Diverse molecular functions of m6A mRNA modification in cancer.

N6-methyladenosine (m6A), the most prevalent chemical modification found on eukaryotic mRNA, is associated with almost all stages of mRNA metabolism and influences various human diseases. Recent research has implicated the aberrant regulation of m6A mRNA modification in many human cancers. An increasing number of studies have revealed that dysregulation of m6A-containing gene expression via the abnormal expression of m6A methyltransferases, demethylases, or reader proteins is closely associated with tumorigenicity. Notably, the molecular functions and cellular consequences of m6A mRNA modification often show opposite results depending on the degree of m6A modification in specific mRNA. In this review, we highlight the current progress on the underlying mechanisms of m6A modification in mRNA metabolism, particularly the functions of m6A writers, erasers, and readers in the context of tumorigenesis.

Introduction

Since the discovery of the DNA double-helix structure in the 1950s, how genetic information is controlled and inherited has been a fundamental question. The discovery that alteration of the chromatin structure and DNA modifications affect heritable phenotypes in addition to the DNA sequence itself opened up a new field of epigenetics[1]. Similarly, many recent studies have proposed various chemical modifications of RNA as another layer of post-transcriptional gene expression regulation termed “epitranscriptomics”[2-4]. Post-transcriptional regulation is critical for the control of gene expression programs that dictate a variety of cellular functions and cell fate decisions. To date, at least 160 different chemical modifications have been identified in multiple RNA species, including messenger RNAs (mRNAs), transfer RNAs (tRNAs), ribosomal RNAs (rRNAs), noncoding RNAs (ncRNAs), and viral RNA genomes[4,5]. Although the majority of these modifications map to noncoding RNAs, increasing evidence implicates multiple mRNA modifications as components of another layer of gene expression regulation[2,6].

Discovered in the 1970s, N6-methyladenosine (m6A) is the best-characterized RNA modification and particularly is involved in almost all stages of the mRNA life cycle, including splicing, export, translation, and stability[7-11]. It is the most prevalent mRNA modification, with approximately one-fourth of the eukaryotic mRNAs harboring at least one m6A-modified base[3,12]. The m6A modification is found in multiple organisms and associated with various cell functions, including meiosis in yeast[13,14], plant development[15], mouse spermatogenesis[16], mouse embryogenesis[17], and various cancers[18-22].

In this review, we discuss the current understanding of m6A mRNA modification regulation at the molecular level and its various cellular effects. In particular, we highlight the emerging understanding of m6A mRNA modification in cancer.

Mechanism of dynamic m6A modification

The discovery of methyltransferases (also known as m6A writers) and demethylases (also known as m6A erasers) provided evidence that m6A modification is a dynamic and reversible event[23]. In addition to the combined action of m6A writers and erasers on m6A modification regulation, m6A reader proteins contribute to the regulation of the fate of m6A-containing RNAs. The m6A modification is the methylation of the sixth position of nitrogen atom of adenosine, with the cellular methyltransferase substrate S‑adenosylmethionine serving as the methyl donor for m6A formation[24,25]. Methyltransferase-like protein 3 (METTL3, also known as MT-A70) and METTL14 form a heterodimer at a ratio of 1:1, and they functions as a catalytic core complex recognizing the DRACH motif (D = A, G, or U; R = G or A; and H = A, C, or U) and inducing m6A modification of mRNA[12,24]. Growing evidence has revealed that METTL3 plays a central role in introducing m6A onto nascent transcripts cotranscriptionally, while METTL14 supports binding of the METTL3 protein to the target mRNA (Fig. 1)[26,27]. In addition, at least five other proteins are involved in the regulation of m6A mRNA modification, although it often shows a slightly different composition of the protein complex in each study. While they lack methyltransferase activity, they stabilize the METTL3/14 complex and facilitate its localization to the specific RNA sites for m6A modification[28,29]. Wilms tumor 1-associated protein [WTAP, also known as female-lethal(2)d] recruits other proteins to the METTL3/14 complex, thereby affecting the overall levels of m6A modification[30]. RNA-binding motif 15 (RBM15) protein and its paralog RBM15B have been shown to interact with METTL3 in a WTAP-dependent manner[28,31]. It has been suggested that they preferentially bind to U-rich regions in RNA and recruit the METTL3/14-WTAP complex to sites proximal to the m6A consensus motifs[31]. Vir-like m6A methyltransferase associated protein (VIRMA, also known as Virilizer or KIAA1429) was recently found to mediate mRNA methylation near the stop codon in the 3′ untranslated region (UTR), where it plays a role in alternative polyadenylation[32]. In mouse embryonic stem cells (mESCs), Cbl proto-oncogene like 1 (CBLL1, also known as Hakai) protein and zinc finger CCCH-type containing 13 (ZC3H13) proteins have been shown to be required for the nuclear localization of ZC3H13-WTAP-VIRMA-CBLL1, which promotes m6A mRNA modification[29]. In Drosophila, ZC3H13 has also been shown to act as an adapter protein between WTAP and RBM15 in the methyltransferase complex to support efficient methylation[28]. On the other hand, methyltransferase-like protein 16 (METTL16) was recently found to be critical for m6A modification in several pre-mRNAs, U6 small nuclear RNAs (U6 snRNAs), and noncoding RNAs containing a specific stem-loop structure[33-35]. Interestingly, METTL16 has been shown to control S-adenosylmethionine levels by regulating the expression of a S-adenosylmethionine synthetase methionine adenosyltransferase 2 A (MAT2A) by the enhanced splicing of a retained intron[33,34]. When METTL16 is depleted, the level of m6A in a cell decreases by ~20%[33].

