N¹-methyladenosine modification in cancer biology: Current status and future perspectives

Abstract

Post-transcriptional modifications in RNAs regulate their biological behaviors and functions. N¹-methyladenosine (m¹A), which is dynamically regulated by writers, erasers and readers, has been found as a reversible modification in tRNA, mRNA, rRNA and long non-coding RNA (lncRNA). m¹A modification has impacts on the RNA processing, structure and functions of targets. Increasing studies reveal the critical roles of m¹A modification and its regulators in tumorigenesis. Due to the positive relevance between m¹A and cancer development, targeting m¹A modification and m¹A-related regulators has been of attention. In this review, we summarized the current understanding of m¹A in RNAs, covering the modulation of m¹A modification in cancer biology, as well as the possibility of targeting m¹A modification as a potential target for cancer diagnosis and therapy.

1. Introduction

Over 170 distinct chemical modifications have been identified in RNA, which post-transcriptionally and extensively regulate the behaviors and biological functions of RNAs. Methylation is the most frequent internal modification in RNA, including N⁶-methyladenosine (m⁶A), N¹-methyladenosine (m¹A), N⁶,2′-O-dimethyladenosine (m⁶Am), 5-methylcytidine (m⁵C), 5-hydroxymethylcytosine (hm⁵C), N⁷-methylguanosine (m⁷G), and ribose-methylation (N_m). Among them, m¹A is a critical internal RNA modification that controls gene expression.

1) Discovery of m¹A methylation.

m¹A, which occurs on the first nitrogen atom of adenosine in RNA, is a ubiquitous RNA modification (Fig. 1). m¹A was first reported in 1961 [1]. Later, m¹A was found to exist in tRNAs [2], rRNAs [3], [4], mRNAs [5], [6], and long non-coding RNAs (lncRNAs) [7]. Among them, tRNA is the most heavy-modified class of RNA. In tRNA, N¹-methylation can occur on both purines (adenine and guanine), with m¹A occurring at a greater number of positions than m¹G [8]. In mRNA, m¹A is less prevalent compared with m⁶A. The occurrence of m¹A in mRNA is about 6-fold less than that of m⁶A in mRNA [9]. Interestingly, it was reported that the nitrogen atom could transfer from m¹A to m⁶A after the Dimroth rearrangement under an alkaline environment [10], suggesting the potential joint regulation of RNA via the dynamic interaction between m¹A and m⁶A.

Fig. 1.

M¹A methylation and its positions in tRNAs. (A) Chemical structure of the m¹A modification; (B) The m¹A methylation shown on a tRNA at reported modification sites (red). The domain in which the modification has been identified is indicated as A: archaea, E: eukaryotes, and B: bacteria. M indicates m¹A modifications found in mitochondria; (C) Location of m¹A modifications in human tRNA with tertiary structure. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

2) Characterizations of m¹A methylation in RNAs.

m¹A was found to be an ancient RNA modification across Bacteria, Archaea and Eukarya. It has been identified in various RNA transcripts including Table 1: i) tRNA: m¹A modification occurs at positions 9, 14, 22, 57, and 58 of cytoplasmic tRNAs (cyt-tRNAs), and positions 9 and 58 in mitochondrial tRNAs (mt-tRNAs) [8], [11], [12]. The m¹A at position 58 (m¹A58) is the first m¹A modification of the initial transcripts of tRNA in the nucleus [13], and it is the most conserved and common in bacteria, archaea, and eukaryotes [14], especially that of eukaryotic initiator tRNA^iMet [15]; m¹A14 is only found in a limited number of cytoplasmic cyt-tRNA^Phe of mammals [16]; m¹A22 occurs only in bacterial tRNAs [17]; m¹A57 exists as an intermediate to 1-methylinosine (m¹I) by hydrolytic deamination in archaea [18]; and m¹A9 has been found in cyto-tRNA from archaea and mammalian mt-tRNAs such as mt-tRNA^Lys and mt-tRNA^Asp [19], [20]. ii) rRNA: m¹A modification occurs in the bacterial and mitochondrial 16S rRNA, yeast 25S rRNA, and 28S rRNA in humans and mice [3], [21]. Among them, m¹A is conserved in position 947 of the mitochondrial 16S rRNA of vertebrates [22]; m¹A645 in 25S rRNA of budding yeast and m¹A1322 in 28S rRNA of mammals are conserved ³. iii) mRNA: over 900 transcripts containing m¹A modification were identified in eukaryotic cells [5]. The deposition of m¹A can occur along the mRNA, but most heavy-modified in the first splice site and the highly structured regions in the 5′ untranslated regions (UTRs) of mRNAs such as GC-rich regions [5]. Current reports suggest that each modified transcript contains on average one m¹A, and the ratio of m¹A/A is approximately 0.02 % in humans, which is relatively low compared to m⁶A/A in mRNA [5], [9], [23]. In mitochondria, one single m¹A site in ND5 mRNA is identified [24]. vi) lncRNA: recent study reveals that each m¹A methylated lncRNA had an average of 1.04 m¹A modification sites, where HGGAGRA and WGGANGA (H = A, U, or C, R = G or A) might be the m¹A motifs in lncRNA [7]. Nevertheless, m¹A in position 8398 of MALAT1 has been the only validated m¹A modification in lncRNA [24].

