Different mechanisms of epigenetic silencing prevent the transcription of genomic locations. The wrapping of DNA around the histone octamers to form nucleosomes offers itself a barrier for the recruitment and processivity of RNA polymerases [30
]. Highly packed nucleosomes are inaccessible for RNA polymerases and transcription factors and therefore the genes contained in these regions remain silent. Packaging of the genome is influenced by modifications in histone tails and DNA. Alterations in histone and DNA modifications are common in cancer and other diseases and therefore it is expected that small molecules able to inhibit the activity of the enzymes involved in these modifications have therapeutic potential. Several of these inhibitors are currently in clinical trials or have been FDA-approved for the treatment of certain cancers [31
]. These drugs target distinct factors involved in epigenetic regulation, including writers able to modify histones or DNA, erasers that remove those marks and readers able to recognize and bind to the modifications. Several of the epigenetic enzymes targeted by these drugs, such as DNA methyltransferases (DNMTs), histone deacetylases (HDACs), the histone demethylase (HDM) LSD1/KDM1A and the histone methyltransferases (HMTs) EZH2 and G9A, play a role in the transcriptional repression of TEs in cancer cells (Figure 5
). The effects of epigenetic inhibitors targeting these repressors in cancer cells will be discussed next.
6.1. DNA Methyltransferases Inhibitors
DNA methylation plays a key role in gene silencing, X chromosome inactivation, genome stability, and imprinting. Although most of the interest has been focused in the role of DNA methylation in silencing the expression of tumor suppressor genes in cancer, DNA methylation is likely to play an important role in repeat element regulation since most of the genomic CpG sites subjected to DNA methylation are located at these elements [32
The DNMTs inhibitors (DNMTi) azacytidine (AZA) and decitabine (DAC) are the most successful and have the longest history of epigenetic drugs used in cancer treatment to date (Figure 5
). These compounds are cytidine analogs that are incorporated into DNA during replication and form covalent complexes with DNMTs depleting in this way the pool of active enzymes in the cell. AZA and DAC are currently approved for the treatment of myelodysplastic syndrome and acute myeloid leukemia. Their therapeutic effects in solid tumors appear weaker.
In addition to showing antiproliferative effects in cancer cells, DNMTi also have been described to improve the effects of immune checkpoint inhibitors in mouse models. Combinations of AZA or the HDACs inhibitor (HDACi) entinostat and anti-PD-1/anti-CTLA-4 antibodies improves the treatment outcome in mouse models of colorectal and breast cancer [33
]. In clinical trials for non-small cell lung cancer (NSCLC) a small number of patients that had previously received AZA showed robust and durable responses to immune checkpoint blockade [34
]. Guided by these favorable responses, clinical trials are on course to determine if DNMTi can sensitize patients with different types of cancer to immune checkpoint inhibition.
More mechanistic studies revealed that cancer cell lines treated with low doses of AZA induce the expression of genes involved in innate immunity, including antigen presentation, apoptosis, interferon signaling, and anti-viral defense [34
]. Deep investigation into the molecular mechanisms responsible for the induction of innate immunity genes and potential benefit in combination with immunotherapy, revealed the accumulation of cytosolic dsRNA that triggers a type I interferon response after treatment of ovarian and colorectal cancer cell lines with DNMTi [21
]. These dsRNAs were likely originating from increased expression of multiple DNA hypermethylated HERVs.
Fine mapping of transcriptional responses to DNMTi in the lung cancer cell line NCI-H1299 revealed that treatment with DAC had no significant direct effects on gene promoter driven transcription. Instead it induced the transcription of thousands of non-annotated transcription start sites (TSSs) that overlapped with TEs [36
]. The most enriched elements experiencing non-annotated transcription initiation after DAC treatment were solo LTRs belonging to the LTR12C family. Combinations of HDACi (vorinostat or pracinostat) and DAC had the most prominent effects in inducing the expression of these elements. Loss of DNA methylation likely facilitated the binding of transcription factors like GATA2 to these elements, inducing histone acetylation and consequent engagement in transcription. Surprisingly, the LTRs induced by the treatments displayed unidirectional transcription and it is unclear how they might contribute to generate dsRNAs. Interestingly, those elements were found also enriched in H3K9me3 suggesting that they could be also induced by inhibitors of writers of this mark.