Fig. 1

An overview of cotranscriptional m6A mRNA modification.

Introduction of m6A modification is currently suggested to occur cotranscriptionally in the nucleus. Individually transcribing mRNAs illustrate the different modes of cotranscriptional m6A modification, including different compositions of associated DNA- and RNA-binding proteins with distinct methylation sites. The thick line represents the coding sequence, and the thin line represents the UTR. The dashed box indicates the heat shock condition.

To date, two mammalian m6A demethylating enzymes have been identified, namely, the fat mass and obesity-associated protein (FTO) and a-ketoglutarate-dependent dioxygenase alk B homolog 5 (ALKBH5) protein[36,37]. FTO was the first identified m6A demethylase originally found to be associated with increased body mass and obesity in humans[36,38]. Demethylation of m6A by FTO generates an intermediate product, N6-hydroxymethyladenosine (hm6A), which is then further oxidized to N6-formyladenosine (f6A)[39]. However, the potential functions of these intermediate products remain unclear. While several studies have provided evidence that depletion of FTO increases the level of total m6A, another recent report suggested FTO preferentially demethylates 2′-O-dimethyladenosine (m6Am), which is found adjacent to the 7-methylguanosine (m7G) cap in mRNA, thereby influencing mRNA stability[40]. Most recently, FTO was also shown to demethylate N1-methyladenosine (m1A) in tRNAs[41]. ALKBH5, the second identified m6A demethylase, preferentially recognizes the m6A mark for demethylation in a consensus sequence-dependent manner; thus, it is considered as a better candidate for global m6A demethylation[37].

Molecular functions of m6A in mRNA metabolism

Gene expression is the result of orchestrated transcriptional and post-transcriptional regulation. Recently, an increasing number of studies have suggested m6A mRNA modification as a layer of gene expression regulation previously unrecognized. Various m6A reader proteins are involved in many processes of overall mRNA metabolism (Fig. 2).

Fig. 2

Molecular details for m6A-mediated mRNA metabolism.

Multiple m6A reader proteins dynamically regulate m6A-containing mRNA metabolism, including alternative splicing, mRNA export, structural switch, translation, and mRNA stability, depending on the specific m6A-bound reader protein. The thick line represents the coding sequence, and the thin line represents the UTR. The dashed box indicates the heat shock condition.

Cotranscriptional m6A modification

In general, m6A modifications of mRNAs are enriched near translation stop codons in the 3′ UTR[3,12,42]. However, this characteristic varies among different mRNAs and depends on the tissue. There are several lines of evidence indicating that the m6A modification is a cotranscriptional event (Fig. 1)[18,26,27]. One report showed that METTL3 binds to chromatin in a transcription-dependent manner and cotranscriptionally methylates nascent transcripts[26]. In a case of acute myeloid leukemia (AML), METTL3 can be recruited to the promoter region independent of METTL14 by binding to CCAAT/enhancer-binding protein zeta (CEBPZ)[18]. METTL3 can induce m6A modification cotranscriptionally within the coding region of the associated transcripts, ultimately resulting in translation enhancement[18]. Moreover, it has been shown that cotranscriptional modification of m6A is dependent on the activity of RNA polymerase II (RNAP II)[27]. A low rate of transcriptional activity induces increased levels of m6A modification throughout the gene body, resulting in reduced levels of translation[27]. On the other hand, in the case of heat shock stress, METTL3 can be recruited with DGCR8 to the chromatin of heat shock responsive genes in the region of the transcription ending site, where it subsequently methylates nascent mRNAs, leading to the degradation of the target mRNAs as a consequence[26]. Considering the accumulating evidence that the m6A modification is mainly found around the translation stop codon in mRNAs[3,12,42] and that VIRMA preferentially mediates mRNA methylation near the stop codon in the 3′ UTR[32], further studies are required to clarify such discrepancies in the methylation mechanism. Moreover, despite consistent results showing that m6A modification is a cotranscriptional event, the molecular consequences of this modification vary among different studies. Therefore, further research is required to determine the regulating factors that lead to these discrepancies.

