The RNA Modification N⁶-methyladenosine and Its Implications in Human Disease

Abstract

Impaired gene regulation lies at the heart of many disorders, including developmental diseases and cancer. Furthermore, the molecular pathways that control gene expression are often the target of cellular parasites, such as viruses. Gene expression is controlled through multiple mechanisms that are coordinated to ensure the proper and timely expression of each gene. Many of these mechanisms target the life cycle of the RNA molecule, from transcription to translation. Recently, another layer of regulation at the RNA level involving RNA modifications has gained renewed interest of the scientific community. The discovery that N⁶-methyladenosine (m⁶A), a modification present in mRNAs and long noncoding RNAs, can be removed by the activity of RNA demethylases, launched the field of epitranscriptomics; the study of how RNA function is regulated through the addition or removal of post-transcriptional modifications, similar to strategies used to regulate gene expression at the DNA and protein level. The abundance of RNA post-transcriptional modifications is determined by the activity of writer complexes (methylase) and eraser (RNA demethylase) proteins. Subsequently, the effects of RNA modifications materialize as changes in RNA structure and/or modulation of interactions between the modified RNA and RNA binding proteins or regulatory RNAs. Disruption of these pathways impairs gene expression and cellular function. This review focuses on the links between the RNA modification m⁶A and its implications in human diseases.

The N⁶-methyladenosine RNA modification

The modification N⁶-methyladenosine (m⁶A) is present in several classes of RNAs, including the poly(A) fraction of mRNA, where it is the most abundant internal RNA modification [1]. m⁶A was first identified among poly(A) RNAs in the 1970s [2], [3], [4], [5], [6], [7], [8], [9]. In the poly(A) fraction of RNA, the m⁶A modification is present in a well-defined RNA motif, RRACH (R can be either A or G, and H can be A, C or U). The frequency of the methylation consensus motif far exceeds the m⁶A content of mRNA [1], suggesting that the primary sequence is likely necessary, but not sufficient for the m⁶A modification to occur. Indeed, high-throughput sequencing experiments have demonstrated that the m⁶A modification is not randomly distributed in mature transcripts, but is enriched at specific transcript landmarks such as the last transcribed exon, near the 5′ end, or in long exons [10], [11], [12], [13]. Furthermore, a correlation between m⁶A modification and tandem alternative polyadenylation (APA) site usage has been described. In a restricted group of mRNAs, knockdown of the m⁶A methylation complex, whose structure and function will be detailed below, resulted in a switch from proximal to distal APA site usage [10]. In a different study, sequencing of intact transcripts after m⁶A immunoprecipitation, revealed a strong bias toward the presence of m⁶A in transcript isoforms with shorter 3′UTRs [14], lacking regulatory elements.

The presence of m⁶A in the poly(A) fraction of RNA is widely conserved among eukaryotes [1]. In the yeast (Saccharomyces cerevisiae), cell fate decisions in response to starvation, such as dividing under a pseudo-hyphal foraging program or undergoing meiosis to form protective spores, are dependent on m⁶A [15], [16]. In Arabidopsis thaliana, loss of components of the methylation complex, and consequent loss of m⁶A, results in embryonic lethality [17], [18]. Later in development, loss of m⁶A yields plants with altered growth patterns and reduced apical dominance, with flowers that show defects in their floral organ number, size, and identity [19]. In the fruit fly (Drosophila melanogaster), loss of components of the methylase complex, and consequent loss of m⁶A, results in impaired neuronal function, flightless animals, and a sex bias toward maleness [20], [21]. In zebrafish (Danio rerio), the protein YTH N⁶-methyladenosine RNA binding protein 2 (Ythdf2), an m⁶A reader, facilitates removal of maternal mRNAs during the maternal-to-zygotic transition (MZT), an important developmental step. Loss of ythdf2 decelerates the decay of m⁶A-modified maternal mRNAs and impedes zygotic genome activation [22]. Knockdown of components of the methylase complex in zebrafish embryos resulted in multiple developmental defects including smaller head and eyes, smaller brain ventricle, and curved notochord [23]. These observations suggest that in eukaryotes, m⁶A has crucial functions in development. In mice, genes encoding for methyltransferase like 3 (METTL3) and METTL14 are essential genes, and m⁶A is required for embryonic stem cells (ESC) to exit pluripotency [11], [24]. The landscape of m⁶A modification is very similar in mice and humans [11], [12]. Comparison of the m⁶A landscape of human, chimpanzee, and rhesus monkeys identified sets of m⁶A modification sites consistent with patterns of stabilizing or directional selection, with evolutionary gains of m⁶A signal correlating with evolution of higher mRNA levels [25].

