T cells and continues to be hypomethylated upon Th1 cell differentiation; however, upon Th2 cell differentiation, which do not produce IFNγ, the *IFNγ* promoter is methylated via *de novo* DNA methylation by Dnmt3a [86,87]. Moreover, in mouse, *Dnmt3a* deletion in T cells can lead to a complete failure of naïve T cell differentiation into Th2, Th17, and iTreg lineage cells, due to their inability to methylate DNA (de novo) by Dnmt3a at the *Ifnγ* promoter region [88]. Indeed, *Dnmt3a* expression is stimulated upon TCR activation and is recruited to the *Ifnγ* promoter region to carry out methylation in Th2 cells [89]. In addition, deregulated de novo methylation patterns resulted in reduced histone silencing mark (H3K27me3) and increased transcriptionally active histone mark (H3K4me3) upon re-stimulation in the presence of IL-12 [81,88]. Furthermore, Th2 cells produce high

amounts of IL-4 as a result of DNA hypomethylation at the *IL4* gene loci and transcriptional activation, whereas in naïve T cells, the *IL4* gene loci are hypermethylated [88]. Finally, during differentiation of naïve CD4<sup>+</sup> T cells into memory CD4<sup>+</sup> T cells a global loss of DNA methylation was observed, suggesting a role of DNA methylation in memory CD4<sup>+</sup> T cell formation [51]. These data suggest that CD4<sup>+</sup> T cells differentiation into Th1, Th2, Th17, and memory subtypes require DNA methylation changes at gene promoters and enhancers of critical genes such as *IFNγ* and *IL-4* (Figure 3) [36,81–83,88].

Strong evidence suggests that the MBD proteins together with the nucleosome remodeling deacetylase (NuRD) complex are essential in regulating DNA methylation-dependent differentiation of T cells [90–92]. For instance, loss of either MBD2 or NuRD complex can result in polarization of CD4+ T cells to Th2 cell type. Aoki et al. suggested that the NuRD– MBD2 complex may be required for the demethylation of gene loci encoding cytokines specific for Th2 differentiation [91]. Mechanistically, the chromodomain-helicase-DNAbinding protein 4 (Chd4) subunit of the NuRD–Mbd2 complex forms a complex with Gata3 that both activates Th2 cytokine transcription and represses the Th1 cytokine, IFN-γ, by forming a transcriptional activation complex at Th2 cytokine gene loci and a transcriptional repressive complex at the Tbx21 (encoding T-bet) gene locus in Th2 cells, respectively (Figure 3) [90]. TET proteins have also been linked to the differentiation and function of CD4<sup>+</sup> T cells (Figure 3). A study analyzing 5-hydroxymethyl-cytosine (5hmC) patterns in CD4<sup>+</sup> peripheral T cells found a positive correlation between 5hmC alterations at gene bodies of transcription factors, including *Tbx21* and *Gata3*, which drive differentiation into Th1 and Th2 subtypes and their expression levels, respectively [93–95]. Similarly, another study suggested similar Th1/2-specific 5hmC alterations during differentiation of human CD4<sup>+</sup> T cells [93]. In addition, a Tet2 knock-out (KO) mouse model was reported to have Th1 and Th17 cells producing low IFNγ and IL-17, respectively [94]. Overall, these studies suggest that not only DNA methyltransferases (DNMT1 and DNMT3A/B) are required for regulating differentiation of CD4<sup>+</sup> T cells into various subtypes but also DNA readers and DNA demethylases such as MBD2 and TET proteins, respectively [22,93–95].

#### Regulatory T Cells (Tregs)

Tregs can be natural (nTreg), i.e., derived from the thymus, or Ag-induced (iTreg), i.e., differentiated from naïve T cells by TGF-β and IL-2 in the periphery (Figure 3) [31,36,37]. Tregs typically act as pro-tumor, are immunosuppressive, and are associated with poorer prognosis in several cancer types [35,96]. Tregs block the activation of CD8<sup>+</sup> T cells through expressing cytotoxic T lymphocyte antigen 4 (CTLA4), which is an inhibitory molecule for CD8<sup>+</sup> T cells [31,96]. In addition, inflammation enhances Treg function because prostaglandin E2 (PGE2) causes differentiation of Tregs. Tregs were also reported to block killing by NK cells, and thus downregulate both adaptive and innate anti-tumor immunity [31,97].

