The Emerging Roles of RNA m⁶A Methylation and Demethylation as Critical Regulators of Tumorigenesis, Drug Sensitivity, and Resistance

RNA methylation research started gathering momentum less than 10 years ago when RNA methylation was found to be reversible (1). After the establishment of methylated RNA immunoprecipitation sequencing technology (2, 3), a plethora of RNA methylation sites have been discovered. Because it occurs in approximately 25% of mRNAs at the transcriptome-wide level and regulates mRNA expression and protein synthesis, RNA methylation has been termed “epitranscriptomics” (2, 4–11).

The most common and most important mRNA methylation is the N⁶-methyladenosine (m⁶A) modification. RNA m⁶A modification regulates pre-mRNA splicing, mRNA decay, stability, and protein translation, and thereby plays a critical role in normal physiology and diseases, including cancer (3, 12–14). As m⁶A modification plays critical roles in tumorigenesis and drug resistance, inhibitors of m⁶A modification proteins are emerging as novel anticancer agents (15–19).

Methyltransferase-like 3 (METTL3) and METTL14 proteins form an RNA m⁶A methyltransferase complex with cofactors WTAP, VIRMA (KIAA1429), RBM15, RBM15B, ZC3H13, and HAKAI (20–26). As the only components with methyltransferase activity, METTL3 and METTL14 induce RNA GAC, AAC, and GAC m⁶A methylation, respectively, and synergistically induce mRNA m⁶A modification (Fig. 1A; refs. 21, 27, 28). VIRMA, RBM15, and RBM15B recruit METTL3/METTL14/WTAP to target mRNA positions for m⁶A methylation (23, 24, 26). ZC3H13 is required for nuclear localization of the ZC3H13-WTAP-VIRMA-HAKAI complex and is, therefore, essential for m⁶A methylation (Fig. 1A; refs. 25, 26).

The RNA m⁶A writers, partly through the m⁶A methyltransferase complex, alter pre-mRNA splicing and mRNA decay (29, 30), reduce (31, 32) or enhance mRNA stability (33, 34), and promote mRNA nuclear-cytoplasmic export (14, 35) and protein translation (12, 36, 37).

While the METTL3/METTL14 m⁶A writer complex mainly induces mRNA methylation, METTL16 binds U6 small nuclear RNA, other noncoding RNAs, and pre-mRNAs to induce m⁶A modification and regulate pre-mRNA splicing (38, 39).

Demethylation of RNA m⁶A is regulated by the demethylases FTO and ALKBH5, an FTO homologue (Fig. 1A; refs. 1, 40). FTO preferentially binds to intronic regions of pre-mRNAs and promotes exon inclusion (41). Although FTO depletion leads to accumulation of mRNAs with m⁶A modification, only a fraction of the mRNAs exhibit increased mRNA expression and reduced protein translation (42, 43).

ALKBH5 colocalizes with nuclear speckles, which are nucleoplasm domains enriched in pre-mRNA splicing factors. ALKBH5 induces mRNA m⁶A demethylation, suppresses mRNA nuclear-cytoplasmic export, decreases nascent RNA synthesis, and increases global RNA stability and expression (40). ALKBH5 depletion results in aberrant splicing, exon skipping, and rapid degradation of transcripts with elevated m⁶A (44).

M⁶A reader proteins include YTH domain–containing proteins, YTHDC1–2 and YTHDF1–3; insulin-like growth factor 2 mRNA binding proteins, IGF2BP1–3; and the “royal family” protein SND1. YTHDC1 interacts with and recruits histone H3 lysine 9 dimethylation (H3K9me2) demethylase, KDM3B, to m⁶A-associated chromatin regions, leading to H3K9me2 demethylation and transcriptional activation (45). YTHDC1 binds m⁶A-modified pre-mRNAs and mRNAs, and facilitates exon inclusion, splicing, mRNA nuclear-cytoplasmic export (46–48), and the decay of chromosome-associated regulatory RNAs (Fig. 1A; refs. 49, 50). In comparison, YTHDC2 binds m⁶A-modified RNAs, decreases mRNA stability, but enhances mRNA translation efficiency (Fig. 1A; refs. 51, 52).