Writing the m⁶A modification: the methylase complex

The addition of the methyl group to the position N⁶ of adenosine is catalyzed by a large protein complex [26] (Figure 1). The core of this complex is a stable heterodimer of METTL3 [27] and METTL14 [28]. Although both proteins have methyltransferase domains, crystal structure studies of the heterodimer demonstrate that only METTL3 has catalytic activity, while METTL14 stabilizes METTL3 and its interaction with the RNA molecule [29], [30], [31]. A physical interaction between METTL3 and RNA polymerase II (RNAPII) has been observed under conditions that induce slow progression or frequent pausing of RNAPII, suggesting that m⁶A deposition occurs co-transcriptionally. Attenuated transcription leads to increased levels of m⁶A, which is detrimental to translation [32]. UV-induced DNA damage leads to increased levels of m⁶A and localization of METTL3 and METTL14 to sites of UV-induced DNA damage [33]. Accumulation of m⁶A at sites of UV-induced DNA damage requires ADP-ribose polymerase (PARP) and is required for early recruitment of DNA polymerase kappa (Pol κ), an enzyme implicated in both nucleotide excision repair and trans-lesion synthesis [33]. WT1-associated protein (WTAP) can interact with the METTL3/METTL14 heterodimer and is required for proper localization of the methylase complex to specific loci in the nucleus and efficient m⁶A modification [23], [28], [34], [35]. Additional proteins identified as interactors of METTL3/METTL14 include KIAA1429 [35] and RNA binding motif protein 15 (RBM15) [36]. While the molecular function of KIAA1429 remains unknown, RBM15, which is required for the function of X inactive specific transcript (Xist), has been shown to target the methylation complex to specific RRACH motifs to promote methylation [36]. The activity of the methylase complex can be inhibited by sequestration of METTL3 by the transcription factor zinc finger protein 217 (ZFP217), which in mouse ESCs (mESCs) is enriched at the promoters of m⁶A-modified RNAs [37].

Figure 1.

The m⁶A regulatory pathway

Schematic representation of known components of the m⁶A regulatory pathways. The methylase complex (contained in the blue box) can modify adenosines in an RRACH context (R = G/A, H = G/A/C). The function of the methylase complex can be inhibited by ZNF217. The methylase complex interacts with RNAPII during transcription and recruits Pol κ to sites of DNA damage. Deposition of m⁶A can alter the RNA secondary structure. The demethylases (contained in the red box) can remove the methyl group. Two types of reader proteins (included in the green boxes) can interact with modified RNA through direct recognition of the methyl group or interaction with an RNA secondary structure induced by m⁶A modification. ALKBH5, alkB homolog 5; eIF3, eukaryotic initiation factor 3; FTO, fat mass and obesity-associated protein; HNRNPA2B1, heterogeneous nuclear ribonucleoprotein A2/B1; HuR, human antigen R; METTL3, methyltransferase like 3; Pol κ, DNA polymerase κ; PBM15, RNA binding motif protein 15; RNAPII, RNA polymerase II; WTAP, WT1-associated protein; YTHDC1, YTH domain containing 1; YTHDF, YTH N⁶-methyladenosine RNA binding protein; ZFP217, zinc finger protein 217.