Table 1.

Overview of m¹A modifications in human RNAs.

RNA	Position	Functions
tRNA	9	Formation of cloverleaf structure; Stabilization of tRNA structure [35]
mt-tRNA	9	Formation of cloverleaf structure; Stabilization of tRNA structure [19], [36]
tRNAPhe	14	Unknown [16]
tRNA	58	Promotion of translation; Stabilization of tRNAiMet[15], [29]
mt-tRNA	58	Unknown [12]
mt-16S rRNA	947	Stabilization of mitochondrial ribosome [22]
28S rRNA	1322	Formation of ribosome 60S subunit [37]
mRNA	All segments	Promotion of translation [9]
ATP5D mRNA	A71 at Exon 1	Suppression of translation elongation [38]
mt-mRNA	CDS	Prevent the effective translation of modified codons [6]
ND5 mt-mRNA	1374	Unknown [24]
MALAT1 lncRNA	8398	Unknown [24]

3) Regulators of m¹A methylation in mammalian RNAs.

Similar to the dynamic modification of m⁶A, m¹A is installed by methyl-transferases called “writer” (TRMT6, TRMT61A, TRMT61B, TRMT10C, and NML), removed by demethylases called “eraser” (ALKBH1, ALKBH3, ALKBH7, FTO), and recognized by m¹A-binding proteins called “reader” (YTHDF1, YTHDF2, YTHDF3, YTHDC1) [21], [24], [25], [26], [27]. Here, m¹A-related enzymes in human are focused on (Fig. 2). i) Writer: TRMT61A contains a methyl donor (S-adenosyl-l-methionine, SAM) binding pocket and functions as the catalytic subunit. TRMT61A forms a functional heterotetramer complex with TRMT6, which lacks SAM binding motif and is essential for tRNA-binding [28], [29]. TRMT6/61A complex locates in the cytosol and seems to recognize targets in a structure-dependent manner. TRMT6/61A can methylate both cyto-tRNA m¹A58 (T-loop) and mRNAs with a T-loop-like structure, where with a GUUCRA tRNA-like motif⁶. TRMT61B methylases mitochondrial 16S rRNA [22], whilst recombinant TRMT10B can weakly methylate the A9 of tRNA^Asp(GTC) in vitro [30]. TRMT61B likely recognizes its tRNA and rRNA targets by a similar mechanism, where a weak consensus sequence (YMRAW) is revealed surrounding targets [22]. TRMT10C installed mt-tRNA m¹A and/or m¹G at position 9 coupling with SDR5C1 [31]. NML, also known as RRP8, locates in nuclei and methylases the m¹A on 28S rRNA [21]. ii) Eraser: ALKBH1 demethylates most m¹A in cyto-tRNAs, except tRNA^Phe(GAA) and tRNA^Sec(UCA), and m¹A58 in tRNAs is a major substrate of ALKBH1 [15]. ALKBH3 performs demethylation of both m¹A and m³C in tRNAs, as well as the m¹A demethylation in mRNA [5], [27]. ALKBH7, which locates in mitochondria, can demethylate the m¹A and m²²G in mitochondrial pre-tRNA^Leu1 and pre-tRNA^Ile in the nascent polycistronic mitochondrial RNA [32]. FTO, the first discovered RNA m⁶A demethylase, is found to bind multiple RNA species and perform demethylation including m⁶A and m⁶Am in mRNA and snRNA, and m¹A in tRNA [33]. iii) Reader: Four YT521-B homology (YTH) domain-containing proteins including YTHDF1, YTHDF2, YTHDF3, and YTHDC1 are recently identified as readers for m¹A modification, in spite of their roles of m⁶A readers [24], [26]. Their binding affinity with m¹A sites is weaker than that of m⁶A. YTH domain is presented in 174 different proteins and is evolutionarily conserved across the eukaryotic species [34], suggesting that the YTH domain is one of the sufficient structures for the recognition of methyl-group in RNAs. Research on their functions acting as m¹A readers remained limited.