In addition to the effects of DNMTi in stimulating innate immunity pathways these inhibitors might also enhance the effects of immune checkpoints inhibitors through the induction of expression of tumor-specific antigens [34
]. Recent evidence suggests that the induction of TEs expression contributes to generate immunogenic peptides. First, in the NCI-H1299 cancer cell line DNMTi induces the expression of several LTRs located in the proximity of coding genes generating fusion transcripts that encode novel protein isoforms [36
] (Figure 3
A). These chimeric proteins, partially encoded by repetitive elements, might become immunogenic if processed and presented by cancer cells. Second, glioblastoma cell lines can express and present several neoantigens derived from TEs after treatment with AZA [37
]. These potentially immunogenic peptides are derived not only from HERVs, but also from other classes of TEs including LINEs and SINEs and might not only contribute to stimulate immune responses against tumors but could be used to develop personalized anti-cancer vaccines. Therefore, DNMTi might offer combinatorial opportunities with different immuno-oncology strategies, including immune checkpoint inhibitors, vaccines, and perhaps CAR-T cells targeting peptides originating from TEs.
A major drawback of epigenetic therapies that has hampered their progress in the clinic is the inability to select which cancer patients are more likely to respond to therapy. Finding novel markers to predict sensitivity to DNMTi might help to find clinical applications for these drugs in solid tumors. Several lines of evidence suggest a correlation between the expression of TEs, enrichment in innate immunity-related gene signatures and sensitivity to DNMTi treatment. Chiappinelli et al. observed that ovarian serous cancers with high HERV expression were enriched in innate immunity signatures [21
]. Additionally, enrichment of innate immunity signatures correlated with sustained response to immune checkpoint inhibitors in melanoma patients. In a similar way, Kong et al. described that tumors with great loss of DNA methylation at TEs showed TEs reactivation and presence of immune infiltrates [37
]. Finally, low LINE-1 expression has been correlated with global DNA hypermethylation and responsiveness to AZA in colorectal cancer cell lines [38
]. This raises the idea that tumors with low innate immunity signature and low HERV expression might benefit from DNMTi treatment to reactivate expression of TEs and promote anti-tumor immunity. Therefore, the level of expression of certain repeats, their methylation status, or the enrichment in innate immunity signatures in tumors could be used as a predictor of response to DNMTi.
6.2. LSD1 Inhibitors
Histone lysine specific demethylase LSD1 (KDM1A) was the first discovered histone demethylase [39
]. LSD1 was characterized as a transcription co-repressor that works primarily by demethylating mono and dimethylated lysine 4 of histone H3 (H3K4me1/2) [40
]. In very specific cases LSD1 might also display H3K9 demethylation activity contributing to gene activation [41
]. In addition LSD1 can demethylate a number of non-histone proteins [42
LSD1 has been described to be involved in several types of cancer. High LSD1 expression correlates with malignancy in a range of solid tumors [43
]. In distinct solid tumors and hematologic malignancies LSD1 inhibits differentiation, and enhances proliferation and invasiveness [43
]. Accordingly, inhibitors of the catalytic activity of LSD1 (LSD1i) have shown antiproliferative effects in preclinical models of cancer and are being tested in clinical trials [42
Having LSD1 multiple targets and acting both as a coactivator and corepressor has complicated the understanding of how LSD1i block proliferation of cancer cells. Inhibition of transcriptional programs governed by oncogenic transcription factors and induction of differentiation pathways have been reported in solid cancer cell lines and mouse models of hematologic cancers [46
]. Moreover, several studies report that LSD1i treatment can stimulate immune responses against tumors. LSD1i induced the expression of cytotoxic T cell attracting chemokines in triple negative breast cancer (TNBC) cell lines [53
]. In a mouse model of breast cancer combinations of LSD1i with anti-PD-1 antibodies significantly suppressed tumor growth and pulmonary metastasis, and increased T cell infiltration [53
]. In pediatric high-grade glioma (pHHG) cell lines LSD1i induced an immune-related gene signature and promoted tumor regression and augmented natural killer (NK) cells infiltration of tumors in a mouse model of this disease [54
Recent data suggests that LSD1i might potentiate anti-tumor immune effects through the reactivation of HERVs. Treatment of cancer cell lines with the LSD1i GSK-LSD1 resulted in the induction of interferon-related genes and expression of a subset of HERVs in several cancer cell lines [55
]. While interferon-related genes were not direct targets of LSD1, certain HERVs were found occupied by LSD1, gained H3K4 methylation, and were induced after LSD1 depletion. Sense and antisense transcripts originating from these HERVs as well as accumulation of dsRNAs could be observed in response to LSD1 knock down. As expected, the responses to LSD1 depletion were dependent on of TLR3 and MDA5. Furthermore, authors demonstrated that inhibition of LSD1 in cancer cells increased their immunogenicity. Inoculation of wild type and LSD1 depleted B16 cells into immunocompetent mice showed that LSD1 ablation reduced tumor growth and promoted T cell infiltration into tumors in an MDA5 dependent fashion. Further combination of LSD1 depletion with PD-1 blockade showed striking anti-tumor effects.
