Targeting DNA methylation - DNA methyltransferase (DNMT) inhibitors
Inhibitors of DNA methylation were among the first epigenetic drugs tested for use in treatment of cancer [
112]. The most widely studied DNMT inhibitors include azacitidine (5-azacitidine) and decitabine (5-aza-2’-deoxycytidine), which act as analogues of cytosine. These molecules get incorporated into DNA and covalently bond with DNA methyltransferase, thus preventing its function [
113,
114] and leading to its degradation [
115]. These drugs were initially used as cytotoxic chemotherapeutics in the late 1960s [
116], but were found to be highly toxic [
117‐
120]. Subsequently, recent studies have discovered that low doses of DNMT inhibitors had greater efficacy in sustaining decreased DNA methylation and associated re-expression of silenced genes in leukemic and epithelial tumor cells [
121]. These lower doses were also able to reduce tumorigenicity and target CSC populations within the tumor. In lung cancer, Liu et al. showed that inhibition of DNMT1 was able to decrease proliferation and tumorigenic ability of lung CSCs [
34].
Multiple studies have also demonstrated the role of DNMT inhibitors in differentiation therapy. Pinto et al. showed that azacitidine could induce primary AML cells from patients to differentiate into less or non-malignant cells [
122,
123]. Prostate cancer derived-CSCs that were treated with decitabine showed decreased expression of stemness genes Octamer-binding transcription factor 4 (
OCT40029 and Nanog homeobox (
NANOG), leading to overall reduction in tumor growth [
124]. In addition, low doses of SGI-110, a newer DNMT inhibitor, was recently reported to be capable of reprogramming ovarian CSCs to a more differentiated state [
125]. Treatment with SGI-110 also decreased tumor-initiating ability and re-sensitized these cells to platinum, suggesting a potential use of DNMT inhibitors in combination with other chemotherapeutic agents in preventing recurrence of ovarian cancer [
125]. Both azacitidine and decitabine have been approved by the FDA for treatment of myelodysplastic syndrome (MDS) [
126]. Clinical trials for other indications such as AML and colorectal cancer are still ongoing. SGI-110 is also in phases of clinical trials for treatment of various cancers such as AML, MDS, liver cancer and platinum-resistant ovarian cancer.
Targeting histone deacetylation - Histone deacetylase(HDAC) inhibitors
An important histone tail modification is acetylation, which is regulated by histone acetyltransferases (HATs) and histone deacetylases (HDACs). HATs are responsible for adding an acetyl group onto lysine residues of histone tail, which neutralizes the positive charge, resulting in a more “open” chromatin structure [
127]. In contrast, HDACs remove the additional acetyl group, leading to increased binding affinity between DNA and histones, which is generally associated with gene repression [
128]. Very often, deregulated gene silencing in cancers has been associated with aberrant histone deacetylation. For instance, in leukemia, this can be mediated through aberrant recruitment of HDACs by fusion proteins such as Acute myeloid leukemia protein 1 Eight twenty-one protein (AML1-ETO) and Promyelocytic leukemia protein retinoic acid receptor alpha (PML-RARα), which leads to abnormal gene silencing and subsequent leukemogenesis [
129,
130]. Besides, HDACs can also acetylate non-histone proteins, including tumor suppressor p53 and oncogene B-cell lymphoma 2 (BCL2), resulting in inhibition of p53-dependent transcription [
131] and upregulation of pro-survival protein, BCL2 [
132]. Hence, the use of HDAC inhibitors in returning histone acetylation patterns to a normal state has been found to be effective in inducing apoptosis and differentiation as well as inhibit proliferation of tumor cells [
129,
133]. These HDAC inhibitors can be divided mainly into two classes—the pan HDAC inhibitors and the class-specific inhibitors [
134], and they all function via chelating the zinc atom in the active site of the enzyme [
127].
Two HDAC inhibitors, vorinostat (subseroylanilide hydroxamic acid) and romidepsin (depsipeptide), have been approved for treatment of cutaneous T-cell lymphoma [
135,
136]. Both drugs were found to produce durable response and efficacy in patients with cutaneous T-cell lymphoma in Phase 2 multi-center trials [
135‐
138]. However, besides cutaneous T-cell lymphoma, monotherapy of vorinostat and romidepsin in treatment of various solid tumors have had little success in clinical trials [
139‐
150]. Apart from these two compounds, many other HDAC inhibitors have also been developed and tested in clinical trials, the details of which have been well-reviewed elsewhere [
115,
151‐
153]. Monotherapies of these compounds, including panobinostat [
154,
155], entinostat [
156,
157], belinostat [
158,
159] and pracinostat (SB939) [
160], are being tested against various haematological malignancies and solid tumors.