YTHDF1, YTHDF2, and YTHDF3 form complexes with m⁶A-modified mRNAs, and the YTHDF-bound mRNAs are subjected to increased decay (53–55). In addition, YTHDF1 and YTHDF3 increase m⁶A-modified mRNA translation by interacting with translation initiation factors eIF3, eIF4A3, and eIF4A3 (Fig. 1A; refs. 55, 56).

IGF2BP1, IGF2BP2, and IGF2BP3 recognize m⁶A through their K homology domains, and facilitate m⁶A-modified mRNA stabilization and protein translation (57–59). The “royal family” protein SND1 also binds m⁶A-modified mRNAs and induces mRNA stabilization (Fig. 1A; ref. 60).

METTL3 has been predominantly reported to exert tumorigenic effects (Fig. 1B). METTL3 is highly expressed in human lung and colon adenocarcinoma tissues (61), and the high expression correlates with poor patient prognosis (62). In lung and colon cancer cells, METTL3 induces oncogenic SOX2, EGFR, TAZ, ABCA3, and FOXN3 mRNA m⁶A methylation, prevents SOX2 mRNA degradation (62), promotes EGFR and TAZ mRNA translation (61), and enhances ABCA3 and FOXN3 mRNA degradation (29). METTL3, therefore, induces lung and colon cancer cell proliferation, survival, and invasion in vitro and tumor growth and metastasis in mice (Fig. 1B; refs. 29, 61, 62). In human gastric cancer tissues, METTL3 and m⁶A are upregulated and high levels of METTL3 independently predict poor patient prognosis (33, 63). In gastric cancer cells, ZMYM1 suppresses E-cadherin gene transcription and HDGF activates GLUT4 and ENO2 expression, and METTL3 stabilizes ZMYM1 and HDGF mRNAs through inducing m⁶A modification. METTL3 thereby induces gastric cancer cell glycolysis, proliferation, and epithelial-to-mesenchymal transition in vitro and tumor growth and metastasis in mouse models (Fig. 1B; refs. 33, 63).

Similarly, METTL3 is significantly upregulated in human breast and bladder cancer tissues (64, 65). Through inducing HBXIP, c-Myc, and the NF-κB pathway mRNA m⁶A modification and upregulation, METTL3 promotes breast and bladder cancer cell proliferation, invasion, and survival in vitro and tumor progression in mice (Fig. 1B; refs. 64, 65). In human hepatocellular carcinoma tissues, high METTL3 expression predicts poor patient prognosis (66, 67). Mechanistically, METTL3 downregulates m⁶A-modified SOCS2 mRNA expression, upregulates Snail mRNA translation, and thereby induces hepatocellular carcinoma cell proliferation, migration, and epithelial-to-mesenchymal transition in vitro and tumor growth and lung metastasis in vivo (Fig. 1B; refs. 66, 67). In acute myeloid leukemia, METTL3 and METTL14 enhance the translation of c-Myc, PTEN, MYB, and BCL2 mRNAs, decrease AKT protein phosphorylation, and induce leukemia cell differentiation block, proliferation, cell-cycle progression, survival, and in vivo progression (Fig. 1B; refs. 68–70).

While METTL3 promotes tumorigenesis in other cancers, its role in glioblastoma is controversial. METTL3 is upregulated in human glioblastoma tissues, and this upregulation correlates with poor patient prognosis (71). METTL3 or METTL14 knockdown induces glioblastoma cell apoptosis in vitro and tumor growth inhibition in mice through demethylating and reducing SOX2 mRNA expression (71). However, METTL3 and METTL14 depletion enhances glioblastoma stem cell growth, self-renewal, and tumorigenesis by upregulating oncogenic KLF4 and ADAM19 and downregulating tumor suppressive CDKN2A and BRCA2 expression (Fig. 1B; refs. 16, 72). Future studies should be performed to explore the mechanisms for differential mRNA selectivity of the m⁶A writers in glioblastoma tumor cells versus stem cells.