Fig. 2.

M¹A modification of RNAs in mammalian cells.

2. Functions of m¹A in RNAs

The methyl group in m¹A carries a positive electrostatic charge under physiological conditions [39], which can disrupt the Watson-Crick base pairing with uridine. The electro-chemical interaction of m¹A-modified RNAs may have impacts on the processing, structural formation, and biological functions of RNAs, as well as their interactions with potential partner proteins.

1) Alteration of pre-RNA processing: m¹A58 can be found in both mt-tRNA and mt-dsRNA. Within the mitochondrial Leu1 pre-tRNA region, m¹A58 is partially methylated and can be demethylated by ALKBH7, whilst the m¹A9 level is not affected. Knockdown of ALKBH7 reduces the total mt-tRNA level through degradation of pre-tRNA and nascent mt-mRNA (mt-dsRNA) [32], suggesting that m¹A in mt-RNA regulates the nascent polycistronic mt-RNA processing.

2) Alteration of RNA structure: The distribution of m¹A in RNAs shows marked features, commonly around the GC-rich regions, such as the T-loop of tRNAs, within sequence GUUCNANNC in mRNAs [6], [9], and around sequences HGGAGRA and/or WGGANGA (H = A, U, or C, R = G or A) in lncRNAs [7]. m¹A is considered to alter the secondary and tertiary structures of RNA via its positive charge [7], [9]. It is worth noting that the GC-rich regions tend to form T-loop-like or hairpin structures within RNAs, which is almost invariable for the m¹A deposition. In addition, m¹A writer TRMT6/TRMT61A catalyzes m¹A at T-loop-like elements of RNA [24]. Therefore, m¹A is likely to play role in maintaining or stabilizing the T-loop-like structure, instead of changing the linear structure into a T-loop-like structure. The “loop” structure could be further strengthened by the presence of m¹A through its positive charge to impair Watson-Crick base pairing. In some cases, m¹A induces the correct folding of tRNAs. For instance, in mt-tRNA^Lys, m¹A9 shifts the dynamic equilibrium in favor of the cloverleaf structure and contributes to the correct formation of DHU loop [19], [36]; in the initiator methionine tRNA (tRNA^iMet), m¹A58 maintains the tRNA-like structure for tRNA^iMet through the unique A54–A58 interaction [29], [40]. m¹A modification is likely an early maturation event for tRNA.

3) Regulation of translation: The m¹A can affect both the initiation and elongation process of translation via modifications in tRNA, mRNA and rRNA. The regulatory effects on translation depend on the deposition of its targets. i) tRNA: m¹A modification in either tRNA or mt-tRNA shows a positive role in translation. For instance, overexpression of m¹A erasers ALKBH1, ALKBH3 or FTO significantly suppresses the translation initiation and translation efficiency [15], [33], [41], [42]. On one hand, m¹A58 stabilizes tRNA^iMet to enhance the translation initiation, and ALKBH1-catalyzed demethylation attenuated the translation initiation [15], [29]. On the other hand, the m¹A-modified tRNAs are preferentially recruited to active polysomes to promote translation [15]. In mt-tRNA, m¹A modifications in mt-tRNAs promote the translation of mitochondrial transcripts [43]. For instance, the level of m¹A58 in mt-tRNA^Lys strongly increases protein synthesis [44]. ii) mRNA: The modifications of m¹A have diversified effects on protein synthesis. m¹A located in 5′UTR of mRNA is associated with increased translation initiation and translation efficiency [9]. m¹A can disrupt RNA base-pairing and induce local RNA duplex melting, which destabilizes the secondary structure in the 5′ UTR to promote the initiation step [45]. In contrast, m¹A in the CDS of both mRNA and mt-mRNA interferes with translation [6], [24], [32]. The inhibitory effect of m¹A within the CDS is probably through a mechanism involving ribosomal scanning or the binding of releasing factor to modulate translation [45]. Our recent study revealed that knockdown of ALKBH3 can increase the binding between ATP5D mRNA and eRF1, which resulted in enhanced termination efficiency [38]. iii) rRNA: m¹A in rRNA is likely associated with the proper formation of the 80S initiation complex. Lacking yeast m¹A methylase Rrp8 causes the incompetent loading of the 60S on the 43S-preinitiation complex, which may be due to the conformational alteration of 25S rRNA in the absence of m¹A645 [37].