Another mechanism of action of HDAC inhibitors for cancer treatment is via differentiation or reprogramming of cancer cells. As therapy resistance is a major hurdle in cancer treatment and is often associated with CSCs and epigenetic control [
161], HDAC inhibitors possess the ability to induce differentiation of CSCs from their quiescent state, thereby re-sensitising them to other chemotherapy agents. Valproic acid, an antiepileptic drug, has been found to be a powerful HDAC inhibitor [
162]. Gottlicher et al. demonstrated that valproic acid could trigger differentiation of transformed hematopoietic progenitor cells and leukemic blasts from AML patients [
162]. Furthermore, Travaglini et al. found that valproic acid was able to epigenetically reprogram breast cancer cells into a more “physiologic” phenotype, thus improving sensitivity to other forms of breast cancer therapy [
163]. In addition, entinostat, a selective inhibitor of class I HDACs, was recently reported to reverse EMT phenotype and decrease the population of tumor-initiating cells in triple-negative breast cancer (TNBC) [
164]. These tumor-initiating cells possessed CSC properties and were responsible for driving metastasis and drug resistance in TNBC, thus contributing to poor patient prognosis. Hence, this study demonstrated the utility of HDAC inhibitors in preventing CSC invasiveness and tumor metastasis. Overall, these studies demonstrate the potential use of epigenetic modulators towards the differentiation and therapeutic sensitization of CSCs.
Targeting histone methylation – Histone methyltransferase (HMT) inhibitors
A class of enzymes called histone lysine methyltransferases (HKMTs) mediate the addition of a methyl group to the nitrogen atom of the lysine side chain [
165]. Despite catalysing a common chemical reaction, this family of HKMTs demonstrate large structural diversity of its active sites, allowing these enzymes to have high substrate specificity [
127]. For example, DOT1L (KMT4) is a unique HKMT as it is currently the only known enzyme that methylates lysine 79 of histone H3 (H3K79) [
166]. Similarly, methylation of H3K27 is only mediated by the catalytic subunit EZH2 (KMT6) of PRC2 [
127]. In contrast, some methylation marks can be catalysed by several proteins, such as H3K9 methylation. These post-translational methylation of histones have important roles in regulation of gene expression, differentiation, DNA damage repair as well as in tumorigenesis [
167,
168]. Aberrant histone methylation can be due to gene mutations, over-expression or deregulated control of epigenetic modulatory enzymes involved. Thus, HKMTs are potential therapeutic targets, and the structural differences between members of the family also enable greater selectivity in inhibition of these proteins by small molecule compounds [
169].
HKMT inhibitors have only recently gained more attention as cancer therapeutics, resulting in a rapidly increasing number of these small molecule inhibitors being developed [
170‐
172]. In fact, several DOT1 like histone H3K79 methyltransferase (DOT1L) and EZH2 inhibitors have progressed to being tested in clinical trials as cancer interventions [
173]. H3K79 methylation by DOT1L is associated with transcriptional activation of genes under its regulation [
174,
175], and overexpression or aberrant DOT1L activity has been found in cancer, such as leukemia with mixed lineage leukemia (MLL) gene translocation. The MLL fusion protein can recruit DOT1L into a transcription complex, which subsequently methylates H3K79 [
176‐
180]. This leads to dysregulation and overexpression of many MLL-target genes, including Homeobox A9 (HoxA9) and Meis homeobox 1 (Meis1), which are key regulators of hematopoietic stem cell differentiation that contributes to leukemogenesis [
165]. Therefore, DOT1L is an attractive target for therapy, resulting in the first selective DOT1L inhibitor EPZ-4777 to be synthesised with anti-tumor effects against murine models of MLL-rearranged leukemia [
181]. Further optimisation of the drug led to the development of EPZ-5676, the first HKMT inhibitor to enter clinical trials. This compound has been shown to be highly potent and selective for DOT1L. Treatment with EPZ-5676 in a MLL-rearranged leukemia xenograft model showed durable and complete tumor regression [
182]. EPZ-5676 is currently under clinical studies (Phase I) for MLL-fusion leukemia, AML, MDS and myeloproliferative disorders.
