The central dogma declares that DNAs are transcribed into RNAs which are translated into proteins thereafter [1]. However, with the advent of the human gene program in June 2000, only a few genomes of humans encode for proteins via the transcription of mRNA. Scientists started to dig into the function and mechanism of noncoding RNAs (ncRNAs), including micro RNAs (miRNA), small nucleolar RNAs (snoRNAs), long noncoding RNAs (lncRNA), small nuclear RNAs (snRNAs), and circular RNAs (circRNA) ^[2][3][4][2,3,4]. In contrast to mRNA, ncRNAs cannot encode proteins, although they are involved in epigenetic modifications and gene-transcription regulation at the RNA level [3].

The concept of N6-methyladenosine (m6A) was first reported by several revolutionary studies in the 1970s ^[5][6][5,6]. The process of m6A methylates adenosine at the N6 position, which is common in both coding and noncoding RNAs, acting as the most common process of methylation of RNAs until now [7]. In 1997, scientists successfully cloned the first m6A writer, methyltransferase-like protein 3 (METTL3), which methylates nearly all m6A in mRNA [8]. In 2011, the fat mass and obesity-related protein (FTO) was identified as the first m6A demethylase, suggesting that m6A modification is reversible and dynamic, which injected new vitality into epigenetics [9].

2. m6A-Modification and ncRNA

2.1. m6A-Modification in miRNA

miRNAs are short ncRNAs with an average length of 22 nt. A mature miRNA derives from a primary miRNA (pri-miRNA) which is cleaved by RNA Polymerase II (Pol II) from DNA. Then, the pri-miRNA is processed into a precursor miRNA (pre-miRNA) with the assistance of a microprocessor complex, consisting of Drosha, a ribonuclease III enzyme, and DiGeorge Syndrome Critical Region 8 (DGCR8) ^[10][11][62,63]. The lack of METTL-3 is proven to affect miRNA biogenesis by reducing the binding between DGCR8 and pri-miRNA ^[12][64]. As a result, the number of mature miRNAs decreases, and pri-miRNAs without cleavage accumulate. In further studies, METTL-3 was found to accelerate the maturation of pri-miR221/222 via interplaying with DGCR8, leading to the oncogenesis of bladder cancer ^[13][65]. In addition, METTL-3 was revealed to promote the modification of miR-181d-5p in some patients with resistance to 5-Fluorouracil (5-FU), which was attributed to the interaction between METTL-3 and DGCR8 ^[14][66]. Besides DGCR8, METTL-3 was reported to engage in the axis of METTL3-miR-25-3p-PHLPP2-AKT as an outcome of smoking. Cigarette smoke condensate (CSC) triggers the overexpression of METTL-3, which catalyzes the methylation thereafter being decoded by an m6A reader, NF-κB associated protein (NKAP). Methylation could affect the maturity of miR-25-3p, then prohibit PH domain leucine-rich repeat protein phosphatase 2 (PHLPP2), which provokes the oncogenic signaling of AKT-p70S6K and finally causes pancreatic cancer ^[15][67]. In hepatocellular carcinoma (HCC), METTL-14 can also interact with DGCR8. Downregulation of METTL-14 inhibits the expression of miRNA 126 (a cancer suppressor) and gives rise to the proliferation of HCC ^[16][68]. Recently, FTO was discovered to increase the expression of ADP ribosylation factors like GTPase 5B (ARL5B) in breast cancer cells by inhibiting miR-181b-3p. MiR-181b-3p is an miRNA that takes part in the FTO/miR-181b-3p/ARL5B axis ^[17][69]. Yes-associated protein (YAP) is an important transcriptional regulator controlled by m6A modification on miRNA in cancer progression. Overexpression of ALKBH5 reduces the expression of miR-107. However, silencing depends on Human antigen R (HuR), which can reverse this process and decrease YAP activity by intervening in the axis of miR-107/LATS2 depending on HuR, which indicates ALKBH5 is a suppressing factor in the oncogenesis of nonsmall-cell lung cancer (NSCLC) ^[18][19][70,71]. RALY (HNRNPCL2) is newly found as an RNA-binding protein associated with the aggressiveness of colorectal cancer (CRC). As a part of the Drosha complex, RALY processes the maturation of a group of miRNAs, such as miR-483, miR-676, and miR-877. Consequently, these miRNAs downregulate relative genes in the mitochondria and contribute to the growth of CRC cells ^[20][72]. HNRNPA2B1 interacts with DGCR8 to promote the generation of pre-miRNA ^[21][33]. In tamoxifen-resistant breast cancer cells, HNRNPA2B1 was reported to be upregulated, which led to the hypothesis that HNRNPA2B1 facilitates endocrine resistance by affecting the expression of miRNA. A follow-up study revealed that the overexpression of HNRNPA2/B1 reduced miR-29a-3p, miR-29b-3p, and miR-222. On the contrary, miR-1266-5p, miR-1268a, and miR-671-3p were upregulated in MCF-7 cells ^[22][73]. The result manifests the overexpressed HNRNPA2B1 and contributes to the reduction of certain miRNA, which leads to tamoxifen resistance in breast cancer cells. The evidence above confirms that m6A-modification plays an important part in miRNA in the occurrence and development of cancer. However, further studies are still needed.