4) Regulation of RNA stability: The presence of m¹A might be involved in the structural thermostability of tRNAs. In yeast and mammals, tRNA m¹A58 is a major modification and is critical for the stability of tRNA^iMet. ALKBH1 deficiency improves the cellular level of tRNA^iMet by stabilizing tRNA^iMet [15]; hypomodified tRNA^iMet can be polyadenylated and then degraded by exonuclease Rex1p and exosome in yeast [46]. On the contrary, deficiency of m¹A writers has the possibility of inducing thermo-sensitivity of tRNAs [8]; knockdown of ALKBH1 fails to rescue tRNA from cleavage [47]; overexpression of ALKBH3 induces the formation of tRNA fragments (tRFs) [27]. It is reported that m¹A together with other post-transcriptional modifications can enhance the melting temperature of tRNAs [48]. Interestingly, certain nucleotide modifications including m¹A can be induced at high temperatures, and the biosynthesis of these modifications confers significant thermal stabilization to thus modified tRNAs [48]. However, m¹A58 in the mt-tRNA^Leu(UUR) does not appear to affect its stability [44]. In mitochondrial 16S rRNA, m¹A947 is likely to stabilize it [22]. Besides, m¹A is associated with the mRNA destabilization mediated by reader proteins YTHDF2 and YTHDF3 [49], [50], [51]. Together, the regulation of m¹A in the stability of RNA is far better studied. It is likely that m¹A plays a critical role in the stabilization of tRNAs, but its effect on the stabilization of other RNAs such as mt-tRNA and mRNA remains to be elucidated.

In the nucleus, m¹A modifications in pre-tRNA (as well as tRNA) and 28S rRNA are installed by m¹A writers TRMT6/TRMT61A and NML, respectively. m¹A-modified tRNA can be removed by erasers ALKBH1, ALKBH3 and FTO. In the mitochondrion, mt-tRNA and a subset of mt-mRNA are methylated by TRMT10C. TRMT61B methylates mt-16S rRNA and possibly mt-tRNA. m1A in mt-tRNA is erased by ALKBH7. mt, mitochondrial. In the cytoplasm, mRNA is methylated by TRMT6/TRMT61A and demethylated by ALKBH3. m¹A readers recognize m¹A and execute specific functions.

3. Modulation of m¹A modification in cancer biology

m¹A is associated with diverse cellular functions. Dysregulation of m¹A leads to diseases such as tumorigenesis, cardiovascular diseases, pulmonary diseases and Alzheimer’s disease [52], [53], [54]. Herein, the roles of m¹A in cancer biology are summarized Table 2.

Table 2.

Regulation of m¹A modifications in cancer biology.

Biological function	m1A regulator	RNA target	Effect	Cancer/Cell type	Mechanism
Proliferation	TRMT6, TRMT61A, ALKBH3	mRNA	Promotion	GI, HCC, Glioma	Modulation of PI3K/AKT pathway [56], [57], [61]
	TRMT6, TRMT61A	tRNAiMet	Promotion	Glioma	Upregulation of m [1]A58 in tRNAiMet[58]
	ALKBH3	tRNA	Promotion	PAAD, OC	Enhancing protein synthesis by demethylating tRNA [41]Generation of tRNA-derived small RNAs by demethylating tRNA [27]
Invasion	ALKBH3	mRNA	Promotion	BC, OC	Promoting translation of m1A-modified CSF-1 mRNA [51]
	ALKBH3	tRNA	Promotion	PAAD, OC	Generation of tRNA-derived small RNAs by demethylating tRNA [27]
	YTHDF3	mRNA	Suppression	Trophoblast	Promoting the mRNA decay of IGF1R mRNA [50]
Cancer metabolism	TRMT6, TRMT61A	tRNA	Promotion	HCC	Increasing PPARδ translation to stimulate cholesterol synthesis [62]
	ALKBH3, YTHDF1	mRNA	Promotion	OC	Increasing expression of ATP5D mRNA to promote the glycolysis [38]
Senescence, cell death	ALKBH3	mRNA	Promotion	NSCLC, BLCA	Increasing the expression of p27 and p21 [65], [66].
Tumor microenvironment	YTHDF3	mRNA	Promotion	RAW264.7	Regulating the macrophage M1 and M2 polarization [70],

