The majority of human transcription products are non-coding RNAs (ncRNA) that are ubiquitously expressed in a broad spectrum of tissues (17). These RNAs engage in intricate gene expression processes, such as RNA splicing and protein translation, although they have little or no capacity for protein-coding. Recently, by virtue of the advancement in detection methods, numerous noncoding transcripts that were previously overlooked and merely regarded as intermediaries of protein synthesis, have been characterized to be critical for the posttranscriptional regulation of transcriptome expression (17). In general, ncRNAs mainly contain long noncoding RNA (lncRNA), microRNA (miRNA), circular RNA, rRNA, tRNA, and snRNA. Considering the vital role of the first three ncRNA listed above, this review describes their functions in multiple biological and pathological processes.

The dysregulation of lncRNAs has been proven to play a non-negligible role in tumorigenesis. In colorectal cancer (CRC) cells, increased expression of lncRNA RP11 in the nucleus and chromatin is mediated by m6A methylation. RP11 is capable of directly binding to the mRNA of Siah1 and Fbxo45 and then stimulates mRNA degradation, leading to the post-translational stabilization of ZEB1 (19). As revealed by Ni et al., m6A methylation participates in the regulation of YAP signaling in the progression of CRC. lncRNA GAS5 inhibits CRC progression via mediating the phosphorylation and degradation of YAP. However, the interaction of GAS5 and the m6A reader YTHDF3 results in the degradation of the former, contributing to the suppression of cell proliferation and metastasis abilities (28) ( Figure 2A ). It is of note that lncRNA can also interact with m6A writers such as METTL3. In gastric cancer, LINC00470 associates with METTL3 to weaken the stabilization of PTEN mRNA. YTHDF2 accounts for the detection of m6A sites within PTEN mRNA (38). In addition, lncRNA NEAT1 is demethylated by ALKBH5, consequently upregulates and prompts the invasion and metastasis of GC cells (29). In addition, In osteosarcoma, m6A demethylation on lncRNA PVT1 is mediated by ALKBH5 and associated with malignant properties (30).

Patil et al. demonstrated that the knockdown of METTL3 impaired the silencing of gene transcription in human cells. This process was modulated by lncRNA XIST (4). An enhanced abundance of m6A modification on XIST residues is mediated by RBM15/15B, which recruits the WTAP-METTL3 complex, and subsequently, YTHDC1 preferentially binds to m6A residues on XIST. Methyltransferase METTL16 can interact with lncRNA MALAT1, a cancer-related lncRNA (7, 48, 49). The crosslinking sites mainly occur at the 3’ UTR of MALAT1, wherein the triple helix element is usually recruited. m6A methylation disrupts the local RNA architecture of MALAT1 (50) and impedes the binding of RBPs to MALAT1. The aberrant expression of IGF2BP2 has been associated with insulin resistance, diabetes, and even neoplasia (51). IGF2BP2 interacted with lncRNA DANCR modified by m6A, resulting in the stability enhancement of DANCR. IGF2BP2 and DANCR jointly contribute to the proliferation and stemness-like properties of pancreatic cancer cells (31).

Based on genome-wide microarray analysis, a novel upregulated oncogenic lncRNA FAM225A was found to be significantly related to recurrence and distant metastasis in nasopharyngeal carcinogenesis (NPC) (32). Enhanced stabilization of FAM225A likely resulted from the enrichment of m6A modifications within its transcripts. Most FAM225A is located in the cytoplasm and absorbs miR-590-3p and miR-1275, resulting in the upregulation of their common target integrin β3(ITGB3). ITGB3 was shown to account for malignant phenotype progression in NPC cells (32). Yang et al. found that lncRNA 1281 could sequester pluripotency-related let-7 family miRNAs in mouse embryonic stem cells (mESC) (33). Sufficient m6A modification on lncRNA 1281 transcripts was necessary for this direct RNA-RNA interaction, which maintained mESC markers and proper differentiation ( Figure 2B and Table 1 ). Studies have revealed that lncRNA 00958 acts as an oncogenic gene in gliomagenesis (52), whereas its role in hepatocellular carcinoma (HCC) was not revealed until very recently. Zuo et al. disclosed that METTL3 induced the upregulation of lncRNA 00958, which exerted lipogenesis and the unfavorable survival of HCC patients (34). LncRNA 00958 could sponge miR-3619-5p whose target was hepatoma-derived growth factor (HDGF). The inhibitory effects of miR-3619-5p in the dissemination and invasion of HCC cells have also been described previously (53). In addition, m6A modification is also involved in nontumor pathologic or physiological processes, including heart and brain ischemia reperfusion injury (54). LncRNA MALAT1 targets miR-26b in the regulation of PTGS2, and has been reported to exacerbate inflammatory response in myocardial ischemia infarction patients, while m6A modification may also enhance MALAT1 expression (55). The lncRNA-BLACAT2 is a sponge of miR-193b-5p and has been associated with the progression of gastric cancer while silencing BLACAT2 inhibited cancer cells migration and invasion. A further study showed that METTL3 was a direct target of miR-193b-5p (39). Furthermore, similar to ceRNA, the lncRNA pseudogene Olfr29-ps1 sponged miR-214-3p and influenced IL6-mediated m6A modification, which jointly modulated the differentiation of myeloid-derived suppressor cells (40). Likewise, METTL3-mediated LINC00958 upregulation also played a similar ceRNA role over miR-378a-3p to promote YY1 expression in BC tumorigenesis (41). METTL3 induced LncRNA MEG3 stability and suppressed the progression of HCC by targeting miR-544b/BTG2 signaling (35). Furthermore, m6A methylation on lncRNA PCAT6 contributed to PCAT6 upregulation in an IGF2BP2‐dependent manner in prostate cancer (36). Yuan et al. also showed that METTL3 promoted osteogenic ossification through the upregulation of lncRNA XIST, and further investigation confirmed that lncRNA XIST regulated osteogenic differentiation of primary ligament fibroblasts via miR-302a-3p, which targets ubiquitin-specific protease 8 (USP8) (37).