A master regulator switch for Tregs is FOXP3, which is required for its functions (Figure 3). DNA methylation of *FOXP3* together with intergenic CD3G/CD3D regions were utilized as a biomarker for TILs and Treg quantification in several tumor tissues [98]. This DNA methylation-based quantification of immune cells was even comparable to flow cell cytometry and outperformed IHC techniques. Using differential methylation analysis between nTreg, naive CD4<sup>+</sup> T cells, activated CD4<sup>+</sup> T cells, and iTreg, Lal et al. found a unique CpG site at the enhancer of *Foxp3* that was unmethylated in nTreg compared to other Tregs that were heavily methylated at this locus [99]. Demethylation by DNMTi (Aza) promoted acetylation of histone 3, and interaction with TIEG1 and Sp1, which ultimately led to upregulation of *Foxp3*. To study Tregs in non-small-cell lung cancer (NSCLC) using a co-culture system, Ke et al. showed demethylation of *FOXP3* in the promoter region increased FOXP3 expression in Tregs, which led to downregulation of immune response in the TME (Figure 3) [100].

Treg-specific demethylated region (TSDR) is a CpG dinucleotide dense region which is within the conserved non-coding sequences 2 (CNS2) located in the first intron of the *FOXP3* gene [101]. DNA demethylation at the TSDR region can discriminate between Tregs and

other cell types [102]. Interestingly, using ChIP analysis, Wang et al. showed that MBD2 binds to the TSDR site of the *FOXP3* locus in Tregs [103]. Knocking down Mbd2, in vitro and in vivo, reduced the number of Tregs and impaired Treg-suppressive function (Figure 3). Surprisingly, this was due to increased methylation (>75%) of the TSDR in the Mbd2-/- Tregs because: (i) WT Tregs had a complete TSDR demethylation; and (ii) expressing Mbd2 in Mbd2-/- Tregs rescued the TSDR demethylation. TET proteins are essential for stable Foxp3 expression because they were shown to demethylate the CNS2 region as well as another non-coding sequence, CNS 1, in the *Foxp3* gene (Figure 3) [104,105]. Deletion of Tet2/3 in CD4<sup>+</sup> T cells of mice led to hypermethylation of CNS1 and 2 in Tregs. Moreover, deletion of *Tet1/2* also resulted in hypermethylation of CNS2 [104,105]. Overexpression of the TET1 catalytic domain in CD4<sup>+</sup> T cells also resulted in partial demethylation of CNS2 and differentiation of CD4<sup>+</sup> into iTregs in vitro [106]. TET2 protein may function via interacting with the MBD2 protein because loss of MBD2 resulted in hypermethylation of TSDR in CNS2 [103]. In TME, higher demethylation at the TSDR FOXP3 locus in adjacent normal tissues in colon cancer patient samples were associated with distant metastases and worse recurrence-free survival. The poor survival rates could be due to abnormal recruitment of nTregs in TME [101]. Collectively, these studies show a potential role of DNA methylation in controlling the effector function of Tregs through regulating the expression of the master switch FOXP3 of Tregs.

#### 2.2.2. CD8<sup>+</sup> T Cells

CD8<sup>+</sup> T cells control tumor growth and kill tumor cells directly in an Ag-specific manner within the TME [31,36,37]. The CD8<sup>+</sup> T cells, upon recognizing an Ag, can undergo activation and clonal expansion, thereby carrying out effector functions, such as cytokine production (IFNγ, TNFα), and these processes are regulated by DNA methylation (Figure 4) [31,36,37,78].

Epigenetic mechanisms that govern these processes are largely unknown. A study was conducted to delineate these mechanisms and compared Ag-specific naive and effector CD8<sup>+</sup> T cells after stimulating them with an acute CMV viral infection [107]. The DNA methylome was rewired globally upon effector differentiation of CD8<sup>+</sup> T cells, and a negative correlation between DNA methylation at gene promoters and gene expression was observed. The DMRs were associated with transcription binding sites and promoters of genes that control effector CD8<sup>+</sup> T cell function. For instance, DMR at promoters of *Gzmb*, which encodes a serine protease granzyme B essential for cytolytic function, and *Zbtb32*, which encodes a transcription factor induced in activated lymphocytes, was demethylated and had high expression in the effector CD8<sup>+</sup> T cells compared to naïve cells. In contrast, *Ccr7*, *Ccr2*, *Ccr9*, and *Tcf7*, essential for naïve T cell development and homeostasis, were methylated and had reduced expression. Another study examined *Dnmt3a* KO CD8<sup>+</sup> T cells and found effector functions to be normal; however, *Dnmt3a* KO T cells developed into fewer terminal effector cells and more memory precursors in a T-cell intrinsic manner. This was due to ineffective repression of *Tcf1* expression by Dnmt3a in *Dnmt3a* KO T cells [108]. The role of Dnmt1 in regulating T cell activation and production of Ag-specific effector and memory CD8<sup>+</sup> T cells after a viral infection was also investigated. Dnmt1 was knocked-out at the time of activation and *Dnmt1*-/- had marked reduction (>80%) in Ag-specific clonal expansion in effector CD8<sup>+</sup> T cells but only moderately affected memory CD8<sup>+</sup> T cells. Even in reduced T cell expansion, the infection was effectively controlled. Thus, Dnmt1 may be required for proliferation of Ag-specific CD8<sup>+</sup> T cells but not differentiation into effector and memory CD8<sup>+</sup> T cells [109].

