N6-methyladenosine (m6A) as a regulator of carcinogenesis and drug resistance by targeting epithelial-mesenchymal transition and cancer stem cells

Abstract

Emergence of drug resistance to chemotherapeutic agents is the principal obstacle towards curative cancer treatment in human cancer patients. It is in an urgent to explore the underlying molecular mechanisms to overcome the drug resistance. N6-Methyladenosine (m6A) RNA modification is the most abundant reversible RNA modification and has emerged in recent years to regulate gene expression in eukaryotes. Recent evidence has identified m6A is associated with cancer pathogenesis and drug resistance, contributing to the self-renewal and differentiation of cancer stem cell, tumor epithelial-mesenchymal transition (EMT) and tumor metastasis. Here we reviewed up-to-date knowledge of the relationship between m6A modulation and drug resistance. Furthermore, we illustrated the underlying mechanisms of m6A modulation in drug resistance. Lastly, we discussed the regulation of m6A modulation in EMT and cancer stem cells. Hence, it will help to provide significant therapeutic strategies to overcome drug resistance for cancer patients by changing m6A-related proteins via targeting cancer stem cells and EMT-phenotypic cells.

1. Introduction

Cancer is the most vital obstacle to extending life expectancy in the world. In 2020, there were 1,806,590 new cancer cases and 606,520 mortalities to occur in America [1]. Despite promising advances in the treatment for malignant tumors including chemotherapy and radiotherapy, emergence of resistance to chemotherapeutic agents is still the principal obstacle to achieving cures in human cancer patients, leading to high cancer mortality and risk of recurrence. Therefore, elucidation of the molecular mechanisms of therapeutic resistance has been of an urgent medical requirement in exploring effective treatment to guide clinical decision. Drug resistance is ambiguous and multifactorial due to the interaction among cancer cells, cancer stem cells, and the tumor microenvironment [2], which are classified into intrinsic and acquired resistance [3]. Among major mechanisms of drug resistance in cells includes reduced drug influx, elevated drug efflux via ATP-binding cassette (ABC) transporters, alterations in cellular drug targets, drug sequestration within intracellular or extracellular organelles, disruption of cancer cell apoptosis, promotion of drug metabolism and detoxification systems, and repair of drug-induced DNA damage, altered autophagy activity [[4], [5], [6], [7], [8], [9], [10]]. For example, overexpression of efflux transporters exporting drugs from the cell via ABC transporters or downregulation of drug influx transporters importing drugs into the cell decrease the drug effective concentration, leading to anticancer drug resistance [[11], [12], [13], [14]]. When cells able to repair DNA damage or failure to recognize the lesions and signal to the apoptotic machinery, resistance takes place [5,15]. In addition, cancer stem cells (CSCs) and Epithelial-Mesenchymal Transition (EMT) significantly contributes to drug resistance [16,17]. Thus, the molecular knowledge of drug resistance related to CSCs and EMT is now considered as an important focus for cancer research. Recently, accumulated evidence has demonstrated the critical role of m6A modulation in carcinogenesis and drug resistance in multiple cancers [18,19], as well as the vital role of m6A modulation in CSCs and EMT regulation. Hence, focusing on the researches of molecular mechanisms underlying m6A-mediated drug resistance and its modulation in CSCs and EMT is emerging important, which could be helpful for the early diagnosis and therapy of cancers.

2. The regulation and function of m6A modulation

Epigenetic modulation including DNA and RNA methylation, histone modification, noncoding RNA modification, and chromosome remodeling, regulates gene expression in several organisms during early development, contributing to cell fate determination and physiology [20,21]. Recently, RNA modification has received increasing attention in biosciences. Approximately 150 types of RNA modifications have been confirmed, including messenger RNAs (mRNAs), transfer RNAs (tRNAs), ribosomal RNAs (rRNAs), microRNAs (miRNAs), and long non-coding RNAs (lncRNAs) [21], which regulate most steps of RNA metabolism [22,23]. As the most abundant reversible RNA modification in eukaryotic mRNAs, N6-methyladenosine (m6A), referring to the addition of a methyl group at position N6 of adenosine [24], occurs in >25% of human mRNAs and is essential for the regulation of diverse RNA activities including RNA splicing, translocation, stability, and translation [23,25,26]. There are almost 0.1% to 0.4% of adenosines in mRNA with m6A modification in mammals [27,28]. Notably, with the enrichment in the 3′ untranslated regions (3’ UTRs) and near stop codons [29,30], m6A often appears at the RRACH sequences (R = A, G, or U; R = A or G; and H = A, C, or U) [31,32], and interactions with novel RNA-binding proteins affect cellular processes [33]. A variety of coding and noncoding RNAs also accept m6A modification [34]. Aberrant modulations of m6A are related to the growth, differentiation, and metabolism of cells [35] and play critical roles in the pathogenesis of a variety of human diseases including spermatogenesis and carcinogenesis [34,36,37]. Recently, mounting evidence has identified m6A is involved in cancer pathogenesis, contributing to the self-renewal and differentiation of CSC, tumor EMT and metastasis, promotion or inhibition of cancer cell proliferation, and resistance to radiotherapy or chemotherapy [[37], [38], [39], [40]].