1) Proliferation: m¹A regulators including TRMT6, TRMT61A and ALKBH3 have been found to promote the proliferation of cancer cells from gastrointestinal cancer (GI) [55], [56], hepatocellular carcinoma (HCC) [57], glioma [58], prostate cancer [41], [59], and colorectal cancer (CRC) [60]. In GI, HCC and glioma, both m¹A regulators and m¹A-modified mRNAs can modulate the PI3K/AKT pathway to promote the proliferation of cancer cells [56], [57], [61]. In addition to mRNA, m¹A can promote proliferation via the regulation of tRNAs. For instance, TRMT6/TRMT61A-elevated m¹A58 tRNA^iMet promotes the proliferation and malignant transformation of glioma [58]; ALKBH3 promotes cell proliferation through enhancing protein synthesis by demethylation of tRNA in prostate cancer [41]. Remarkably, our recent study shows that ALKBH3 can generate tRNA-derived small RNAs (tDRs) via demethylation of m¹A-tRNA, which further promotes cancer cell proliferation and invasion [27].

2) Invasion: Only a few studies have reported the regulation of m¹A on cell invasion. In ovarian and breast cancers, ALKBH3 induces the increased m¹A-modified CSF-1 mRNA to improve its translation initiation and cancer cell invasiveness [51]; ALKBH3 can also promote cancer cell invasion via destabilizing tRNAs [27]. YTHDF3, acting as m¹A reader, is found to suppress the invasion and migration of trophoblast, via promoting the mRNA decay of IGF1R [50].

3) Cell metabolism: In HCC, TRMT6/TRMT61A elevates a subset of m¹A-tRNA to increase PPARδ translation, resulting in the stimulation of cholesterol synthesis to promote the oncogenesis of HCC [62]. Our recent study reveals that ALKBH3 promotes the glycolysis of cancer cells by modulating the expression of m¹A-modified ATP5D mRNA [38]. In addition, ALKBH7 is found to regulate obesity by facilitating the utilization of short-chain fatty acids [63]. However, whether m¹A modification is involved in the ALKBH7-mediated regulation of obesity remains to be elucidated.

4) Senescence and cell death: m¹A might likely play role in the regulation of cellular senescence and cancer cell death, since ALKBH3 has been reported to regulate the cell cycle [64]. Knockdown of ALKBH3 results in the senescence induction and cell cycle arrest in lung cancer and urothelial carcinomas, by increasing the expression of cell cycle arrest proteins p27 and p21, and modulating NADPH oxidase and tweak/Fn14/VEGF signals [65], [66]. However, whether ALKBH3 mediates this function as an RNA or DNA methyltransferase was not elucidated.

5) Tumor microenvironment (TME): Studies shows that m¹A modification is critical to the shaping of immune microenvironment and the formation of TME complexity. In ovarian cancer (OC), colon cancer (CC), HCC and oral squamous cell carcinoma (OSCC), m¹A regulators and m¹A modification patterns correspond to the ever-changing immune microenvironment during cancer development [60], [67], [68], [69]. The m¹A-related phenotypes are correlated to the immune cell infiltration in TME, where different m¹A patterns are identified in immune-desert, immune-inflamed and immune-excluded phenotypes [60], [67]. In addition, eight m¹A regulators are positively correlated with activated mast cells, plasma cells and M1 macrophages in abdominal aortic aneurysm (AAA). Among them, YTHDF3 promotes macrophage M1 polarization but inhibits macrophage M2 polarization [70].

Increasing evidence highlights the important role of m¹A modification in the tumorigenesis of various cancers. Remarkably, some of these correlations might be due to the downstream effects or “side effects” of the global manipulation of m¹A-related regulators. To avoid such interference, site-specific manipulation of m¹A meets the urgent needs of m¹A methylation study. However, only two programmable m¹A tools have been reported [38], [71] and their usage in functional study remains to be popularized.