γ α

*γ* γ **Figure 4.** Role of DNA methylation in regulating differentiation and activation of naïve CD8+ T cells into effector cells, including cytotoxic and memory T cell subtypes. DNA methylation changes during differentiation and activation can lead to formation of subtypes of CD8+ T cells. For instance, cytotoxic CD8+ T cells are formed due to whole genome remodeling and expression, and repression of various genes in naïve CD8+ T cells. The genes that are essential for activation, proliferation, and effector functions are demethylated and highly expressed, such as *IL-2*, *IFNG* or *IFNγ*, and *GZMB*, whereas genes that are not required are methylated and repressed (e.g., *TCF7*). Although methylation and gene silencing are usually positively correlated, there are examples of genes that could be methylated and expressed, such as *HAVCR2*, depending upon the precise location of the methylation. In memory CD8+ T cell differentiation, effector genes remain demethylated, whereas methylation at naïve T cell-associated genes are gained and repressed, such as in the case of *TCF1*. These methylation levels are maintained by DNMT1, DNMT3A, and TET2. The green boxes indicate the cytokines released from differentiated cells. These immune cells and released cytokines can further lead to tumor suppression and elimination. However, CD8+ T cells can become exhausted in the TME, highly expressing exhaustive markers such as CD39, CD103, PD-1, and CTLA-4. The exhausted CD8+ T cells are non-functional and produce low amounts of effector cytokines (e.g., IFNγ).

Memory CD8<sup>+</sup> T cells, which are formed from a subset of effector CD8<sup>+</sup> T cells after Ag/pathogen clearance, remain in the blood and lymphoid organs for a long time, giving long-term immunity. These memory CD8<sup>+</sup> T cells also resemble naïve T cells as they have pluripotency and can travel to lymph nodes and the spleen. A study comparing memory CD8<sup>+</sup> T cells with terminal effector cells found that memory cells formed from effector cells gain de novo DNA methylation patterns at naïve CD8<sup>+</sup> T cells-associated genes while becoming demethylated at the loci that are effector-specific genes [110]. *Dnmt3a* KO in effector T cells resulted in reduced DNA methylation and quicker re-expression of naïve T cell genes, decreasing the time for memory T cell development. Therefore, in memory CD8<sup>+</sup> T cells, DNA methylation repression at the naïve-related genes can be reversed and effector genes remain demethylated without the need for memory cells to differentiate, allowing them to become faster effector CD8<sup>+</sup> T cells upon Ag/pathogen re-exposure.

Long-lived memory CD8<sup>+</sup> T cells can be identified with a few markers, such as CD127hi and KLRG1low. CD127low and KLRG1hi are typically markers for short-lived effector CD8<sup>+</sup> T cells. Moreover, transcription factors, including T-bet, Eomes, Blimp-1, Bcl-6, Irf4, and Runx3, define the fate of activated CD8<sup>+</sup> T cells and these are further regulated by DNA methylation. In a mouse model with Tet2-deficient CD8<sup>+</sup> T cells infected with lymphocytic choriomeningitis virus (LCMV), CD8<sup>+</sup> T cells differentiated more into long-

lived memory cells having gp33-specific memory markers, KLRG1low CD127hi, and less into effector short-lived effector cells (CD127low and KLRG1hi) [111]. These memory-like cells had markers of central memory cells expressing CD27, CD62L, and CXCR3, and high expression of transcription factor Eomes compared to wild-type Tet2. Furthermore, these memory cells also had superior pathogen control upon re-challenge. Global methylation analysis revealed several DMRs that gained 5mC/5hmC in Tet2-deficient cells versus WT CD8<sup>+</sup> T cells. These DMRs were present in transcriptional regulator genes known to be vital for effector and memory CD8<sup>+</sup> T cell differentiation. Pharmacological inhibition of TET2 by 2-HG also showed similar results to genetic *Tet2* KO, such as a decrease in 5hmC and an increase in Eomes and CD62L expression [112]. The role of MBD2 in the differentiation of naïve CD8<sup>+</sup> T cells into effector and memory cells was determined following LCMV infection. In contrast to Tet2-deficient CD8<sup>+</sup> T cells, Mbd2-deficient mice had a reduced number of Ag-specific memory CD8<sup>+</sup> T cells and an effective primary effector CD8<sup>+</sup> T cell response leading to a rapid viral clearance. Essentially, generation of precursor memory CD8<sup>+</sup> T cells (IL-7Rα high) was delayed and the MBD2 KO memory cells were phenotypically defective with altered memory cell markers (e.g., IL-7Rα, KLRG-1, CD27) and cytokine production, and were unprotective against re-challenge (Figure 4) [113]. These studies suggest a key role of MBD2 and TET proteins in regulating the differentiation of CD8<sup>+</sup> T cells into memory versus effector cells. Together, the above studies show the crucial role of DNA methylation in differentiation of naïve CD8<sup>+</sup> T cells into effector cytotoxic CD8<sup>+</sup> T cells and memory CD8<sup>+</sup> T cells (Figure 4).

