YTH domain-containing proteins

YTH domain-containing proteins are the best-known family of m⁶A-interacting proteins, or m⁶A readers. This protein family has been associated with almost all cellular processes linked to the m⁶A modifications to date, including transcriptional regulation, pre-mRNA splicing, RNA nuclear export, translation regulation, and RNA stability and decay (Boulias & Greer, 2023).

Proteins of this family bind the m⁶A modification specifically through their conserved YTH-domain (Fig 4C). Within the YTH domain, two crucial aromatic residues form the m⁶A binding pocket. Removal of either residue leads to loss of m⁶A binding (Theler et al, 2014; Xu et al, 2014; Zhu et al, 2014). The YTH domain is characteristically flanked by two intrinsically disordered regions at the N- and C-terminus (Sikorski et al, 2023).

YTHDC1 is the only exclusively nuclear YTH-domain-containing m⁶A reader protein. In contrast to the other YTH domain-containing m⁶A readers, YTHDC1 shows a sequence preference towards the canonical DRACH motif, with a preference for C after the m⁶A residue and purines at the n-1 and n-2 positions around the methylation site (Table 1). Because of its nuclear localization, YTHDC1 is involved in almost all nuclear processes that involve m⁶A-modified RNAs, including chromatin remodeling, transcriptional regulation, X-chromosome inactivation, mRNA processing, and nuclear mRNA export (Widagdo et al, 2022).

YTHDC1 can be recruited to co-transcriptionally m⁶A-modified nascent RNAs and recruits the histone H3K9me3 demethylase KDM3B by an unknown mechanism. This leads to the demethylation of H3K9 and, thereby, the removal of a repressive histone mark (Li et al, 2020).

Furthermore, YTHDC1 has the propensity to initiate phase separation through its arginine-rich C-terminal disordered domain (Cheng et al, 2021; Lee et al, 2021). One study showed that m⁶A modifications deposited on enhancer RNAs are bound by YTHDC1 and form m⁶A–enhancer RNA/YTHDC1 condensates. These condensates can undergo mixing with BRD4 coactivator condensates and facilitate the formation of transcriptional activator condensates (Lee et al, 2021). Together with splicing factor hnRNPG, YTHDC1 can also prevent premature transcription termination by binding to co-transcriptionally installed m⁶A-marks on 5′-ends of RNAs thereby preventing binding of the integrator complex (Xu et al, 2022). During pre-mRNA splicing, YTHDC1 is furthermore involved in alternative splicing regulation. YTHDC1 is recruited to m⁶A-modified exons and recruits splicing factor SRSF3 via direct interactions between its C-terminus and the C-terminus of SRSF3, resulting in exon inclusion. At the same time, binding of YTHDC1 inhibits the binding of splicing factor SRSF10 and thereby interferes with SRSF10-associated exon skipping. In the absence of an m⁶A modification within a given exon, SRSF10 can bind and promote exon skipping (Fig 4D) (Xiao et al, 2016). Finally, at the 3′-end of mRNAs, YTHDC1 can interfere with alternative polyadenylation leading to longer 3′UTRs (Chen et al, 2022).

YTHDC1 stays associated with mature m⁶A-modified mRNA to facilitate their nuclear export through the direct interaction with splicing factor and nuclear export adapter SRSF3. SRSF3, on the other hand, can interact with the nuclear export receptor NXF1, which facilitates the nuclear export of m⁶A-modified mRNA through protein–protein interactions (Roundtree et al, 2017).

The m⁶A marks on lncRNA Xist are also recognized and bound by YTHDC1. This interaction with Xist is crucial for X-chromosomal inactivation and gene silencing (Patil et al, 2016). Mechanistically, YTHDC1 binds to m⁶A-modified highly conserved AUCG tetraloops in the A-repeats, which leads to partial melting of the hairpin and modulation of the RNA structure (Jones et al, 2022).

The YTHDF paralogs YTHDF 1–3 are preferentially localized to the cytosol. Two contradictory models have been proposed regarding the function of these three different proteins. The canonical model assigns each YTHDF paralog-defined functions and targets, whereas a more recent model proposes that all three proteins have redundant functions and targets in mRNA degradation (Zaccara & Jaffrey, 2020).

Within the canonical model, YTHDF1 enhances the translation of m⁶A-modified transcripts by binding m⁶A marks at the stop codon and 3′ UTRs of mRNAs. YTHDF1 then directly interacts with and recruits translation initiation factor eIF3 to promote cap-dependent translation (Wang et al, 2015).

YTHDF2, on the other hand, is involved in mediating mRNA decay (Wang et al, 2014). To promote m⁶A-modified mRNA decay, YTHDF2 binds the m⁶A marks on mRNAs and either directly recruits the CCR4–NOT deadenylation complex (Du et al, 2016) or interacts with HRSP12, which in turn recruits the RNaseP/MRP endoribonuclease (Park et al, 2019).

