INTRODUCTION
Kidney cancer accounts for 2%–3% of all cancers worldwide. Renal cell carcinoma (RCC) is the most preva-lent subtype of kidney cancer, accounting for approximately 90% of all cases,1 and is a heterogeneous disease with near-ly 20 subtypes. The most prevalent subtype is clear cell (cc)RCC (75%), followed by papillary (p)RCC (10%) and chromophobe (ch)RCC (5%)2; as a result, RCC research is mostly limited to ccRCC. The most common genetic event of ccRCC is the inactivation of the von Hippel-Lindau (VHL) tumor suppressor gene,3 which induces the stabilization of HIF1a and HIF2a to increase transcription of the hy-poxia response genes VEGF and PDGF and leads to angiogenesis.4 pRCC is further classified by genotype: type1 pRCC occurs due to a mutation in the c-Met proto- oncogene, which encodes a tyrosine kinase, and type 2 pRCC is associated with mutations in the gene encoding fu-marate hydratase.5 In other RCC subtypes, the involvement of VHL and c-Met are unclear. Therefore, these targets alone are insufficient for treating all forms of RCC, high-lighting the critical need to identify novel therapeutic targets relevant to several subtypes of RCC. In this review, we will summarize the epigenetics in RCC and discuss the potential use of epigenetic modifiers as RCC biomarkers.
EPIGENETIC REGULATION
Regulation of gene expression and activity is an essential process for cell survival and adaptation amidst various envi-ronmental conditions. Chromatin structure must be correctly regulated by epigenetic regulators for appropriate gene expression. These regulators are classified as epigenetic writers, readers, and erasers according to their role. Epigenetic writers are enzymes that add functional groups (epigenetic marks) to DNA and histones that are recognized by epigenetic readers. Most epigenetic marks are reversible by action of epigenetic erasers, which can remove the marks and reverse its effects. Writers and erasers can contain read-er domains that allowing a ‘reading-writing’ or ‘reading- erasing’ process. The interaction system of the 3 epigenetic regulators controls gene transcription, and numerous reports have demonstrated that defects of this system induce the ini-tiation and progression of several cancers, including RCC.6,7
EPIGENETIC WRITERS IN RCC
Epigenetic writers enable several epigenetic modifications, including methylation, acetylation, phosphorylation, and ubiquitination. This review will focus on methylation and acetylation, which are the most widely studied mod-ifications among epigenetic marks. Both DNA and histone proteins are liable to methylation, while acetylation is re-stricted to histones.8 These 2 modifications regulate cellular gene expression patterns by cycling between activation or suppression of transcription. Here, we describe 3 epigenetic writers: DNA methyltransferases (DNMTs), histone methyl-transferases (HMTs), and histone acetyltransferase (HATs).
1. DNA Methyltransferases
DNA methylation is a process during which methyl groups bind to DNA to affect gene expression without changing DNA sequences. This modification most often in-volves the covalent binding of methyl to the 5'-carbon of cytosine residues. In mammals, the DNMT family of en-zymes, which mediates DNA methylation, consists of DNMT1, DNMT2, DNMT3A, DNMT3B, and DNMT3L. Maintenance of DNA methylation during DNA replication is accomplished by DNMT1, whereas de novo DNA methylation is catalyzed by DNMT3A and DNMT3B. DNA methylation can occur in promoters, intergenic regions, gene bodies, and enhancers. Promoter methylation typically leads to transcriptional repression and is usually associated with CpG dinucleotides. The importance of gene body methylation has been unestablished but is generally asso-ciated with a permissive transcriptional state. Although both DNMT3A and DNMT3B participate in de novo DNA meth-ylation, but their target regions are different.9,10 DNMT3A primarily methylates promoters and enhancers, whereas DNMT3B is associated with actively transcribed genomic loci.