#### **3. Role of DNA Methylation in Regulating T Cell Exhaustion**

If an Ag is exposed to CD8<sup>+</sup> T cells for a long time, CD8<sup>+</sup> T cells can become nonfunctional or exhausted, resulting in reduced effector functions, such as decreased cytokine production (IFNγ and TNF-α) and/or loss of cytotoxicity (e.g., low granzyme B production). Exhausted T cells generally have high surface expression of multiple inhibitory molecules, such as PD-1, TIM3, LAG3, TIGIT, and 2B4, and transcription factors associated with high PD-1 expression are T-bet, Eomes, and YY1 [114–116]. YY1 is a key transcription factor that can regulate the inhibitory molecules PD-1, LAG3, and TIM3 expression, and was shown to have downregulated IL-2 via EZH2 activation, features characteristic of exhausted T cells [114–116]. In human patient tumors treated with immune checkpoint inhibitors (ICPi), around 72% of TILs were found to be dysfunctional. These TILs showed different stages of differentiation and interestingly, had higher proliferation rates compared to effector T cells, ruling out the possibility that exhausted T cells have low proliferation rates [114–116].

CD8<sup>+</sup> TILs become exhausted and lose their effector functions in the TME due to numerous factors, such as immunosuppressive mechanisms by tumor cells. Analyzing the transcriptome and methylome of CD8<sup>+</sup> TILs in the TME of colorectal cancer simultaneously, Yang et al. confirmed tumor-reactive TILs have an exhausted tissue-resident memory signature [117]. They showed tumor-reactive markers CD39 and CD103 of CD8<sup>+</sup> TILs were demethylated and CD8<sup>+</sup> TILs had an exhausted phenotype, including high expression of CTLA4, HAVCR2, LAYN, and TOX [117,118]. To delineate changes in methylation from naïve to cytotoxic CD8<sup>+</sup> T cell phenotype and cytotoxic to exhausted CD8<sup>+</sup> T cell phenotype, promoter methylation of these cells was compared. Naïve CD8<sup>+</sup> T cells showed the most promoter demethylation compared to cytotoxic and exhausted T cells; however, essential cytotoxic CD8<sup>+</sup> T cell effector genes, including *PRF1*, *GZMB*, *IFNG*, *CCL4*, *CCL3*, *CST7*, and *NKG7*, went through hypermethylation to hypomethylation from naïve to cytotoxic CD8<sup>+</sup> T cell differentiation, respectively [117]. For exhausted T cells, two inhibitory checkpoint markers, *PDCD1* (encoding PD-1) and *CTLA4*, were demethylated within cytotoxic CD8<sup>+</sup> T cells. Moreover, *LAG3* and *LAYN* were also differentially methylated from naïve to cytotoxic CD8<sup>+</sup> T cell transition [117]. Therefore, these studies determined that aberrant DNA methylation at certain gene loci could result in T cell exhaustion (Figure 4) [116–118].

Interestingly, DNA methylation could determine if T cell exhaustion can be reversed. In chronic LCMV infection, the PD-1 gene promoter of the effector CD8<sup>+</sup> T cells remained

unmethylated, whereas the exhausted T cells showed complete demethylation [116,119]. Furthermore, studies analyzing the chromatin states using transposase-accessible chromatin using sequencing (ATAC-seq) have determined two chromatin states that define exhaustion: one in which T cell factor 1 (TCF1) transcription sites are closed and another in which transcription sites for eE2F, ETS, and KLF family proteins are opened (Figure 4) [120]. Low TCF1 expression is associated with the low effector function of CD8<sup>+</sup> T cells and nonrenewal of CD8<sup>+</sup> effector T cells [121]. DNA methylation can, therefore, regulate the state of exhaustion of CD8+ T cells, which, due to the reversable nature of DNA methylation patterns, provides new opportunities for therapeutic intervention.

#### **4. Role of m6A RNA Methylation in Immunity**

m6A has various functions, including mRNA stability, translation, splicing, and phase separation, and also takes part in cell differentiation and development [23–30]. These essential functions indicate that m6A RNA methylation can potentially regulate immunity. Although the role of m6A RNA methylation in immunity has not been fully elucidated, few studies have reported its role in both innate and adaptive immune response [122–131].