Analysis of gene expression data in cutaneous melanoma, TNBC, and pHHG patients revealed inverse correlations between LSD1 expression and CD8+ T cell infiltration, prognosis, levels of cytotoxic T cell attracting chemokines expression, PD-L1 levels, and innate immunity signatures [53
]. These data suggest that LSD1 expression may be an informative biomarker for prognosis and treatment.
From the mechanistic point of view there are still many unresolved questions about the effects of LSD1i in the activation of TEs. A fine dissection of the sites of cryptic transcription induced by these inhibitors is granted to fully understand their effects. This will allow to determine whether DNMTi and LSD1i regulate the same or different types of elements and whether inhibition of both will further stimulate anti-tumor immunity. In addition, further work is needed to explore the effects of LSD1 inhibition in immune cells.
6.3. EZH2 Inhibitors
EZH2 is the catalytic subunit of the PcG repressor complex 2 (PRC2) that trimethylates lysine 27 on histone 3 (H3K27me3), a mark involved in gene repression. Several lines of evidence have implicated EZH2 in the development and progression of a variety of cancers. EZH2 overexpression has been shown to correlate with aggressiveness and advanced disease in several cancer types [56
]. Additionally, gain of function mutations that increase the catalytic activity of EZH2 have been described in follicular lymphomas, diffuse-large B cell lymphomas, and melanomas [59
]. Given the evidence for EZH2 involvement in cancer, the development of EZH2-specific catalytic inhibitors (EZH2i) has been an active area of investigation for quite some years. A major current goal is the search for molecular markers predictive of patient response. In this regard, EZH2i have been reported to cause antiproliferative effects in several cancer cell lines including lymphoma cell lines with EZH2-activating mutations [63
] and cancer cell lines with inactivating mutations in subunits of the chromatin-remodeling complex SWI/SNF such as ARID1A-deficient ovarian cancer [66
], SMARCA4- and SMARCA2-deficient ovarian cancer [67
], and SMARCB1/INI1-deficient rhabdoid tumor and synovial sarcomas cell lines [68
]. While several clinical trials are testing these potential dependencies in patients, early in 2020 the EZH2i tazemetostat received for the first time FDA approval for the treatment of epithelioid sarcomas with SMARCB1/INI1 deletions.
In addition to the described antiproliferative effects, several studies have shown that the treatment with EZH2i improves the responses to diverse immunotherapy approaches in mouse models of ovarian, bladder cancer, and melanoma [70
] and several clinical trials are taking place to test the efficacy of such combinations in cancer patients [73
]. Among the molecular mechanisms responsible for such actions EZH2 has been described to be involved in regulating both cancer cell immunogenicity, through the silencing of immuno-attractant cytokines and antigen presentation-related genes, and T cell anti-tumor responses [70
Recent evidence suggests that EZH2 plays a role in silencing TEs in cancer cells. Lower levels of EZH2 and higher levels of innate immunity-related genes expression have been reported in chemotherapy-resistant small-cell lung cancer (SCLC) cell lines compared to parental sensitive cell lines [75
]. Reduced expression of EZH2 in resistant cell lines caused antisense transcription of several HERVs located in the 3′UTR regions of ISGs that generated dsRNAs by paring the sense transcripts generated from ISGs in response to IFNs (Figure 3
B). Similar results were obtained after treatment of parental cell lines with the EZH2i GSK-126. Another study showed that taxane-resistant TNBC cell lines had reduced levels of metabolites of the methionine cycle and decreased S-adenosyl-L-methionine (SAM) production compared to parental sensitive cell lines [76
]. These metabolic changes correlated with decreased levels of DNA methylation at TEs and a relocation of the H3K27me3 mark to these regions to maintain TEs repressed. Treatment with EZH2i UNC1999 or GSK343 induced the expression of HERVs and accumulation of dsRNA in taxane-resistant cell lines but not in the parental sensitive lines. Importantly, a similar epigenetic switch from DNA methylation to H3K27 trimethylation at TEs has been also described in mouse embryonic stem cells (mESCs) cultured under DNA-hypomethylating conditions [77
]. These results illustrate how the environment can change the epigenetic mechanisms silencing the expression of TEs and create new vulnerabilities.