m6A promotes alternative splicing

Multiple model organism studies have shown that dynamic m6A modification alters mRNA splicing. In Drosophila, mutation of IME4 (a METTL3 homolog) influences sex determination by modulating female-specific splicing of the Sex-lethal (Sxl) gene[43,44]. In addition, the Drosophila orthologs of VIRMA and/or ZC3H13 have been shown to regulate alternative splicing of pre-mRNAs involved in sex determination[28]. m6A demethylases were also reported to be involved in splicing machinery[37,45,46]. FTO regulates mouse pre-adipocyte differentiation by regulating the alternative splicing of the genes involved in adipogenesis[45]. ALKBH5 regulates splicing by removing m6A from pre-mRNAs and allows the production of a subset of mRNAs containing relatively long 3′ UTRs in mouse germ cells[37,46]. While m6A writers and erasers regulate alternative splicing by modulating the levels of m6A modification, m6A reader proteins directly regulate splicing[8,47]. The m6A-bound YTHDC1 associating with splicing factor SRSF3 has been shown to block the binding of SRSF10 to m6A-modified RNA, promoting exon inclusion in the selected transcripts[8,48]. Moreover, m6A modification influences mRNA structural changes, which allows heterogeneous nuclear ribonucleoprotein C (hnRNPC) and hnRNPG binding[9,47]. While hnRNPC binds opposite strand U-rich sequences after the disruption of RNA base pairing by m6A modification[9], hnRNPG preferentially binds to purine-rich motifs, including m6A sites[47]. Binding of either hnRNPC or hnRNPG influences the alternative splicing of m6A-modified transcripts[9,47]. Finally, METTL16 induces the m6A modification of U6 snRNA, which base pairs with 5′ splice sites of pre-mRNAs during splicing, suggesting that METTL16 plays an important role in mRNA splicing[34,35].

m6A facilitates mRNA export

mRNA export is also influenced by m6A modification. ALKBH5-deficient cells exhibit increased levels of cytoplasmic m6A-containing mRNA, suggesting that the m6A modification accelerates mRNA export[37]. Another report showed that YTHDC1 facilitates the export of m6A-modified mRNAs via its interaction with nuclear RNA export factor 1 (NXF1)[10].

m6A alters RNA structure

It has been well established that gene expression is largely affected by the secondary and tertiary structures of mRNA[49]. Introduction of m6A modification promotes the destabilization of A/U pairings, resulting in alterations to the thermostability of RNA duplexes and changes in the RNA secondary structure[50]. Another study demonstrated that RNA structural changes caused by the introduction of m6A also alter the interaction between RNAs and proteins[9,47].

m6A regulates translation efficiency

Many m6A reader proteins are reported to be crucial for the efficient translation of methylated mRNAs. Members of the YT521-B homology (YTH) domain-containing protein family have been identified as direct m6A readers, including YTHDF1, YTHDF2, YTHDF3, YTHDC1, and YTHDC2[7,11,51-56]. Among these proteins, YTHDF1, YTHDF3, and YTHDC2 have been shown to promote target mRNA translation[11,51-53]. YTHDF1 selectively binds to m6A sites near the stop codon and cooperates with translation initiation factors to promote the translation of the target mRNAs[51]. YTHDF3 cooperates with YTHDF1 in the regulation of translation by interacting with a common set of ribosomal proteins[52]. YTHDC2 has been suggested to play a role in enhanced translation levels while reducing target mRNA abundance[53]. Furthermore, increased levels of YTHDF2 translocate to the nucleus under heat shock stress and bind m6A in the 5′ UTR of a subset of stress-induced mRNAs, protecting them from FTO-mediated demethylation and promoting their cap-independent translation[57]. Eukaryotic translation initiation factor 3 (eIF3) is also considered an m6A-binding protein. mRNAs containing m6A modification in the 5′ UTR can be recognized by direct binding of eIF3 to the methylated region, which in turn recruits the 43 S complex to initiate translation in a cap-independent manner in the absence of the cap-binding protein eIF4E[58]. However, the mechanism of eIF3 in the recognition of m6A is not yet clearly understood. Interestingly, most recent studies have suggested that the m6A writer protein METTL3 also functions as a reader protein in the cytoplasm, promoting the translation of a large subset of target mRNAs[21,22]. These studies revealed that 3′ UTR m6A modification near the stop codon significantly increases translation through mRNA looping, governed by the interaction between METTL3 at the 3′ UTR and the translation initiation factor eIF3 subunit h (eIF3h) at the 5′ end[21,22].