Unlike METTL3, METTL14 plays both oncogenic and tumor suppressive roles. METTL14 is overexpressed in human pancreatic cancer tissues, and high levels of METTL14 correlate with poor patient prognosis (73). Through promoting PERP mRNA m⁶A methylation, degradation, and downregulation, METTL14 induces pancreatic cancer cell proliferation, migration, and invasion in vitro and tumor growth and metastasis in mouse models (Fig. 1B; ref. 73).

However, more literature suggests that METTL14 exerts tumor suppressive function. METTL14 is downregulated in human colorectal cancer tissues, and the downregulation independently predicts poor patient prognosis (74). By promoting SOX4 mRNA m⁶A methylation and YTHDF2-dependent SOX4 mRNA degradation, METTL14 suppresses PI3K/AKT signaling, colorectal cancer cell proliferation, migration, invasion, and epithelial-to-mesenchymal transition in vitro and tumor metastasis in vivo (74). In human hepatocellular carcinoma tissues, m⁶A modification is decreased and low levels of METTL14 are a marker for tumor metastasis and poor patient prognosis (75). Mechanistically, through interacting with the microprocessor protein, DGCR8, and inducing pri‐miR126 methylation, METTL14 increases miR126 expression and thereby inhibits hepatocellular carcinoma metastasis (75). In human endometrial tumors, m⁶A methylation levels are decreased because of METTL14 R298P mutation (76). The reduction in m⁶A methylation results in decreased expression of the negative AKT regulator, PHLPP2, and increased expression of the positive AKT regulator, mTORC2, leading to endometrial cancer cell proliferation and tumorigenicity (Fig. 1B; ref. 76).

KIAA1429 and WTAP are highly expressed in human hepatocellular carcinoma tissues, and the upregulation independently predicts poor patient prognosis (77, 78). By upregulating GATA3 pre-mRNA and ETS1 mRNA m⁶A modification and degradation, KIAA1429 and WTAP block p21/p27-dependent cell-cycle arrest, and induce hepatocellular carcinoma cell proliferation in vitro and tumor growth and metastasis in mouse models (77, 78).

FTO demethylates m⁶A-modified tumor suppressive BNIP3 mRNA and induces its degradation, leading to breast cancer cell proliferation and colony formation in vitro and tumor metastasis in vivo (79). FTO is upregulated, while BNIP3 is downregulated in human breast cancer tissues, and a high level of FTO and a low level of BNIP3 are markers for poor patient prognosis (Fig. 1C; ref. 79). In a subset of human acute myeloid leukemia samples, FTO is highly expressed because of MLL and MLL-fusion oncoproteins. By reducing m⁶A-modified mRNAs, including ASB2 and RARA mRNAs, FTO inhibits retinoic acid–induced leukemia cell differentiation, promotes leukemia cell proliferation and survival in vitro, and leukemogenesis in vivo (Fig. 1C; ref. 80).

Similarly, in human cervical squamous cell carcinoma tissues, FTO is highly expressed and a high level of FTO is a marker for poor patient prognosis (81). By demethylating m⁶A-modified β-catenin mRNA and upregulating its expression, FTO mediates DNA damage response and renders cervical cancer cells resistant to chemoradiotherapy (81). In addition, FTO is overexpressed in human melanoma tissues (82). FTO promotes melanoma cell proliferation, migration, and invasion in vitro and tumor progression in vivo via demethylating the critical tumorigenic PD-1, CXCR4, and SOX10 mRNAs and blocking their decay (82). In glioblastoma stem cells, FTO inhibition reduces oncogenic ADAM19, EPHA3, and KLF4 mRNA expression and subsequently glioblastoma stem cell proliferation and tumor growth (Fig. 1C; ref. 16).

Contrary to the above, FTO functions as a tumor suppressor in ovarian cancer. FTO is downregulated in human ovarian cancer tissues and ovarian cancer stem cells (83). By reducing PDE1C and PDE4B mRNA m⁶A methylation and stability and enhancing cAMP signaling, FTO inhibits ovarian cancer stem cell self-renewal and suppresses ovarian cancer tumorigenesis (Fig. 1C; ref. 83).