#### *4.1. Role of m6A RNA Methylation in Innate Immune Response*

Certain DNA and RNA molecules can be detected by the innate immune system as non-self entities via PPRs, such as TLRs. For instance, a study investigated the mammalian innate immune response of DCs through stimulation with DNA, RNA, and modified RNAs, including m6A-modified RNA [128]. Although DNA containing methylated CpG were not stimulatory, RNA could be stimulatory or not stimulatory depending upon modification on RNA [123,129,130]. Modified RNA, including m6A modification exposed to DCs, did not activate their TLR3, TLR7, and TLR8, and led to lower cytokines and activation markers, compared to DC stimulated with unmodified RNA that activated TLRs [123]. Unmodified RNA that is present in bacteria could trigger innate immune response to bacterial infection, whereas highly modified RNA, such as mammalian RNA, would not, indicating a role of RNA modifications in selectively triggering the immune system against pathogens. Indeed, DC are activated via m6A RNA modifications and lack of METTL3 can result in lack of DC maturation [123,128,129]. Regulators of m6A RNA, METTL14, and ALKBH5 were reported to regulate type I IFN production triggered by dsDNA or HCMV [125,129,130]. Depletion of METTL14 decreased viral replication and induced IFNβ1 mRNA production and stability upon dsDNA and HCMV infection, whereas ALKBH5 depletion had an opposing effect (with the exception of affecting IFNβ1 mRNA stability). This control of IFNβ1 mRNA was due to m6A modification at the coding sequence and the 3' UTR of the *IFNβ1* gene. Another study reported increased interferon-stimulated genes upon METTL3 (m6A writer) or YTHDF2 (m6A reader) deletion. Specifically, following deletion of METTL3 or YTHDF2, mRNA of IFNβ was modified at m6A, increasing its stability [125,129,130]. These studies indicate that m6A can play a role in the negative regulation of anti-viral response by dictating increased turnover of IFN mRNAs. One study established a key link of m6A to cellular antiviral response by showing that m6A induces antiviral immunity as it regulates crucial proteins of innate immunity [131]. Mechanistically, m6A demethylase ALKBH5 is recruited by RNA helicase DDX46 to remove m6A from 3' UTRs of genes encoding TRAF3, TRAF6, and MAVS, thereby reducing export of their transcript out of the nucleus and subsequently preventing production of type I IFNs.

#### *4.2. Role of m6A RNA Methylation in Adaptive Immune Response*

m6A RNA methylation has also been shown to regulate adaptive immune responses. Similar to DNA methylation regulating differentiation of CD4<sup>+</sup> T cells into various subtypes, m6A RNA methylation was shown to regulate differentiation of CD4<sup>+</sup> T cells [124]. The authors utilized a conditional KO mouse model (CD4<sup>+</sup> -CRE conditional Mettl3 flox/ flox) to delete Mettl3 in CD4<sup>+</sup> T cells [124]. After validating Mettl3 deletion, they checked for thymocyte differentiation or cellularity and found no difference compared to WT mouse.

However, the proportion of naïve T cells (CD44lo CD62Lhi) was higher in spleens and lymph nodes compared to WT. When the function of Mettl3-/- CD4<sup>+</sup> T cells was compared to WT, they observed normal sensitivity to TCR signaling; however, T helper polarization had abnormalities. For instance, the KO CD4<sup>+</sup> T cells had a significant reduction in differentiation into Th1 and Th17 cells, but increased differentiation into Th2 cells. In-depth analysis showed that m6A targets the mRNA of the IL-7 protein, which regulates T cell homeostatic proliferation and differentiation to various subtypes upon numerous external stimuli. SOCS proteins are adaptors which bind to cytokine receptors, such as the IL-7 receptor, thereby preventing STAT5 and downstream signaling [126,129]. SOCS proteins are produced immediately in response to acute stimuli but are degraded quickly and have short half-lives [126,129]. The m6A modification was shown to regulate the degradation of the *Socs* genes, via the IL-7-JAK1/STAT5 signaling pathway, and without m6A, Socs mRNA persists, leading to high levels of SOCS proteins and reduced sensitivity to IL-7. This study indicates that m6A not only regulates CD4<sup>+</sup> T cells differentiation but also T cell homeostasis [124]. Using a similar Mettl3 conditional KO mouse model, the authors analyzed the Tregs subset (Mettl3-/- and WT) of CD4<sup>+</sup> T cells and found that Mettl3 -/- Tregs mice developed severe autoimmune disorders compared to WT, suggesting loss of m6A modification can lead to loss of Treg immune suppressive functions [127]. In addition to the writer of m6A, readers have shown potential in regulating immune response. As such, compared to WT, a direct reader of m6A, Ythdf1 KO mice showed better crosspresentation of tumor antigens in DC and better cross-priming with CD8<sup>+</sup> T cells, leading to high Ag-specific CD8<sup>+</sup> T cells in response to tumors [122]. Specifically, binding of Ythdf1 at the m6A of transcripts encoding lysosomal proteases lead to increased translation of these lysosomal proteases' (cathepsins) transcripts in DCs, whereas inhibition of Ythdf1 led to inhibition of these cathepsins, resulting in enhanced cross-presentation by DCs and cross-priming of CD8<sup>+</sup> T cells by DCs. Indeed, mature DCs were reported to have higher expression of writer complex, including METTL3, than naïve DCs [128]. In addition, patient tumor samples that had low YTHDF1 expression had higher tumor-infiltrating CD8<sup>+</sup> T cells [122]. Interestingly, mice with Ythdf1 KO showed a better response to ICPi (anti-PD-L1) therapy than the Ythdf1 WT [122].

