Introduction

Chemical modifications of nucleobases, like methylation of adenine and cytosine, are critical for controls of gene expressions on different levels, which are important for cells to adapt to the environment or to develop to complex organisms from a single cell [1]. RNA methylation has been found in various RNAs including messenger RNA (mRNA) [2, 3], microRNA (miRNA) precursor [4] and long non-coding RNA (lncRNA) [5] transfer RNA [6,7,8,9], ribosomal RNA [10], small nuclear RNA [11]. The N6-methyladenosine (m6A) is one of the most common and abundant internal modifications on RNA molecules, which has been identified in eukaryotic mRNA [12,13,14] and virus nuclear RNA [15, 16] since 1970s. m6A modification is involved in almost all the stages of RNA life cycle, including RNA transcription, exporting through nuclear, translation and degradation [17,18,19,20], and has gained more and more attractions.

The adenosine methylation occurs preferentially within the degenerate consensus RRACH (R = G or A; H = A, C or U) in gene coding regions and 3′UTRs [21] or broader consensus motif DRACH (where D denotes A, G or U) [22]. The m6A modification has been considered as an epigenetic regulation similar to DNA and histone modifications, performing critical functions in important biological processes [23,24,25]. Accumulating evidences have supported the fact that the m6A RNA methylation impact on the initiation and progression of human cancers through diverse mechanisms [26,27,28]. It has been reported that the m6A RNA methylation is related to different processes, including stem cell differentiation and motility [29], immune response [30, 31], cellular stress [6], tissue renewal and pathology [32] and viral infection [33]. In the following sections, the m6A modification and its functional mechanism, and the m6A-associated modification landscapes in several major cancers are reviewed. The detection methods of m6A modification and the potential clinical application for RNA m6A modification are also discussed.

Survey methodology

In this review article, three steps of journal article searching were performed. First, the functional mechanism of RNA m6A methylation was identified using the search terms “RNA N6-methyladenosine” and terms “m6A writers, erasers, readers”. Moreover, terms “methyltransferase”, “demethylases” and “m6A-binding proteins” were also searched respectively. Then the functions of the RNA N6-methyladenosine modification in the cancer progression were further investigated by searching “RNA N6-methyladenosine” together with “cancer” and searching “RNA N6-methyladenosine” together with certain cancers. Finally, the development of the m6A modification detection was investigated by using the terms “m6A modification detection”. All relevant studies were included regardless of the year of publication. The research articles that were published before August 2018 based on the above search criteria were collected. In this publication survey, the PubMed databases were explored.

Functional mechanism of RNA m6A methylation

m6A modifications are dynamic and reversible, which is created by proteins called m6A “writers” and can be reversed by m6A “erasers” (Fig. 1). Some other proteins recognizing and binding m6A-containing mRNA serve as “readers” and regulate downstream molecular mechanisms accordingly.

Fig. 1
figure 1

Mechanism of m6A-based modification of RNA. m6A RNA methylation is regulated by its “writers,” “erasers,” and “readers.” Writers refer to the m6A methylase complex including METTL3, METTL14 and WTAP. Erasers are m6A demethylases involving FTO and ALKBH5. Readers are proteins that bind to m6A modifications, including YTH-domain containing proteins (YTHDF1, YTHDF2, YTHDF3, YTHDC1, YTHDC2), eIF3 and nuclear m6A reader protein, which can exert various functions in almost all stages of RNA life cycle. ALKBH5 a-ketoglutarate-dependent dioxygenase alkB homolog 5, eIF3 eukaryotic initiation factor 3, FTO fat mass and obesity-associated protein, m6A N6-methyladenosine, METTL3 methyltransferase like 3, METTL14 methyltransferase like 14, WTAP Wilms’ tumour 1-associating protein, YTH YT521B homology