4. m¹A modification as a potential diagnostic and therapeutic target

The relevance between m¹A modifications and cancer progression and prognosis has been of attention recently. Increasing reports suggest that the levels of m¹A methylation, m¹A-related regulators, as well as the m¹A-related RNAs might be novel biomarkers for cancer prognosis. In addition, modulating the m¹A-related regulators and/or m¹A modification on transcripts is likely becoming a breakthrough in cancer therapies.

1) Targeting m¹A methylation for cancer prognosis:

i) Total m¹A level: Level of m¹A can be detected in spontaneous urine samples. Zheng et al., first reported that the m¹A content in CRC patients is higher than that in healthy controls. And m¹A levels are positively associated with the Duke stage for CRC [72]. Later, m¹AScore is generated to quantify the individual patient’s m¹A modification pattern. In CC and OSCC, patients with a high m¹AScore show more lymphatic invasion, shorten survival and worse response after immunotherapy [60], [68]. However, a high m¹AScore in OC patients is usually accompanied by a better survival advantage, a lower mutational load and marked therapeutic benefits from chemotherapy and immunotherapy [67]. The m¹A level shows the potential predictive ability for the prognosis of cancers, but the relevance between m¹A level and cancer progression and/or prognosis is likely related to tumor species.

ii) m¹A-related regulators: m¹A-related regulators are commonly high expressed in multiple cancers. Among them, m¹A writer TRMT6, TRMT61A and eraser ALKBH3 are associated with poorer prognosis in a variety of cancers [41], [51], [65], [66], [73], [74], [56], [57], [59], [60], [61]. Expression of TRMT61B and TRMT10C, which are m¹A writers for mitochondrial RNAs, could be used as an independent risk factor for the prognosis of gastric cancer and HCC, respectively [55], [57], [75]. ALKBH7, which is high expression in 17 cancers and low expression in 5 cancers compared to paired normal tissues, might be a potential biomarker for pan-cancer prognosis and prediction of therapeutic outcomes [76]. Remarkably, ALKBH1 is the only m¹A regulator that is related to the good prognosis of patients with pancreatic cancer (PAAD) [77]. In glioblastoma, ALKBH1 expression of patients is not associated with overall survival rates [78]. For FTO, YTHDF1, YTHDF2, YTHDF3 and YTHDC1, high expression of them is closely associated with poor clinical outcomes in both CRC and HCC [57], [60], [79]. However, all of them are not restricted in the regulation of m¹A, but also m⁶A and other modifications. Therefore, their prognostic potential may be associated with m⁶A as well.

iii) m¹A-related RNAs: m¹A-related RNAs might become effective prognostic predictors for cancer diagnosis. On one hand, Zheng et al., reveal that m¹A-regulating genes are correlated with advanced clinical stages and poor prognosis of PAAD [77]. On the other hand, tRFs derived from m¹A-regulated tRNAs are correlated with the dysregulation of TRMT6/61A expression, and long 3′ tRFs are highly enriched in bladder cancer [80]. Our study further confirms that the plasma level of 5′-tRF-GlyGCC acts as a novel biomarker for colorectal cancer diagnosis [81].

2) m¹A modification for cancer therapy.

i) Inhibitors targeting m¹A-related regulators: Due to the positive roles of m¹A-related regulators in tumorigenesis, developing selective inhibitors targeting m¹A-related regulators is one of the efficient ways for cancer therapy. So far, inhibitors targeting TRMT6/TRMT61A, ALBKH3 and FTO are reported (Table 3). For TRMT6/TRMT61A, thiram (tetramethyl thiuram disulfide) is recently identified as a potential candidate compound that selectively inhibits the interaction between TRMT6 and TRMT61A. Thiram treatment suppresses the self-renewal of HCC cells in vitro and also decreases tumor growth in vivo [62]. However, thiram was approached by FDA as an antimicrobial drug and shows liver damage in chickens at high doses [82]. The pre-clinical investigation of HCC therapy needs further evaluation. For ALKBH3, HUHS015 and its derivative Compound 7I (1-(5-fluoro-1H-benzimidazol-2-yl)-3-methyl-4-phenyl-1H-pyrazol-5-ol) are evaluated as prostate cancer antigen-1 (PCA-1/ALKBH3) inhibitors. Compound 7I shows potent antiproliferative effects on DU145 cells in vivo [83], [84]. For FTO, as the first discovered RNA methylase, over ten FTO-targeted small molecule inhibitors are developed, including Rhein, CHTB, N-CDPCB, Meclofenamic acid 2 (MA2), R-2-hydroxyglutarate (R-2HG), FB23-2, CS1, CS2, 18,077 and 18,097 [85], [86]. Among them, seven FTO inhibitors, MA2, R-2HG, FB23-2, CS1, CS2, 18,077 and 18,097 show anti-cancer effects both in vitro and in vivo. MA2, which can bind at the surface area of the FTO active site, is an FDA-approved nonsteroidal anti-inflammatory drug (NSAID) [87]. MA2 shows an inhibitory effect on tumor growth in glioblastoma-bearing mice [88], which likely becomes the most potent small molecule FTO inhibitor for anti-tumor therapy.