m6A regulates mRNA stability

An increasing number of studies have demonstrated that m6A modification influences mRNA stability. Various structural and functional studies suggest that all three YTHDF reader proteins (YTHDF1, YTHDF2, and YTHDF3) may share the same subset of target mRNAs[51,52]. However, accumulating evidence suggests that YTHDF2 is the major factor involved in the degradation of m6A-containing mRNA either through exoribonucleolytic decay or the endoribonucleolytic cleavage pathway[55,56]. YTHDF2 has been shown to selectively recognize m6A sites and recruit the CCR4-NOT deadenylase complex directly, which in turn recruits exosomes (3′-to-5′ exoribonuclease) to initiate mRNA decay[56]. Other recent studies revealed that YTHDF2 promotes the translocation of m6A-containing mRNA from the translation machinery to processing bodies (P bodies), where cellular proteins participating in mRNA degradation are enriched[7,59]. In addition, a very recent study revealed the YTHDF2-mediated endoribonucleolytic cleavage of m6A-containing mRNAs[55]. Mechanistically, heat-responsive protein 12 (HRSP12, also known as reactive intermediate imine deaminase A homolog, UK114 antigen homolog, and 14.5 kDa translational inhibitor protein) bridges m6A-bound YTHDF2 to an endoribonuclease, RNase P/MRP, triggering the endoribonucleolytic cleavage of an m6A-containing mRNA[55]. Another study suggested that YTHDC2 recruits the 5′ to 3′ exoribonuclease XRN1 for subsequent m6A-containing mRNA degradation[54]. In addition to the YTH proteins, a variety of other RNA-binding proteins are involved in the regulation of m6A-containing mRNA stability. Fragile X mental retardation protein (FMRP) can bind to the sequence motifs YGGA (Y = C or U) and GAC, which likely overlap with the DRACH motif involved in m6A modification, resulting in stabilization of the m6A-containing mRNA through the competition of FMRP with YTHDF2[60]. In another case, stress granule protein (G3BP1) has binding affinity for m6A-methylated transcripts, promoting their demethylation and resulting in stabilization of the target mRNAs[61]. Insulin-like growth factor 2 mRNA-binding protein (IGF2BP) 1, 2, and 3 or human antigen R (HuR, also known as ELAVL1) have also been reported to stabilize m6A-containing mRNAs[62,63].

Molecular functions of m6A in various cancers

Interest in m6A modification has been extended to many human diseases as well as to its molecular function. In particular, an increasing number of studies are examining the role of m6A-mediated gene expression regulation in cancers. In general, many different signaling pathways converge onto translation machinery to satisfy the increased anabolic demands of cancers. Given the crucial function of m6A modification in regulating mRNA metabolism, it is reasonable to speculate that m6A modification plays an important role in human carcinogenesis. Nonetheless, the molecular details of how m6A modification affects the cellular phenotype of cancer are still being investigated. The physiological effects of m6A mRNA modification in cancer often lead to opposite results (Table 1); thus, further understanding of a balanced m6A modification is required for the treatment of cancer. Here, we highlight recent insights into the biological functions of m6A mRNA modification and the underlying molecular mechanisms of m6A regulatory proteins in various cancers (Fig. 3 and Table 1).

Table 1

Cellular effects of m6A mRNA modification in cancer.

	Positive regulation of m6A in cancer			Negative regulation of m6A in cancer
	Molecular function	Target	Reference	Molecular function	Target	Reference
Lung cancer	Translation	A subset of mRNAs	[21,22]	mRNA level change	A subset of mRNAs	[68]
		EGFR, TAZ, MAPKAPK2, DNMT3A mRNA	[67]
	Protein level change	BAX, BCL-2	[66]	mRNA stabilization	MZF1 mRNA	[69]
Acute myeloid leukemia	Translation	A subset of mRNAs	[18]	mRNA stabilization	ASB2, RARA mRNA	[19]
		MYC, BCL2, PTEN mRNA	[71]
	Translation	MYB, MYC mRNA	[73]	mRNA decay	A subset of mRNAs	[72]
	mRNA stabilization		[73]
Hepatocellular carcinoma	mRNA level change	SOCS2 mRNA	[75]	NR	NR
	mRNA decay	EGFR mRNA	[76]
	mRNA level change	SON, SREBBP mRNA	[77]
Breast cancer	mRNA level change	HBXIP mRNA	[79]	mRNA level change	NANOG, KFL4 mRNA	[80,81]
Gastric cancer	mRNA stabilization	SEC62 mRNA	[83]	Unknown	Unknown	[85,86]
		HDGF mRNA	[84]
Bladder cancer	mRNA level change	AFF4, MYC mRNA	[20]	NR	NR
	Protein level change	AFF4, MYC, IKBKB, RELA	[20]
	Translation	ITGA6 mRNA	[87]
Glioblastoma	mRNA stabilization	SOX2 mRNA	[92]	mRNA level change	Nascent FOXM1 transcript	[89]
				mRNA level change	A subset of mRNAs	[91]
Colorectal cancer	mRNA stabilization	SOX2 mRNA	[93]	NR	NR
Renal cell carcinoma	NR	NR		Unknown	Unknown	[95]
Endometerial Cancer	NR	NR		Translation	PHLPP2 mRNA	[96]
				mRNA decay	PRR5, PRR5L, mTOR mRNA	[96]
Cervical cancer	NR	NR		Unknown	Unknown	[97]
Pancreatic cancer	NR	NR		Protein level change	YAP	[98]