Full size image

m6A writers

The RNA methylation modification is catalyzed by the m6A writer complex [34, 35]. Methyltransferase like 3 (METTL3, also known as MT-A70) and methyltransferase like 14 (METTL14) both belong to the class I methyltransferase family30 with methyltransferase domains and efficiently catalyses methyl group transfer by forming a core catalytic complex [36, 37]. METTL3 appears to be a predominantly catalytic enzyme with a function reminiscent of N6-adenine methyltransferase systems [38], while METTL14 was reported to be a pseudomethyltransferase to stabilize METTL3 and recognize target RNA [39]. Wilms’ tumour 1-associating protein (WTAP) is a major regulatory component of the m6A methylation complex by interacting with METTL3 and METTL14 and helping their localization into nuclear speckles [40]. The METTL3 catalytic process has been considered as a common m6A pathway and modifies most m6A sites in mRNA [41]. Generally, the m6A writer complex can exert their catalytic function with or without other factors.

Additional subunits, like KIAA1429 (also known as vir-like m6A methyltransferase associated, or VIRMA) [42], RBM15 and its paralog RBM15B [43] are also associated with the methylation complex. Another methyltransferase METTL16 was recently found to be a m6A-forming enzyme in mRNA and mediate m6A formation in snRNA U6 by regulating methionine adenosyltransferase 2A intron retention in response to intracellular S-adenosylmethionine levels [44]. Meanwhile, the SpoU-TrmD RNA methyltransferase superfamily (including TrmH and TrmD) [45] and NSun6 [46] were reported to catalyze the transfer of the methyl group to tRNA.

m6A erasers

m6A can be demethylated by m6A demethylases like fat mass and obesity-associated protein (FTO) [47, 48] and a-ketoglutarate-dependent dioxygenase alkB homolog 5 (Alkbh5) [49, 50], which are described as m6A erasers. Both FTO and ALKBH5 belong to the AlkB family of nonheme Fe(II)/2-oxoglutarate dioxygenase. FTO is first shown as a human obesity susceptibility gene and was found to influences RNA processing through dynamic demethylation of m6A [51, 52]. ALKBH5 has catalytic domain that can demethylate both single stranded RNA (ssRNA) and single-stranded DNA (ssDNA), and specially catalyze the demethylation of m6A in ssRNA, supporting the reversible m6A modification on RNA [50, 53].

m6A readers

The m6A functions mainly with recruiting m6A-binding proteins. It can be recognized by proteins containing a YT521B homology (YTH) domain or by the translation initiation factor eukaryotic initiation factor 3 (eIF3) [41, 54]. The YTH domain characterized by 14 invariant residues within an α-helix/β-sheet structure recognizes m6A by a conserved aromatic cage sandwiched between a Trp and Tyr residue and with the methyl group pointed to another Trp residue [55, 56].

There are five proteins that contain the YTH domain in human cells, including YTH domain family (YTHDF1–3), YTH domain-containing 1 (YTHDC1) and YTH domain-containing 2 (YTHDC2). Among these proteins, YTHDF1–3 are cytoplasmic m6A-specific binders and highly similar to each other with a C-terminally located YTH domain [57]. Structural basis analysis revealed that YTHDF1 can recognize m6A without sequence preference and facilitates the translation of m6A-modifed transcripts [58]. Compared with YTHDF1, YTHDF2 was identified to selectively bind to mRNAs (also non-coding RNAs) with a conserved core motif of G(m6A)C, accelerating the decay of m6A-modifed RNAs [57, 59, 60]. YTHDF3, another member of the cytoplasmic YTH family, was reported to facilitate the translation of targeted mRNAs and affect methylated mRNA decay in synergy with YTHDF1, by binding to m6A-modifed mRNAs and interacting with 40S and 60S ribosome subunits [61, 62].

In contrast to the YTH domain family members, YTHDC1 is predominantly located in nuclear harboring a distinctly selective binding pocket of GG(m6A)C sequence. It prefers guanosine and disfavors adenosine at the position preceding the m6A nucleotide modification [58, 63]. YTHDC2 is a nucleocytoplasmic protein containing a DEAD-box RNA helicase domain. Recent studies have shown that YTHDC2 has a RNA-induced 3′ → 5′ RNA helicase activity and regulates m6A transcripts by recruiting the 5′/3′ exoribonuclease XRN1 via Ankyrin repeats [64, 65].

eIF3 has been considered as a m6A-binding protein as well. In the absence of the cap-binding factor eIF4E, mRNAs containing 5′ UTR m6A can be translated in a cap-independent manner by recruiting the 43S complex to initiate translation via directly binding to eIF3 [66]. However, unlike the YTH proteins which recognize m6A through a specific well-characterized YTH domain, the exact mechanism of eIF3-m6a recognition is not clearly understood yet. It probably relies not only on the sequence motif but also on adjacent RNA structures [67].

