m⁶A: Signaling for mRNA splicing

ABSTRACT

Among myriads of distinct chemical modifications in RNAs, dynamic N6-methyladenosine (m⁶A) is one of the most prevalent modifications in eukaryotic mRNAs and non-coding RNAs. Similar to the critical role of chemical modifications in regulation of DNA and protein activities, RNA m⁶A modification is also observed to be involved in the regulation of diverse functions of RNAs including meiosis, fertility, development, cell reprogramming and circadian period. The RNA m⁶A modification is recognized by YTH domain containing family proteins comprising of YTHDC1-2 and YTHDF1-3. Here we focus on the nuclear m⁶A reader YTHDC1 and its regulatory role in alternative splicing and other RNA metabolic processes.

There are more than 100 distinct chemical modifications in RNAs and one third of such modifications comprise the addition of methyl groups to the RNA base. Among them, N6-methyladenosine (m⁶A) modification is the most prevalent one.¹ The advance of high throughput sequencing technology has enabled us to identify the location and distribution of RNA modifications at single base resolution. RNA m⁶A methylation occurs at the N-6 position of the adenosine residue in the consensus motif of RRACH (R represents purine and H is either A, C or U) catalyzed by METTL3 and METTL14 heterodimer.² WTAP and KIAA1429 serve as regulatory factors in this process.^3,4 The occurrence of RNA m⁶A modification is highly selective despite of abundant RRACH motif along the RNA. Moreover, m⁶A is particularly enriched in the 3′ untranslated regions (UTRs) and within internal exons, suggesting their evolutionarily conserved significance.^5-7 This modification is reversed by demethylases FTO and ALKBH5 demonstrating the reversible and dynamic nature of m⁶A modification.^8,9 FTO was discovered as the first RNA m⁶A demethylase in year of 2011 followed by the identification of another m⁶A demethylase ALKBH5 in 2013. Since then, a number of studies have defined the landscape and significance of RNA m⁶A methylome, which largely advances the RNA field.

The biological function of dynamic RNA m⁶A modification is also mediated by another class of proteins capable of recognizing m⁶A, such as YTHDC1. YTHDC1 localizes in the nuclear compartment called YT bodies adjacent to nuclear speckle, the site for RNA processing.¹⁰ Recent studies illustrate that RNA m⁶A modification serves as mark for the recruitment of splicing factors and then affects alternative splicing (AS) events. YTHDC1 has been reported to affect AS patterns in a concentration dependent manner but the underlying mechanism remains unclear.¹¹ Here we shed light on the mechanism of how m⁶A and its reader YTHDC1 coordinately regulate AS events in mammalian systems.

YTHDC1 and other m⁶A readers

YTHDC1 (also known as YT521-B) is the founding member of YTH (YT521-B homology) domain containing protein family. YTH domain-containing family proteins were initially identified as single stranded RNA binding proteins through YTH domain at the C terminus. YTH domain contains 100–150 amino acid residues capable of binding short, degenerated, single-stranded RNA sequence motif.¹² Biochemical and biophysical investigations have shown that the aromatic cage formed by tryptophan residues are critical for the recognition of m⁶A in RNA by YTH domain containing proteins.¹³ Humans have 5 YTH domain containing proteins named YTHDC1–2, and YTHDF1–3.

YTHDC1 is nuclear m⁶A reader reported to bind to m⁶A through its tryptophan residues at 377 and 428 positions by forming aromatic cage.¹⁴ The amino-terminal glutamic acid rich region and the carboxyl-terminal glutamic acid/arginine rich-region of YTHDC1 are critical in its nuclear localization where it involves YT body formation.¹¹ Phosphorylation of YTHDC1 by src kinases c-src and p59^fyn affects its solubility and distribution in the YT body leading to its diffusion in the nucleoplasm and disruption of its dynamic interactions with pre-mRNA protein complexes.¹⁵ Owing to the proximity of YT bodies to nuclear speckle, it is possible that YTHDC1 gets access to nascently transcribed pre-mRNA facilitating the recruitment of other YTHDC1-associated splicing factors.

YTHDF1, YTHDF2 and YTHDF3 are cytoplasmic m⁶A readers.¹⁴ YTHDF1 has been shown to promote translation by interacting with the translation machinery where it can selectively recognize the m⁶A modified RNAs from the pool of RNAs and facilitate their loading on to the ribosome.¹⁶ YTHDF2 has been shown to regulate the stability of cytoplasmic mRNAs by targeting them to RNA decay site, such as processing bodies, and is also implicated in promoting translation initiation of selective transcripts by interacting with the translation initiation factors under heat shock stress.^17,18