EZH2 is a member of PRC2, along with proteins embryonic ectoderm development protein (EED) and SUZ12, and is responsible for catalysing H3K27 mono-, di- and tri-methylation [
183‐
185]. Overexpression of EZH2 has been found in various cancers of the breast, lung, prostate and haematological malignancies [
186‐
191], and is associated with poor disease prognosis. Studies have also shown the role of EZH2 deregulation in tumor progression, metastasis [
192,
193] and maintenance of CSC self-renewal properties [
194]. In glioblastoma multiforme (GBM), inhibition of EZH2 by S-adenosylhomocysteine hydrolase (SAH) inhibitor 3-deazaneplanocin A (DZNep) was able to reduce self-renewal and tumor-initiating capabilities of GBM CSCs in vivo via affecting transcriptional regulation of oncogene
MYC [
193]. However, DZNep affects methylation of other histone residues [
195], leading to the development of more specific EZH2 inhibitors. The earliest SAM-competitive and selective EZH2 inhibitor to advance into clinical trials for treatment of rhabdoid tumors and lymphomas is EPZ-6438 (E7438) [
196,
197]. A more recent drug, GSK2816126 (GSK126) has also entered clinical studies for relapsed/refractory diffuse large B-cell lymphoma (DLBCL), multiple myeloma and transformed follicular lymphoma [
198,
199]. Both drugs have shown high potency and selectivity in inhibiting tumor growth in preclinical studies [
197,
198,
200].
H3K9 methyltransferases, such as euchromatic histone lysine methyltransferase 2 (G9a/EHMT2) and euchromatic histone lysine methyltransferase 1 (GLP/EHMT1), catalyse mono- and di-methylation of the lysine residue, while tri-methylation of H3K9 is mediated by Suppressor of variegation 3–9 homolog 1 (SUV39H1) and Suppressor of variegation 3–9 homolog 2 (SUV39H2) [
201]. Upregulation of G9a activity has been linked to several types of cancer, including ovarian, lung, liver and bladder cancers [
202‐
208]. Hence, several substrate-competitive inhibitors of these HKMTs have been developed. BIX-01294 is the first specific inhibitor of G9a and GLP, and studies have reported its ability to decrease H3K9me2 levels in mammalian cells [
209‐
211]. Kim et al. reported that BIX-01294 was able to induce cell death in colon and breast cancer cells via EHMT dysfunction [
212]. However, due to the increased toxicity levels of BIX-01294 at higher concentrations, the use of this drug is limited. This led to the recent development of a more potent, specific and selective EHMT inhibitor, UNC0638 that was found to decrease local H3K9me2 and DNA methylation levels [
213]. Further development generated UNC0642, which possessed better pharmacokinetic properties and higher efficacy in inhibiting colony formation ability of pancreatic adenocarcinoma cells [
214].
Methylation of H3K9 by SUV39H1 is associated with silencing of tumor suppressor genes, including E-cadherin and p15INK4B, in AML [
215]. Overexpression of SUV39H1 has also been correlated with poor prognosis in multiple myeloma patients [
216]. Treatment of multiple myeloma cells with chaetocin, a small molecule inhibitor of SUV39H1 showed anti-tumor effects at low doses of the drug [
216]. Similarly, chaetocin was found to decrease H3K9me3 levels and induce differentiation of AML cells at non-toxic doses [
217]. Furthermore, chaetocin was able to repress cell proliferation and induce apoptosis in hepatocellular carcinoma (HCC) cultures and xenografts [
218], implying a potential tumorigenic role of EHMTs in HCC progression and development.
Targeting histone demethylation - Histone demethylase (HDM) inhibitors
Methylation of lysine on histones is also regulated by histone lysine demethylases (KDMs). This group of epigenetic erasers function in removing the methyl groups from lysine side chains on histones [
219,
220]. As proper functioning of both HKMTs and KDMs is required to maintain stable histone methylation levels, small molecule inhibitors have also been developed to target KDMs. KDMs can be grouped into two families - the lysine-specific demethylase (LSD) family and Jumonji domain-containing (JmjC) family [
221]. The LSD family are flavin adenine dinucleotide (FAD)-dependent amine oxidase that demethylates mono- and di-methyl lysine residues, while JmjC enzymes utilise 2-oxoglutarate and iron to oxidatively release methyl groups from all three methylation states at lysine residues [
172,
222].