The modulation of intracellular m6A level is dynamic and reversible, which is regulated by three types of proteins: methyltransferases as "writers”, demethylases as "erasers”, and specific m6A-binding proteins as "readers” [41,42]. RNA can be methylated under the action of methyltransferases and demethylated under the action of demethylases (Table 1 and Fig. 1).

Table 1.

m6A-related enzymes in m6A modulation.

Category	Subfamily	m6A-related enzymes	Functions	References
Writers	Methyltransferase complex	METTL3	Major catalytic enzyme with methyltransferase activity	[45]
		METTL14	Stabilize and form a stable complex with METTL3	[43]
		WTAP	Recruit METTL3 and METTL14 to help their localization in nuclear spots and promote catalytic activity	[50]
	Other methyltransferases	VIRMA	Mediate m6A in 3′ UTR and near stop codon by recruiting the MTC to modulate region-selective methylation	[53]
		RBM15/RBM15B	Mediate m6A methylation of lncRNA XIST to regulate X-inactivation and gene silencing	[51]
		ZC3H13	Control m6A level by bridging WTAP to the mRNA-binding factor Nito	[52]
		METTL16	Catalyze m6A43 of U6 snRNA and related to 5′-splice site recognition by U6 snRNA during pre-mRNA splicing	[54]
		METTL5	Related to 18S rRNA m6A modification and combine with TRMT112 to gain metabolic stability	[55]
Erasers	AlkB family	FTO	Oxidize m6A into N6-hydroxymethyladeosine and N6-formyladenosine to exert demethylation activity	[[56], [57], [58]]
		ALKBH5	Catalyze the direct removal of m6A modification of nuclear RNA, regulating nuclear RNA export, RNA metabolism, and gene expression	[56,57,59])
Readers	YTH domain family- DF family	YTHDF1	Induce the translation of m6A-modulated mRNAs	[61]
		YTHDF2	The first identified m6A reader and function on the promotion of mRNA decay	[60]
		YTHDF3	Serve as a hub to fine-tune the accessibility of RNA to YTHDF1 and YTHDF2	[62]
	YTH domain family- DC family	YTHDC1	Induce RNA splicing and export	[63]
		YTHDC2	Increase the translation efficiency of target RNA and reduce the abundance of them	[64]
	IGF2BPs	IGF2BP1	Induce RNA expression by modulating mRNA stability, mRNA localization, and translational control via recognization of RRA*CH sequence	[65]
		IGF2BP2I	Induce RNA expression by modulating mRNA stability, mRNA localization, and translational control via recognization of RRA*CH sequence	[65]
		IGF2BP3	Induce RNA expression by modulating mRNA stability, mRNA localization, and translational control via recognization of RRA*CH sequence	[65]
	HNRNP family	HNRNPA2B1	Regulatethe alternative splicing of transcripts	[66]
		HNRNPC	Affiliate in the pre-mRNAs processes	[67]
		HNRNPG	Involved in pre-mRNA alternative splicing	[68]

Fig. 1.

Mechanism of m6A modulation. The m6A methylation is modulated by the writers, erasers, and readers. Writers refer to the m6A methylase complex including METTL3, METTL14, WTAP, VIRMA, RBM15, and ZC3H13. Erasers refer to the m6A demethylase complex including FTO and ALKBH5. Readers recognize m6A and determine target RNA fate.

2.1. Writers

Installation of m6A is modulated by Writers. The typical methyltransferase complex contains methyltransferase-like 3 (METTL3), METTL14 and their cofactors Wilms tumor suppressor-1-associated protein (WTAP) [[43], [44], [45], [46]]. Besides, writers also include vir-like m6A methyltransferase associated (VIRMA, also termed as KIAA1429), RNA-binding motif protein 15/15B (RBM15/RBM15B), and zinc finger CCCH domain-containing protein 13 (ZC3H13) [23]. METTL3 and METTL14 binds S-adenosylmethionine (SAM) or S-adenosylhomocysteine (SAH) in the catalytic site, and co-located in nuclear spots to form stable complexes, catalyze the covalent transfer of a methyl group to adenine with the assistance of WTAP [47]. The consensus sequence motif for m6A modification by the METTL3 and METTL14 writer complex is GRm6ACH (R is A or G, H is not G) and is commonly GGm6ACU [48,49]. METTL3 is a major catalytic enzyme with methyltransferase activity and is implicated in mRNA biogenesis, decay, and translation [45]. Although METTL3 is the catalytic subunit, its activity is strongly dependent on METTL14. METTL14 stabilizes and forms a stable complex with METTL3, helping recognize domain of DNA N(6)-adenine MTase of the target [43]. WTAP recruits METTL3 and METTL14 to help their localization in nuclear spots and to promote catalytic activity [50]. RBM15/RBM15B mediates the m6A methylation of lncRNA XIST to regulate the X-inactivation and gene silencing [51]. ZC3H13 (zinc finger CCCH domain-containing protein 13) controls m6A level through bridging WTAP to the mRNA-binding factor Nito [52]. VIRMA mediates m6A in 3′ UTR and near stop codon by recruiting the MTC to modulate region-selective methylation [53]. In recent year, many new writers have been found such as METTL16, METTL5. In details, METTL16 catalyzes m6A modification at position 43 (m6A43) of U6 snRNA and plays an important role in 5′-splice site recognition by U6 snRNA during pre-mRNA splicing [54]. METTL5 is associated with 18S rRNA m6A modification and combines with TRMT112 to gain metabolic stability [55].