There are several additional factors that have been shown to serve as the m⁶A readers. HNRNPA2B1 binds to the consensus m⁶A modified site both in vivo and in vitro. Upon binding to m⁶A, HNRNPA2B1 regulates AS events in a subset of its targeted RNAs and miRNA processing in a m⁶A-dependent manner, however, the underlying mechanism needs further investigation.¹⁹ ELAVL1, a ubiquitously expressed RNA stabilizer, also serves as a potential m⁶A reader identified by m⁶A containing RNA bait.⁵ ELAVL1/HUR binding to m⁶A occurs only when its binding motif is adjacent to m⁶A, and meanwhile this binding is spatiotemporally regulated. METTL3/METTL14-mediated m⁶A modification in developmental regulator-associated RNA transcripts has been shown to interfere the m⁶A binding of ELAVL1/HUR leading to destabilization of the target transcripts.²⁰ eIF3 is another m⁶A reader, and the findings from in vivo and in vitro experiments revealed that eIF3 facilitates recruitment of 43S preinitiation complex to 5′ UTR of mRNA and promotes translation through selectively binding to m⁶A residues in 5′UTR.²¹

Alternative splicing regulated by m⁶A reader YTHDC1

Alternative splicing, the process producing multiple mRNA variants from a single gene transcript and ultimately diversifying the mammalian proteome, was discovered along with splicing.^22,23 Almost 95% of human genes undergo AS allowing them exert different regulatory functions from chromatin modification to signal transduction.²⁴ AS is regulated in a tissue specific way with striking variations depending on developmental or differentiation cues. Many factors such as protein regulators (trans-acting splicing factors) and RNA sequence elements (cis-regulatory elements) play a coordinated role in AS regulation.²⁵ m⁶A-dependent AS regulation is a well-observed phenomenon.^5,6 This is supported by our previous studies showing that methyltransferase METTL3, its regulator WTAP and m⁶A demethylases FTO and ALKBH5 all influence alternative splicing.^3,9,26 Here we will discuss the potential significance of ubiquitous RNA m⁶A modification in AS regulation through its nuclear reader YTHDC1.

Nuclear m⁶A reader YTHDC1 has been implicated in the regulation of splicing by maintaining a dynamic interaction network of different family of proteins. In our recent study by Xiao et al,²⁷ we found that YTHDC1 interacts with splicing factors SRSF1, SRSF3, SRSF7, SRSF9, and SRSF10, suggesting its potential role in the course of pre-mRNA splicing. Analysis of AS event using RNA-seq following percent spliced in (PSI) technique revealed that YTHDC1 and SRSF3 promote exon inclusion, whereas SRSF10 facilitates exon skipping. SRSF1, SRSF7 and SRSF9 failed to show any significant effect on the splicing pattern of their bound targets, indicating that they may function in other aspects of RNA metabolism. We investigated the AS change patterns of the representative target genes by RT-PCR analysis and found that YTHDC1/SRSF3 and SRSF10 differentially regulate the splicing in a m⁶A-dependent manner, which is supported by the observation that the binding regions of YTHDC1, SRSF3 and SRSF10 were significantly enriched in coding sequences and 3′UTR, in agreement with the m⁶A distribution pattern. Moreover, YTHDC1 clusters showed a significant enrichment in the longer exons with enriched m⁶A modifications, indicating a preferential regulation of AS in longer exons by YTHDC1. Depletion of RNA m⁶A methyltransferase METTL3 had a similar effect on the splicing of the target exons as that of YTHDC1/SRSF3 depletion suggesting a m⁶A-dependent splicing regulation.

Comprehensive analysis of m⁶A distribution along the mRNA revealed that m⁶A peaks do not significantly coincide with exon-exon junctions. As such, m⁶A may not directly influence the mRNA binding of the splicing factors.⁶ We found a significant representation of binding clusters of YTHDC1, SRSF3 and SRSF10 in exonic regions near splicing sites. Additionally, the motif enriched in YTHDC1 binding clusters coincided well with m⁶A consensus motif of RRACH, whereas the motifs in SRSF3 and SRSF10 binding clusters were not fully consistent with m⁶A motif even though they were quite similar to a significant extent. Also it was revealed that both SRSF3 and SRSF10 binding motifs are in a close proximity with the binding region of YTHDC1, however, SRSF3 binding motif is closer to YTHDC1 binding region than that of SRSF10. These findings provide a logical explanation that YTHDC1 modulates the recruitment of these factors to their target motifs in a m⁶A-dependent manner, and at the same time creates spatial competing relationship favoring binding of either SRSF3 or SRSF10 only to its target RNA. This was further validated by the observation that both SRSF3 and SRSF10 interact directly with N-terminal of YTHDC1 through their RS domain located at the carboxy terminus, and YTHDC1 is competitively bound by SRSF3 and SRSF10. The competitive binding of SRSF3 and SRSF10 to YTHDC1 provides a plausible explanation for the antagonistic effect of SRSF3 and SRSF10 on inclusion of target exons, and therefore, provides a novel regulatory mechanism for cells to favor required exon inclusion/skipping event. The interaction of YTHDC1 with SRSF3 may sequester YTHDC1 leading to the decrease of its availability for interaction with SRSF10 and consequently predominant recruitment of SRSF3 to its target site. It is also likely that the in vivo interaction of YTHDC1 and SRSF10 may induce their conformational changes masking their own RNA-binding motifs consequently blocking the binding of YTHDC1 and SRSF10 complex to their target sites.