Upregulated expression of LSD1 (KDM1A) has been found in various human cancers, including AML, ovarian, lung, bladder and colorectal cancers [
223‐
225]. Hence, small molecule inhibitors of LSD1 that target the enzyme cofactor FAD have been developed, the first of which is tranylcypromine [
226]. Further studies have led to the synthesis of more selective derivatives of tranylcypromine, such as ORY-1001 [
227] and GSK2879552 [
228]. They function by irreversibly changing FAD, leading to the formation of a tetracyclic adduct [
229]. LSD1 is important for normal hematopoiesis; loss of LSD1 has been found to inhibit differentiation and impair hematopoiesis [
230]. This suggests a potential role of aberrant LSD1 activity in affecting stemness properties in tumor cells. The inhibitor ORY-1001 has been shown to decrease the population of AML stem cells and improve survival of mice with acute lymphoblastic leukemia (ALL) in preclinical studies [
227,
228]. GSK2879552 has also been found to influence differentiation in small cell lung cancer (SCLC) [
231]. These compounds are currently in phase 1 studies for relapsed or refractory AML (ORY-1001) and SCLC (GSK2879552).
Similarly, JmjC demethylases are amenable to pharmacological intervention as well. Ubiquitously transcribed tetratricopeptide repeat X chromosome (UTX), also known as KDM6A, is responsible for demethylating H3K27 [
232‐
234], and loss of UTX activity has been found in multiple human malignancies, including multiple myeloma, esophageal squamous cell carcinoma and renal carcinoma [
166]. However, no inhibitors of JmjC enzymes have advanced beyond biochemical studies [
127]. Nevertheless, as UTX is a component of the mixed lineage leukemia protein 2 (MLL2) H3K4 methyltransferase complex, and interacts with SWI/SNF chromatin remodelling complex [
235‐
237], it is still an important epigenetic target and its role in epigenetic modulation still warrants further study.
Targeting epigenetic readers – BET inhibitors
While epigenetic modulatory enzymes are obvious targets for therapy, epigenetic readers are also important components of the epigenetic machinery as they directly or indirectly regulate gene expression. One such group of readers called bromodomain and extra-terminal (BET) proteins modulate gene expression by recognising acetylated histones. Increased BET activities have been associated with NUT midline carcinoma (NMC), glioblastoma and various haematological malignancies, through aberrant transcription of disease-associated genes and oncogenes such as
MYC [
238]. Hence, BET proteins appear to be attractive therapeutic targets for controlling dysregulated gene expression.
JQ1 is a selective BET inhibitor of BRD family of proteins, including Bromodomain-containing protein 4 (BRD4) [
239]. In preclinical studies, JQ1 was able to cause tumor regression in NMC mouse models, inhibit proliferation, induce apoptosis and differentiation in cancer cells [
239‐
242]. Another BET inhibitor, I-BET762 (GSK525762A), functions by binding to the acetyl-binding pocket of BET proteins [
243,
244]. Studies have shown that I-BET762 treatment was able to induce terminal differentiation of patient-derived malignant cells [
245] and activate apoptosis in neuroblastoma and prostate cancer via inhibition of Myc-driven pathways [
246,
247]. This compound is currently in phase I trials for solid tumors and relapsed or refractory haematological cancers.
OTX015 is another BET inhibitor that has progressed into clinical trials for various haematological malignancies. This compound has been found to possess anti-proliferative effects via directly influencing
MYC expression and activity [
248,
249]. Similarly, CPI-0610 has also entered clinical testing for lymphoma, multiple myeloma and myelodysplastic or myeloproliferative neoplasms. I-BET151 is a pan-BET inhibitor, similar to JQ1, and has been found to block proliferation and induce apoptosis in myeloma cells via repressing Myc activity [
250]. Antitumor effects have also been observed in NMC, MLL, ALL, lung cancer and brain cancer [
238].
Combination therapy with epigenetic modulators
While epigenetic drugs have been tested preclinically and clinically as single agents, further studies have revealed the increased efficacy of these drugs when used in combination with other therapies. One common combination of different epigenetic therapies is that of DNMT and HDAC inhibitors. Pathania et al. reported that combining azacitidine (DNMT inhibitor) and butyrate (HDAC inhibitor) was capable of significantly decreasing breast cancer CSC population [
251]. In addition, combination of azacitidine and HDAC inhibitor entinostat at low doses in a phase I/II clinical trial showed sustained and favourable responses in treatment-resistant non-small cell lung cancer (NSCLC) patients [
252]. Azacitidine and valproic acid co-treatment was also able to promote tumor regression in
Patched mutant mouse models of medulloblastoma [
253]. Besides DNMT-HDAC inhibitor combination therapy, studies have demonstrated synergistic effects of other epigenetic drug combinations. For example, inhibiting both EZH2 and G9a histone methyltransferases showed greater efficacy in blocking cell proliferation as compared to single drug treatment [
254]. Furthermore, the DOT1L inhibitor EPZ-5676 could interact synergistically with DNA hypomethylating agents, such as azacitidine and decitabine, in MLL-rearranged leukemia cells [
255].