2.2. Erasers

The demethylation of m6A is mediated by m6A demethylases which remove methyl groups. Demethylases compose of fat-mass and obesity associated protein (FTO) and α-ketoglutarate dependent dioxygenase alkB homolog 5 (ALKBH5), both belonging to AlkB family proteins [56,57]. FTO is located in both nucleus and cytoplasm, and oxidises m6A into N6-hydroxymethyladeosine and N6-formyladenosine to exert demethylation activity [58]. ALKBH5 catalyzes the direct removal of m6A modification of nuclear RNA, in turn regulating nuclear RNA export, RNA metabolism, and gene expression [59].

2.3. Readers

In addition, m6A modulation is recognized and interacts with m6A-binding proteins (readers), such as YTHDF1 and IGF2BP1 . There are three classes of m6A “reader” proteins including YTH (YT521-B homology) domain family, insulin-like growth factor 2 mRNA binding proteins (IGF2BPs) family, and heterogeneous nuclear ribonucleoproteins (HNRNP) family. YTH domain group are classified to two subgroups, YTH domain family protein 1-3 (YTHDF1-3, DF family) and YTH domain containing protein 1-2 (YTHDC1-2, DC family). YTHDF2 is the first identified m6A reader and functions on the promotion of mRNA decay [60]; YTHDF1 elevates the translation of m6A-modulated mRNAs [61]; and YTHDF3 serves as a hub to fine-tune the accessibility of RNA to YTHDF1 and YTHDF2 [62]. YTHDC1 induces RNA splicing and export [63] and YTHDC2 increases the translation efficiency of target RNA but reduces the abundance of them [64]. In addition, IGF2BPs family consists of IGF2BP1, IGF2BP2 and IGF2BP3 and plays a critical role in promoting RNA expression by modulating mRNA stability, mRNA localization, and translational control via identification of the consensus RRA*CH sequence [65]. HNRNP family consists of heterogeneous nuclear ribonucleoproteins A2/B1 (HNRNPA2B1), heterogeneous nuclear ribonucleoproteins C (HNRNPC) and heterogeneous nuclear ribonucleoproteins G (HNRNPG), and interacts with a variety of mRNA transcripts to promote their stability and storage in an m6A-dependent manner under normal and stress conditions [65]. In details, HNRNPA2B1 regulates the alternative splicing of transcripts [66]; HNRNPC affiliates in the pre-mRNAs process [67]; and HNRNPG is involved in pre-mRNA alternative splicing [68].

3. EMT and drug resistance

EMT is a highly plastic and dynamic cellular process whereby epithelial cells acquire mesenchymal characteristics with a transition from their epithelial apico-basal polarity to a front-rear polarity and fibroblastic-like phenotype in morphology [69]. EMT processes properties of reducing intercellular adhesion strength and increasing cell motility, invasion and stem cell. During this process, a multitude of epithelial cells identified markers, such as zonula occludens-1 (ZO-1), and E-cadherin, are downregulated whereas mesenchymal cells identified markers, such as Fibronectin, Vimentin and N-cadherin, are elevated [70,71]. In addition, several signaling pathways are involved in EMT including TGF-β, fibroblast growth factor (FGF), hepatocyte growth factor (HGF), Wnt, Notch, and Hedgehog [72]. And EMT-transcription factors (EMT-TFs) regulate EMT process via interaction with chromatin modifying enzymes, including the zinc finger proteins of the SNAIL family, such as Snail1 (Snail), Snail2 (Slug), and Snail3 (Smuc); zinc finger and E-box binding proteins of the ZEB family, such as ZEB1 and ZEB2 (SIP1); bHLH proteins Twist1, Twist2, E47, and E2.2; the homeobox proteins goosecoid (GSC) and Six1; the forkhead-box protein FoxC2 and the Krüppel-like factor KLF8 [73].