Upregulated under hypoxic conditions, ALKBH5 increases Nanog and FOXM1 mRNA and protein expression by m⁶A demethylation, and induces breast cancer stem cell and glioblastoma stem-like cell proliferation and tumorigenesis (84, 85). Similarly, ALKBH5 promotes acute myeloid leukemia stem cell and leukemia cell proliferation, leukemia initiation, and progression through enhancing AXL and TACC3 mRNA stability in an m⁶A demethylation–dependent manner (86, 87). In addition, ALKBH5 is highly expressed in human acute myeloid leukemia cells, compared with normal lymphocytes, and high levels of ALKBH5 predict poor patient prognosis (Fig. 1C; refs. 86, 87).

In contrast, ALKBH5 functions as a tumor suppressor in hepatocellular carcinoma and pancreatic cancer. ALKBH5 is downregulated in human hepatocellular carcinoma and pancreatic cancer tissues, and decreased ALKBH5 expression is an independent marker for worse survival in patients (88–90). Through IGF2BP1- and YTHDF2-dependent mechanisms, ALKBH5 induces mRNA m⁶A demethylation, reduces LYPD1 and upregulates PER1 and WIF-1 mRNA expression, and activates ATM-CHK2-p53/CDC25C pathway signaling. Consequently, ALKBH5 reduces hepatocellular carcinoma and pancreatic cancer cell proliferation, migration, and invasion in vitro and tumor growth and metastasis in vivo (Fig. 1C; refs. 88–90).

YTHDF1 is highly expressed in human colon and gastric cancer tissues, and this upregulation is associated with poor patient prognosis (91, 92). Directly upregulated by c-Myc oncoprotein, YTHDF1 enhances Wnt receptor frizzled 7 (FZD7) mRNA translation and induces Wnt/β-catenin pathway activation in an m⁶A-dependent manner, leading to colon and gastric cancer cell proliferation in vitro and tumor progression in mouse models and resistance against anticancer drugs, including fluorouracil and oxaliplatin (Table 1; refs. 91, 92).

YTHDF2 mediates c-Myc, CEBPA, and SOCS2 mRNA m⁶A modification to enhance c-Myc and CEBPA mRNA stability and to induce SOCS2 mRNA degradation, leading to leukemia and lung cancer cell proliferation and oncogenesis (19, 66). By promoting OCT4 mRNA m⁶A methylation at the 5′ untranslated region and OCT4 mRNA translation, YTHDF2 induces hepatocellular carcinoma stem cell stemness, tumor progression, and metastasis (93). In human hepatocellular carcinoma tissues, high levels of YTHDF2 expression positively correlate with OCT4 mRNA m⁶A methylation and overexpression, and are associated with poor patient prognosis (93).

YTHDC2 is highly expressed in human colon cancer tissues, and the upregulation is positively associated with tumor metastasis (94). YTHDC2 unwinds the Twist1 mRNA 5′ untranslated region to enhance its protein translation, and thereby induces colon cancer metastasis (94).

IGF2BP1, IGF2BP2, and IGF2BP3 are highly expressed in human cancer tissues, such as neuroblastoma tissues, and high levels of IGF2BP proteins correlate with poor patient prognosis (58, 95). Through their role as m⁶A readers, IGF2BP proteins enhance the stability and expression of target mRNAs, including c-Myc, leading to liver and cervical cancer cell proliferation and tumorigenesis (58, 96). In addition, highly expressed in human cervical cancer tissues compared with normal counterparts, IGF2BP3 and YTHDF1 bind the 5′ untranslated region of PDK4 mRNA to enhance PDK4 mRNA stability and translation, and induce cervical cancer cell glycolysis and tumor progression in mouse models (97).