Several studies have shown that abnormal DNA methyl-ation occurs in RCC. For example, hypermethylation of tu-mor suppressor genes by abnormally high expression of DNMT or hypo-methylation of oncogenes due to low levels of DNMT may contribute to the generation, development, and progression of ccRCC.11 In addition, the hypermethy-lation of KLF5, an epidermal growth factor response ele-ment, by DNMT1 is strongly associated with a poor prog-nosis in ccRCC patients.12 While DNMT1 plays an onco-genic role in RCC, DNMT3A exhibits tumor suppressive role. HIF2a, an important gene in ccRCC, is silenced via methylation by DNMT3A, which limits the proliferation ca-pacity of cells under hypoxic conditions.13
2. Histone Methyltransferases
The methylation of histone lysine residues can be wide-spread and can result in various transcriptional effects, de-pending on the extend and position of the modifications. Lysine is modified with one or more methyl groups (mono-, di-, and trimethyl) at various positions. Methylation of H3K4, H3K36, and H3K79 is considered an active chroma-tin mark, whereas H3K9, H3K27, and H4K20 methylation are associated with transcriptional repression. However, methylation of H3K4 usually occurs with trimethyl marks found in active promoters, whereas monomethyl marks tar-get active enhancers.14
In general, methyl marks are enzymatically added to his-tone lysine residues by lysine methyltransferases (KMT), a group of highly specific enzymes that target lysine residues with a defined number of methyl groups, leading to corre-sponding transcriptional effects (Table 1). For example, en-zymes KMT1A (SUV39H1) and KMT1B (SUV39H2) convert only H3K9me1 to H3K9me3, while the related enzyme KMT1C (G9a) generates only H3K9me2. KMT1C has been suggested as a potential therapeutic target for ccRCC be-cause it catalyzes the methylation of H3K9 to negatively regulate essential tumor suppressors during hypoxia.15 KMT4 (DOT1L) is the only enzyme involved in the methyl-ation of H3K79 and can lead to mono-, di-, or trimethyl conversion. Methylation by KMT4 is important in a variety of biological processes, including telomere maintenance and cell cycle regulation. Notably, higher KMT4 expression in ccRCC patients has been associated with poor clinical outcomes.16 Analysis of ccRCC-related data from The Cancer Genome Atlas revealed a relevant somatic mutation of SETD2, which normally methylates H3K36 and acts as a tumor suppressor. DNA hypo-methylation resulting from the loss of SETD2 is associated with ccRCC oncogenesis.17 However, another study reported that SETD2 expression it-self, without H3K36 methylation, is related to the ag-gressiveness of ccRCC.18 Through tissue screening of global methylated H3K among ccRCC patients, mRNA expression levels of 4 HMTs (MLL1, MLL2, SMYD2, and NSD2) were significantly increased compared to normal renal cells.19 EZH2, which catalyzes methylation of H3K27, is highly expressed in RCCs, contributes to short survival times in these patients, and is a rational target for RCCs that are resistant to sunitinib, a common ccRCC chemothe-rapeutic.20,21
Table 1.
Lysine methyltransferases and their associated substrates and effects on chromatin
3. Histone Acetyltransferases
Lysine acetylation induces decompaction of chromatin, providing access for transcription factors to increase gene expression. Acetylation marks and HATs are mainly found in association with active promoters and enhancers.22 Numerous lysine residues in the tails of histones H2A, H2B, H3, and H4 can be acetylated, and these modifications play an important role in DNA repair and replication, as well as in transcription.23,24
Histone acetyltransferases, also known as lysine acetyl-transferases (KATs), are grouped based on their amino acid sequence, with the GNAT, MYST, and p300/CBP families being most characteristic (Table 2). The GNAT family con-sists of KAT2A (GCN5) and KAT2B (PCAF), which are coactivators for various transcription factors. In mammals, the MYST family includes 5 enzymes (KAT8 [MYST1], KAT7 [MYST2], KAT6A [MYST3], KAT6B [MYST4], and KAT5 [TIP60]) with the MYST domain that are di-rectly involved in protein proliferation, differentiation, and apoptosis.25 The CBP/P300 family comprises CBP and KAT3B (P300), enzymes with catalytic domains and bro-modomains that also act as coactivators.26 In ccRCC, over-expression of CBP not only induces carcinogenesis27 but al-so plays an important role in the transcriptional activation of the G250MN protein, an isoenzyme of the carbonic an-hydrase family, which is involved in increased expression of HIF1a in ccRCC.28
Table 2.