Collectively, the above studies show the essential role of m6A RNA methylation in regulating innate and adaptive immune responses. The role of RNA methylation in immunity is still at its infancy and requires further research for discovery of novel therapeutic targets for its translational potential.

#### **5. Targeting Methylation in the Treatment of Human Disease**

Alterations in methylation have been strongly associated with the initiation and progression of cancer [132]. Compared to normal control tissues in tumors, DNA hypomethylation occurs at global and gene-specific levels, which results in genomic instability and activation of silenced oncogenes [133]. In contrast, DNA hypermethylation occurs at the promoter regions of tumor suppressor genes (TSGs), which leads to their silencing [133]. With our increasing understanding of the role of methylation in cancer and immunity, further efforts are now aimed at its translational potential to develop new therapeutic strategies that can alter the methylation landscape. Towards these goals, both DNA hypoand hyper-methylation can serve as viable targets which, unlike genetic changes, are both dynamic and reversible.

#### *5.1. Targeting DNA Hypermethylation*

Several DNA hypomethylating agents have been developed that target DNA hypermethylation. However, among these DNA methyltransferase inhibitors (DNMTi), 5-azacytidine (Vidaza®) and 5-aza-2′deoxycytidine (Decitabine, Dacogen®) have been approved by the Food and Drug Administration (FDA) [16]. Because multiple hematologic malignancies are linked to abnormal DNA methylation patterns, DNMTi were first tested in these cancers. Among these, myelodysplastic syndromes (MDS) comprising a group of

hematologic disorders derived from abnormal progenitor cells were the first to be evaluated. Patients with MDS have hypoproliferative bone marrow and a risk of developing different forms of acute leukemia [51]. The inhibitor 5-azacytidine was first tested on MDS patients, where it showed improved response rates, lower transformation to acute leukemia, and prolonged survival [134], and 5-aza-2′deoxycytidine showed similar clinical outcomes [135]. Both 5-azacytidine and 5-aza-2′deoxycytidine have also shown success in a clinical setting for acute myeloid leukemia (AML) and chronic myelomonocytic leukemia (CMML) [16].

Following the clinical success of DNMTi with hematologic malignancies, DNMTi were also tested in solid tumors [136–138]. Although DNMTi showed a good response in patients with ovarian cancer and non-small cell lung cancer, the response was highly variable and less effective in other solid tumors [136–138]. DNMTi has shown the greatest potential in combination with cytotoxic agents or immunotherapies. With cytotoxic agents, DNMTi appear to sensitize tumors and increase the efficacy of conventional cytotoxic agents, even for patients who were previously resistant to the cytotoxic agents alone [139]. Recently, studies have established that malignant cells escape host immune recognition by acquiring an immune evasive phenotype through epigenetically downregulating essential molecules for cancer and immune interactions [35]. For instance, these mechanisms include suppression of tumor associated antigens (TAAs), reducing the expression of many components of antigen processing and presentation machinery (APM), and decreasing co-stimulatory molecules, stress-induced ligands, and death receptors [35]. DNMTi and histone deacetylase inhibitors (HDACi) reverse the immune evasive phenotype, for example, by upregulating the expression of TAAs and APM components on tumor cells, which helps the immune system to recognize and eliminate tumor cells [35,140–142]. Additionally, T cell exhaustion can also be reversed using DNMTi in mouse models, resulting in enhanced anti-cancer immunity [143,144]. DNMTi can also trigger a state of "viral mimicry" by activating dsRNAs, thereby increasing type I interferon responses [35,145]. In addition, DNMTi and HDACi increased cytotoxic activity of CD8 T cells and NK cells, and increased these anti-tumor cells' immune infiltration in the TME while reducing pro-tumor macrophage infiltration in a murine ovarian cancer model [75]. These anti-cancer effects were further elevated in triple combination with ICPi (anti-PD-1), which reduced the tumor burden and provided longest overall survival. Collectively, the above studies indicate priming of the immune system by DNMTi (and HDACi), thereby increasing the efficacy of ICPi therapy.