2.2. m6A-Modification in lncRNA

lncRNAs are ncRNA transcripts with at least 200 nucleotides that interact with RNA and proteins to regulate transcription and epigenetic modifications in carcinogenesis ^[23][74]. X inactive-specific transcript (Xist) is a regulatory site located on the X chromosome which can produce a lncRNA 15 to 17 kb in length ^[24][75]. Knockdown of METTL14 reduces m6A modification on Xist and enhances the expression of Xist which consequently promotes the metastasis and aggressiveness of CRC. Intriguingly, Xist methylated with m6A can be recognized by YTHDF2, resulting in the degradation of Xist ^[25][76]. The evidence further confirms the function of Xist is negatively correlated to METTL14 and YTHDF2. In gastric cancer (GC) cells, THAP7-AS1 is a lncRNA which obviously upregulated compared with normal gastric cells. METTL3-mediated m6A modifies the activation of THAP7-AS1 by SP1. SP1 is a transcription factor that carries out the function of oncogenesis and supports the characteristic of THAP7-AS1 in facilitating lymphatic metastasis. Notably, this process is not only modulated by METTL3 but also with the engagement of IGF2BP1 to decipher the signal of m6A methylation ^[26][77]. In NSCLC, ABHD11-AS1, a lncRNA, is upregulated. MeRIP-Seq proved ABHD11-AS1 had specific methylation sites which were prone to be installed with m6A by METTL3. As a result, METTL3 induces the upexpression of ABHD11-AS1, thus promoting the proliferation of NSCLC tissue ^[27][78]. Another study analyzed the interaction between LNCAROD, an oncogenic lncRNA, and METTL3/METTL14. METTL3/METTL14-induced m6A methylation stabilized LNCAROD in head and neck squamous cell carcinoma (HNSCC) tissue, which augmented malignant cell multiplication and invasion in vivo and in vitro ^[28][79]. In nasopharyngeal carcinoma (NPC), WTAP fine-tuned by KAT3A-mediated H3K27 acetylation is required to stabilize the m6A methylation of DIAPH1-AS1, a lncRNA, and plays an essential role in the growth and metastasis of NPC ^[29][80]. As for m6A erasers, FTO is reported to demethylate m6A on LINC00022, a lipogenesis-related lncRNA, leading to the upregulation of LINC00022 in esophageal squamous cell carcinoma (ESCC) cells ^[30][31][81,82]. Though progress has been made in discovering the regulation of ESCC, the mechanism of m6A modification in lncRNA cancer susceptibility candidate 15 (CASC15) is poorly identified. A study revealed FTO mediated the demethylation process on CASC15. The absence of FTO impaired neoplastic proliferation and prompted apoptosis of ESCC cells conducted by CASC15 ^[32][83], which demonstrated the interaction between erasers and lncRNA in oncogenesis. Additionally, in pancreatic cancer, IGF2BP2 was reported to combine with LncRNA-PACERR. LncRNA-PACERR is a stimulator of protumor macrophages to increase the stability of KLF12 and c-myc ^[33][84] which denotes the critical role of m6A regulators in the polarization of tissue-associated macrophages (TAMs). Of note, besides direct interplay with m6A regulators, lncRNA can encode a peptide with 71 amino acids called RNA-binding regulatory peptide (RBRP) to indirectly bind to IGF2BP1. Therefore, IGF2BP1 recognizes m6A on RNAs and causes tumors ^[34][85].

2.3. m6A-Modification in circRNA

circRNAs are covalently closed RNAs produced through the back-splicing of exons in eukaryotes. Unlike linear RNAs, circRNAs are unique both in structure and biomedical functions, such as transcription, splicing, and translation ^[35][86]. Currently, studies have proved that m6A modification can regulate circRNAs via m6A writers, erasers, and readers and vice versa ^[36][37][38][87,88,89]. There are four main aspects that m6A interacts with circRNAs: (1) m6A regulates the biogenesis of circRNAs. In NSCLC, circIGF2BP3 is methylated with m6A by METTL3 and circulated in a YTHDC1-dependent manner to impair the immune response of cancer ^[39][90]. (2) m6A prompts the nuclear exportation of circRNAs. circNSUN2, a circRNA upregulated in liver-metastatic colon cancer, can be exported from the nucleus to cytosol with the presence of YTHDC1. Additionally, METTL3 is found to facilitate the transfer of circNSUN2 ^[40][91]. (3) m6A modulates the translation of circRNAs. Different from other ncRNAs, circRNAs are capable to encode proteins driven by m6A modification. The process is mediated by YTHDF3 and enhanced by METTL3/14. M6A-driven circRNA translation was widespread ^[41][92]. (4) m6A is associated with the degradation of circRNA. In gefitinib-resistant cells, m6A-modified circASK1 is downregulated due to the increase of endoribonucleolytic cleavage induced by YTHDF2 ^[42][93].