YTHDC1 forms YT bodies adjacent to nuclear speckle supporting the assumption of its involvement in the regulation of splicing. We found that YTHDC1 not only interacts with SRSF3 and SRSF10 but also regulates the patterns of their nuclear speckle localization in a m⁶A-dependent manner. YTHDC1 promotes SRSF3 but inhibits SRSF10 in their nuclear speckle staining. Additionally, SRSF10 nuclear speckle staining pattern was diminished by SRSF3 and vice versa. Interestingly, the antagonistic effect of YTHDC1 on the regulation of nuclear speckle-staining patterns of SRSF3 and SRSF10 is m⁶A-dependent. These observations provide further evidence that the differential splicing outcomes regulated by YTHDC1 are in a m⁶A dependent manner.

The observations that SRSF3 and SRSF10 competitively bind to YTHDC1 and YTHDC1/SRSF3 antagonize SRSF10 in the regulation of both nuclear speckle staining and AS events suggested that YTHDC1 is involved in the regulation of RNA binding ability of SRSF3 and SRSF10. This was further proved by PAR-CLIP (photoactivatable ribonucleoside crosslinking and immunoprecipitation) in combination with RNA 3′ end biotin-labeling assay. We found that YTHDC1 indeed affects the RNA binding ability of its interactors SRSF3 and SRSF10. YTHDC1 enhances SRSF3 but inhibits SRSF10 in their RNA binding ability. Consistent with the antagonistic effect of SRSF3 and SRSF10 on their own nuclear speckle staining pattern, a similar role in RNA binding ability was also observed between them. Interestingly, SRSF10 inhibited the RNA binding ability of YTHDC1 whereas SRSF3 did not show such effect.

The presence of RNA m⁶A modification was found to be critical in the regulation of RNA binding ability of YTHDC1, SRSF3 and SRSF10, as evidenced by the PAR-CLIP experiments in control and METTL3 knockdown conditions. Our findings strongly support the assumption that YTHDC1 first binds to m⁶A modification and then recruits SRSF3 to its binding motif of its target RNA. This also explains how the competitive binding of YTHDC1 by SRSF3 and SRSF10 influences their RNA binding ability and hence leads to differential splicing outcomes. Since SRSF3 is more abundant than SRSF10, it is likely that under normal physiological condition, formation of SRSF3-YTHDC1 complex is dominant preferably favoring exon inclusion. Moreover, the effect of YTHDC1 on the RNA binding ability of SRSF3 and SRSF10 and further AS depends on its binding ability to m⁶A modification.

These discoveries provided substantial evidence regarding the significance of RNA m⁶A modification and its reader YTHDC1 in the regulation of AS through differentially modulating the RNA binding ability of splicing factors SRSF3 and SRSF10.

Future perspective

m⁶A as the most prevalent modification on mRNAs has long been identified but significant advances were only gained after the discovery of m⁶A demethylase FTO followed by the identification of another m⁶A demethylase ALKBH5 together with the invention of improved m⁶A detection approach. These studies concurred that RNA m⁶A modification is a reversible mark and has a significant role in the regulation of gene expression.^5,6,8,9 The next generation sequencing of m⁶A containing RNAs immunoprecipitated by m⁶A antibody revealed the presence of m⁶A in thousands of mRNA and hundreds of non-coding RNA (ncRNA) transcripts.^5,6 Given the significant enrichment of m⁶A in 3′ UTR, it is likely that m⁶A could affect the mRNAs targeted by miRNA and many RNA binding proteins with preferentially binding to m⁶A-modified regions. Further Investigations shall decipher the significance of m⁶A modification in 3′UTR. The distribution of m⁶A along the 5′ UTR has been implicated in the translational control but the underlying mechanism needs to be clarified.¹⁸ m⁶A modifications are also enriched in the intronic region, providing another possibility that the process of intron removal and then gene expression by recruiting splicing factors may be influenced or regulated. RNA m⁶A modification is also the most prevalent modification found in various different types of transcripts ranging from microRNAs to long noncoding RNAs, suggesting that m⁶A may also regulate ncRNA processing via its readers including YTHDC1.²⁷ m⁶A has been shown to modulates the local structure of coding and non-coding transcripts and hence regulate RNA-HNRNPC interaction affecting gene expression.²⁸ This suggests the role of prevalent m⁶A modification in gene expression regulation through structural remodeling of RNAs influencing their interaction with RNA binding proteins. So far only a handful of mRNA m⁶A readers are identified and have been implicated in splicing, stability and translation.^{5,13,15-20,26} Given the ubiquitous nature of m⁶A modification throughout different types of transcripts, there shall be more m⁶A binding proteins that are worthy of further investigation. Identification of such readers and further elucidation of their effect on the RNA metabolism will definitely expand our current understanding of the potential significance of m⁶A modification.

Disclosure of potential confllicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by NSFC31430022, CAS Strategic Priority Research Program XDB14030300 and CAS Precision Medicine Initiative KJZD-EW-L14.