RNA m⁶A modification proteins suppress or enhance cancer cell sensitivity and resistance to cancer therapies, depending on individual m⁶A modification protein, cancer type, and therapy (Table 1). In response to DNA double-strand breaks, PARP recruits METTL3 to DNA damage sites, and METTL3 recruits YTHDC1 to cause m⁶A methylation of DNA damage–associated RNAs. The METTL3–m⁶A–YTHDC1 axis induces DNA-RNA hybrid accumulation, DNA polymerase κ localization to DNA damage sites for nucleotide excision repair, and recruitment of RAD51 and BRCA1 for homologous recombination–mediated repair (13, 98). METTL3, therefore, significantly suppresses cancer cell sensitivity to DNA damage–based radiotherapy and chemotherapy, such as cisplatin, in vitro and in mouse models (Table 1; ref. 13). In pancreatic cancer cells, METTL3 knockdown sensitizes pancreatic cancer cells to anticancer agents, such as irradiation, gemcitabine, 5-fluorouracil, and cisplatin (Table 1; ref. 99). On the contrary, METTL3 is downregulated in human sorafenib-resistant hepatocellular carcinoma tissues, and METTL3 sensitizes hepatocellular carcinoma cells to sorafenib therapy by promoting FOXO3 mRNA m⁶A methylation and enhancing FOXO3 mRNA stability (Table 1; ref. 34).

FTO is highly expressed in human cervical squamous cell carcinoma tissues, mediates DNA damage response, and renders cervical cancer cells resistant to radiotherapy and cisplatin (Table 1; ref. 81). FTO is also overexpressed in human melanoma tissues and induces melanoma cell resistance to IFNγ and anti-PD-1 treatment in mice, suggesting that FTO inhibition is likely to overcome melanoma cell resistance to immunotherapy (Table 1; ref. 82).

In contrast, in PARP inhibitor–resistant BRCA-mutated ovarian cancer cells, both FTO and ALKBH5 are downregulated to increase FZD10 mRNA m⁶A modification, FZD10 mRNA stability, and Wnt/β-catenin pathway activation (100). FZD10 depletion or treatment with Wnt/β-catenin inhibitors sensitizes resistant ovarian cancer cells to PARP inhibitors in vitro and in a mouse model, suggesting that FTO and ALKBH5 downregulation renders BRCA-mutated ovarian cancer cells resistant to PARP inhibitors by modulating FZD10 expression (Table 1; ref. 100).

ALKBH5 also modulates cancer cell resistance to immunotherapy. ALKBH5 upregulates MCT4 expression, lactate content, tumor-infiltrating regulatory T cells (Treg), and myeloid-derived suppressor cells in melanoma tumor microenvironment (17). In human melanoma tissues, high levels of ALKBH5 expression correlate with poor response to immunotherapy (Table 1; ref. 17). In contrast, through demethylating and thereby upregulating PER1 and WIF-1 mRNAs in an YTHDF2-dependent manner, ALKBH5 activates ATM–CHK2–p53/CDC25C and suppresses Wnt signaling (Table 1; refs. 89, 90). These effects enhances pancreatic cancer cell sensitivity to gemcitabine treatment both in vitro and in mouse models (89, 90).

YTHDF1 enhances m⁶A-modified FZD7 mRNA translation into protein, leading to Wnt/β-catenin pathway hyperactivation in colon cancer cells (92). YTHDF1, therefore, induces colon cancer cell resistance against anticancer agents, including fluorouracil and oxaliplatin (Table 1; ref. 91). In addition, YTHDF1 depletion in dendritic cells in mice results in elevated antigen-specific CD8⁺ T-cell antitumor response, and leads to greater anticancer efficacy of PD-L1 checkpoint inhibitors (101).

Using an in silico screening, the S-adenosyl-L-methionine (SAM) binding site was selected as the target area for potential METTL3-METTL14 RNA m⁶A methyltransferase complex ligands (102). Four compounds representing piperidine and piperazine derivatives have been found to bind to the METTL3 protein region containing Asp295, Phe534, Arg536, and Asn539 (Table 2). These compounds function as METTL3-METTL14 RNA m⁶A methyltransferase complex activators, inducing mRNA m⁶A modification in cells (102). Their potential anticancer effects are untested.

After screening a library of 4,000 analogues and derivatives of the adenosine moiety of SAM through in silico high-throughput docking, Bedi and colleagues have identified one N-substituted amide of ribofuranuronic acid analogues of adenosine as an efficient METTL3 inhibitor (103). The binding mode of the METTL3 inhibitor has been validated by protein crystallography and the METTL3 inhibitor showed good ligand efficiency. However, its anticancer efficacy has not been tested (Table 2; ref. 103).