A recent study showed that GAS5-AS1, the antisense RNA of GAS5, was downregulated in HCC (56) and non-small cell lung carcinoma (NSCLC) (57). In cervical cancer (CC) cells, the lncRNA GAS5-AS1 attenuates m6A modification of GAS5 through antisense pairing with the GAS5 3’ UTR, and thus, epigenetically enhanced GAS5 stability in a YTHDF2-dependent fashion. Whereas knockdown GAS5-AS1 resulted in larger tumors, enhanced metastasis, and advanced prognosis in CC patients (42). The lncRNA GAS5-AS1 also interacts with ALKBH5 to induce GAS5 upregulation ( Figure 2C and Table 1 ). Analogously, the lncRNA ARHGAP5-AS1 can recruit METTL3 to methylate ARHGAP5 mRNA in gastric cancer cells, thus, increased levels of m6A-ARHGAP5 were predictive of enhanced chemoresistance and poor prognosis (46). Zhang et al. revealed that FOXM1 maintained self-renewal and tumorigenic properties of glioblastoma stem-like cells (GSCs) (43). ALKBH5-mediated demethylation of nascent FOXM1 transcripts and upregulation of FOXM1 expression was facilitated by lncRNA FOXM1-AS. GSCs properties could be destroyed significantly when ALKBH5 or FOXM1-AS expression was blocked. He et al. determined that lncRNA KCNK15-AS1 was significantly decreased in cancer cells compared to normal pancreatic ductal epithelial cells (44) and its expression was correlated with the degree of m6A methylation. Simultaneously, ALKBH5 could demethylate KCNK15-AS1 and enhance the repression effects of KCNK15-AS1 on cancer cell viability. Furthermore, METTL3 was found to be responsible for the m6A methylation on the ABHD11-AS1 transcript and enhanced its stability in NSCLC tumorigenesis (45).

Recently, the lncRNA HOTAIR was reported to act as a plasma-derived biomarker of NSCLC, and m6A methylation was found to be co-expressed with HOTAIR (47). Moreover, many other lncRNAs, such as ANRIL, NEAT1, PVT1, TUG1, and DICER1-AS1, probably undergo m6A modification (58). These findings were conducive to improving understanding of the underlying mechanisms of human diseases and designing applicable therapeutic strategies.