Some studies did not identify molecular mechanisms or targets, but only measured m6A levels and their effects on cancer, which are marked as “unknown”. “Translation” indicates the m6A-mediated translation enhancement. “Protein level change” and “mRNA level change” indicate their steady-state levels without specifying the translation efficiency or mRNA stability.

NR not reported.

Fig. 3

m6A-mediated mRNA regulation in tumorigenesis.

A number of studies have identified the molecular mechanism of m6A-mediated mRNA regulation and their effects on tumorigenesis. To date, m6A-mediated regulation of translation or mRNA stability has been demonstrated, while the relevance of pre-mRNA splicing or mRNA export remains unclear for specific cancer types.

Lung cancer

Lung cancer causes the greatest number of cancer-related deaths worldwide. There are two main histological types of lung cancer: small-cell lung cancer and non-small-cell lung cancer (NSCLC). Approximately 85% are classified as NSCLC, which statistically shows just a 15.9% 5-year survival rate[64,65]. Nevertheless, therapeutic efforts have improved only slightly over the last few decades. Therefore, it is urgent to explore new treatments and deepen our understanding of the underlying mechanisms of lung cancer occurrence and development. The relevance of m6A modification in lung cancer has been extensively studied, and several lines of evidence show that METTL3 is highly expressed in NSCLC cells and is associated with cell proliferation, invasion, and viability[21,22,66-68]. Two recent studies from the same group revealed intriguing effects of METTL3 in lung cancer progression. These studies showed that cytoplasm-localized METTL3 functions as an m6A reader protein that enhances translation of a large subset of oncogenic mRNAs without affecting mRNA abundance[21,22]. Mechanistically, the 3′ UTR near the stop codon-bound METTL3 directly interacts with eIF3h. This interaction mediates mRNA looping to facilitate the recycling of ribosomes at the termination codon in a similar way to canonical eukaryotic mRNA looping mediated by the interactions between eIF4E (a cap-binding protein), eIF4G (a translation initiation factor), and PABP (a poly(A)-binding protein)[21,22]. Indeed, ectopic expression of METTL3, but not a mutant that fails to interact with eIF3h, promotes cell proliferation, invasion, and oncogenic transformation[21]. Other studies have shown that METTL3 mRNA can be targeted by microRNAs (miRNAs)[66,67]. Exogenously expressed miR-600 targets the 3′ UTR of METTL3 mRNA, resulting in the inhibition of METTL3 expression[66]. Depletion of METTL3 inhibits the survival and proliferation of A549 and H1299 cells and leads to increased levels of the pro-apoptotic regulator BAX and decreased levels of the anti-apoptotic regulator BCL-2, suggesting that the altered expression ratio of BAX/BCL-2 triggers the mitochondrial apoptotic pathway[66]. In addition, knocking down METTL3 decreases the phosphorylation of AKT, thus affecting cell growth and apoptosis via the alteration of the PI3K/AKT/mTOR pathway[66]. Another miRNA, miR-33a, has also been shown to reduce METTL3 expression and, as a result, inhibits NSCLC cell proliferation[67]. On the other hand, METTL3 is SUMOylated by small ubiquitin-related modifier 1 (SUMO1), which modifies METTL3 at lysine residues and represses its methyltransferase activity without altering its stability, localization, or interaction with two other writer proteins, METTL14 and WTAP[68]. The SUMOylation of METTL3 reduces m6A levels and subsequently changes the mRNA expression profiles, ultimately promoting the development of NSCLC[68]. Besides, the m6A demethylase FTO has also been shown to play a critical role in lung squamous cell carcinoma (LUSC), one of the most common NSCLCs. FTO knockdown effectively inhibits cell proliferation and invasion while promoting apoptosis of L78 and NCI-H520 cells[69]. In contrast, overexpression of FTO, but not its mutant form, facilitates the acquisition of malignant phenotypes[69]. Mechanistically, FTO increases the stability of myeloid zinc finger 1 (MZF1) mRNA by reducing its m6A level, leading to high levels of protein expression, which has an oncogenic function[69]. MZF1 is a member of the SCAN-zinc finger transcription factor family, which contributes to cell proliferation, migration, and metastasis through the regulation of diverse target genes.