Several RNA m⁶A methyltransferase inhibitors display anticancer effects. Two distinct small-molecule METTL3 inhibitors have recently been discovered through a structure-guided medicinal chemistry platform (104). They reduce METTL3 target mRNA abundance and suppress leukemia cell proliferation in vitro, as well as leukemia progression in mice (Table 2; ref. 104).

STORM Therapeutics has recently discovered STC-15, a highly potent and selective METTL3 inhibitor, which has shown in vitro and in vivo anticancer efficacy against chemoresistant leukemia and solid tumors. STC-15 is expected to progress into clinical trials in patients with acute myeloid leukemia in 2021. However, preclinical data on STC-15 have not been published.

Rhein, CHTB, and N-CDPCB have been identified as FTO inhibitors via structure-based in silico screening (105–107). Rhein competitively binds the FTO catalytic domain, CHTB binds FTO surface area, and N-CDPCB binds FTO between its L1 loop and antiparallel β-sheet, all disrupting FTO function and suppressing m⁶A demethylation (Table 2; refs. 105–107). In addition, the NSAID, meclofenamic acid 2, can function as an FTO inhibitor. It binds FTO's active surface and selectively inhibits FTO-mediated m⁶A demethylation (15). Meclofenamic acid 2 reduces glioblastoma stem cell proliferation and sphere formation in vitro, and suppresses glioblastoma progression in mice when injected locally to tumors (Table 2; ref. 16).

Four small-molecule FTO inhibitors, R-2-hydroxyglutarate (R-2HG), FB23-2, CS1, and CS2 have shown anticancer effects when administered systematically. R-2HG and FB23-2 reduce oncogenic c-Myc and CEBPA and activate tumor suppressive RARA and ASB2 expression, by increasing mRNA m⁶A modification. In acute myeloid leukemia cell lines and primary leukemia blast cells, R-2HG and FB23-2 substantially repress cell proliferation, promote differentiation, and induce apoptosis (18, 19). Importantly, combination therapy with R-2HG and other anticancer agents synergistically reduces leukemia and glioma cell growth, induces apoptosis in vitro, and blocks leukemia progression in mice (19). Treatment with FB23-2 also significantly suppresses leukemia progression in mice (Table 2; ref. 18).

CS1 and CS2 are the most potent small-molecule FTO inhibitors, effective at nmol/L concentrations (108). In addition to dramatically attenuating leukemia stem/initiating cell self-renewal, CS1 and CS2 reprogram immune response by reducing the expression of immune checkpoint genes, including LILRB4, sensitize leukemia cells to T-cell cytotoxicity, and overcome immune evasion. CS1 and CS2, therefore, show potent antileukemic efficacy in mouse models (108). In addition, CS1 and CS2 suppress glioblastoma, breast, and pancreatic cancer cell proliferation, and display potent anti-breast cancer efficacy in mouse models (Table 2; ref. 108).

The small-molecule ALKBH5 inhibitor, ALK-04, was identified through in silico screening of compounds, using the X-ray crystal structure of ALKBH5 protein and studying structure–activity relationship (17). Combination therapy with ALK-04 and GVAX/anti-PD-1 immunotherapy synergistically reduces melanoma tumor growth in mice (Table 2; ref. 17). The results provide a rationale for ALKBH5 inhibitor and immunotherapy combination therapy against melanoma.

Through compound library screening, the small molecule, BTYNB, has been identified as a potent and selective inhibitor of IGF2BP1 protein and c-Myc mRNA interaction (109). BTYNB destabilizes oncogenic c-Myc, E2F1, and eEF2 mRNAs, reduces c-Myc and E2F1 target gene expression, and cellular protein synthesis, and thus inhibits ovarian cancer and melanoma cell proliferation and tumor progression (Table 2; refs. 109, 110).