6.4. HDAC Inhibitors
Histone acetylation plays an important role in gene expression. Acetylated histones are recognized by epigenetic readers that participate in the recruitment of Pol II to gene promoters. Levels of acetylation at particular genomic locations are the result of an equilibrium between histone acetyl transferases (HATs) and histone deacetylates (HDACs). There are 18 human HDACs grouped into four classes based on their primary homology to yeast HDACs. Among these, class I and II HDACs play a major role in the lysine deacetylation of N-terminal histone tails. Class I comprises HDAC1, 2, 3, and 8. Class II is further divided into two subclasses: IIa (HDAC4, 5, 6, 7, and 9) and IIb (HDAC6 and 10). Four inhibitors (vorinostat, romidepsin, belinostat, and panobinostat), which target several HDACs among the different classes, have been FDA approved for the treatment of several hematological malignancies [78
]. Other HDACi such as entinostat, are currently undergoing clinical trials (Figure 5
Evidence obtained in clinical trials suggest that treatment with HDACi improves the efficacy of immunotherapeutic approaches [79
]. Effects of HDACi might be exerted at several levels of the tumor microenvironment. In cancer cells, HDACi induce the reactivation of genes involved in antigen presentation and the production of immunomodulatory chemokines [79
]. As discussed above, recent data suggest that these effects might be at least partially mediated by the reactivation of endogenous retroviruses. Both DNMTi and HDACi stimulated cryptic transcription of LTR12C retroviral elements in the lung cancer cell line NCI-H1299 [36
]. Induction of LTR12 transcription by HDACi mocetinostat and entinostat that target specifically HDACs 1, 2, and 3 was also reported in cancer cells derived from many tumor species [80
]. However, in this study authors suggest that the targeted LTR12 elements act as promoters that drive the expression of proapoptotic genes after HDACi treatment rather than a source of dsRNAs. The effects of HDACi on TEs might not be limited to cancer cells since exposure of primary CD4+ T cells to a high dose of vorinostat has been reported to also induce expression of LTR12 elements [81
6.5. G9A Inhibitors
G9A is a mammalian histone methyltransferase that catalyzes the methylation histone H3 at lysine 9 (H3K9), participates in gene repression, and is required for the establishment of the silencing of newly integrated proviruses in mESC [82
]. G9A is frequently overexpressed in cancer and correlates with poor prognosis. In agreement, blocking its methyltransferase activity has been reported to block the proliferation of certain cancer cell lines [16
]. Several inhibitors of G9A catalytic activity (G9Ai) have been developed, however, their potential for clinical application is limited by their poor pharmacokinetics in vivo.
Despite being less explored than the previously discussed inhibitors, combination of G9Ai and DNMTi caused synergistic antitumor effects that correlated with the upregulation of HERVs and viral defense genes in ovarian cancer cell lines with high levels of G9A expression [83
]. HERVs upregulated by G9Ai or DNMTi alone were mostly different being CpG-rich HERVs more likely to be repressed by DNA methylation [83
]. Surprisingly, a large number or HERVs were induced only by the combination of both treatments. This might be explained by an epigenetic switch in which these HERVs become repressed by G9A-H3K9 mediated methylation upon loss of DNA methylation. In addition, different ovarian cancer cell lines responded upregulating different HERVs and no correlation was found between the number of HERVs upregulated and the magnitude of induction of viral defense-related genes. These findings suggest that the mechanisms that repress HERVs are to some extent cell specific and that not all HERVs are equal with respect to their abilities to activate the viral defense pathway. In a similar way, a recent study described the upregulation of HERVs, accumulation of dsRNAs, and induction of innate immunity-related genes by the dual G9a/DNMT inhibitor CN-272 in bladder cancer cell lines. Treatment with this inhibitor improved the antitumor effects of anti-PD-L1 therapy in a transgenic mouse model of aggressive metastatic, muscle-invasive bladder cancer [85
In addition to G9A, other H3K9 methyltransferases have been involved in the repression of HERVs. For example, the methyltransferase SETDB1 is known to play a role in the silencing of retroelements in mESCs [82
]. Knock out of SETDB1 in acute myeloid leukemia (AML) cell lines triggered the expression of retrotransposable elements, production of dsRNAs, and the activation of viral defense-related genes [77
]. These results are in agreement with the described function of SETDB1 as a corepressor of transcription factors KRAB-ZFPs that play a major role in the recognition and transcriptional silencing of transposable elements [86