Lysine methyltransferases and their associated substrates
4. Epigenetic Therapy: Targeting Writer in RCC
Many studies evaluating the therapeutic benefits of drug targeting epigenetic writer are underway (Table 3, Fig. 1). For example, decitabine (5-AZA-2'-deoxycytidine) is a rep-resentative DNMTi that inhibits cell growth and NF-kB sig-naling in various cancers, including RCCs.29,30 The dinu-cleotide of decitabine, guadecitabine, exhibits improved ab-sorption capacity for DNA of rapidly dividing tumor cells, leading to improved efficacy.31 Both of these drugs are cur-rently in a clinical trial for their activity against RCCs (NCT04049344,32 NCT00561912, NCT03165721). Azacitidine, another drug currently in clinical trials, is coadminis-tered with interferon-α2b or bevacizumab (NCT00217542 and NCT00934440, respectively), while MG98 is being evaluated in a phase II clinical trial as a monotherapy for RCCs.33
Fig. 1.
Graphical abstract for DNA and histone modification. This figure shows the regulatory mechanisms of DNA and histones by the epigenetic writer, reader, and eraser, as well as a list of inhibitors for each enzyme used in renal cell carcinoma therapy.

Table 3.
Clinical trial status of epigenetics-based therapeutic agents in renal cell carcinoma
Although the development of HMTi has not advanced as far as that of other therapies, increasing evidence indicates that HMT is an important druggable target for cancer therapy. In RCCs, the use of HMTi leads to potent anti-tumorigenic effects.34–37 Specifically, the EZH2 inhibitor ta-zemetostat is in a phase II clinical trial as a monotherapy for RCC (NCT02601950). Numerous studies of HATi have been performed, such as P300 inhibitor,38,39 but no clinical trials are currently underway in RCCs.
EPIGENETIC READERS IN RCC
Various histone and DNA modifications suggested by ep-igenetic writers are recognized by epigenetic readers, which contain specialized domains that identify and bind to these modifications. We will focus on protein domains that can recognize methylated DNA and methylated or acetylated histones, the 2 most widely studied histone modifications.
1. DNA Methylation Readers
Methyl-CpG binding domain (MBD) proteins recognize and bind to methylated DNA to recruit factors for chromatin silencing. These proteins are classified into 3 groups according to domains: MBD, ZnF, and SRA (SET and RING fin-ger-associated) families (Fig. 2). The MBD family includes MeCP2 and proteins MBD1–6; MeCP2. Among them, MBD1 and MBD2 bind to methylated DNA and interact with various inhibitory complexes through their transcriptional repressive domain (TRD) to silence the bound loci. MBD3 and MBD4, which do not contain a TRD, cannot bind to methylated DNA, although MBD4 enhances DNA repair through glycosylase activation.40 Because MBD is rarely reported in RCC, further research on its possible role in carcinogenesis is needed.
Fig. 2.
Methyl-CpG binding domain proteins (MBPs). These epigenetic readers are classified as into MBD, ZnF, or SRA families based on the presence of specific binding motifs. MBD: methyl-CpG binding domain, ZnF: zinc finger, SRA: SET and RING finger-associated.

