2. m6A Modification and m6A Regulators
N
6-methyladenosine modification is a dynamic and reversible process. In this process, methyltransferases are responsible for the addition of methyl, while removals of m
6A are performed by demethylases. The fate of m
6A-modified RNA depends on the protein that recognizes and binds to it. All of the methyltransferases, demethylases and m
6A-modified RNA bound proteins that take part in the m
6A-modification-associated process belong to m
6A regulators.
2.1. Methyltransferases
Methyltransferases are also called “writers”. They catalyze the installation of m
6A in the form of a complex, the m
6A methyltransferase complex (MTC), which consists of methyltransferase-like 3 (METTL3), methyltransferase-like 14 (METTL14), Wilms = tumor-1-associated protein (WTAP), RNA binding protein 15 (RBM15), RNA binding protein 15B (RBM15B), vir like m
6Amethyltransferase associated (VIRMA) (also known as KIAA1429), zinc finger CCCH-type containing13 (ZC3H13) and HAKAI. METTL3, METTL14 and WTAP constitute the catalytic core of MTC. METTL3 is the main subunit of MTC, with catalytic activity, and can bind S-adenosylmethionine (SAM) and transfer methyl to the N
6 position of adenosine
[30]. In addition, cytoplasmic METTL3 can bind to the m
6A-modified 3′-untranslated region (UTR) of target mRNA, recruit eukaryotic translation initiation factor 3 subunit H (EIF3H) and enhance the translation efficiency
[31,32][31][32]. METTL14 forms a stable heterodimer with METTL3 at a 1:1 ratio, stabilizes METTL3 and recognizes and binds to target RNA through its C-terminal RGG repeats
[30,33][30][33]. With the help of WTAP, the METTL3-METTL14 heterodimer can be located in the nuclear speckle
[34]. In addition to the catalytic core of MTC, other components are associated with WTAP and are helpful for the recruitment and localization of the METTL3-METTL14 heterodimer
[35,36,37,38,39][35][36][37][38][39]. In a WTAP-dependent manner, RBM15/RBM15B correlates with METTL3 and recruits MTC to U-rich regions immediately adjacent to RRACH motifs to catalyze the deposition of m
6A
[37,40,41][37][40][41]. VIRMA is also a WTAP-associated factor and recruits MTC to 3′-UTR and near the stop codon for m
6A catalysis
[39]. ZC3H13 connects RBM15/RBM15B with WTAP through its C-terminal structure domain
[36,38][36][38]. The knockdown of ZC3H13 leads to translocation from the nucleus to cytoplasm for a large proportion of WTAP, VIRMA, HAKAI, METTL3 and METTL14, indicating the important role of ZC3H13 for the nuclear localization of MTC
[38]. HAKAI is an E3 ubiquitin-protein ligase, while its role in m
6A catalysis is unclear. It is worth noting that m
6A deposition depends on transcription. METTL3 and METTL14 form a heterodimer in the cytoplasm and then the heterodimer enters the nucleus with the help of a nuclear localization signal in METTL3. METTL4 can recognize and bind to histone H3 trimethylation at Lys36 (H3k36me3), promote the binding of MTC with RNA polymerase II and transfer MTC to actively transcribing RNAs to install m
6A cotranscriptionally
[42].
In addition to MTC, there are other methyltransferases, such as methyltransferase-like protein 16 (METTL16), cap-specific adenosine methyltransferase (CAPAM), methyltransferase-like protein 5 (METTL5)/tRNA methyltransferase activator subunit 11-2 (TRMT112) complex and zinc finger CCHC-type containing 4 (ZCCHC4). METTL16 catalyzes m
6A deposition in the A43 of U6 small nuclear RNA (snRNA), and is involved in the splicing of RNA
[43,44][43][44]. The m
6A 43 is considered to affect the interaction between snRNA and pre-mRNA and thus regulate the splicing of pre-mRNA
[44]. Methionine adenosyl transferase 2A (MAT2A) encodes SAM synthetase. The 3′-UTR hairpins of MAT2A mRNA are substrates of METTL16. The m
6A modification of 3′-UTR hairpins influences the splicing of MAT2A pre-mRNA and maintains SAM homeostasis
[43,45][43][45]. If 2′-O-methyladenosine (Am) is the first transcribed nucleotide of eukaryotic capped mRNAs, CAPAM can recognize it and deposit m
6A on it to form a m
7GPPPm
6Am motif
[46,47][46][47]. The METTL5/TRMT112 complex and ZCCHC4 are responsible for the methylation of the A1832 of 18S and A4220 of 28S rRNA, respectively
[48,49,50][48][49][50].
2.2. Demethylases
Demethylases are termed as “erasers”. The m
6A demethylation occurs on nascent transcripts. FTO is the first identified demethylase, and alkB homolog 5 (ALKBH5) is the second one. Ferrous iron and α-ketoglutarate are cofactors of FTO and ALKBH5
[51]. Both FTO and ALKBH5 can remove m
6A modification on single RNA and DNA
[8,52][8][52]. In addition, FTO also can act as demethylase for N
6,-2′-O-dimethyladenosine (m
6Am) near the N
7-methylguanosine (m
7G) cap
[53].