EMT has been identified to occur in some important biological processes and its functions includes three types: during embryonic development, adult tissue regeneration, and during cancer progression [74]. Especially, EMT has been increasingly recognized to have a critical role in cancer drug resistance, and targeting EMT to overcome drug resistance has been the subject of some clinical trials [75]. Furthermore, EMT-derived tumor cells acquire stem cell properties and exhibit marked therapeutic resistance [76]. As early as 1990, the relationship between EMT and drug resistance in cancer was proposed. Up to date, observations have found that drug resistance in a variety of cancers is associated with EMT such as osteosarcoma [77], colorectal cancer (CRC) [78], prostate cancer (PCa) [79], ovarian cancer [80,81], lung cancer [82,83], and breast cancer [84,85]. Besides, EMT-TFs directly confer drug-resistance in cancer [86,87]. For instance, in cisplatin-resistant cell lines, both morphological and phenotypic hallmarks of EMT were identified; gene expression profiling identified several EMT-TFs, including Snail1/2, which are further validated as key players in drug resistance [86].

4. Cancer stem cell and drug resistance

Stem cells are associated with organ development during embryogenesis and tissue homeostasis or tissue regeneration in adult organisms [88]. Stem cells balance cellular fates through asymmetric and symmetric divisions in order to self-renew or to generate downstream progenitors. Therefore, stem cell-based therapy is used to provide a hopeful option in the fight against cancers. Stem cells from different sources exhibit different capacities of proliferation, migration, and differentiation, which determine their application in anti-tumor therapy. Among stem cells, it divides into Embryonic stem cells (ESCs), pluripotent stem cells (iPSCs), adult stem cells mostly hematopoietic stem cells (HSCs), mesenchymal stem cells (MSCs), neural stem cells (NSCs), and CSCs. Notably, CSCs are a small population of stem cell-like cancer cells, with the capacity to self-renew and differentiate into heterogeneous tumor cells, playing a major role in the maintenance and propagation of the tumor [89]. Inhibition of crucial CSC markers such as ABC transporters, NANOG, Wnt, CD44, CD133, CD55, ALDH1, OCT4, SOX2 or KLF4 sensitizes cancer cells to chemotherapy.

Being more resistant to therapeutic agents than non-stem cancer cells, CSCs may survive after chemotherapy and radiotherapy, thereby causing cancer post-treatment relapses [90]. Extensive research has reported that CSCs may contribute to drug resistance in vitro and in vivo. The mechanisms in CSCs responsible for drug resistance are still poorly understood. Recently, CSCs regulation by m6A is an emerging explain.

5. m6A modulations in cancer therapeutic resistance

Recent evidence has indicated that m6A modulations display a significant role in tumorigenesis and drug resistance in an increasing number of tumors such as melanoma and lung cancer [91]. m6A modifications modulate drug-target interaction and drug-mediated cell death signaling to regulate cancer drug resistance [92].

For instance, METTL3 induces the m6A modification of YAP in non-small cell lung cancer (NSCLC) via recruiting YTHDF1/3 and eIF3b to the translation initiation complex, in turn promoting the growth, invasion and metastasis of cancer cells and enhancing the resistance to cisplatin via targeting MALAT1-miR-1914-3p-YAP axis [93]. Moreover, the sum enrichment scores for m6A modulation of the lung cancer resistant cell line (H1299) displays in a higher level than that of the sensitive cell line (A549). Functional enrichment analysis revealed that m6A methylation might influence afatinib response by modifying cell cycle, and the further analysis in protein-protein interaction (PPI) network showed that the functionally m6A-modified genes have significantly higher average degree and clustering coefficient than other genes, especially are over-represented in the putative drug resistance-associated genes and the FDA-approved drug targets [94]. Pan et al. also demonstrated that METTL3 promotes adriamycin resistance by targeting miR-221–3p/HIPK2/Che-1 axis in MCF-7 breast cancer cells in a m6A-dependent manner [95]. In glioblastoma (GBM), METTL3, required to maintain the features of GBM stem cells, could be upregulated by temozolomide (TMZ) and mechanistically, METTL3 was regulated by SOX4/EZH2 axis to contribute to TMZ-resistance [96]. It suggested that METTL3 is a critical oncogene in multiple cancers and promotes m6A modification to enhance tumorigenesis by regulating RNA–miRNA interactions.

In addition, m6A demethyltransferase, ALKBH5, is reported to induce JAK2 m6A demethylation through JAK2/STAT3 signaling pathway, consequently attributing to cisplatin resistance of epithelial ovarian cancer (EOC) [97]. Interestingly, one group developed a chemotherapy benefit predictive classifier (m6A score) based on m6A regulators including ZCCHC4, IGF2BP3, ALKBH5, YTHDF3, METTL5, G3BP1, and RBMX, and fond that m6A score is an independent factor for evaluating chemotherapy resistance in small-cell lung cancer [98].