Acute myeloid leukemia (AML)

AML is one of the most prevalent hematopoietic malignancies. It is often derived from genetic mutations and aberrant regulation of epigenetic modification, including DNA methylation and histone modification[70]. Recently, many studies have pointed to m6A mRNA modification as a new role for a gene expression regulator associated with AML[18,19,71,72]. As previously described, promoter-bound METTL3 induces m6A modification within coding regions of a subset of nascent transcripts independent of METTL14[18]. In this way, the genes necessary for AML growth enhance their translation efficiency by relieving ribosome stalling at the GAN (GAG, GAT, GAC, and GAA) codons during translation elongation[18]. Another study revealed that increased expression levels of METTL3 promote the translation of MYC proto-oncogene (c-MYC), B-cell lymphoma 2 (BCL2), and phosphatase and tensin homolog (PTEN) mRNAs by increasing the levels of m6A modification, thereby altering phosphoinositide 3-kinase (PI3K) and protein kinase B (PKB, also known as AKT) signaling, an intracellular signaling pathway important in regulating the cell cycle, to control cell differentiation and self-renewal[71]. METTL14 was also shown to function in a similar way by promoting translation of its target mRNAs, the proto-oncogenes MYB and MYC, through m6A modifications, which in turn leads to block the myeloid differentiation[73]. Notably, in addition to m6A writers, differentially expressed eraser or reader proteins seem to contribute to various AML subtypes through the modulation of m6A modification in a target mRNA-specific manner. Elevated expression of FTO enhances cell transformation and leukemogenesis by downregulating both the mRNA and protein expression of targets, such as ASB2 and RARA mRNAs, by reducing the m6A levels in their UTRs[19]. On the other hand, YTHDF2 overexpression plays a crucial role in disease initiation and propagation in human and mouse AML by destabilizing a subset of mRNAs, including tumor necrosis factor receptor TNFRSF2 mRNA[72].

Hepatocellular carcinoma (HCC)

HCC is a major type of primary liver cancer and is a highly progressive malignant tumor associated with a low survival rate[74]. It was recently reported that METTL3 levels are increased in human HCC, leading to increased m6A modification of the tumor suppressor SOCS2 mRNA[75]. Increased levels of m6A in SOCS2 mRNA can be targeted by YTHDF2, leading to its rapid degradation, which is associated with the efficient proliferation of HCC cells[75]. Besides, overexpression of YTHDF2 has been shown to suppress cell proliferation and tumor growth in HCC cells[76]. Mechanistically, the m6A-modified 3′ UTR of epidermal growth factor receptor (EGFR) mRNA is recognized by YTHDF2 and undergoes degradation, which in turn impairs mitogen-activated protein kinase kinase (MEK) and extracellular signal-regulated kinases (ERK)[76]. Similarly, another report showed that YTHDF2 mRNA can be targeted by miR145, leading to an increase in overall m6A levels in HCC cells, which is associated with HCC malignancy[77].

Breast cancer (BrC)

Of all malignant tumors in women, BrC is highly metastatic and has the highest cancer-related mortality[78]. One interesting report suggested a potential positive feedback loop between mammalian hepatitis B X-interacting protein (HBXIP) and METTL3[79]. High expression levels of HBXIP elevate METTL3 expression through the suppression of let-7g, and increased METTL3 upregulates HBXIP expression through m6A modifications of mRNA. This positive feedback loop leads to the acceleration of cell proliferation in BrC. On the other hand, a decrease in m6A modification also promotes BrC tumorigenesis. In BrC stem cells, hypoxic stress induces overexpression of ALKBH5 and/or ZNF217, leading to inhibition of the methylation of pluripotency markers NANOG and KLF4 mRNAs[80,81]. Increasing the expression of NANOG and KLF4 mRNA by inhibiting m6A modification promotes the specification of BrC stem cells[80,81]. Another report also showed that m6A levels increased by METTL14 overexpression or ALKBH5 knockdown inhibited BrC growth and metastasis[82].

Gastric cancer (GC)

GC is a prevalent tumor occurring in the digestive system. One clear mechanism showed that the preprotein translocation factor SEC62 mRNA can undergo m6A modification by METTL3[83]. In turn, IGF2BP1 recognizes m6A and facilitates the stabilization of SEC62 mRNA. Moreover, miR4429 has been suggested to target METTL3 and prevent the m6A modification of SEC62 mRNA, thus destabilizing SEC62 mRNA[83]. Downregulated SEC62 inhibits GC cell proliferation and promotes apoptosis[83]. Another report showed that METTL3 transcription is elevated in GC by a histone acetyltransferase, P300, which mediates H3K27 acetylation at the METTL3 promoter region, which in turn induces the methylation of hepatoma-derived growth factor (HDGF) mRNA[84]. The methylated HDGF mRNA is then recognized and stabilized by IGF2BP3. Overexpressed HDGF protein can be secreted and promotes tumor angiogenesis, while nuclear HDGF stimulates the expression of glucose transporter type 4 (GLUT-4) and enolase 2 (ENO2) mRNAs, resulting in increased levels of glycolysis and subsequently causing tumor growth and liver metastasis[84]. On the other hand, it has been suggested that FTO and ALKBH1 play crucial roles in GC progression and metastasis, although the relevance of m6A in these processes is unclear[85]. It has been shown statistically that lower ALKBH1 protein expression correlates with larger tumor size, while lower FTO protein expression is associated with shorter overall survival in patients with GC[85]. Another report revealed that the downregulation of m6A modification by METTL14 knockdown leads to the acquisition of oncogenic phenotypes through the alteration of Wnt and PI3K-AKT signaling pathways, although the exact upstream regulatory mechanism is unclear[86].