m⁶A is a dynamic mRNA modification: the demethylases

Discovery of RNA m⁶A demethylases raised the possibility that m⁶A can be dynamically regulated in the cell. Two enzymes capable of removing m⁶A have been identified to date: fat mass and obesity-associated protein (FTO) [38] and alkB homolog 5 (ALKBH5) [39], enzymes of the non-heme Fe(II)- and α-ketoglutarate-dependent dioxygenase AlkB family of proteins (Figure 1). While m⁶A methylation is broadly conserved across eukaryotes, orthologues of FTO and ALKBH5 have not been identified in S. cerevisiae, suggesting that in some organisms demethylation is passive [40]. FTO was first identified in genome-wide association studies and was linked to increased body mass index (BMI) [41], [42], [43]. FTO localizes to the nucleus, and in mice, mRNA levels of Fto are higher in the brain, particularly in the hypothalamic nuclei governing energy balance. Furthermore, Fto mRNA levels in the arcuate nucleus are regulated by feeding and fasting [44]. Initially, 3-methylthymine in single-stranded DNA (ssDNA) was proposed as the substrate of FTO [44]. A later study demonstrated that FTO preferred 3-methyluracil in single-stranded RNA (ssRNA) over 3-methylthymine in ssDNA [45]. Preference for a single-stranded substrate was confirmed by protein structure analysis [46]. In 2011, He and colleagues reported that FTO preferred m⁶A in vitro, and manipulation of FTO in vivo led to changes in cellular m⁶A levels [38]. Oxidative demethylation of m⁶A in RNA by FTO leads to the formation of N⁶-hydroxymethyladenosine and N⁶-formyladenosine, further expanding the range of chemical modifications present in RNAs [47]. Recently, FTO was described to show higher affinity for the modification N⁶,2′-O-dimethyladenosine (m⁶A_m) as a substrate. m⁶A_m is found on the first nucleotide adjacent to the 7-methylguanosine cap. Transcripts with a m⁶A_m modification are more stable due to resistance to the mRNA-decapping enzyme, decapping mRNA 2 (DCP2) [48]. Alkbh5 is expressed in most tissues, and is particularly abundant in the testes. While Alkbh5 null mice are viable and appear anatomically normal, they exhibit impaired fertility due to the apoptosis of meiotic metaphase-stage spermatocytes [39]. A study of human asthenozoospermia patients found a correlation between elevated METTL3 and m⁶A levels, and sperm motility [49]. ALKBH5 localizes to the nucleus, and loss of ALKBH5 impairs mRNA export, RNA metabolism, and assembly of mRNA processing factors in nuclear speckles [39]. Structural studies of ALKBH5 revealed a region that excludes binding of dsDNA [50], [51], [52], [53]. Although the RRACH motif is critical for m⁶A methylation, it is not required for demethylase activity [54]. Rather, the conformational changes in the RNA structure induced by m⁶A modification provide a platform for interactions between the demethylases and the target RNA [54].

The effect of m⁶A on RNA biology: the m⁶A readers

How the cell ‘perceives’ the presence of m⁶A on the RNA determines the downstream effect of this modification. The presence of m⁶A creates a binding site for proteins that preferentially recognize modified bases, and binding of m⁶A readers to target transcripts can determine multiple aspects of RNA metabolism. A number of proteins have been identified as direct m⁶A readers (Figure 1). Several proteins with a YTH domain were shown to bind m⁶A-modified RNAs. For instance, YTHDF2 selectively binds m⁶A-modified mRNAs, and recruits them to mRNA decaying sites, controlling mRNA stability [55]. YTHDF2 promotes RNA degradation through deadenylation, mediated by the CCR4-NOT complex [56]. YTHDF1, another m⁶A reader, interacts with the translation machinery and actively promotes translation of target mRNAs [57]. YTHDF3 cooperates with YTHDF1 to enhance translation, and impacts YTHDF2-mediated mRNA decay [58], [59]. Therefore, these three proteins, YTHDF1, 2, and 3, coordinate to provide spatiotemporal control over RNA metabolism. The protein YTH domain containing 1 (YTHDC1) is a nuclear m⁶A reader that regulates splicing and is required for the function of Xist, a long non-coding RNA (lncRNA) involved in X chromosome inactivation [36], [60]. The eukaryotic initiation factor 3 (eIF3) complex, through a multisubunit interface, interacts with m⁶A-modified 5′UTR to directly recruit the 40S preinitiation complex to the 5′UTR of mRNAs to stimulate translation initiation. This mechanism allows mRNA with m⁶A-modified 5′UTRs to be translated in a cap-independent way [61]. In the nucleus, heterogeneous nuclear ribonucleoprotein A2/B1 (HNRNPA2B1) binds m⁶A-modified transcripts and regulates splicing and microRNA (miRNA) maturation [62]. In addition, m⁶A has been shown to affect RNA secondary structure, modulating the ability of RNA binding proteins (RBPs) to bind to RNA in the neighborhood of m⁶A modification sites. This mechanism, termed “m⁶A-switch”, facilitates the binding of HNRNPC and HNRNPG to targets, to regulate mRNA abundance and splicing [63], [64], [65], [66], [67] (Figure 1). Binding of human antigen R (HuR) to RNA is constrained by distance to m⁶A modification sites, and is blocked by m⁶A on some transcripts [68].