3. Clinical Applications of m6A for ncRNA in Cancer

3.1. Role of ncRNA Methylation in Predicting Prognosis of Cancer

Noncoding RNA methylation can be used as a marker to predict prognosis. For example, overexpression of miR-25-3p caused by METTL3 indicates a poor survival of pancreatic cancer ^[15][67]. METTL3 increases the biogenesis of miR-143-3p, which was negatively correlated with the overall survival (OS) rate of lung cancer ^[43][95]. M6A may upregulate the expression of LNCAROD by enhancing its stability. LNCAROD increases the expression of pyruvate kinase isoform PKM2 to activate glycolysis in hepatocellular carcinoma cells, eventually elevating the malignancy of HCC. Thus, LNCAROD indicates a poor prognosis for HCC patients ^[44][96]. The m6A level of the long noncoding RNA NEAT1 was a powerful predictor of eventual death ^[45][97]. Furthermore, METTL3 increases the stability of Lung Cancer Associated Transcript 3 (LCAT3). LCAT3 is known as a long noncoding RNA. Upregulation of LCAT3 represents the poor prognosis of lung adenocarcinoma (LUAD) patients ^[46][98]. In addition, Circ3823 facilitates metastasis and angiogenesis of CRC via the circ3823/miR-30c-5p/TCF7 axis. In CRC patients, Circ3823 indicated a worse prognosis ^[47][99]. METTL3 promotes the maturation of pri-miR221/ 222 and was associated with poor prognosis in bladder cancer patients ^[13][65]. METTL-14 inhibits the expression of miRNA 126, and downregulation of METTL-14 is an adverse prognosis factor in hepatocellular carcinoma ^[16][68]. METTL14 downregulates the expression of long noncoding RNA XIST depending on YTHDF2. Decreased expression of METTL14 was associated with an unfavorable prognosis for CRC patients ^[25][76]. In gastric cancer, lower expression of METTL14 is associated with poor prognosis. Mechanically, reduced expression of METTL14 upregulated the expression of circORC5. circORC5 could sponge miR-30c-2-3p and downregulate AKT1S1 and EIF4B. Meanwhile, upregulation of circORC5 represents poor prognosis too ^[48][100]. The FTO/miR-181b-3p/ARL5B axis modulates the migration and invasion of breast cancer cells. Overexpression of FTO predicts poor prognosis and advanced TNM stage [P = 0.001] ^[17][69]. Higher expression of LINC00460 represents poor disease-free and five-year overall survival. LINC00460 interacts with IGF2BP2 and DHX9 to enhance the stability of high-mobility group AT-hook 1 (HMGA1) mRNA ^[49][101]. LINC00266-1 encodes RBRP which could interact with IGF2BP1. A high level of RBRP was correlated with poor prognosis ^[34][85]. RAL Y is known as a novel m6A reader. Overexpression of RAL Y is associated with a poor prognosis of colorectal cancer ^[20][72]. With the advancement of detection technologies, noncoding RNA can be utilized as a reliable indicator to evaluate the prognosis of various cancer.

3.2. m6A Inhibitors as Potential Treatment and Diagnostic Target for Cancer

The complex interplay between lncRNA and cancer provides a promising idea for a therapeutic invention for cancer. Moreover, ncRNA may become novel noninvasive diagnosis biomarkers for early detection. Rhein is known as the first cell-active FTO inhibitor ^[50][102] and was demonstrated to inhibit the occurrence of breast cancer ^[51][103]. However, the selectivity of Rhein was not optimistic ^[52][104]. Meclofenamic acid (MA) is a more specific FTO inhibitor than ALKBH5 ^[53][54][105,106]. MA is able to restrict the growth and self-renewal of glioblastoma stem cells ^[55][107]. Furthermore, MO-I-500 was found to be a selective FTO inhibitor, which restrains the survival and colony formation of breast cancer cells ^[56][108]. R-2-hydroxyglutarate (R-2HG) is a metabolic product of mutant IDH1/2 which exert its function by inhibiting the m6A demethylase activity of FTO to suppress the progression of leukemic cells ^[57][109]. Moreover, FB23-2 is also an FTO inhibitor that significantly restrains the progression of AML cells ^[58][110]. In addition to FTO inhibitors, plenty of ALKBH5 inhibitors have been discovered. MV1035 is one of the inhibitors of ALKBH5 which suppress the migration and aggressiveness of glioblastoma ^[59][111]. As an inhibitor for METTL3, STM2457 is effective in the inhibition of AML ^[60][61][62][112,113,114].