Bladder cancer (BlC)

BlC is the most prevalent urogenital cancer. Recent studies suggest that increased levels of m6A modification are correlated with BlC[20,87]. One study identified the mRNAs of AF4/FMR2 family member 4 (AFF4), two key regulators of the NF-κB pathway (IKBKB and RELA), and MYC as direct METTL3 targets for m6A modification[20]. METTL3 depletion led to a reduction in AFF4 and MYC mRNA and protein expression, while only the protein expression was reduced for IKBKB and RELA. METTL3 downregulation in BlC drastically reduced cell proliferation, invasion, and survival in vitro and tumorigenicity in vivo[20]. Considering the results indicating that (1) MYC is a well-known oncogene that triggers the expression of target genes to benefit cell proliferation, cell survival, and stemness maintenance and (2) AFF4 and NF-κB are known to regulate MYC expression, through which NF-κB signaling enhances the proliferation and survival of cancer cells during the development and recurrence of BlC, it can be speculated that m6A modification by METTL3 affects the AFF4/NF-κB/MYC signaling network to regulate BlC progression[20]. In addition, upregulated METTL3 promotes the translation of integrin alpha-6 (ITGA6) mRNA via the recognition of m6A in the 3′ UTR by the m6A reader proteins YTHDF1 and YTHDF3[87]. As a result, the upregulated ITGA6 protein promotes BlC cell adhesion, migration, and invasion, similar to multiple other types of cancer, in which ITGA6 overexpression promotes tumorigenesis and metastasis[87].

Glioblastoma (GBM)

GBM is a primary malignant brain tumor prevalent in adults[88]. GBMs have heterogeneous characteristics and contain cells with stem-like properties[89]. These self-renewing GBM stem-like cells (GSCs) contribute to tumor initiation and therapeutic resistance[90]. Intriguingly, the expression levels of both METTL3 and ALKBH5 are elevated in GSCs, with opposite results on m6A-mediated tumor formation in a target-specific manner[89,91,92]. High METTL3 expression levels exhibit oncogenic function through efficient m6A modification in the 3′ UTR of sex-determining region Y (SRY)-box 2 (SOX2) mRNA, which is stabilized by binding of HuR[92]. Silencing METTL3 expression reduces SOX2 expression and, as a result, inhibits GBM tumor growth and prolongs the survival of mice[92]. In contrast, ALKBH5 is highly expressed in GSCs and demethylates FOXM1 nascent transcripts, leading to FOXM1 overexpression, stem-like cell proliferation, and tumorigenesis[89]. The elevated levels of the transcription factor FOXM1 play critical roles in regulating GSC proliferation, self-renewal, and tumorigenicity[89]. Similarly, another study suggested a tumor-suppressive function for the m6A modification in GSCs[91]. Reduction of m6A modification by the depletion of METTL3 or METTL14 or the chemical inhibition of FTO upregulates the mRNA expression of critical oncogenes such as ADAM19, EPHA3, and KLF4 and downregulates the mRNA expression of many tumor suppressors, including CDKN2A, BRCA2, and TP53I11 mRNAs, resulting in overall enhanced GBM stem cell growth, self-renewal, and tumorigenesis[91].

Colorectal cancer (CrC)

In CrC, METTL3 and YTHDF1 expression is significantly upregulated[93,94]. High levels of METTL3 expression have been shown to significantly upregulate m6A methylation in the coding sequences of SOX2 mRNA, a well-known CrC marker that is involved in maintaining the properties of tumor-initiating cells[93]. Methylated SOX2 mRNA is subsequently recognized by IGF2BP2, preventing mRNA degradation. Indeed, knocking down METTL3 reduces the SOX2 expression level, inhibiting CrC development and metastasis[93]. On the other hand, c-MYC has been suggested to promote YTHDF1 transcription[94]. A statistical analysis suggests that patients with high YTHDF1 expression have significantly poorer overall survival[94]. Moreover, knocking down YTHDF1 results in the inhibition of cell proliferation and sensitization of cells to anticancer drugs such as fluorouracil and oxaliplatin[94].

