Roles of METTL3 in cancer: mechanisms and therapeutic targeting

There are more than 170 modifications in RNA, among which N⁶-methyladenosine (m⁶A) is the most prevalent internal modification in messenger RNA (mRNA) [1‐4]. Over 7000 human transcripts harbor at least one m⁶A site, which is found within the consensus motif RRACH (where R = A/G, H = A/C/U), and most of the m⁶A sites are enriched in the coding sequence (CDS) and the 3′ untranslated region (3′UTR) of mRNA, especially around the stop codons [5, 6]. Although m⁶A was discovered more than 40 years ago [7], it failed to spark enthusiasm in this field until the identification of FTO as an m⁶A demethylase in 2011 [8], which reveals that m⁶A can be dynamically regulated and might play vital roles in development and diseases. Since then, FTO and ALKBH5, both belonging to the AlkB family of Fe(II)/a-ketoglutarate(a-KG)-dependent dioxygenases [9], were classified as m⁶A “eraser” proteins that remove m⁶A modifications from RNA (Fig. 1a and Table 1). In contrast to being removed by “eraser” proteins, m⁶A can be recognized by a set of RNA-binding proteins called m⁶A “reader” proteins that can specifically recognize and bind to m⁶A-modified transcripts. The list of m⁶A “reader” proteins is increasing (Table 1), including the YTH domain family proteins (YTHDC1/2, YTHDF1/2/3) [28, 36, 37, 41‐43], the insulin-like growth factor 2 mRNA-binding proteins (IGF2BP1/2/3) [38], and the heterogeneous nuclear ribonucleoproteins (hnRNPC, hnRNPG), which were shown to mediate the regulation of RNA stability, translation efficiency, RNA splicing, and RNA exporting [44‐50].

The enzymes catalyzing the formation of m⁶A are referred to as m⁶A “writer” proteins, first being purified as a protein complex in 1994 by Bokar et al. [51], and were further characterized in recent years as a multicomponent m⁶A methyltransferase complex (MTC) comprised of a METTL3-METTL14 heterodimer core and other binding partners (Fig. 1a), such as WTAP, ZC3H13, VIRMA, and RBM15/15B [10‐14]. In addition to the MTC, other m⁶A writers have also been identified in recent years, including METTL16, METTL5, and ZCCHC4 (Fig. 1a and Table 1), which are responsible for the deposition of m⁶A into structured RNAs, such as U6 snRNA, 28S rRNA, and 18S rRNA, and in some cases, the introns of mRNA [15‐22]. The MTC core component METTL3-METTL14 heterodimer catalyzes most of m⁶A methylations in mRNA, with METTL3 being the only catalytic subunit that uses S-adenosylmethionine (SAM) as the methyl donor [52‐54].

The full-length METTL3 has 580 amino acids and is comprised of a zinc finger domain (ZFD) and a methyltransferase domain (Fig. 1b), both of which are needed for the enzymatic activity. Huang et al. characterized the ZFD solution structure using nuclear magnetic resonance (NMR), showing that the domain contains two tandem CCCH-type zinc fingers (ZnF1 and ZnF2) connected by an anti-parallel β-sheet (Fig. 1c), which is responsible for target recognition, specifically for binding to single-stranded RNAs containing 5′-GGACU-3′ consensus sequence [55]. The structure of the methyltransferase domain of METTL3, named MT-A70, has been determined using X-ray crystallography in a complex with the corresponding domain of METTL14 by three independent groups [52‐54]. It is demonstrated that METTL14 only plays a structural role for RNA-binding and stabilization of the complex, while METTL3 is the catalytically active subunit, with a co-factor binding pocket for SAM or S-adenosylhomocysteine (SAH) [52, 53]. The crystal structure reveals that the MT-A70 domain of METTL3 is formed by a Rossman fold comprising a central, curved, eight-stranded β-sheet flanked by four α- helixes, as well as an interface loop to interact with METTL14 and two gate loops that have important roles in adenosine recognition (Fig. 1d). The conserved DPPW motif (residues 395–399) of the enzyme is located in the gate loop 1, which undergoes a significant conformational change together with gate loop 2 upon SAM/SAH binding, resulting in the closure of the co-factor binding pocket [52].