In the nucleus, YTHDF2 is also involved in the clearance of R-loop structures. R-loops are three-stranded nucleic acid structures, consisting of a DNA:RNA hybrid and a single-stranded DNA, that are formed at the transcription bubble during transcription. RNA in R-loop structures can be m⁶A-modified by the METTL3/METTL14 methyltransferase, which leads to the recruitment of reader YTHDF2 to promote mRNA degradation, and R-loop clearance (Abakir et al, 2020; Kang et al, 2021).

Lastly, YTHDF3 is a modulator of YTHDF1 and YTHDF2 functions. YTHDF3 enhances translation through direct interaction with YTHDF1 (Shi et al, 2017). Similarly, YTHDF3 also interacts with YTHDF2 and seems to be involved in mRNA decay through this interaction (Shi et al, 2017). In addition, YTHDF3 binds and recognizes stress-induced newly methylated mRNA and drives their translocation into stress granules during oxidative stress (Anders et al, 2018). It is important to point out that a more recent study claims that m⁶A marks on mRNAs have only a limited effect on their translocation into stress granules (Khong et al, 2022).

The more recent model proposes that all three paralogs share similar targets and mediate mRNA degradation via association with CNOT, which is a scaffolding subunit of the CCR4vNOT deadenylase complex (Zaccara & Jaffrey, 2020).

The detailed functions and an attempt to consolidate the two existing models have recently been reviewed (Sikorski et al, 2023).

In contrast to the other members of the YTH domain family, YTHDC2 comprises several well-folded domains, including the YTH domain. YTHDC2 can interact with the small ribosomal subunit close to the mRNA entry and exit sites via its YTH and R3H domains. Therefore, YTHDC2 is thought to facilitate efficient translation by bridging m⁶A-marked mRNAs with the ribosome (Kretschmer et al, 2018). Furthermore, it can recruit 5′-3′ exonuclease XRN1 via direct interaction through its ankyrin domain and potentially promote mRNA decay (Kretschmer et al, 2018).

In summary, YTH domain-containing proteins bind m⁶A-modified RNA via similar mechanisms but lead to different downstream effects. How substrate specificity is achieved outside of the modification itself still needs to be better understood and will need further investigation to elucidate distinct functions and functional redundancies.

Non-YTH domain m⁶A readers

Non-YTH domain-containing m⁶A readers recognize and bind m⁶A-modified RNAs via alternative modes of recognition and are often dependent on the cellular context.

One such m⁶A reader is the translation initiation factor eIF3. eIF3 can directly read m⁶A marks in the 5′ UTRs of mRNA and recruits the 43S pre-initiation complex to the translation start site. Through this mechanism, eIF3 can initiate m⁶A-dependent/cap-independent translation initiation under stress conditions (Meyer et al, 2015).

Insulin-like growth factor 2 mRNA-binding proteins 1, 2, and 3 (IGF2BP1/2/3) are a family of distinct m⁶A readers, which specifically bind and recognize m⁶A modifications within the GG(m⁶A)CU motif (Huang et al, 2018). IGF2BPs bind the m⁶A modifications via their K homology (KH) domain, though how exactly the binding specificity is achieved is still unclear. In contrast to YTHDF proteins, IGF2BP proteins promote the stability of their mRNA targets in the cytosol and promote their translation.

A novel pair of m⁶A reader proteins is FMR1 and its paralogs FXR1 and FXR2. These proteins have been known to bind RNAs with the consensus motifs ACUG/U or U/A/GGGA, which show strong overlap with the METTL3/METTL14 DRACH consensus sequence (Ascano et al, 2012). Indeed, FXR1 was recently identified as a sequence context-specific m⁶A reader (Edupuganti et al, 2017). Furthermore, FXR1 has been described to bind to m⁶A marks in nascent RNAs and recruit DNA 5-methylcytosine dioxygenase TET1 to active chromatin loci leading to DNA demethylation and reprogrammed chromatin accessibility (Deng et al, 2022).

IMP1, a recently described m⁶A reader, interacts with the modification via a dedicated hydrophobic platform, enabling a high-affinity interaction (Table 1). This interaction is sequence-independent but is embedded into the methylation-independent sequence preference of IMP1, GGAC, which has significant similarity with the METTL3/METTL14 DRACH motif (Nicastro et al, 2023). The methyl group within the binding sequence results only in a small change in the on-rate from k_on = 1.3 × 10⁵ ± 0.1 × 10⁵ M⁻¹s⁻¹ to k_on = 8.7 × 10⁴ ± 0.1 × 10⁴ M⁻¹s⁻¹ but a more significant decrease in the off-rate from k_off = 2.7 × 10⁻³ ± 0.4 × 10⁻³ s⁻¹ to k_off = 3.2 × 10⁻⁴ ± 0.7 × 10⁻⁴ s⁻¹, highlighting how the modification can specifically increase the lifetime of a protein–RNA complex.

Lastly, proline-rich coiled-coil 2A (PRRC2A) is a recently discovered m⁶A reader critical for male spermatogenesis. It recognizes m⁶A marks within DRACH sequences, but little is known yet on the molecular details that govern this interaction (Wu et al, 2019; Tan et al, 2023).