Many studies have reported that miRNAs are abnormally expressed in different pathological processes, including diabetes, neurodegenerative diseases, and carcinomas. Pri-miRNA transcript can be methylated by METTL3 in the nucleus, and contribute to the recognition of pri-miRNAs by the microprocessor protein DGCR8 (59). Consistently, METTL3 deficiency can markedly decrease the expression of mature miRNAs. In bladder cancer, for example, upregulation of METTL3 accelerated the processing of pri-miR221/222, which promoted the proliferation, migration, and invasion of cancer cells in an m6A-dependent manner (60). Mature miR221/222 was also confirmed to have a carcinogenic role in other carcinomas (61), such as prostate cancer (62) and thyroid cancer (63). Analogously to its impact on bladder carcinoma, METTL3 facilitated the pri-miR-1246 maturation process and paved the way for the enhanced metastasis of CRC cells via the MAPK signaling pathway (64). Analogously, METTL3 upregulated miR-1246 expression and contributed to NSCLC cell growth (65). Likewise, miR-25-3p maturation was impeded by m6A modification mediated by METTL3 in pancreatic ductal adenocarcinoma (16). Furthermore, METTL3 drove the development of obstructive renal fibrosis by promoting miR-21-5p maturation (66). In vitro, the knockdown of METTL3 reduced the expression of miR-221-3p via m6A methylation of pri-miR-221-3p mRNA (67). It is of note that cigarette smoking has been shown to stimulate the METTL3 promoter resulting in its enhanced energetic transcription. Wang et al. provided evidence that METTL3-mediated miR-143-3p upregulation enhanced VASH1 repression, which thereafter triggered brain metastasis and the angiogenesis of lung cancer (68). Consistently, miR-873-5p maturation induced by exogenous METTL3 activity could protect mouse renal tubular epithelial cells (mRTECs) against colistin-induced nephrotoxicity (69). Interestingly, Yang et al. suggested that miR24-2 not only indirectly facilitated METTL3 transcription, but also enhanced miR6079 expression by promoting m6A methylation on pri-miR6079 in liver cancer cells (70).

Downregulation of the methyltransferase METTL14 was shown to be responsible for the aberrant m6A modification observed in HCC metastasis. miR126 is a downstream target of METTL14 and was markedly decreased in cancer tissues compared to the adjacent tissue (71). Consistently, Chen et al. revealed that downregulation METTL14 could induce pri-miRNA-375 processing arrest and reduced overall levels of miRNA-375 in CRC (72). Thus, miRNA-375 was revealed as an anti-oncogene regulated by METTL14 through the YAP/SP1 pathway and m6A methylation. miR-200a is also subject to METTL14 modulation ( Figure 3A and Table 2 ). As a reader of the m6A label in pri-miRNAs, HNRNPA2/B1 widely directed the maturation of pri-miRNA in LCC9 breast cancer cells (83). This processing was m6A dependent and promoted endocrine resistance in LCC9 cells, resulting in the poor survival of patients with advanced tumors.

The transcriptional regulator SRF is regulated epigenetically and responsible for cells contractility and proliferation (84). A recent study indicated that IGF2BP1 promotes the expression of SRF in a conserved and m6A-dependent manner, IGF2BP1 can impair the miRNA-directed decay of the SRF mRNA, resulting in enhanced SRF-dependent transcriptional activity (85). A sequence pairing mechanism contributes to miRNAs regulation of m6A methylation of mRNAs. Yang et al. reported that miRNA-145 expression might be considered a negative prognostic marker in liver cancer. The 3′UTR region of YTHDF2 mRNA contains direct binding sites for miR-145, and the overexpression of mir-145 increased m6A deposition and downregulated YTHDF2 expression ( Figure 3B and Table 2 ) (73).

In hemorrhoid vascular endothelial cells, a decrease in miR-4729 expression was confirmed to be responsible for vascular cell proliferation. miR-4729 was responsible for silencing METTL14 expression, reducing TIE1 mRNA stability, and inhibiting angiogenesis (74). Additionally, miR-103-3p directly targets METTL14 to inhibit osteoblast activity (75). METTL3 has also been reported to be a target of miR-186 in hepatoblastoma (76). Overexpression of miR-186 drastically ameliorated the metastatic phenotype induced by METTL3 upregulation. Moreover, as demonstrated in bladder cancer and CRC, the oncogenic role of METTL3 substantially relied on the degradation of tumor suppressor mRNAs targeted by miRNAs. Du et al. confirmed that in NSCLC cells, miR-33a bound to the 3’-UTR of METTL3 mRNA, and thus, resulted in a decrease in m6A deposition and proliferation of cancer cells accompanied by the downregulation of METTL3 (79). Moreover, in gastric cancer, miR-338-5p has also been reported to target METTL3 and repress the m6A-mediated translation of CDCP1 (77). Furthermore, Wei et al. indicated that miR-600 repressed METTL3 expression and eliminated the oncogenic activity induced by METTL3 on NSCLC progression (80). MiR-4443 directly targeted METTL3 and regulated the expression of FSP1 via m6A methylation (78). It has also been reported that WTAP downregulation in GSCs was caused by miR-29a upregulation, which could interfere with the malignant potential of GSCs (81). Cai et al. demonstrated that in breast cancer cells, HBXIP increased METTL3 expression by interfering with let-7g, a miRNA binding to the 3′UTR of METTL3, which subsequently interfered with METTL3 expression (82). Interestingly, METTL3 also upregulated HBXIP expression following the marked increase of overall m6A methylation, favoring the positive feedback loop HBXIP/let-7g/METTL3/HBXIP, which strongly promoted tumor cell growth and metastasis ( Figure 3C and Table 2 ) (82).