The expression of METTL3 is dysregulated in cancer via different mechanisms (Fig. 2a). It was demonstrated in pancreatic cancer that cigarette smoke condensate induces hypomethylation of METTL3 promoter and subsequently the recruitment of transcription factor NFIC to induce METTL3 overexpression [56]. An intestinal microbial metabolite, butyrate, was suggested to downregulate the expression of METTL3 and inhibit the development of colorectal cancer; however, the detailed mechanism is unclear [57]. Wang et al. showed in gastric cancer that P300 mediates histone H3 acetylation at lysine 27 (H3K27ac) and promotes METTL3 transcription [58]. It was also reported that microRNA miR-24-2 might promote METTL3 transcription; however, the detailed mechanism remains elusive [59]. Several other microRNAs, including miR-186, miR-4429, miR-600, and let-7g [60‐63], were proposed to suppress METTL3 by targeting METTL3 mRNA. In addition, SUMOylation of METTL3 protein was reported to repress its methyltransferase activity without affecting the protein stability or localization, although the mechanism remains unclear [64].

The methyltransferase activity of METTL3 is guided in diverse manners (Fig. 2b–e). It is suggested that m⁶A deposition could be governed in cis via the sequence code and structure at the modified site [65]; however, the mechanism underlying was unknown. Recently, it was shown that histone H3 trimethylation at lysine 36 (H3K36me3) could guide m⁶A RNA modification co-transcriptionally through direct interaction with METTL14 and the subsequent recruitment of the MTC complex, allowing for the selective deposition of m⁶A in CDS and 3′UTR where H3K36me3 is often found [66]. Increasing efforts have also been paid to trans regulators; among which, zinc finger protein 217 (ZFP217) was reported to sequester METTL3 and counteracts m⁶A deposition on stemness associated transcripts [67], while the TGFβ signaling factor SMAD2/3 could recruit METTL3/14 complex to a subset of transcripts involved in early cell fate decisions [68]. Another transcriptional factor, CAATT-box-binding protein CEBPZ, was demonstrated to directly recruit METTL3 to chromatin [68]. A large-scale computational screening aimed at identifying RNA-binding proteins as cell-specific trans regulators of m⁶A, and validated experimentally that TRA2A and CAPRIN1 could interact with METTL3 [69]. Fish et al. found that the RNA-binding protein TARBP2 recruits METTL3 and deposits m⁶A on the introns of the target mRNA, regulating RNA splicing and stability [70]. Two long non-coding RNAs (ARHGAP5-AS1 and LINC00470) have also been revealed to guide METTL3 to specific targets [71, 72], among which, the natural anti-sense transcript ARHGAP5-AS1 mediates METTL3 to deposit m⁶A marks on ARHGAP5 mRNA, promoting the mRNA stabilization and inducing chemoresistance [71].

Accumulating evidence in recent years has demonstrated that METTL3 plays critical roles in cancer as an m⁶A methyltransferase, either as an oncogene or a tumor suppressor, as summarized in Table 2.

Elevated expression of METTL3 was detected in lung cancer cells, which was associated with cancer cell growth, survival, and invasion [133]. In-depth investigation found that METTL3 promoted the translation of certain oncogenic transcripts such as EGFR and TAZ, independent of its catalytic activity, but through recruiting eukaryotic translation initiation factor 3 (eIF3) to the translation initiation complex [133]. In a following report from the same group, the authors further revealed that the N-terminal fragment of METTL3 (1–200 amino acids) interacted directly with the eIF3 subunit h (eIF3h) and that the mutated METTL3 (A155P) abrogated the interaction and the ability to promote mRNA translation was severely compromised [134]. The authors thus proposed a novel “mRNA looping” mechanism, in which METTL3 binds to the m⁶A-modified 3′UTR of target mRNA and then recruits eIF3h as well as translation initiation factors such as CBP800 and eIF4e, thus facilitating the ribosome recycling and promoting translation efficiency [133, 134].

A recent work by Hua et al. also presented a similar mechanism, in which METTL3 promoted translation of the receptor tyrosine kinase AXL, independent of its methyltransferase activity. The authors also showed that upregulated METTL3 in ovarian carcinoma was significantly associated with tumor grade and overall survival rate [135]. It is noteworthy that METTL3 only binds to approximately 22% of the m⁶A sites [14], suggesting a mechanism of selectivity of METTL3 towards its targets for translational regulation, which needs to be further elucidated in the future.