Eight proteins (Kaiso, ZBTB4, ZBTB38, ZFP57, KLF4, EGR1, WT1, CTCF) with a ZnF motif compose the ZnF family. Kaiso, ZBTB4, and ZBTB38, the most well-studied proteins in this family, bind to methylated DNA41 and are known to repress transcription in a DNA methylation-de-pendent-manner. However, a recent study suggested that some proteins containing the ZnF motif initially bind to un-methylated regions. ZFP57 is essential for the recognition of methylated hexanucleotides (TGCCGC) in the imprinting control region and for protecting DNA methylation imprints during early embryogenesis in mammals.42
The SRA domain proteins consist of UHRF1 (ubiquitin- like with plant homeodomain [PHD] and RING finger-do-main 1) and UHRF2. UHRF1 recognizes hemi-methylated DNA and recruits DNMT1.43 Although UHRF2 has an ami-no acid sequence similar to that of UHRF1, UHRF2 functions differently to maintain methylation and primarily rec-ognizes 5hmc and 5mc.44 UHRF1, which is overexpressed in most cancers, has been proposed as a universal oncogenic biomarker; in ccRCC, UHRF1 inhibits p53-mediated cell apoptosis by inactivating p53.45
2. Histone Methylation Readers
Methylated histone lysine residues are recognized by sev-eral domains: Ankyrin, ADD (ATRX-Dnmt3-Dnmt3L), bro-mo-adjacent homolog, chromo-barrel, chromodomain, dou-ble chromodomain, malignant brain tumor, PHD, Proline- Threonine-Proline-Threonine (PWWP), tandem Tudor do-main, and WD4046 (Fig. 3). Each domain is highly specific and recognizes lysines in a particular methylation state. For example, the PHD finger interacts with H3K4me3, an active promoter mark, whereas the WD40 domain binds to various trimethylated lysines related to repressive marks, and the ADD domain of ATRX recognizes H3K9me3 only when H3K4me2 and H3K4me3 are absent. The specificity of these domains is controlled by interactions with amino acids near the modified lysine and the structure of the binding pocket. In RCC, ankyrin repeat and single KH domain 1 was significantly increased compared to normal renal cells and was essential for controlling cell proliferation in ccRCC through interaction with 3 tumor-suppressive miRNAs (miR− 29a, miR-205, and miR-196a).47 Nuclear expression of PHD finger 2 in ccRCC is also greatly increased, suggesting that epigenetic control of lipid synthesis via CEBPa is important.48
Fig. 3.
Histone methylation readers. Proteins that recognize methylated histone lysine residues contain diverse recognition domains as indicated by the outer boxes. ADD: ATRX-Dnmt3-Dnmt3L, BAH: bromo-adjacent homolog, TTD: tandem tudor domain, PWWP: Proline-Threonine-Proline-Threonine, PHD: plant homeodomain, MBT: malignant brain tumor, DCD: double chromodomain.

3. Histone Acetylation Readers
The decompaction of chromatin induced by histone acety-lation is specifically recognized by the bromodomain and tandem PHD domains. However, the specificity of these do-mains is lower than that of methylation readers. Bromodo-mains predominate in studies of histone acetylation readers because bromodomains were the only known acetylation reader until 2008, when the PHD domain was discovered.49 Approximately 60 types of proteins with broad functions in humans contain bromodomains.50 For example, Bromodo-mains and Extraterminal (BET) family proteins, which con-tain 2 bromodomains, include the readers BRD2, BRD3, BRD4, and BRDT. Among these, BRD4 is highly expressed in various cancers and is considered an oncogene51; for ex-ample, upregulated BRD4 in RCC plays a critical role in RCCs progression (Fig. 4).52 In support of this relationship, the combined chemical inhibition of BRD4 and PI3K-AKT pathway components via SF2523 can effectively block ccRCC growth.53
4. Epigenetic Therapy: Targeting Reader in RCC
The treatment of RCC by targeting DNA and histone methylation readers has been rarely studied, whereas the re-search on BETs, histone acetylation reader, has been actively performed. Although the study of JQ1, which is one of the famous BET inhibitors, has been conducted by many study groups, JQ1 is known to not only have a short half- life but also the concentration required when used in mono-therapy exceeds the physiological safety level.54 As new BET inhibitors, TET-010 and BMS086158 are currently being evaluated in clinical trials (NCT01987362 and NCT024 19417) (Table 3, Fig. 1). However, because drugs targeting epigenetic readers are relatively less researched than other enzymes, more research on readers will be needed.