Table 3.

Inhibitors targeting m¹A-related regulators.

Target	Compound	Mechanism	Effect on cancer
TRMT6/TRMT61A	Thiram	Inhibits TRMT6-TRMT61A interaction	HCC: Suppress self-renewal and tumor growth [62]
ALKBH3	HUHS015/Compound 7I	Suppress ALKBH3 activity	PC: Inhibit proliferation of DU145 cells [83], [84]
FTO	18077/18097	Bind to the active site of FTO	BC: Inhibited cell cycle process and migration of cancer cells [86]
	CHTB	Bind to a novel site of FTO and inhibit demethylase activity [89]	−
	CS1/CS2	Block the catalytic pocket of FTO	AML: Inhibits cancer cell proliferation, cancer stem cell self-renewal and immune evasion [90]
	Dac51	Suppress FTO activity	Melanoma: Promotes T cell response and enhances the anti-PD-1 therapy [91]
	Entacapone	Compete with the catalytic site of FTO	HCC: Interfere with gluconeogenesis through FOXO1 [92]
	FB23/FB23-2	Directly bind to FTO and inhibit demethylase activity	AML: Suppress proliferation and promote differentiation/apoptosis [93].
	FTO-04	Block the catalytic activity of FTO	GBM: Prevents neurosphere formation and self-Renewal [94].
	N-CDPCB	Competitive bind to a novel site of FTO and inhibit demethylase activity [95]	−
	MA/MA2	Inhibition of the catalytic activity of FTO	GBM: Inhibit GSC cell growth and self-renewal [88]
	MO-I-500	Inhibition of the catalytic activity of FTO	BC: Inhibit cell survival via decreasing FTO and IRX3 proteins [96]
	R-2HG	Suppress FTO activity	AML, GBM: Inhibit proliferation/survival of FTO-high cancer cells [97]
	Rhein	Competitive bind to FTO catalytic domain	AML: Prevent or override tyrosine kinase inhibitor resistance [98] BC: Decrease tumor growth [99]
	Saikosaponin-d	Occupy the substrate-binding site of FTO	AML: suppress cell proliferation, promote apoptosis/cell-cycle arrest [100]

ii) Editing m¹A-modified RNAs: m¹A modification on target RNA regulates its expression. Therefore, targeted editing of m¹A in specific RNA becomes a possible approach for cancer therapy. CRISPR proteins recognizing RNA as substrates, such as Cas13 and CasRx, are powerful tools to establish editing systems targeting RNA modification. Xie et al., developed a targeted m¹A demethylation tool called “REMOVER” via combining the RNA-targeting capability of dCasRx (catalytically inactive RNA-targeting RfxCas13d enzyme) and m¹A eraser ALKBH3. “REMOVER” can demethylate m¹A of specific RNA such as MALAT1 and modulate their stability⁷¹. However, no further application of “REMOVER” has been reported yet. Recently, we establish a targeted m¹A demethylation system “dm¹ACRISPR” by fusing PspCas13b with ALKBH3 [38], which was based on our previous method to demethylase m⁶A [101]. Targeted demethylation of m¹A in ATP5D mRNA, which is a key regulator for glycolysis, by “dm¹ACRISPR” significantly suppresses the energy metabolism and tumorigenesis of cancer cells [38], indicating the possibility of novel cancer therapy via modulating specific m¹A modification. Similar to CRISPR/Cas9 which has shown great progress and potential in tumor therapy, site-specific manipulation strategy on m¹A exhibits anticipated potential as well. In addition, the m¹A modification pattern is closely associated with the immune cell infiltration in TME. Therefore, combining the targeting m¹A-editing and immune response therapy such as PD-L1, might be another promising approach for cancer therapy in the future.