Another protein, heterogeneous nuclear ribonucleoprotein A2/B1 (HNRNPA2B1), has been identified as a nuclear reader of m6A, binding to RRAC containing sites on nuclear RNAs [68]. HNRNPA2B1 regulates the alternative splicing of exons, which is highly correlated with METTL3 [68]. Moreover, HNRNPA2B1 interacts with the pri-miRNA microprocessor complex component DGCR8 and modulates the nuclear processing of pri-miRNA [68].

m6A-mediated regulation of human cancer progression

Errors in the m6A pattern are associated with most types of cancers. Increasing studies show that m6A levels are related to cancer progression. As an example, m6A regulates the cancer-related RNA fate and functions including alternative splicing of pre-mRNA and the processing of miRNA and lncRNA [4, 69,70,71]. Relations between m6A and cancer progression in line with different cancer types are reviewed as follows.

Acute myeloid leukemia

Acute myeloid leukemia (AML), characterized by recurring chromosomal aberrations, gene mutations, and epigenetic modifications, is one of the most common cancers of the hematopoietic system [72, 73]. It has been proved that m6A modifications substantially contribute to the phenotype of leukemia cells [39, 74]. A strong association of mutations and/or copy number variations of m6A regulatory genes with the presence of TP53 mutations was concluded in AML patients by Cancer Genome Atlas Research Network datasets analysis [75]. Genetic alterations in m6A modifiers including METTL3, METTL14, YTHDF1, YTHDF2, FTO, and ALKBH5 together with TP53 and/or its regulator/downstream targets confer a worse survival with prognostic roles [75]. m6A demethylase FTO showed a significantly high expression in AMLs with t(11q23)/MLL-rearrangements, t(15;17)/PML-RARA, FLT3-ITD and/or NPM1 mutations and exerts its oncogenic role by regulating reducing the m6A levels of hematopoiesis regulators ASB2 and RARA transcripts [74]. Then METTL14, a m6A methyltransferase, was also found highly expressed in normal hematopoietic stem/progenitor cells (HSPCs) and t(11q23), t(15;17), or t(8;21) translocation subtype of AML and is decreased during myeloid differentiation. METTL14 suppresses HSPCs differentiation and promotes leukemogenesis by regulating the stability and translation of its mRNA targets via m6A modification [39]. It is very interesting that the elevated expressions of both writer and eraser may balance the perturbation of RNA m6A methylation contributing to AML progress. METTL14 appears to account for the stability of its mRNA target MYC through regulating m6A abundance mainly on the 3′-terminal exon [39], while FTO affects mRNA stability by inhibiting YTHDF2-mediated RNA decay via decreasing m6A abundance on the mRNA 5′-terminal and internal exons [76]. These data indicate that aberrant m6A modifiers on different transcript regions lead to distinct effects, probably due to the recognition of different readers. The underlying mechanisms need to be elucidated in further studies.

Cervical cancer

Cervical cancer is one of the most prevalent gynecological malignancies worldwide with poor prognosis [77]. In recent years, researchers have uncovered the important roles of RNA methylation in cervical cancer development. Dot blot assay results showed that the m6A levels were significantly decreased in cervical cancer samples compared with the adjacent normal tissues. And the reduction of m6A levels was significantly correlated with tumor progression (including tumor stage, tumor size, differentiation, lymph invasion) and cancer recurrence [78]. The m6A levels of certain genes mRNA also relate to the therapy resistance of cervical cancer. For example, m6A demethylase FTO, which shows higher expression in cervical squamous cell carcinoma tissues than respective adjacent normal tissues, enhances the chemo-radiotherapy resistance of cervical cancer by regulating the m6A levels of β-catenin mRNA transcripts [79]. The critical functions of m6A modifiers and their substrates in the regulation of chemoradiotherapy resistance may bear potential clinical implications for cervical cancer treatment.