2.3. m
6
A RNA Binding Proteins
The fate of m
6A-modified RNA depends on the protein that binds to it. This kind of RNA binding protein is referred to as a “reader”. Readers include YT521-B homology (YTH) family proteins, insulin like growth factor 2 mRNA binding proteins (IGF2BPs), heterogeneous nuclear ribonucleoproteins (HNRNPs) and eukaryotic translation initiation factor 3 (EIF3).
3. m6A Modification and Its Effect on Various RNAs in ESCA
A proper m
6A level, mainly relying on the appropriate expression and function of m
6A regulators, is necessary for sustaining normal bioprocesses. The disruption of the dynamic balance between the installation and removal of m
6A modification will lead to the development of diseases, including cancer. An aberrant m
6A level associated with the dysregulation of m
6A regulators has been reported in a variety of cancers, such as gastric cancer, hepatocellular carcinoma, bladder cancer, etc.
[54,55,56][54][55][56]. Nucleotide sequence changes could also result in the gain or loss of m
6A sites and contribute to carcinogenesis
[57,58][57][58]. For example, a base transition from the G to A of rs5746136 in SOD2 led to an increased m
6A modification level of SOD2 and an increased binding of HNRNPC with SOD2 through an “m
6A switch” mechanism followed by the upregulation of SOD2. The overexpression of SOD2 inhibited the proliferation, migration and invasion of bladder cancer cells, which suggested that SOD2 acted as a tumor suppressor gene for bladder cancer. Thus, the A allele of rs5746136 in SOD2 was associated with a reduced risk of bladder cancer
[57]. m
6A could deposit on various type of RNA, such as coding RNA and non-coding RNA, participate in all steps of RNA metabolism and post-transcriptionally regulate the expression of the gene (
Table 1)
[59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83][59][60][61][62][63][64][65][66][67][68][69][70][71][72][73][74][75][76][77][78][79][80][81][82][83].
Table 1.
The m
6
A methylation modified various RNAs in ESCA.
3.1. m
6
A Modification in ESCA
Most studies revealed that the m
6A level was elevated in ESCC tissues and cell lines
[73,84,85,86][73][84][85][86]. However, Cui et al.
[83] found lower m
6A levels in ESCC cell lines and tissues compared to their counterparts. Some studies indicated that the m
6A level could act as a diagnostic and prognostic marker. In gastric cancer, the m
6A level might be used to distinguish patients from healthy individuals and predict the prognosis of patients treated with immunotherapy
[87,88][87][88]. In lung cancer, a higher m
6A level in circulating tumor cells than in whole blood cells might be associated with tumor metastasis
[89]. Inconsistent results of the m
6A level in ESCC suggested that further large sample studies are needed to draw a reliable conclusion.
3.2. The Effect of m
6
A Modification on mRNA in ESCA
The fate of m
6A-modified RNA relies on the m
6A reader that binds to it. YTH family proteins, consisting of YTH m
6A-binding protein 1 (YTHDF1), YTHDF2, YTHDF3, YTH domain-containing 1 (YTHDC1) and YTHDC2, contain a specific YTH domain, through which, they can recognize and bind to target RNA in an m
6A-dependent way
[90,91][90][91]. YTHDF1 promotes translation initiation and protein synthesis
[10]. On the contrary, YTHDF2, the first discovered m
6A reader, enhances the degradation of m
6A-modified mRNA, either by delivering them to the mRNA decay site or recruiting the CCR4-NOT deadenylase complex to initiate mRNA degradation
[9,92][9][92]. YTHDF3 interacts with YTHDF1 or YTHDF2, playing opposite roles by promoting mRNA translation or enhancing mRNA degradation
[93,94][93][94]. YTHDC1 regulates the splicing of the exon and promotes the translocation of m
6A-modified mRNA from the nucleus to the cytoplasm
[95,96,97][95][96][97]. YTHDC2 elevates the translation efficiency of m
6A-modified mRNA; accordingly, the abundance of target mRNA is reduced
[98,99][98][99]. IGF2BPs, including IGF2BP1, IGF2BP2 and IGF2BP3, enhance mRNA stability
[100]. HNRNPs contain HNRNPA2B1, HNRNPC and HNRNPG. HNRNPC and HNRNPG regulate the alternative splicing of mRNA in an m
6A-dependent way
[13,101][13][101]. EIF3 can act as reader of m
6A in the 5′-UTR of mRNA
[17]. EIF3 participates in almost all steps of translation initiation, which is a rate-limiting process. EIF3 promotes the formation of the 43S pre-initiation complex (PIC), bridges 43S PIC and mRNA bound to the EIF4F complex and takes part in the AUG start codon scanning process
[102,103,104,105][102][103][104][105].
It is generally acknowledged that YTHDF1 plays a facilitating role in translation initiation and protein synthesis
[10]. That was true in the study from Zhao et al.