Moreover, FTO is found to be upregulated in human melanoma, and to induce melanoma tumorigenicity. Knockdown of FTO increases melanoma cells sensitivity to interferon gamma (IFNγ) and melanoma sensitivity to anti-PD-1 treatment via adaptive immunity [99]. Mechanistically, FTO deletion induces RNA degradation through the m6A reader YTHDF2, via the promotion of m6A methylation in the critical protumorigenic melanoma cell-intrinsic genes including PD-1 (PDCD1), CXCR4, and SOX10 [99]. Similarly, high expression of FTO enhances leukemia cells resistance to tyrosine kinase inhibitor (TKI) and suppression of FTO renders resistant cells sensitive to TKIs, through controlling mRNA stability of proliferation/survival transcripts bearing m6A and protein synthesis [100]. In line with these findings, FTO mRNA level is higher in cervical squamous cell carcinoma (CSCC) tissues in comparison with cervical normal tissues. Meanwhile, in vitro and in vivo experiments demonstrated that FTO contributes to the chemo-radiotherapy resistance through decreasing m6A mRNA methylation of β-catenin to regulate its expression and consequently elevating excision repair cross-complementation group 1 (ERCC1) activity [101]. Remarkedly, upregulation of ALKBH5 is reported to inhibit cell migration, proliferation and invasion of pancreatic ductal adenocarcinoma (PDAC) and to facilitate the sensitivity of PDAC cells to gemcitabine chemotherapy. Furthermore, ALKBH5 is downregulated in gemcitabine-treated patient-derived xenograft (PDX) model, and its target genes are found to be changed expression by Global m6A profile [102]. Thus, FTO/ALKBH5-dependent m6A demethylation contributes to the drug resistance through the regulation of some specific protein levels in signaling pathways, consequently influencing the regulation of mRNA stability, and degradation and translation efficiency.

Notably, m6A modification of FZD10 mRNA is reported to enhance PARP inhibitors (PARPi) resistance via the promotion of Wnt/β-catenin pathway in BRCA-mutated EOC cells. At the same time, the research group also revealed that the downregulated level of FTO and ALKBH5 is necessary for FZD10 m6A modification, which is related to the enhance of homologous recombination activity. In vitro and in vivo research further confirmed that combined inhibition of PARP and Wnt/β-catenin exerts synergistic suppression of PARPi-resistant cells [19]. Another group described that CBX8 has a high expression level in chemoresistant colon cancer (CC) tissues, and increases cancer cell stemness and chemoresistance through targeting KMT2b and Pol II to increase LGR5 expression in a noncanonical manner. Mechanistically, m6A methylation promotes CBX8 mRNA stability to upregulate CBX8 expression [103]. Concordantly, DDX3 is found to be overexpressed in cisplatin-resistant OSCC lines and chemoresistant cancer tissues to contrast their sensitive lines and tissues. Furthermore, DDX3 modulates ALKBH5 to reduce m6A methylation in FOXM1 and NANOG nascent transcript and increases CSC population through upregulating the expression of FOXM1 and NANOG, resulting in chemoresistance. A PDX model of chemoresistant OSCC confirmed that ketorolac salt restores cisplatin-mediated cell death and facilitates a significant reduction of tumor burdens [104]. It is worth noting that m6A methyl embedded in different areas of RNA could influence RNA splicing, RNA stability, and translation in cancer drug resistance.

Taken together, m6A modification has been demonstrated to play as a decision maker contributing to drug resistance of cancers, dependent on the irregular expression of “writers”, “erasers”, and “readers” at different layers (Fig. 2 and Table 2).

Fig. 2.

Mechanism of m6A modulation in drug resistance. m6A modifications modulate drug-target interaction and drug-mediated cell death signaling to regulate cancer drug resistance dependent on the irregular expression of “writers”,“erasers”, and “readers”.

Table 2.

m6A modulations in cancer therapeutic resistance.

Cancers	m6A-enymases	Drug resistance	Mechanisms	References
NSCLC	METTL3	Increase resistance to DDP	MALAT1-miR-1914-3p-YAP axis via recruiting YTHDF1/3 and eIF3b	[93]
Lung cancer	m6A modulation	Increase resistance to afatinib	m6A-modified genes increase	[94]
Melanoma	FTO	Increase resistance to IFNγ and anti-PD-1 treatment	Inhibit m6A methylation of protumorigenic melanoma cell-intrinsic genes including PD-1 (PDCD1), CXCR4, and SOX10	[99]
Leukemia	FTO	Increase resistance to TKIs	Regulate mRNA stability of proliferation/survival transcripts bearing m6A and protein synthesis	[100]
CSCC	FTO	Increase chemo-radiotherapy resistance	Decrease m6A methylation of β-catenin and then increase ERCC1 activity	[101]
PDAC	ALKBH5	Decrease resistance to gemcitabine	Change its target genes via Global m6A profile	[102]
BRCA-mutated EOC cells	FTO and ALKBH5	m6A modification of FZD10 mRNA increase resistance to PARPi	Wnt/β-catenin pathway induces m6A modification of FZD10 mRNA via downregulation of FTO and ALKBH5	[19]
Colon cancer	m6A methylation	CBX8 increases chemoresistance	m6A methylation promotes CBX8 mRNA stability to increase CBX8	[103]
OSCC	ALKBH5	DDX3 increase resistance to cisplatin	DDX3 modulates ALKBH5 to reduce m6A methylation in FOXM1 and NANOG nascent transcript and increases CSC population	[104]