5. Challenges and perspectives

m¹A methylation is one of the most prevalent modifications that are highly conserved in both eukaryotic and prokaryotic RNAs, suggesting its critical role in living body. Maintaining the structure/folding of RNA is considered to be the most important function of m¹A, especially in tRNAs that regulate protein biosynthesis. There are over 90 modifications in tRNAs, and about 13 modifications per tRNA are presented on average [8], [102]. Modifications in tRNA greatly limit the functional study of one particular modification in tRNAs. Therefore, whether m¹A would cooperate with other modifications in tRNA to shape the tRNA folding, whether there are other unrevealed m¹A sites in tRNA, and what the exact function of m¹A in tRNAs are questions that need to be elucidated. In addition to tRNA, m¹A is presented in a low number of mRNAs, typically at low stoichiometries. The low stoichiometry of m¹A has important implications for how m¹A could affect mRNA. It is commonly believed that m¹A regulates the expression of targets, and it acts as a gene expression regulation mechanism in response to changes in the external environment. However, this is worth thinking about the impact of m¹A, which with low stoichiometry in mRNAs, on the regulation of target expression.

Despite the to-be-elucidated function of m¹A on RNAs, the regulatory effects of m¹A in cancer development have been confirmed in the past decade. It is confirmed that the m¹A level, m¹A modification pattern, and the expression of m¹A-related regulators are closely associated with tumorigenesis. Therefore, it is promising to target m¹A and its regulators for cancer diagnosis and therapies. Targeting epigenetic modification is a new and challenging development direction for cancer diagnosis and therapy. For instance, there is no epigenetic drug targeting RNA modification has been approached currently, although thousands of candidates are under clinical evaluation. So far, there are twelve potential chemicals targeting m¹A-related regulators that show suppressive effects on cancer cells, meaning that the anti-cancer drugs targeting m¹A are still in the initial stage of research. In addition to small molecule inhibitors, targeted m¹A editing is a novel approach to cancer therapy. CRISPR/Cas9 system is a powerful tool for genetic disease treatments. Similar to CRISPR/Cas9, RNA-targeting CRISPR shows great capability in transcript editing, with further advantages such as no DNA mutation inheritance caused by the off-target effect. However, challenges about CRISPR delivery and its editing effect in solid tumors remain to be solved.

6. Conclusions

In this review, we summarized the functions and the modulation of m¹A modification in cancer biology. Acting as one of the most abundant internal modifications of RNAs, the specific biological functions of m¹A modification in RNAs, including tRNA, rRNA, mRNA and lncRNA, remains largely unknown. Increasing evidence suggests a critical role of m¹A modification in cancer development. On one hand, m¹A-related regulators can modulate the progression of cancers; on the other hand, m¹A level and/or m¹A-related regulators might be novel biomarkers or therapeutic targets for cancers. A full understanding of the mechanism underlying m¹A modification is distant. However, with the in-depth study of m¹A modification, including its functions and association with cancer progression, targeting m¹A modification might provide new potential options for both cancer diagnosis and treatment.

Funding

The authors’ research on RNA modifications is currently funded by the National Key Research and Development Program of China (No. 2022YFC2601800), the National Natural Science Foundation of China (Grant Nos. 32161143017, 82173833, 82173126, and 81973343), The International Cooperation Project of the Science and Technology Planning Project of Guangdong Province, China (No. 2021A0505030029), the Open Program of Shenzhen Bay Laboratory (No. SZBL202009051006), the Fundamental Research Funds for the Central Universities (Sun Yat-sen University) (Nos. 19ykzd24, 19ykpy130), the Guangdong Provincial Key Laboratory of Chiral Molecule and Drug Discovery (2019B030301005), the Guangdong Basic and Applied Basic Research Foundation (No. 2020A1515010290), and Shenzhen Bay Scholars Program.

CRediT authorship contribution statement

Jiexin Li: Conceptualization, Investigation, Writing – original draft. Haisheng Zhang: Investigation, Writing – original draft. Hongsheng Wang: Conceptualization, Investigation, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.