Other cancers

Similar to the cancers discussed above, modulation of m6A modification plays a critical role in renal cell carcinoma, endometrial cancer, and cervical cancer[95-97]. In renal cell carcinoma, depletion of METTL3 promotes cell proliferation, cell invasion, and migration, and induces G0/G1 arrest[95]. Conversely, upregulation of METTL3 results in significant suppression of tumor growth[95]. Moreover, knocking down METTL3 promotes the acquisition of an epithelial phenotype and represses the manifestation of a mesenchymal phenotype, while overexpression of METTL3 reverses epithelial–mesenchymal transition progression[95]. Furthermore, the observation that increased phosphorylation levels of PI3K/AKT/mTOR due to METTL3 knockdown suggests that these METTL3-mediated pathways may also be involved in renal cell carcinoma progression[95]. A report revealed that METTL14 is frequently mutated and METTL3 expression is significantly reduced in endometrial cancer[96]. Mechanistically, m6A mRNA modification affects the YTHDF1-dependent translation enhancement of the negative AKT regulator PHLPP2 and YTHDF2-dependent destabilization of the mRNAs of positive AKT regulators PRR5, PRR5L, and mTOR. Thus, either METTL14 mutation or decreased METTL3 expression leads to m6A reduction in these target mRNAs and, as a result, promotes cell proliferation and tumorigenicity of endometrial cancer through AKT activation[96]. In cervical cancer, downregulation of m6A modification enhances cell proliferation, while upregulation inhibits tumor development[97]. However, the exact mechanism remains unknown. Last, YTHDF2 is upregulated in pancreatic cancer and has two roles in cancer development: 1) YTHDF2 promotes cell proliferation, since it was observed that knocking down YTHDF2 results in the activation of the AKT/GSK3β/Cyclin D1 pathway, leading to G1 arrest, and 2) the YTHDF2-mediated decay of yes-associated protein (YAP) may influence the epithelial–mesenchymal transition, since overexpression of YAP results in decreased expression of epithelial markers and increases in mesenchymal markers[98].

Concluding remarks and future perspectives

Considering the increasing number of studies revealing that m6A modification plays a critical role in almost all stages of mRNA metabolism[10,48,56,62], we can easily speculate that aberrant regulation of these modifications affects many cellular phenotypes. Nevertheless, the molecular mechanisms and cellular effects of m6A mRNA modifications are not yet fully understood, since they do not always function in the same way. For instance, although it is well known that the m6A modification sites in mRNAs are mainly enriched in the 3′ UTR near the stop codon[3,12,42], several recent findings showed that cotranscriptional methylation occurs in coding sequences (Fig. 1)[18,27]. In addition, it is still unclear why some mRNAs are not methylated. Considering that the m6A modification is reversible, the demethylases FTO and/or ALKBH5 may play critical roles in balancing the methylation of specific mRNAs in a cell type-dependent manner.

In recent years, m6A modification studies in various cancers have been conducted. Remarkably, an increasing number of studies have revealed that altered expression levels of m6A methyltransferases, demethylases, and reader proteins aberrantly regulate m6A modification on target mRNAs, resulting in abnormal expression of cancer-associated genes. In particular, increased methyltransferase expression levels were detected in most cancers, suggesting that higher m6A modification levels are closely related to tumorigenesis. However, the molecular functions and cellular consequences of m6A modification differed in each study, depending on the degree of methylation in the specific target mRNAs (Table 1). For instance, increased levels of m6A modification by higher levels of METTL3 or METTL14 expression promoted the translation or stabilization of c-MYC, BCL2, PTEN, or MYB mRNAs in AML[71,73]. In contrast, FTO also showed an elevated level of expression, which downregulated both the translation and abundance of ASB2 and RARA mRNAs through demethylation[19]. Taken together, the coordinated functions of methylation and demethylation of specific targets seem to be critical for tumorigenesis.

Interest in m6A modification resurged quite recently. To date, most of the m6A studies in cancer have been demonstrated based on the discovery of the m6A modification itself rather than the underlying mechanisms with reader proteins (Fig. 3 and Table 1) because efforts to define the molecular mechanism and the biological relevance have been carried out in parallel. To date, only a single m6A reader-dependent molecular mechanism has been demonstrated in most cancer types (Fig. 3). In addition, cancer-related studies on other outcomes of m6A-dependent mRNA regulation, such as pre-mRNA splicing or mRNA export, remain insufficient. Considering that multiple reader proteins recognize m6A, it might be possible to crosstalk between readers on a single or a multiple m6A modification in an mRNA for the tight gene expression regulation. Therefore, to develop novel tumor therapies based on the regulation of m6A modifications, more thorough mechanistic and functional studies are required for each cancer type.