Glioblastoma

Glioblastoma is characterized with stem-like cells at the apex of cellular hierarchies, which are self-renewing, resistant to conventional therapy, and sustaining long-term tumor growth leading to tumor recurrence [80]. It has been reported that m6A demethylase ALKBH5 is highly expressed in glioblastoma stem-like cells and can predict poor patient prognosis for glioblastoma patients [81]. Integrated transcriptome and m6A-seq analyses revealed that the nascent transcripts of transcription factor FOXM1 are substrates of ALKBH5 and can be demethylated by ALKBH5, causing an enhanced FOXM1 expression and reinstating the inhibition of tumor growth [81, 82]. Another study indicated that RNA methyltransferase METTL3 and METTL14 significantly promote the growth and self-renewal of glioblastoma stem-like cells by regulating the mRNA expression of glioblastoma related genes ADAM19 through mRNA m6A enrichment [83]. These data suggest a promising therapeutic tool for glioblastoma by targeting the m6A mRNA methylation.

Breast cancer

Breast cancer is a leading cause of cancer-related death and has a high incidence for women worldwide [84]. Recent studies have reported that the demethylation of NANOG mRNA in breast cancer cells can be induced by the exposure of hypoxia [85]. NANOG, which is a pluripotency factor promoting the breast cancer stem cell specification, is highly expressed due to the demethylation of m6A in its mRNA mediated by ALKBH5 [85]. Except for NANOG, another pluripotency factor KLF4 is also stimulated. In hypoxia condition, KLF4 is m6A demethylated in its mRNA and promotes the specification of breast cancer stem cell [86]. m6A RNA methylation can be also affected by anticancer chemicals. Sulforaphane, an isothiocyanate, is a dietary phytochemical and can promote genetic instability by diminishing m6A RNA methylation in breast cancer cells [87].

Liver cancer

Diverse and reversible m6A modification on RNAs emerges as a new layer of regulation in liver cancer. m6A modifiers YTHDF1 and METTL3 were reported to be significantly up-regulated in liver cancer and the overexpression of YTHDF1 and METTL3 are associated with poor prognosis of patients [88, 89]. By using transcriptome sequencing RNA m6A sequencing and m6A methylated RNA immuno-precipitation quantitative reverse-transcription polymerase chain reaction, METTL3 was identified to promote liver cancer progression through YTHDF2-dependent posttranscriptional silencing of cytokine signaling 2 by m6A modification [88].

Other kinds of cancers

Except for those mentioned above, m6A was also identified to regulate the progression of a variety of cancers, including nasopharyngeal carcinoma [90], prostate cancer [91], colorectal cancer [92], pancreatic cancer [93], lung cancer [94], gastric cancer [95], and renal cell carcinoma [96]. m6A in circulating tumor cells from lung cancer patients were discovered with a significant increase compared with whole blood cells, demonstrating that methylated RNA in body fluids may serve as biomarkers for lung cancer [94, 97]. In nasopharyngeal carcinoma, the alternative splicing of suppressor of variegation 3–9 homolog 1, a histone methyltransferase, is promoted by a natural compound baicalin hydrate via the enhancement of m6A RNA methylation [90]. m6A writer METTL3 acts as a tumor suppressor in renal cell carcinoma, promoting cell proliferation, migration and invasion and inducing G0/G1 arrest [96], whereas METTL3 only shows no morphologic and proliferative effects in pancreatic cancer but affects the sensitivity to anticancer reagents such as gemcitabine, 5-fluorouracil, cisplatin and irradiation [93]. METTL3 imparts radioresistance in glioma stem-like cells through SOX2-dependent enhanced DNA repair and enhanced sensitivity to γ-irradiation was observed when METTL3 was silenced [98]. m6A readers YTHDF1 and YTHDF2 have ontogenetic roles in promoting cell proliferation and migration by regulating m6A levels in colorectal cancer [92] and prostate cancer [91].