NSCLC: non-small cell lung cancer; CSCC: Cervical squamous cell carcinoma; PDAC: pancreatic ductal adenocarcinoma; EOC: epithelial ovarian cancers; OSCC: Oral squamous cell carcinoma.

6. m6A modulation and EMT

It is noticed that m6A modification of mRNAs increases in cancer cells during EMT, which is critical for cancer cell metastasis. In recent year, RNAMethyPro, a EMT-associated gene expression signature of seven m6A regulators was developed and used to predict survival at pan-cancer level by Kandimalla. A network-based analysis found an intimate functional interplay between m6A regulators and EMT-associated factors via druggable targets such as XPO1 and NTRK1 [105]. METTL3 is identified as an oncogenic factor with higher expression in cancer tissues. For instance, among liver cancer patients, the expression of METTL3 and YTHDF1 are upregulated and predictive of poor overall survival (OS). Additionally, m6A-sequencing and functional research identified that YTHDF1 regulates m6A-increased translation of Snail mRNA, a critical transport factor in m6A-regulated EMT [106]. Consistently, METTL3 is found to be frequently upregulated in gastric cancer (GC) tissues compared with normal gastric tissues in big crowd data sets such as The Cancer Genome Atlas, Kaplan-Meier plotter, and Gene Expression Omnibus database, as well as in the expression of messenger RNA and protein levels, serving as a poor prognostic factor for GC patients [107,108]. Also, deletion of METTL3 suppressed proliferation and migration of cancer cells through the downregulation of total RNA m6A methylation and α-smooth muscle actin levels [108]. Furthermore, METTL3 is verified to be highly expressed in ovarian carcinoma tissues and in OVCAR3 and COV504 cell lines, which promotes cancer cell proliferation, invasion, motility and focus formation in vitro, and induces tumor formation in vivo [109]. Exploring its mechanism, it is found that METTL3 leads to the promotion of TGF-β-induced EMT and the increase of receptor tyrosine kinase AXL and Snail. And m6A promotes TGF-β1 mRNA decay but impaires TGF-β1 translation progress [109,110]. It is noticed that during TGF-β-induced EMT, METTL3 expression and m6A RNA modification are also increased in A549 and LC2/ad lung cancer cells. In addition, METTL3 knockdown represses cells morphological conversion, induces the alters of EMT-related marker genes such as CDH1/E-cadherin, FN1/Fibronectin and VIM/Vimentin, in turn promoting cell migration. Mechanistically, METTL3 contributes to the m6A modification, total mRNA level and mRNA stability of a critical EMT-related transcriptional factor JUNB [111]. Interestingly, zinc finger MYM-type containing 1 (ZMYM1) is a bona fide m6A target of METTL3 dependent on HuR protein, and can promotes EMT program and metastasis via the inhibition of E-cadherin promoter by recruiting the CtBP/LSD1/CoREST complex [107]. Furthermore, lncRNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) is increased in renal fibrosis in patients with obstructive nephropathy (ON) and it mediates TGF-β1-induced renal fibrosis via MALAT1/miR-145/focal adhesion kinase (FAK) pathway. And METTL3 is the main methyltransferase of m6A modification on MALAT1 [112].

Conversely, METTL3 mRNA and protein expression are downregulated in renal cell carcinoma (RCC) tissues compared with adjacent normal tissues, and decreased in RCC cell lines (CAKI-1, CAKI-2 and ACHN) compared with HK-2 cell. Meanwhile, high expression of METTL3 predicts a better survival outcome for RCC patients and suppresses cancer cell growth, proliferation, migration, invasion, and G0/G1 arrest through mediating EMT and PI3K-Akt-mTOR signaling pathways, suggesting METTL3 acts as a tumor suppressor in renal cell carcinoma [113].

In NSCLC, ALKBH5 is uncovered to inhibit tumor growth and metastasis by decreasing YTHDFs-mediated YAP expression and miR-107/LATS2-mediated YAP activity in an HuR-dependent manner. In details, YTHDF3 combines YAP pre-mRNA, YTHDF2 induces YAP mRNA decay via the AGO2 system, and YTHDF1 promotes YAP mRNA translation by interacting with eIF3a, all in an m6A-independent manner [114].