m⁶A and cancer

Studies on the role of METTL3 in ESCs uncovered several pluripotency transcription factors as targets of the m⁶A methylase complex, raising the possibility that pluripotency is regulated through m⁶A-dependent pathways. These studies demonstrated that loss of m⁶A leads to increased mRNA half-life of transcription factors such as NANOG, blocking the capacity of pluripotent cells to progress through differentiation [11], [24]. The phenotype observed on teratoma assays suggested that loss of m⁶A could also play a role in cancer progression [11], [24]. Indeed, multiple studies have implicated both m⁶A writers and erasers in cancer. Manipulation of m⁶A levels in glioblastoma stem-like cells (GSCs) severely impacted growth, self-renewal, and tumor development [69], [70]. Reduction of m⁶A levels, through knockdown of Mettl3 and/or Mettl14, resulted in enhanced in vitro growth and self-renewal of GSCs and promoted the ability of GSCs to form brain tumors in vivo. At the molecular level, knockdown of Mettl3 and/or Mettl14 resulted in the upregulation of several oncogenes, including the genes encoding ADAM metallopeptidase domain 19 (ADAM19), EPH receptor A3 (EPHA3), and Kruppel-like factor 4 (KLF4), and downregulation of tumor suppressors such as genes encoding cyclin dependent kinase inhibitor 2A (CDKN2A), breast cancer 2, early onset (BRCA2), and tumor protein (TP53l11). Knockdown of Adam19, an m⁶A modified transcript, dramatically reduced growth and self-renewal of GSCs [69]. Increased levels of m⁶A in glioblastoma cells, achieved through overexpression of Mettl3, or treatment of cells with the FTO inhibitor ethyl ester form of meclofenamic acid (MA2), inhibited tumor progression [69]. High expression of ALKBH5 in glioblastoma patients predicts poor prognosis [70]. ALKBH5 is highly expressed in cell lines or patient-derived primary glioblastoma cultures enriched for GSCs, and knockdown of ALKBH5 impairs GSC self-renewal in vitro and decreases proliferation and tumorigenesis of GSCs in vivo. Analysis of gene expression and m⁶A landscape of ALKBH5 knockdown cells uncovered forkhead box M1 (FOXM1) as both a target of ALKBH5 and potential regulator of genes dysregulated upon ALKBH5 knockdown. FOXM1 is a transcription factor with a critical role in the self-renewal and tumorigenesis of GSCs. Knockdown of ALKBH5 leads to a gain of m⁶A on FOXM1 nascent transcript and a decrease in association with HuR, resulting in a decrease of FOXM1 expression. Interestingly, ALKBH5 interaction with FOXM1 is promoted by FOXM1-AS, a lncRNA transcribed in the opposite orientation of FOXM1, with a 457-nt overlap at the 3′UTR (Figure 2A) [70]. These studies suggest that RNA demethylases are a potential target for anti-glioblastoma therapy [69], [70]. In hepatocellular carcinoma (HCC), downregulation of METTL14 results in lower m⁶A levels, and enhances the metastatic capacity of HCC cells [71]. The processing of the metastasis-associated miRNA miR126 requires METTL14 activity and is involved in HCC metastasis [71] (Figure 2B).

Figure 2.

Links between m⁶A regulation and cancer

Schematic representation of m⁶A pathway components and targets involved in cancer progression. Blue lines represent methylation, red lines represent demethylation and green lines indicate m⁶A reader proteins. A.FOXM1-AS promotes interaction of ALKBH5 with FOXM1. HuR interacts with demethylated FOXM1 pre-mRNA and enhances expression of FOXM1. B. Downregulation of METTL14 disrupts miRNA processing. C. Over-expression of FTO driven by leukemic oncogenic proteins leads to downregulation of target genes. D. Hypoxia leads to upregulation of ALKBH5 and ZFP217, resulting in loss of m⁶A levels in NANOG mRNA. E. METTL3 promotes translation of mRNA targets in the cytoplasm. ALKBH5, alkB homolog 5; ASB2, ankyrin repeat and SOCS box containing 2; atRA, all-trans-retinoic acid; BCSC, breast cancer stem cell; EGFR, epidermal growth factor receptor; eIF3, eukaryotic initiation factor 3; FOXM1, forkhead box M1; FOXM1-AS, FOXM1 antisense transcript; GSC, glioblastoma stem-like cell; HCC, hepatocellular carcinoma; HIF1α, hypoxia-inducible factor 1α; HuR, human antigen R; LCC, lung cancer cell; METTL3, methyltransferase like 3; RARA, retinoic acid receptor alpha; TAZ, tafazzin; ZFP217, zinc finger protein 217.