Detection of m6A modification

To better understand the important roles of m6A modification, more and more methods are developed for profiling precise m6A sites and locations. The crucial mechanism of m6A modification are further identified.

An affinity-based sequencing approach methylated RNA immunoprecipitation sequencing (MeRIP-Seq, also known as ‘m6A-seq’) has been developed and applied to survey the global mRNA m6A localizations [21, 99, 100]. In MeRIP-seq, mRNAs are first fragmented into approximately 100-nucleotide long fragments and then are immunoprecipitated with anti-m6A antibody followed by high throughput sequencing to form immunoprecipitation sequencing (IP) samples. In order to measure the background mRNA abundance for the IP experiments, input samples are additionally used, which are generated by sequencing the un-immunoprecipitated mRNA fragments. Finally, the reads enrichment in IP and input are assessed to predict m6A sites. Algorithms needed for MeRIP-seq-based m6A peak detection were subsequently developed. For example, the R/Bioconductor package exomePeak [101], the open source R package MeTCluster [102] and the graphical model-based MeTPeak [103]. By combining m6A-specific methylated RNA immunoprecipitation with high throughput deep sequencing, MeRIP-Seq has the potential to detect the transcriptome-wide distribution of m6A modifications. If the high-throughput assay is not available or necessary, a simple method based on high resolution melting analysis was introduced by Dontsova’s group, which allows a rapid and easy screening for the m6A presence at specific RNA position using total RNA sample and qPCR machine [104].

The mass spectrometry (MS) methodology, which has been used to identify posttranscriptional modifications of proteins, can also be developed to characterize the modified RNA. A promising alternative approach for RNA methylation is top-down MS, based on the RNA ionization in electrospray ionization, backbone cleavage in collisionally activated dissociation and electron detachment dissociation affected by specific nucleobase methylations [105]. However, samples for MS approaches need to be well prepared. By establishing a combined strategy for cell lysis, nucleic acids digestion, and nucleosides extraction in one-tube, Yuan’s lab successfully determined both DNA and RNA methylation in circulating tumor cells using liquid chromatography-electrospray ionization-tandem MS [97].

Different computational tools were also developed for the characteristics of m6A-associated variants, providing a resource for deeper analysis of m6A modifications and cancer disease at the epitranscriptomic layer of gene regulation. m6A-Driver, an algorithm developed by Huang’s group, are used for predicting m6A-driven genes and associated networks [106]. The m6A-driven network was built by integrating the Protein–Protein Interactions network and the predicted differential m6A methylation sites from MeRIP-Seq data using an algorithm called Random Walk with Restart [106]. These m6A-driven genes are likely to be actively modulated by m6A methylation under a specific condition, underlying the dynamic interactions between the methyltransferases and demethylase at the epitranscriptomic layer of gene regulation. Recently, Ren’s group reported a web server called ‘m6ASNP’ (http://m6asnp.renlab.org) [107] and a comprehensive database ‘m6Avar’ (http://m6avar.renlab. org) [108], which can be dedicated to identifying the m6A-associated genetic variants and their targeting m6A modification sites. These tools provided a useful resource for annotating genetic variants involved in m6A modification.

Conclusions

Given that m6A patterns in RNA transcripts play important roles in a variety of cancers, the m6A modification has great potential for clinical application by serving as diagnostic/prognostic targets or implicating an effective treatment strategy. There are also relevant studies focused on the inhibitors of RNA modifiers, giving clues for the rational design of potent and specific m6A inhibitors in medicine use. A natural product rhein was selected to competitively bind to the FTO active site by structure-based virtual screening and biochemical analyses and showed inhibitory effects on the activity on m(6)A demethylation [109]. Another FTO inhibitor, meclofenamic acid, which was approved by US Food and Drug Administration as a nonsteroidal anti-inflammatory drug, can compete with FTO for binding the m6A-containing nucleic acid [110]. Studies on RNA m6A patterns and their selective inhibitors will pave the way for research into epitranscriptomics in chemical biology and shed light on the discovery and development of m6A-specific probes and drugs.