Notably, an increasing number of lncRNAs are confirmed to be involved in m6A modulation targeting EMT. For example, lncRNA RP11-138 J23.1 (RP11) is upregulated in colorectal cancer (CRC) tissues and induced by m6A methylation. Moreover, RP11 promoted the metastasis of CRC cells via the increase of EMT-related transcription factor Zeb1 at post-translational level through the inactivation of two E3 ligases mRNA Siah1 and Fbxo45 [115]. In contrast, lncRNA KCNK15-AS1 is downregulated in pancreatic cancer tissues in comparison with adjacent normal tissues, and represses the migration and invasion of MIA PaCa-2 and BxPC-3 cells. And ALKBH5 is shown to demethylate and decrease KCNK15-AS1 to regulate cancer cell motility [116].

7. m6A modulation and cancer stem cells

Recent evidence has demonstrated the close relationship between m6A modulation and CSCs. In the following we reviewed m6A modulation targeting CSCs in the classification of different cancers.

7.1. Breast cancer stem cell (BCSC)

In MDA-MB-231 breast cancer cells, ablation of ALKBH5 decreases BCSCs numbers, resulting in the decrease of capacity for tumor initiation [117], and significantly decreases metastasis from breast to lungs in immunodeficient mice [118]. It has been demonstrated that exposure of breast cancer cells to hypoxia increases the percentage of BCSCs and NANOG mRNA and protein expression in an HIF- and ALKBH5-dependent manner [117,118]. In details, the upregulation of ALKBH5 elevates BCSCs numbers and phenocopys the effect of hypoxia via inhibiting NANOG and KLF4 mRNA methylation to enhance its expression dependent on ZNF217 [117,118]. NANOG and KLF4 are key pluripotency factors that mediates BCSC specification.

7.2. CRC

Knockdown of METTL3 in CRC cells is identified to suppress stem cell frequency, cell self-renewal and migration in vitro and inhibited CRC tumorigenesis and metastasis in cell-based and PDX models. Mechanistically, SRY (sex determining region Y)-box 2 (SOX2) transcripts, the downstream gene of METTL3, are methylated and recognized by IGF2BP2, leading to prevent SOX2 mRNA degradation [119]. Furthermore, knockdown of YTHDF1 also drastically represses the tumorigenicity of CRC cell in vitro and murine xenograft tumor growth in vivo, via the inhibition of Wnt/β-catenin pathway activity. In addition, the mRNA and protein expression of YTHDF1 is upregulated in CRC tissues, as well as in TCGA, GEO CRC databases [120].

7.3. Glioblastoma stem-like cells (GSCs)

One group used Methylated RNA immunoprecipitation-seq (MeRIP-seq) and found that m6A modification peaks are enriched at metabolic pathway-related transcripts in GSCs compared with neural progenitor cells [121]. Moreover, METTL3 expression is increased in GSCs and decreased during differentiation, while METTL3 transcripts is elevated in GBM tumors [122]. In addition, silencing METTL3 or overexpressing dominant-negative mutant METTL3 inhibited GSC growth and self-renewal in U87/TIC cells, and prolonged mice survival in an intracranial orthotopic mouse model [121,122]. Moreover, METTL3 silencing elevates GSCs sensitivity to γ-irradiation and inhibits DNA repair with accumulation of γ-H2AX and RasV12 mediated transformation of mouse immortalized astrocytes [122]. Mechanistically, METTL3 downregulation depresses m6A modification levels of serine- and arginine-rich splicing factors (SRSF), resulting in the decreased expression of SRSF protein and nonsense-mediated mRNA decay (NMD) of SRSF transcripts dependent on YTHDC1 [121]. At the same time, METTL3 induces SOX2 mRNA m6A-modification and stabilization dependent on the recruitment of HuR [122]. In contrast, another group showed that knockdown of METTL3 or METTL14 induces GSC growth, self-renewal, tumorigenesis and mRNA m6A enrichment changes, and alters mRNA expression of GSCs-related genes (e.g.,ADAM19) [123]. Furthermore, silencing METTL3 increases several aberrant alternative splicing events. For instance, METTL3 alters A-to-I and C-to-U RNA editing events by differentially modulating RNA editing enzymes ADAR and APOBEC3A, and lead to the high expression of lincRNAs (long intergenic non-coding RNAs) with m6A markers [124]. Of note, ALKBH5 is identified to be overexpressed in GSCs, and to induce the proliferation of patient-derived GSCs. And deletion of ALKBH5 and lncRNA FOXM1-AS could regulate FOXM1 axis to inhibit GSC tumorigenesis dependent on the action of ALKBH5 demethylating and upregulating FOXM1 nascent transcripts [125]. Lastly, the activation of FTO is found to induce cancer progression and to short lifespan of GSC-grafted mice [123].