EPIGENETIC ERASERS IN RCC
Epigenetic marks are not permanent, which allows the cell to adapt to environmental changes. For the chromatin to react rapidly, the expression state of the locus must be modified. To accomplish this rapid conversion, epigenetic erasers remove epigenetic marks generated by the writer. Here, we discuss the reversibility of DNA methylation, his-tone methylation, and histone acetylation via epigenetic erasers.
1. DNA Demethylases
The demethylation of mammalian DNA occurs either pas-sively via absence of DNMT or actively via catalysis of the TET enzyme family.55 In the absence of DNMT, methylated cytosine is repeatedly removed during DNA replication. Active DNA demethylation is catalyzed by TET family en-zymes, which include TET1, TET2, and TET3. The core catalytic domain of these proteins consists of a cystine-rich domain, a double-stranded -helix domain, and a Fe(II)-in-teracting HxD motif and is highly conserved in the car-boxy-terminus.56 TET1 and TET3 have an additional CXXC domain with 2 Cys4-type zinc finger motifs in their amino terminals to promote DNA binding.57,58 TET1 clears the im-printing mark in primordial germ cells, whereas in RCCs, the miR124-mediated silencing of TET1 leads to increased MEG3 expression, cell proliferation, and tumor growth.59 TET2 plays an important role in hematopoiesis, and mutations in TET2 result in hematopoietic malignancies.60,61 Moreover, TET2 is mutated in RCCs, leading to sunitinib resistance during immunotherapy.62 Although TET3 constitutively participates in zygotic paternal DNA reprogram-ming,63 TET3 is one of 8 significantly upregulated genes in 55 primary ccRCC tumors and affects patient survival rates.56
2. Histone Demethylases
Numerous histone lysine demethylases (KDMs) remove methyl groups from lysine residues (Fig. 5). These KDMs are classified as LSD and JMJC proteins by their enzymatic mechanisms. The LSD group consists of KDM1A (LSD1) and KDM1B (LSD2). LSD1 specifically removes mono- and dimethylation of H3K4 and H3K9 lysines to generate formaldehyde via methyl group oxidation.64 LSD2 also de-methylates mono- and dimethylated lysines of H3K4 and may act as a maternal imprinting factor in oocytes, as well as an activator of NF-kB.65,66 In RCC, H3K4 pattern analy-sis determined that LSD2, but not LSD1, is highly ex-pressed and that LSD2 inhibition led to suppressed cell via-bility and eventual apoptosis. Accordingly, LSD2 has been proposed as a new prognostic marker for RCC.19 The JMJC family includes approximately 20 lysine-spe-cific demethylases. Unlike LSD proteins, JMJC enzymes can eliminate trimethylation.67 Representative JMJC mem-bers include KDM2, KDM3, KDM4, KDM5, KDM6, KDM7, and KDM8. Although several histone methyl marks (e.g., H3K79) have not yet been correlated with a specific demethylase,68 some KDMs are reported to play an im-portant role in RCCs.69 For example, KDM3A, a demethy-lase of H3K9me1/2 involved in the development of several cancers70,71 is highly expressed during hypoxic conditions, is associated with increased HIF1a, and contributes to RCC development and progression via the hypoxia-mediated an-giogenesis pathway.63 KDM5C demethylates H3K4me1/2, and its altered expression affects breast cancer develop-ment.72 A systemic sequencing study of ccRCC reported mutations of KDM5C in approximately 4% of the 132 ccRCC cases,73,74 KDM6A and KDM6B demethylases target H3K27me2/3, although in RCC, KDM6A carries a somatic null mutation and KDM6B is overexpressed, leading to on-cogene-induced senescence. Thus, both KDM6A/B appear to play important, yet contrasting, roles in ccRCC develop-ment.75,76
3. Histone Deacetylases
Acetylation of histone lysines is reversed by various his-tone deacetylases (HDACs), which inhibits transcription by increasing chromatin compaction. In mammals, HDACs are classified into 2 families, the HDAC family and the Sirtuins (SIRT)/class III family (Fig. 6). The HDAC family consists of 11 enzymes and is further divided into classes I, IIa, IIb, or IV based on homology. Class I HDACs include HDAC1/ 2/3, which are collectively recruited and expressed at chro-matin as a complex,40,77–80 whereas HDAC8 is active in isolation.81 HDAC1 and HDAC2 are essential for the growth and survival of ccRCC, as the genetic removal of HDAC1/2 inhibits apoptosis and downregulation of the epi-thelial marker E-cadherin.82 HDAC3 is also overexpressed in RCCs and is involved in tumorigenesis.83
Fig. 6.