FTO was found to play an oncogenic role in acute myeloid leukemia (AML). FTO is expressed at higher levels in certain subtypes of AML, as it can be upregulated by leukemic oncogenes. Over-expression of FTO in two mixed-lineage leukemia (MLL)-rearranged AML cell lines promoted cell growth/proliferation and viability, while decreasing apoptosis and the global mRNA m⁶A level [72]. Notably, genes with m⁶A sites that are hypo-methylated upon overexpression of FTO tend to be downregulated in FTO-overexpressing AML cells. Overexpression of FTO leads to a reduction of m⁶A levels, downregulation of mRNA and protein levels in critical genes, such as those encoding ankyrin repeat and SOCS box containing 2 (ASB2) and retinoic acid receptor alpha (RARA). ASB2 and RARA are upregulated during normal hematopoiesis and are important regulators of all-trans-retinoic acid (atRA)-induced differentiation of leukemia cells. Demonstrating the importance of ASB2 and RARA as FTO targets, the over-expression of these genes can rescue the effects of FTO over-expression, while knockdown was sufficient to rescue the inhibitory effect of FTO knockdown on AML cell growth and viability. Importantly, this study demonstrates that the inhibition of ASB2 and RARA expression by FTO contributes to the response of AML cells to atRA treatment [72] (Figure 2C). Analysis of datasets from the Cancer Genome Atlas (TCGA) Research Network AML study revealed a strong association between mutations and/or copy number variation of m⁶A regulatory genes and TP53 in AML patients. Furthermore, alterations in m⁶A regulatory genes confer a worse survival in AML patients [73]. In a meta-analysis study, the FTO polymorphism rs9939609 was found to be significantly associated with pancreatic cancer, although no other significant association was found [74].

Hypoxia is an important feature of the tumor microenvironment, and multiple studies have connected hypoxia to m⁶A regulatory pathways. The m⁶A demethylase ALKBH5 is a direct target of hypoxia-inducible factor 1α (HIF-1 α) [75]. In breast cancer cells, hypoxia leads to upregulation of ALKBH5, which then results in decreased m⁶A levels [76]. Upregulation of ALKBH5 also induced the loss of m⁶A in NANOG mRNA, which in turn increased its stability and NANOG protein levels in breast cancer stem cells (BCSCs). Therefore, hypoxia-dependent ALKBH5 upregulation mediates NANOG expression and BCSC specification and/or maintenance [76]. Exposure of breast cancer cells to hypoxia also leads to the upregulation of ZNF217, a transcription factor known to interact with METTL3 and inhibit its function [77] (Figure 2D). In gynecological tumor cell lines, hypoxia leads to a reduction of YTHDC1 protein levels, through changes in splicing that generate mRNA isoforms targeted by nonsense-mediated decay [78]. The m⁶A reader YTHDC2 [79] has been shown to promote translation of HIF-1 α and contributes to colon tumor metastasis [80]. These studies reveal how the tumor microenvironment can impact gene expression through the epitranscriptome, with devastating outcomes.

METTL3 has also been proposed to promote growth, survival, and invasive capacity of lung and cervical cancer cells in a methylation-independent manner [81]. Lin and colleagues described how METTL3 promotes the translation of specific mRNAs, including those encoding epidermal growth factor receptor (EGFR) and tafazzin (TAZ), through interaction with RNA, the translation initiation machinery, and the ribosome [81]. In this study the authors show that although METTL3 knockdown has little effect on the steady state level of methylated mRNAs, the protein expression level of a specific group of genes was reduced. While knockdown of METTL3 resulted in the re-localization of the EGFR and TAZ mRNAs to sub-polysome fractions, tethering of METTL3 to target mRNAs robustly enhanced translation. The catalytic activity of METTL3 is dispensable, as tethering a catalytically-dead METTL3 also enhances translation of target mRNA (Figure 2E). This study, with results in line with those of other studies concerning viruses that replicate in the cytoplasm (see below), uncovers a function for METTL3 in the cytoplasm.