#### *5.2. Targeting DNA Hypomethylation*

In cancer, promoter hypermethylation of TSGs and silencing of TSGs resulting in tumorigenesis have been the focus of the last few decades, resulting in the discovery of DNMTi [146–149]. By comparison, a phenomenon that is relatively underestimated is genome-wide DNA hypomethylation, which occurs in various solid tumors [133,150]. Several studies have also demonstrated that gene-specific and global hypomethylation play a crucial role in the initiation and progression of cancer [7,133]. However, there is still no approved agent that targets DNA hypomethylation. Currently, the most studied approach to target DNA hypomethylation uses SAM. SAM is a natural and universal methyl donor of all methylation reactions [151,152]. As such, SAM donates its methyl group to key cellular components including proteins, nucleic acids (RNA and DNA), lipids, and secondary metabolites to modulate several physiological functions [151–153].

Although studies investigating the effect of SAM on the immune system are still lacking, SAM has been shown to modulate the immune system [154–167]. SAM manipulates methylation levels, which further modulates T cell functions by regulating the TCR signaling pathway, impairing Th1/Th2 cytokines release, and decreasing T cell proliferation and activation in autoimmunity [154]. Moreover, SAM reduces IL-1 levels in rats with cecal ligation and puncture. In macrophages, SAM inhibited LPS-induced gene expression via modulation of H3K4 methylation [155]. Similarly, deregulation of SAM levels can result in

immune disorders, such as in liver inflammatory diseases. Molecular links between SAM and innate immune functions were reported in which low levels of SAM were shown to affect hepatic PC synthesis and may limit stress-induced protective gene expression upon infection [156]. In addition, SAM prevented upregulation of TLR signaling by blocking the overexpression of TLR2/4 and their downstream partners MyD88 and TRAF-6 in the Mallory–Denk body, forming hepatocytes [157].

Interestingly, studies have shown that SAM is essential for T cell activation and proliferation [154–167]. In activated T cells, both the SAM quantity and the rate of SAM utilization increase dramatically via increased transcription of *MAT2A*, which encodes the catalytic subunit of MATII and is vital for SAM biosynthesis [161,162,164,165]. Blockage of SAM synthesis resulted in blocked T cell proliferation [160]. Furthermore, SAM was shown to be indispensable for T cell proliferation and activation by decreasing both caspase-3 activity and apoptosis in ethanol-related activation-induced cell death (AICD) [159]. Furthermore, SAM was shown to lower the suppressive capacity of Tregs (nTreg cells) by methylating the *FOXP3* gene, thereby reducing its protein and mRNA expression in a dosedependent manner. SAM was also found to decrease expression of an immunosuppressive cytokine, IL-10, and increase expression of IFNγ [168].

Aberrant methylome is a common consequence of a disrupted SAM cycle associated with transformation of cells towards tumorigenesis [152,169,170]. SAM, which increases DNA methylation, has been shown to cause significant anti-tumor effects in breast, osteosarcoma, prostate, hepatocellular, gastric, colon, and other cancers [151,152,169–174]. In addition, SAM levels are depleted by cancer cells through various mechanisms, such as increased conversion of SAM to by-products, which reduces the methylation potential of cancer cells [175,176]. A recent study has shown that an essential immune evasive mechanism used by tumor cells is depriving the CD8+ T cells of SAM and methionine (the pre-cursor of SAM) in the TME. This makes CD8+ T cells non-functional and unresponsive to ICPi [175]. Indeed, we showed that SAM in combination with ICPi (anti-PD-1) significantly reduced tumor volume and weight compared to monotherapy in a syngeneic mouse model of advanced melanoma [177]. This effect was partially due to the elevated activation, proliferation, and cytokine production of CD8 T cells. We also observed increased tumor infiltration of CD8 T cells, a higher number of polyfunctional CD8 T cells, and a lower number of exhausted CD8 T cells in the TME. The above studies show a potential of SAM, a co-factor of methylation, in targeting aberrant DNA methylation patterns in the TME as a novel anti-cancer approach that also enhances anti-cancer immunity. Therefore, the effect of SAM on anti-cancer immunity should be studied comprehensively in future studies.

#### *5.3. Targeting m6A RNA Methylation*

The role of DNA methylation in regulating the immune system and cancer has been the focus of research for more than three decades. Regulation of immunity and cancer by m6A RNA methylation is still at its infancy. However, novel studies have shown the potential of targeting RNA methylation in cancer. For instance, FTO inhibition through selective inhibitors, such as Meclofenamic acid (MA), MA2, and R-2-hydroxyglutarate (R-2HG), have shown potent anti-cancer activity in several cancers including AML, glioblastoma multiforme (GBM), and colorectal cancer (CRC) [26,30,178]. In contrast to other RNA demethylase inhibitors, Rhein was identified to be reversibly bound to the FTO catalytic domain via a crystal structure approach and shown to increase m6A RNA methylation levels [178,179]. Rhein is attractive as it is a natural compound and selective against FTO and not ALKBH5 [179]. Rhein has shown significant anti-cancer activity in various cancers; however, comprehensive in vivo evidence is still lacking and would require further indepth studies [180]. Citrate was identified as an ALKBH5 inhibitor via a crystal structure approach; however, the effect of citrate on ALKBH5 demethylase activity in reducing cancer growth and progression is yet to be determined [181].