RNA m⁶A writers, erasers, and readers are critical regulators of pre-mRNA splicing, mRNA decay, stability, and translation in physiology and diseases, including cancer. While METTL3, FTO, YTHDC1–2, YTHDF1–3, and IGF2BP1–3 proteins have generally been shown to promote tumorigenesis, METTL3, METTL14, FTO, and ALKBH5 can promote or suppress cancer cell differentiation, proliferation, survival/apoptosis, migration, invasion, tumorigenesis, and metastasis. Similarly, METTL3, FTO, and ALKBH5 can promote cancer cell sensitivity or resistance to anticancer therapies.

It is currently not clear why both m⁶A writers and erasers are upregulated in human cancer tissues and can function as oncogenes for the same cancer types, such as breast cancer and leukemia, why the m⁶A writers METTL3 and METTL14 show opposite function in the same cancer types, such as colon and liver cancer, and why the same RNA m⁶A writers or erasers can function as oncogenes and tumor suppressors. Possible explanations include: (i) RNA m⁶A writers and erasers target different sets of mRNAs, oncogenes, or tumor suppressors, even in the same cancer type or cancer cell; (ii) RNA m⁶A modification proteins have different target mRNAs in different cancer cells even from the same organ due to heterogeneity in tumorigenic drivers; (iii) cellular context and different expression levels of RNA m⁶A writers, erasers, and their partner readers alter RNA m⁶A modification protein accessibility to oncogenic and tumor suppressive target mRNAs; and (iv) the corresponding proteins of the target mRNAs may have different functions due to mutations or variations in expression levels or subcellular localization.

In future studies, transcriptome-wide m⁶A RNA sequencing should be performed to examine target mRNA selectivity of RNA m⁶A writers, erasers, and partner readers in the same cancer cells and different cancer cells from the same or different organ origins; and to examine target mRNA selectivity of different levels of RNA m⁶A modification proteins in the same cancer cells and different cancer cells from the same or different organ origins. In addition, multifactor analysis of m⁶A writers, erasers, and their partner readers in particular cellular contexts is also likely to better illustrate the diverse functions of RNA m⁶A modification proteins.

Recent discoveries of METTL3 activators, and METTL3, FTO, and ALKBH5 inhibitors are likely to herald epitranscriptomic therapies. Small-molecule METTL3 inhibitors have shown promising anticancer effects in mouse models of leukemia (104), and the METTL3 inhibitor, STC-15, is expected to undergo clinical trials as early as 2021. The anticancer efficacy of the METTL3 activators (102) should be tested in sorafenib-resistant hepatocellular carcinoma models, in which METTL3 downregulation promotes sorafenib resistance (34). On the other hand, the FTO inhibitors, FB23-2, CS1, and CS2, significantly suppress leukemia progression in mouse models (18, 19), and the ALKBH5 inhibitor, ALK-04, significantly enhances the anticancer efficacy of GVAX/anti-PD-1 immunotherapy (17). Future endeavors should focus on developing more potent and selective small-molecule RNA m⁶A writer, eraser, and reader inhibitors and activators through chemical synthesis, structure-based virtual screening, and laboratory-based screening of small-molecule compound libraries. Their anticancer efficacy, safety profile, and pharmacokinetics should be investigated in vitro and in mouse models. Importantly, as RNA m⁶A writers, erasers, and readers play oncogenic or tumor suppressive functions, the anticancer efficacy of individual RNA m⁶A modulator inhibitor and activator needs to be examined in the suitable cancer types and under the appropriate context, such as chemotherapy-naïve or -resistant cancer cells and mouse models. Ultimately, ideal small-molecule RNA m⁶A modification protein inhibitors and activators appropriate for clinical trials are expected to be tested as monotherapy or in combination with other anticancer agents in patients with cancer tissues characterized by corresponding aberrant expression profile of RNA m⁶A modification proteins.

T. Liu reports grants from Cancer Council NSW and NIH during the conduct of the study and Neuroblastoma Australia outside the submitted work. No disclosures were reported by the other authors.

This work was partly supported by Cancer Council NSW, National Health & Medical Research Council Australia, National Natural Science Foundation of China, NIH (R21 CA226959), and Deutsche Forschungsgemeinschaft grants. P.Y. Liu is the recipient of a Cancer Institute New South Wales Fellowship. Children's Cancer Institute Australia is affiliated with UNSW Sydney and Sydney Children's Hospitals Network.