Histone deacetylases. The acetylation of histone lysine residues is reversed by multiple classes of epigenetic erasers with specific recognition sites with chromatin. SIRT: selective internal radiation therapy.

Class IIa HDACs, consisting of HDAC4, HDAC5, HDAC7, and HDAC9, also function by forming a complex but have lower catalytic activity compared to class I and IIb enzymes.84 A study in which an N-terminal spliced var-iant of HDAC9 repressed gene transcription confirmed that its catalytic domain is nonessential in the complex.85 In RCCs, low levels of H3K18ac and H3K27ac, induced by highly enriched HDAC7 in the OCT2 promoter, regulate drug uptake by inhibiting OCT2 transcription.86
HDAC6 and HDAC10 are class IIb enzymes with differ-ent defining domains.87 In RCCs, HDAC6 and HDAC10 are potential prognostic factors, as HDAC6 is increased and HDAC6 is decreased compared to normal kidneys.88,89 Therefore, further study is needed to determine which path-way(s) are targeted by these HDACs and their contribution to RCC development. HDAC11 is the only class IV HDAC, and while its sequence is similar to that of classes I and II HDACs, this enzyme is relatively uncharacterized.90
The SIRT/class III HDAC family includes SIRT1, SIRT2, SIRT3, SIRT4, SIRT5, SIRT6, and SIRT7, which function through NAD+-dependent mechanisms.91 Expression screening of SIRT proteins revealed that levels of SIRT1, SIRT3, and SIRT6 were significantly lower in RCCs than in normal kidneys.92 In particular, the downregulation of SIRT3 leads to increased reactive oxygen species and stabilization of HIF1a in RCC, highlighting its tumor-suppressive role in RCC, as well as in other cancers.93,94
4. Epigenetic Therapy: Targeting Eraser in RCC
The most commonly developed and used drugs in epi-genetics-based therapy are HDAC inhibitors (HDACi) (Table 3, Fig. 1). Six additional HDACi are currently in clinical trials: vorinostat, panobinostat, romidepsin, entinostat, AR− 42, and belinostat. Vorinostat (SAHA) is a potent nonselective HDACi that has been studied as a monotherapy for RCCs (NCT00278395) and in combination therapy with 5 other drugs (NCT00324740, NCT2619253, NCT00324870).95,96 Panobinostat is a pan-HDACi used either alone or in combi-nation with 2 drugs (NCT00550277, NCT01582009, NCT0 1005797). Romidepsin is an U.S. Food and Drug Adminis-tration (FDA)-approved potent inhibitor of HDAC1 and HDAC2 (NCT01638533, NCT00106613). Entinostat is an-other class I HDACi and has been tested in both monothe-rapy and together with 2 other compounds (NCT03552380, NCT03552380, NCT01038778, NCT03024437).97 AR-42 has been reported to have greater efficacy than SAHA, which was first used in 2005.98 Belinostat, also called PDX101, exhibits antiproliferative and HDAC inhibitory ca-pabilities at nanomolar concentrations against various hu-man cancers.99 Although not all HDACi are cleared for RCCs therapy, many of them are FDA-approved for their therapeutic effects against many diseases.100
CONCLUSIONS
The role of epigenetic changes in the development and progression of RCC has become increasingly clear. However, despite the extensively knowledge of many epigenetic modifiers, further study of poorly characterized enzymes, especially in the context of cancer pathogenesis, is needed. Here, we have reviewed the evidence that epigenetic read-ers, writers, and erasers are candidate targets for the treat-ment of subtypes of RCCs. Taken together, this review pro-vides innovative insights that targeting epigenetic regulators would be a good therapeutic strategy.