Although the inhibitors for RNA methylation regulators have been identified, none of them have been tested in a clinical setting. Furthermore, the effect of these pharmacological inhibitors of RNA methylation on the immune system is yet to be determined. Along this line, recently, RNA demethylase FTO was reported to promote tumorigenesis in melanoma and knockdown of FTO-reduced resistance to ICPi (anti-PD-1) therapy [182]. FTO regulates important immune genes (including PD-1, CXCR4, and SOX10 genes) and KD of FTO led to increased mRNA decay of these genes through the m6A reader YTHDF2. Furthermore, KD of FTO sensitized melanoma cells to IFNγ, thereby reducing resistance to anti-PD-1 therapy. Similarly, RNA demethylase ALKBH5 KO showed significant reduction in tumor growth and prolonged mouse survival during ICPi therapy in B16 melanoma and CT26 colon cancer mouse models [183]. This was due to ALKBH5 altering gene expression and splicing that leads to changes in lactate levels in the TME. These metabolic changes result in decreased Treg and MDSCs infiltration in the TME. Interestingly, the authors also tested an ALKBH5 inhibitor and showed similar phenotype to the ALKBH5 KO model. These studies not only show the inhibition of m6A demethylases as a potential anti-cancer target but also their potential in anti-cancer immunity within the TME.

#### **6. Conclusions**

The role of DNA and RNA methylation in regulating the differentiation and activity of immune cells within the TME is key to determining the fate of tumor growth or suppression (Figure 5). A pro-cancer TME has immune cells expressing pro-tumor cytokines that lead to tumor growth and progression, whereas the reverse is seen in the anti-cancer TME. Precise methylation patterns change gene expression, leading to specific immune cell subtypes. For instance, DNA demethylation and high expression of *IL4* and *FOXP3* genes occur in Th2 and Tregs, respectively. In contrast, DNA demethylation and high expression of *IFNγ* and *IL2* genes occur in both Th1 and CD8 T cells, which results in a better anti-cancer immune response. Studies should further investigate the effect of DNA and RNA methylation on transcriptional regulation of immune cells along with tumor cells in a time-dependent manner in order to uncover the complexity of the TME at various stages of cancer growth and progression. As explained earlier, the balance between pro- and anti-cancer immune cells within the TME is key to tumor progression or suppression. However, most studies investigating the role of methylation have focused only on one immune cell subtype. Future studies should investigate various immune subtypes simultaneously. These comprehensive studies will provide deeper insights into the interplay between the immune system and cancer, and allow discovery of novel epi-therapies that can enhance the immune system against cancer and other pathologies. Targeting methylation is a particularly attractive anti-cancer strategy because it is dynamic and reversible. For instance, DNMTi that target DNA hypermethylation can also enhance the efficacy of immunotherapies. Similarly, SAM, targeting DNA hypomethylation, has shown profound effects in combination with ICPi. Along the same line, inhibitors of m6A RNA demethylases have shown potential in enhancing anti-cancer immunity. However, further comprehensive studies are required to delineate the mechanism of action before these inhibitors can be tested in a clinical setting. In addition, SAM, which donates methyl groups to RNA, has shown significant anti-cancer activity in numerous cancer models by regulating DNA methylation. It is yet to be determined if SAM causes inhibition of tumor growth and metastasis through modulating m6A RNA methylation levels. Although the efficacy of epigenetic-based therapeutic strategies targeting tumor and immune cells needs further elucidation, the current state of knowledge provides compelling evidence to suggest that it will be effective in blocking cancer progression and reducing cancer associated morbidity and mortality.

**Figure 5.** Summary of the role of DNA methylation and co-factor (s-adenosylmethionine, SAM) in regulating differentiation, activation, and proliferation of pro- and anti-cancer immune cells. The pro- or anti-tumor effect of the immune cells in the TME is also described. Abbreviations: Th2, CD4+ helper T cell 2; Tregs, regulatory T cell; Th1, CD4+ helper T cell 1; CD8 T, CD8 cytotoxic T cells; Memory T cells, CD8 memory T cells; NK, natural killer cell; DC, dendritic cell; M1, macrophage M1 subtype; M2, macrophage M2 subtype; TAMS, Tumor associated macrophages; ICPi, Immune checkpoint inhibitors; s-adenosylmethionine, SAM.

> **Author Contributions:** Writing—research, review, and editing, A.M.; review, editing, and supervision, S.A.R. All authors have read and agreed to the published version of the manuscript.

> **Funding:** This work was supported by grants from the Canadian Institutes for Health Research, PJT-156225 to SAR.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