m⁶A and viral replication

The presence of m⁶A, and in some cases the exact location of the m⁶A modification, was observed in viral transcripts as early as the 1970’s. Pioneering work revealed the presence of m⁶A in Rous sarcoma virus (RSV), adenovirus type 2, Simian virus 40 (SV40), and influenza RNAs [82], [83], [84], [85], [86], [87], [88]. While the presence of m⁶A in viral RNAs was well established, the role of the modification in viral biology remained unknown. In an early attempt to understand how m⁶A influences RSV biology, the Beemon Lab mutated the consensus motif around methylated adenosines. Although these studies demonstrated that mutations of the RRACH motif resulted in loss of methylation, no effect on viral RNA processing or viral life cycle was found when the virus was modified to prevent m⁶A methylation [89], [90]. A different study, in which the researchers used a competitive inhibitor of methionine transferase, cycloleucine, suggested that m⁶A is required for the formation of late SV40 RNAs in untransformed cells [91], however it is important to note that cycloleucine is not a m⁶A-specific inhibitor, and the results should be interpreted with caution. Recently, work from three separate groups demonstrated that m⁶A plays an important role in HIV-1 biology. Lichinchi and colleagues [92] showed that acute viral infection triggers a massive increase in host and viral levels of m⁶A, and identified 56 host genes with a role in viral infection that are specifically m⁶A modified upon infection. The authors also demonstrate that loss of m⁶A decreased viral replication, while elevation of m⁶A levels through knockdown of ALKBH5 resulted in increased replication of HIV. The authors focused on two m⁶A sites at the stem loop II (important structural motif) region of HIV-1 Rev response element (RRE), and demonstrated that one of the m⁶A sites, A7883 (in the loop region), is required to enhance binding of HIV-1 Rev protein to this structural RNA element, an interaction required for viral RNA to exit from the nucleus. Interestingly, A7883 is well conserved across isolates, suggesting the importance of this site. Further studies will be required to understand if Rev is a direct or indirect reader of m⁶A. Kennedy and colleagues also detected m⁶A, and showed that it is required for HIV-1 replication. The authors showed that the YTHDF1−3 proteins bind to and enhance expression of viral RNAs. Cullen and colleagues also demonstrated that four m⁶A-modified sites are conserved across isolates, again suggesting that maintaining m⁶A sites confers survival benefits [93]. Tirumuru and colleagues showed that the m⁶A reader proteins YTHDF1−3 bind HIV-1 RNA and inhibit HIV-1 post-entry infection by blocking viral reverse transcription and promoting degradation of the gag mRNA. At a later stage, m⁶A is required for production of Gag protein in virus producing cells, as m⁶A might promote translation of viral proteins in infected cells [94]. While there are some conflicting results between these studies, these papers demonstrate the importance of m⁶A on HIV-1 biology, and suggest that targeting this pathway might offer new tools to control viral infection (Figure 3A).

Figure 3.

m⁶A is involved in viral replication

Schematic representation of the intersection between m⁶A pathway and viral replication. Blue lines represent methylation, red lines represent demethylation and green lines indicate m⁶A reader proteins. A. Writers, erasers, and readers in the m⁶A pathway influence HIV RNA transport, stability, and translation, in both nucleus and cytoplasm. B. m⁶A pathway regulates viruses that replicate in the cytoplasm. ALKBH5, alkB homolog 5; FTO, fat mass and obesity-associated protein; HCV, hepatitis C virus; HIV, human immunodeficiency virus; METTL3, methyltransferase like 3; YTHDF, YTH N⁶-methyladenosine RNA binding protein; ZIKV, Zika virus.

Viruses of the Flaviviridae family, which replicate exclusively in the cytoplasm, have been shown to be m⁶A modified [95], [96]. Comparison of m⁶A profiles across multiple Flaviviridae viruses suggests that the m⁶A modification is potentially conserved in this family [96]. Furthermore, m⁶A is conserved across multiple isolates of Zika virus (ZIKV) [95]. Loss of m⁶A or impairment of m⁶A reader proteins of the YTH family result in higher viral particle production of both ZIKV and hepatitis C virus (HCV), while loss of demethylase activity negatively impacts viral particle production. As observed with HIV infection, ZIKV infection leads to changes in the m⁶A profile of the host cell. In the case of ZIKV infection, new m⁶A peaks in host transcripts are enriched at the 5′UTR and the CDS, while the lost m⁶A peaks tend to localize at the exon junction and at the 3′UTR [95]. HCV infection leads to a re-localization of YTH proteins to loci of HCV particle production. YTH proteins interact with the HCV RNA and suppress viral production [96]. Mutagenesis studies on m⁶A sites in the viral E1 gene of HCV suggest that the interaction of the viral RNA with host YTHDF proteins or viral core proteins is regulated by the presence of m⁶A on the viral gene [96] (Figure 3B).