DDX3, a protein of the DEAD-box RNA helicases family, is involved in miRNAs demethylation due to its interaction with AGO2 protein and is involved in miRNA synthesis and function (86). Furthermore, m6A-AGO2 transcripts influence cellular miRNA levels and induced cell senescence (87). The role of miRNA in spermatogenesis was strongly evidenced in the mouse testis (88). ALKBH5 deficiency caused aberrant mammalian spermatogenesis or apoptosis because of the removal of the m6A modification on mRNAs (9) ( Figures 3B, C and Table 2 ).

In addition to mRNAs, pseudogenic RNAs, or lncRNAs, circRNAs have also be defined as ceRNA and compete with miRNAs or RBPs ( Figure 4A ) (104). These competing transcripts probably cross-regulate each other. In primary liver cancer, the highly expressed circRNA_104075 stimulated YAP expression both at the mRNA and protein levels (90). Further study has suggested that removal of the m6A modification on the 3′UTR region of YAP was essential for miR-582-3p binding to YAP. Furthermore, serum circ_104075 possessed high sensitivity and specificity for HCC diagnosis. circRNA-SORE sustains sorafenib resistance in HCC, and m6A modification increases the stability of circRNA-SORE (91). Yang et al. characterized eleven m6A containing circRNAs, among which seven circRNAs were significantly correlated with YTHDF2 (105). Furthermore, Chen et al. demonstrated that circFOREIGN can potentially activate gene expression of immunogenic factors (92). Nevertheless, YTHDF2 abrogated innate immunity by sequestering m6A-circRNA ( Figure 4B and Table 3 ). The study established a patient-derived xenograft (PDX) tumor model for CRC and found that circNSUN2 strongly interfered with liver metastasis of CRC (93). The exon5-exon4 junction site of circNSUN2 was identified as the m6A modification site. The binding of YTHDC1 to the m6A-modified circNSUN2 promoted a complex nuclear to cytoplasmic export analogous to mRNA trafficking ( Figure 4C and Table 3 ) (93). Interestingly, Wu et al. showed that METTL3 stabilized the expression of circCUX1 through the m6A methylation modification, which in turn, promoted radiotherapy resistance of hypopharyngeal cancer via caspase1 (94).

Some circRNAs carrying internal ribosome entry sites (IRESs) have the potential capacity of coding proteins (106). Granados-Riveron reported that a single m6A modification was sufficient to instigate the functional translation of circRNAs in the presence of eIF4G2 and YTHDF3 (107). Sun et al. found that circPVRL3 structure contained RRm6ACH (R = G or A; H = A, C or U), ORF, and IRES (95). Previous studies have documented that the m6A modification could regulate circRNA translation (108). Thus, it is reasonable to believe that circPVRL3 could translate into detectable peptides endogenously under certain cellular stresses. Zhao et al. reported that circE7 is found mainly in the cytoplasm and derives from oncogenic human papillomaviruses (HPVs) (96). CircE7 is highly m6A modified and translates to E7 oncoprotein following its association with polysomes.

circRNAs also play a vital role in regulating the expression of m6A-related proteins. It has been reported that circNDUFB2 is downregulated in NSCLC and inhibits the malignant progression of NSCLC (97). Mechanically, circNDUFB2 was found co-localized with IGF2BPs in the cytoplasm and physically interacted with IGF2BP1/2/3. Interestingly, overexpression of circNDUFB2 has no effect on the mRNA levels but protein levels of IGF2BPs. Further research found that circNDUFB2 reduces IGF2BPs stability via ubiquitin/proteasome-mediated degradation of IGF2BPs (97). circSTAG1 can bind with ALKBH5 in the astrocyte cytoplasm of the chronic unpredictable stress-treated mouse hippocampus, overexpressed circSTAG1 absorbs ALKBH5 and decreases the translocation of ALKBH5 into the nucleus, resulting in enhancement of m6A methylation on fatty acid amide hydrolase (FAAH) mRNA and subsequent degradation of FAAH (98). In addition, circMAPK4 acts as a ham-miR-139-5p sponge to regulate the expression and activity of YTHDF1 (99). circPTPRA can directly bind to the KH3 and KH4 domains of IGF2BP1, and block recognition of m6A-modified MYC and FSCN1 transcripts in BC cells (100).