7.4. Cutaneous squamous cell carcinoma (cSCC)

The cSCC originates from epithelial stem cells through the dysregulation of self-renewal and differentiation. METTL3 is highly expressed in cSCC tissues, and ablation of METTL3 impaires cSCC cell stem-like properties. Indeed, METTL3 deficiency and methylation inhibitor cycloleucine could decrease the m6A levels and the expression of ΔNp63 in cSCC. Exogenous expression of ΔNp63 partially restores the cell proliferation of METTL3-knockdown cSCC cells [126].

7.5. Osteosarcoma stem cells (OSCs)

Three m6A-related enzymes are identified to be changed in OSCs. Meanwhile, differentially methylated genes are abundant in some pathways modulating pluripotency of stem cells, and several candidate genes are related to poor prognosis in osteosarcoma patients [127]. Specifically, inhibition of DDX3 decreases CSCs population via the down-regulation of FOXM1 and NANOG in an m6A methylation way via ALKBH5, subsequently contributing to cisplatin chemoresistance in OSCC [104].

7.6. Ovarian cancer

TRIM29 contributed to the CSC-like features of cisplatin-resistant ovarian cancer by mediating m6A-YTHDF1，resulting in the carcinogenesis [128].

8. m6A modulation and antitumor treatment

8.1. m6A modulation and METTL3 inhibitors

METTL3, in recent years, has been reported to be related to the initiation and maintenance of various cancers through depositing m6A modification on critical transcripts [[129], [130], [131]]. Therefore, targeting METTL3 is identified as an efficient therapeutic way for the treatment of different kinds of cancers. For example, one group uncovered a highly potent and selective first-in-class catalytic inhibitor of METTL3, STM2457, and found it could reduce the growth and induce the differentiation and apoptosis of acute myeloid leukemia (AML), suggesting METTL3 inhibitor as a potential therapeutic strategy against AML [130]. Similarly, in another study, inhibition of methyltransferases including METTL 3 and METTL 14 could promote the response to anti-PD-1 treatment in mismatch-repair-proficient or microsatellite instability-low (pMMR-MSI-L) colorectal cancer and melanoma [132]. Mechanistically, ablation of METTL3 could impaire the YTHDF1-mediated translation of SPRED2, and then lead to the activation of NF-kB and STAT3 through the ERK pathway [133]. Taken together, it demonstrates that depletion of METTL3 may be an important therapeutic strategy for several types of cancers.

8.2. m6A modulation and FTO inhibitors

Emerging evidence demonstrated that some FTO inhibitors, has been developed for the effective targeted therapeutics in a variety of tumors due to the oncogenic functions of FTO [[134], [135], [136]]. For example, Su et al. found that genetic knockout and pharmacological restrain of FTO could significantly suppress leukemia stem/initiating cell self-renewal and immune evasion [134]. Another study reported two FTO inhibitors (FB23 and FB23-2), and identified that FB23-2 inhibited the progression of human acute myeloid leukemia (AML) in vivo and in vitro [135]. It suggested that targeting FTO by small-molecule inhibitors holds potential to treat tumors.

8.2. m6A modulation and immunotherapy

Recently, more and more potential prognostic and predictive biomarkers are emerging for immunotherapy evaluation [137], among which m6A modification also exerted an important role in immune surveillance and immune treatment. A study reported 17 m6A regulators upregulated in esophageal carcinoma tissues, which is correlated with the expression of immunoinhibitors, immunostimulators, MHC molecules and immune infiltration levels [138]. Similarly, another group identified two m6A modification patterns including an immune-activated differentiation pattern and an immune-desert dedifferentiation pattern, and four m6A-regulator pattern on immunity and stemness. In addition, patients with high m6AScore showed more sensitivity to radio-chemotherapy and immunotherapy [139]. More importantly, the inhibition of METTL3 could induce dendritic cell (DC) activation and DC-based T cell response [140].

Furthermore, immune checkpoint inhibitors (ICIs) have been established as potential anti-tumor agents for various tumors [141,142]. It is reported that the therapeutic efficacy of PD-1 checkpoint inhibit is attenuated in METTL3-deficient mice [133].

9. Conclusions

The m6A modulation participates in carcinogenesis, drug resistance, EMT regulation and CSCs by controlling the expression of their targeted genes. Given these, targeting m6A modulation in the regulation of EMT and CSCs is a significant therapeutic strategy in elevating drug sensitivity of human cancers. However, the results of many researchers in recent years are sometimes contradictory and need more studies to confirm. Thereby, more researches need to further explore, providing more accurate treatment strategy for drug resistant cancer patients.

Declarations

Author contribution statement

All authors listed have significantly contributed to the development and the writing of this article.

Funding statement

This work was supported by the Medical Scientific Research Foundation of Zhejiang Province, China (2020KY832), Natural Science Foundation of Ningbo, China (2019A610305) and Ningbo Public Service Technology Foundation, China (202002N3154).

Data availability statement

Data will be made available on request.

Declaration of interest’s statement

The authors declare no competing interests.