m⁶A and metabolic diseases

In mammals, metabolism, as well as other features of physiology, is regulated in a temporal way by the circadian system. At the cellular level, the circadian clock is ‘paced’ by feedback loops at the transcriptional and translational levels that result in periodic gene expression. Modification of RNA with m⁶A, through its effects on RNA metabolism, directly impacts the length of the circadian period, with loss of m⁶A resulting in circadian period elongation [97]. Metabolism, at the organism level, has also been linked to RNA post-transcriptional modifications through FTO. FTO was first associated with obesity in a series of population studies that linked variations in the first intron of FTO to elevated BMI [41], [42], [43]. Although the link between the FTO polymorphism and BMI has been confirmed in multiple studies, how expression of FTO leads to obesity is not well understood. Several studies propose that the polymorphisms in the first intron of FTO control expression of other genes, such as those encoding RPGRIP1 like (RPGRIP1L), iroquois-related homeobox 3 (IRX3), and IRX5 [98], [99], [100], [101]. Studies in mouse models, on both loss of function and overexpression of Fto, revealed a role for FTO in energy homeostasis and body composition [102], [103], [104]. Global loss of Fto resulted in postnatal growth retardation and significant reduction in adipose tissue and lean body mass [102]. A dominant missense point mutation (I367F) in the Fto gene resulted in reduced fat mass, increased energy expenditure, and unchanged physical activity without postnatal growth retardation or perinatal lethality [103]. Ubiquitous overexpression of FTO results in a dose-dependent increase in body and fat mass of mice [104]. Furthermore, FTO plays a critical role in the regulation of adipogenesis through its effect on RNA splicing, particularly for the gene encoding the adipogenic regulatory factor RUNX1 translocation partner 1 (RUNX1T1) [105]. In skeletal muscle cells, FTO-dependent demethylation of mRNA m⁶A methylation is involved in the regulation of skeletal muscle lipid accumulation [106]. In humans, a missense mutation in the coding region of FTO (R316Q), inactivates enzymatic function and results in a rare autosomal-recessive lethal syndrome [107].

Concluding remarks

Organisms face innumerable challenges to their physiology, both during and after embryonic development. At the cellular level, the ability to execute developmental programs and respond to environmental challenges requires precise and coordinated control of gene expression. As a central player in gene expression, the RNA molecule is the target of several of these regulatory pathways, and disruption of any of these mechanisms can result in disease and have catastrophic consequences. This review focused on the link between pathways that rely on m⁶A to modulate gene expression and its implications in human disease. RNA post-transcriptional modifications, through their impact on transcription, splicing, mRNA stability, and rate of translation, regulate essential features of a cell. It is not surprising that dysregulation of these pathways is implicated in human disease. Although some details are still lacking, it is clear that the balance between methylation and demethylation at specific RNA transcripts plays a critical role in human health and represents an attractive target for therapy. Inhibition of the RNA demethylases ALKBH5 and FTO is a potential strategy to target cancer stem cells [69], [70], [76]. Furthermore, atRA treatment of AML cells demonstrates that m⁶A-related pathways can also be targeted as a potential strategy to enhance other available therapies [72]. Because m⁶A plays an important role in the biology of several viruses with considerable impact on human health, manipulating m⁶A regulation has also been explored as a potential anti-viral therapy. Use of 3-deazaadenosine (DAA), a drug that can block methylation, can inhibit replication of HIV-1 as well as RSV and influenza A virus [93], [108], [109], [110], [111]. Although these experiments do not conclusively show that m⁶A inhibition can be used as an antiviral therapy, since this drug is not an m⁶A specific inhibitor, several studies have demonstrated that targeting m⁶A pathway is a viable strategy. As m⁶A regulation is involved in multiple pathways, targeting m⁶A-dependent pathways can potentially lead to unwanted side effects. Therefore, development of drugs that inhibit these enzymes in a context-specific manner, or target the interaction between specific m⁶A targets and the interacting proteins will be critical for the success of m⁶A-targeted therapies.

Competing interests

The author declares no competing interests.

Handled by Yun-Gui Yang