In recent years, an increasing number of studies have reported the use of epigenetic drugs in combination with conventional chemotherapeutics, with underlying mechanisms of re-sensitising resistant CSCs to drug treatment, or to prime cancer cells for subsequent therapies [
134,
256]. For example, low doses of SGI-110 (DNMT inhibitor) was found to drive ovarian CSCs towards a more differentiated phenotype and sensitise them to platinum treatment [
125]. DOT1L inhibitor EPZ-5676 was also able to establish a chromatin state that enhanced the anti-tumor effects of cytarabine or daunorubicin in MLL-rearranged leukemia [
255]. Moreover, pre-treatment with azacitidine was demonstrated to prime colon cancer cell lines to irinotecan therapy [
257]. Indeed, various combinations have been tested in clinical trials with promising results on drug response and anti-tumor efficacy [
258‐
261]. In addition to drug combination synergy, the method of delivery could also improve response to therapy. A recent paper by Li et al. showed that encapsulating decitabine and doxorubicin in nanoparticles was able to better target breast CSCs and inhibit tumor growth [
262].
The use of immunotherapy in cancer has made significant progress over the past two decades, with several immunotherapy drugs being approved by the FDA for the treatment of cancer. These drugs function to overcome the mechanisms of immune tolerance that are employed by cancer cells to evade or limit the body’s immune response. These mechanisms include changes in antigen processing and presentation, creation of an immunosuppressive microenvironment, induction of T-cell death and activation of negative immune regulatory pathways [
263]. One key receptor involved in the immunoinhibitory pathways is the cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), which is expressed on the surface of immune cells and acts as an immune checkpoint. Studies have shown that targeting CTLA-4 receptor induced favourable responses in patients with advanced melanoma [
264], and the FDA-approved CTLA-4 inhibitor, Ipilimumab, is now in clinical trials for prostate and lung cancers. Another immune checkpoint involved in tumor immune-resistance is the interaction between programmed cell death-1 (PD-1) and programmed death-ligand 1 (PD-L1) [
265]. Specific targeting of PD-1 and PD-L1 has been clinically shown to be very effective in treatment of metastatic cancers and melanomas [
266,
267].
However, as most of these immunotherapy strategies are mainly targeted at bulk tumors, which contain more differentiated cells with “differentiation antigens” [
268], CSCs (which have a different set of tumor antigens) would not be successfully eradicated. Hence, more effective targeting of the CSC population can be achieved via CSC-specific immunologic approaches, or by combining immunotherapy with epigenetic therapies that induce CSC differentiation and alter surface protein expression. The latter approach would likely improve the overall antitumor efficacy as both CSC and bulk tumor populations can be targeted simultaneously. For instance, the use of DNA hypomethylating agent (5-aza-2
’-deoxycytidine) in combination with anti-CTLA-4 monoclonal antibody in syngeneic transplantable murine models demonstrated significant reduction in tumor volumes as compared to single agent treatment alone [
269]. The improved efficacy of this combination was attributed to the increased CD3+ T-cell infiltration in the combination cohort tumors and a sustained expression of cancer antigens and MHC proteins due to promoter demethylation. Furthermore, combinatorial drug treatment with immune checkpoint inhibitors (anti-CTLA-4 and anti-PD-1) and epigenetic modulators (5-azacytidine and Entinostat) showed remarkable eradication of CT26 colorectal tumors and 4 T1 mammary tumors in more than 80% of the tumor-bearing mice [
270]. Importantly, 4 T1 tumor-bearing mice that were given combinatorial treatment did not develop metastases as compared to mice given single agent treatment. These findings demonstrate that epigenetic drugs in combination with immunotherapy can enhance the reversal of immune tolerance in cancer cells, including CSCs.
Another way in which cancer cells evade cytotoxic T-cells is by down-regulating human leukocyte antigen (HLA) to avoid tumor antigen presentation [
134]. Hypermethylation of HLA promoters was frequently observed in gastric cancer and esophageal squamous cell cancers [
271,
272]. Treatment with DNMT and HDAC inhibitors were found to be capable of reversing this hypermethylation and increasing HLA expression [
272‐
275], thus priming these cells for immunotherapy. In addition, Li et al. showed that azacitidine treatment was able to enhance immunomodulatory pathways, such as antigen processing/presentation and interferon signaling, in breast, colorectal and ovarian cancers [
276]. These preclinical data highlight the promising potential of combining epigenetic and immunotherapies in improving cancer treatment efficacy, which will be verified in several ongoing clinical trials.