[73], which demonstrated that YTHDF1 upregulated the protein level of ERBB2 through recognizing m
6A-modified ERBB2 mRNA. Conversely, the knockdown of YTHDF1 enhanced the protein level of HSD17B11, which suggested that YTHDF1 decreased the translation efficiency of m
6A-modified HSD17B11 mRNA
[74]. YTHDF2 promoted the degradation of APC mRNA and decreased APC expression via binding to m
6A-modified APC mRNA
[69]. IGF2BP2 enhanced the stability of m
6A-modified TK1 and KIF18A mRNA and upregulated their expression
[66]. In addition, m
6A readers could regulate the stability and expression of their downstream mRNAs through interacting with lncRNA
[63,64,65,68,70][63][64][65][68][70]. For example, LBX2-AS1 and HNRNPC synergized to increase the stability of ZEB1 and ZEB2 mRNA, upregulated their expression and consequently promoted the migration and epithelial mesenchymal transition (EMT) of ESCC cells
[68].
3.3. The Effect of m
6
A Modification on Non-Coding RNAs in ESCA
Although without a coding ability, non-coding RNAs (ncRNAs), such as miRNA, lncRNA and circRNA, serve a critical role in regulating the expression of the gene.
The miRNAs bind to the 3′-UTR of the target mRNA, and then silence or inhibit the expression of corresponding genes. miRNA biogenesis includes the following steps: firstly, primary miRNA (pri-miRNA) is transcribed from DNA; secondly, pri-miRNA is cleaved into precursor miRNA (pre-miRNA), which requires a microprocessor complex composed of drosha ribonuclease III (DROSHA) and DiGeorge syndrome critical region 8 (DGCR8); thirdly, pre-miRNA is cleaved into mature miRNA. In ESCC, m
6A writers and erasers deposit and remove m
6A on pri-miRNA, respectively
[75,77,78,79][75][77][78][79]. For instance, METTL3 elevates the m
6A level of pri-miR-200-5p, whereas ALKBH5 reduces the m
6A level of pri-miR-194-2
[75,78][75][78]. A previous study showed that HNRNPA2B1 played a crucial role in the maturation of miRNA through recognizing m
6A on pri-miRNA and interacting with DROSHA and DDGCR8
[106]. In ESCC, HNRNPA2B1 promoted the proliferation of ESCC cells through binding to a m
6A-modified miR-17-92 cluster and upregulating the expression of a miR-17-92 cluster
[79].
m
6A modification could also be found on lncRNAs, which might be involved in the regulation of gene expression via influencing the interaction of lncRNAs with RNA binding proteins through an “m
6A switch” mechanism or impacting the interaction between lncRNAs and miRNAs
[13,107][13][107]. In ESCC, the overexpression of FTO significantly reduced the enrichment of m
6A at site 2 of the LINC00022 transcript and led to a decrease in the degradation of LINC00022 by YTHDF2
[83]. The study of Wu et al.
[81] showed that lncRNA LINC00278 encoded a micropeptide named Yin Yang 1 (YY1)-binding micropeptide (YY1BM), and the binding of YTHDF1 to m
6A-modified LINC00278 led to an increased translation of YY1BM. LncRNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) is located at nuclear speckles (NSs). The MALAT1-m
6A-enriched sequence and the binding of YTHDC1 to the m6A of MALAT1 were necessary for maintaining the composition of NSs and migratory capability of ESCC cells
[82]. In addition, as mentioned above, lncRNA could interact with m
6A readers to regulate the expression of target mRNAs
[64,65,68,70][64][65][68][70].
circRNAs perform different biological functions based on their diverse distribution in cells. Nuclear circRNAs might affect transcription and splicing
[108,109][108][109]. Cytoplasmic circRNAs might not only absorb miRNAs and alleviate their depression of the target mRNA
[110,111][110][111] but also might interact with RNA binding proteins and enhance their functional impacts
[112,113][112][113]. It is worth mentioning that the exon-derived circRNAs might have a protein-encoding ability
[114,115][114][115]. m
6A regulators regulated the expression, distribution and function of circRNAs through installing, removing and recognizing m
6A on circRNAs in all sorts of cancers
[112,113,116,117,118,119][112][113][116][117][118][119]. In gastric cancer and cervical cancer, METTL14 and ALKBH5 acted as a transmethylase of circORC5 and demethyltransferase of circCCDC134, respectively
[116,117][116][117]. YTHDC1 facilitated m
6A-modified circMET and circNSUN2 exportation from the nucleus to cytoplasm in NONO-TFE3 fusion renal cell carcinoma and colorectal carcinoma, respectively
[112,113][112][113]. IGF2BP2 interacted with circNSUN2 to increase the stability of HMGA2 mRNA and promote the metastasis of colorectal carcinoma
[113]. IGF2BP1 promoted the translation of m
6A-modified circMAP3K4 into circMAP3K4-455aa via recognizing m
6A modification on circMAP3K4 in hepatocellular carcinoma
[118]. Interestingly, m
6A modification on circALG1 enhanced its ability as competitive endogenous RNA (ceRNA) of miR-342-5p by increasing its binding to miR-342-5p in colorectal cancer
[119]. It is a pity that no study has reported how m
6A modification regulates the expression and function of circRNAs during the development and progression of ESCA until now, which provides a direction for future research in ESCA.