Based on the emerging data on the roles and the molecular mechanisms in cancer, m⁶A regulators have attracted growing investigation as therapeutic targets [136]. As discussed above, METTL3 plays both oncogenic and tumor-suppressive roles in human cancers. Although still limited, activators and inhibitors targeting METTL3 have been reported recently. A structure-based virtual screening of compound databases carried out by Selberg et al. discovered four small molecules displaying the potency of enhancing the activity of the METTL3-METTL14 complex, among which the most effective one increased the relative m⁶A level by 21.4 ± 12.9% in the following cellular assays [137]. Simultaneous docking analysis showed that the compound interacted with SAM in close proximity in the active center of METTL3, which might increase the binding affinity of SAM and also lower the energy barrier of the substrate RNA methylation reaction [137].

Another study performed by Bedi et al. used co-factor mimicking approach to screen a library of 4000 adenosine analogs and derivatives and identified seven compounds showing diverse activities in two in vitro assays; among them, one compound (illustrated in Fig. 4a) has the most favorable inhibitory potency (IC₅₀ = 8.7 μM) [138]. This is the first study reporting METTL3 inhibitors; however, several issues remain to be addressed. First, there is a lack of cellular activity data; thus, the actual effect of the compounds on m⁶A level is unknown. Second, the potential use of the compounds needs to be carefully evaluated because of the poor cell permeability and pharmacokinetics of adenosine analogs. Third, the co-factor binding pockets are common to the wide range of Rossmann fold enzymes; therefore, the selectivity of the inhibitors also requires further investigation.

As discussed above, the oncogenic function of METTL3 relies on the heterodimer structure formed with METTL14 in most cases; therefore, it will be a reasonable option to design inhibitors based on protein-protein interaction (PPI) strategy (Fig. 4b), as having been successfully applied in other drug targets [139]. Proteolysis targeting chimera (PROTAC) is another emerging biotechnology to degrade a wide range of proteins specifically [140], which is a bifunctional molecule consisting of a ligand of the target protein and a covalently linked ligand of an E3 ubiquitin ligase (Fig. 4c). Whether this strategy could be applied for degrading METTL3 in cancers warrants further investigation. In addition, structural analysis suggested that the substrate-binding site and SAM-binding site of METTL3 are merged into a large pocket [141]; therefore, the development of bisubstrate inhibitors that simultaneously occupy both of the binding sites might be an alternative strategy for future METTL3 targeting.

Overall, targeting of METTL3 for clinical application is still in its infancy. With the increasing knowledge regarding the regulation, functions, and mechanisms of METTL3 in cancer, it is promising to develop METTL3 targeted agents in the near future.

It is widely accepted that several biological capabilities are regarded as hallmarks of cancer, including uncontrolled proliferation, cell death resistance, angiogenesis, invasion and metastasis, metabolism dysregulation, and immune escape [142]. The accumulating studies in recent years have revealed diverse pathways in cancers that are affected by METTL3, mostly focused on cell proliferation, cell death resistance, invasion, and metastasis (Table 2). However, emerging evidence also suggests that METTL3 plays vital roles in other biological processes, such as angiogenesis [58] and metabolism dysregulation [58, 101]. A recent publication showed that m⁶A modification of viral RNA enables the virus to avoid detection by innate immunity [143]; thus, there is a possibility that METTL3 might also mediate immune escape in cancers. Further studies are warranted to fully uncover the roles and mechanisms of METTL3 in affecting every hallmark of cancer.

It should be emphasized that although this review focuses mostly on METTL3, other components of the MTC, especially METTL14, should be taken into account when studying the m⁶A-dependent function of METTL3 in cancer. It is reasonably thought that proteins in the MTC should exhibit similar roles in the same pathological conditions; however, this is not always the case as reported. For example, METTL14 plays similar roles as METTL3 in some types of cancer, such as AML [122, 144, 145], breast cancer [146], and endometrial cancer [117], but exhibits opposite functions in other cancers, such as liver cancer [147, 148]. How could components in the same protein complex play opposite roles in the same cancer type needs to be carefully investigated and addressed. It is also noticeable that in glioblastoma, bladder cancer, and colorectal cancer, METTL3 was reported to play either oncogenic or/and tumor-suppressive functions by different groups, which may be explained by tumor heterogeneity and/or different model systems used for the study, and more further comprehensive and detailed studies are warranted to gain a better view.

In conclusion, METTL3 affects a broad range of biological processes and plays diverse roles in cancers, either dependent or independent of its methyltransferase activity, while the therapeutic targeting of METTL3 is just in the early stage. Continuing efforts are still needed to design and optimize strategies for targeting of METTL3 for cancer treatment.