RNA modifying enzymes and their function in a chromatin context

Abstract

During the last years, exciting research has connected specific RNA modifications to chromatin, providing evidence for co-transcriptional deposition and function in gene regulation. Here, we review new insights gained from studying the co-transcriptional roles of RNA modifications, and their influence in normal and disease contexts. We also discuss how the availability of novel technical approaches could raise the translational potential of targeting RNA-modifying enzymes for the treatment of disease.

Regulation of chromatin plasticity has an established role in the control of gene expression and pivotal functions in normal and disease contexts. During the last decade, targeting of the epigenetic machinery emerged as a promising new therapeutic strategy, with a rising number of small molecules targeting a wide range of epigenetic regulators in pre-clinical or clinical development¹. These molecules selectively inhibit various chromatin-associated proteins, including histone methyltransferases such as MLL, EZH2 and PRMT1^2–5, bromodomain and extra-terminal family members such as BRD4^6,7 as well as histone demethylases or deacetylases including LSD1, members of the HDAC family and KDM4C^4,8,9. Recent efforts have connected several RNA binding proteins (RBPs) and long noncoding RNAs (lncRNAs) to chromatin regulation, suggesting that RNA acts as an important link between chromatin-associated proteins and transcriptional regulators^10,11. Furthermore, a number of long noncoding RNAs (lncRNAs) have been proposed to function in binding and modulating the activity of chromatin modifiers^12–14. One of these is the lncRNA Xist, which has a fundamental role in the inactivation of one X chromosome in female mammalian cells via recruitment of epigenetic and transcriptional repressors^15,16. Interestingly, Xist was recently shown to alter chromatin accessibility via a mechanism involving its binding to the chromatin remodeller BRG1, an event that leads to expulsion of the SWI/SNF complex from the X chromosome¹⁷.

Current advances in the field of post-transcriptional regulation of gene expression have revealed that the cellular RNA modifications, termed the ‘epitranscriptome’, are an appealing equivalent to DNA and histone modifications, despite clear differences such as copy number and half-life^18–20. Although vastly underexplored, mainly due to the limited availability of detection systems and reagents with sufficient sensitivity and specificity, there are, to date, more than 140 identified RNA modifications, the majority so far detected in noncoding RNAs (rRNA and tRNA). Despite the urgent need of unbiased and quantitative approaches to interrogate RNA modifications, a number of recent technological advances have aided the molecular understanding of selected RNA modifications. These technologies, including widely used RIP and CLIP-based approaches, utilise modification-specific antibodies and/or chemical reagents followed by high-throughput DNA sequencing to identify modifications on RNA nucleotides^22–31. Evidently, publications illustrating the context-specific contributions of modifications have been biased towards highly abundant modifications such as N⁶-methyladenosine (m⁶A), 5-methylcytosine (m⁵C) and pseudouridine (Ψ) for which specific analytical reagents are available^{22–24, 28, 32–34}. Promising recent developments such as sequence-contextual analysis of RNA modifications by mass spectrometry (MS), based on the adaptation of methodologies used for proteomics, open up new opportunities to accurately identify and quantify RNA modifications. However, these approaches are hampered by the lack of fast bioinformatic search algorithms and are currently also limited to the analysis of abundant RNA species^35–37.

During the last 5 years, several studies showed a clear connection between changes in the epitranscriptome and disease phenotypes. In particular, a number of RNA modifying enzymes have been shown to have a role in the development and maintenance of different types of cancer, predominantly dependent on their catalytic activity^38–42. These findings suggest that targeting the catalytic pocket of those enzymes could be a promising therapeutic avenue for various devastating diseases. Moreover, the latest major advances in gene editing and recessive screening have provided genome-wide and unbiased evidence for the essential nature of a large number of putative RNA-modifying enzymes further highlighting the therapeutic potential of these targets^43–47.

In this Perspective, we provide a brief overview of the latest discoveries pertaining disease - associated RNA modifications, with a focus on the underlying regulatory mechanisms and their biological effects in the context of chromatin. We also discuss the emerging potential for targeting epitranscriptomic mechanisms for the treatment of disease.

RNA modifications and the chromatin connection

In-depth characterisation of chromatin function over the last two decades has established the epigenome as a dynamic layer of information, involving chromatin modifying or remodelling enzymes and modification dependent chromatin interactors^38,48,49,50. Interestingly, a number of investigations have revealed unforeseen mechanistic links between chromatin and other enzymatically active proteins including kinases and metabolic enzymes, making the epigenetic landscape even more diverse^51–54. One such example is a novel nuclear function of JAK2, which was shown to phosphorylate histone 3 tyrosine 41 (H3Y41) leading to downstream effects in normal and malignant hemopoiesis⁵¹. In line with these observations, several studies have recently highlighted the possibility that RNA modifying enzymes and their co-factors could efficiently direct gene regulation through co-transcriptional mechanisms (Table 1). These studies suggest that RNA enzymes can be targeted to DNA by chromatin proteins⁵⁵, chromatin modifications⁵⁶ and by sequence specific DNA binding transcription factors⁵⁷. The functional consequence of this recruitment of RNA enzymes to site of transcription is not yet well understood, but one common theme is the regulation of the subsequent translation of the target RNAs in the cytoplasm^42,57.

Table 1. Overview of RNA modifications with chromatin-associated function.

RNA Modification	RNA Enzyme or co-factor	Chromatin-associated Function	Reference
m6A	METTL3	METTL3 and DNA polymerase κ (Pol κ) co-localise to UV-induced DNA damage sites and facilitate DNA repair and cell survival	73
m6A	METTL3	METTL3 and DGCR8 co-localise on chromatin following triggering of acute temperature stress, where they co-transcriptionally m6A-modify mRNAs for RNA degradation	66
m6A	METTL3	METTL3 is recruited by CEBPZ to transcriptional start sites (TSSs) of active genes and induces m6A modification within the coding region of the relevant mRNAs, leading to efficient mRNA translation by block of ribosomal stalling	54
m6A	METTL3/METTL14/WTAP	The interaction between SMAD2/3 and the METTL3-METTL14-WTAP complex promotes co-transcriptional binding of the m6A methyltransferase complex on selected mRNA and prevents their downregulation upon differentiation and cell control	74
m6A	METTL3/METTL14/WTAP	METTL14 dictates the binding of the m6A methyltransferase complex (METTL3, METTL14 and WTAP) to RNA polymerase II and the m6A marking of actively transcribed nascent RNAs via its preferential binding to H3K36me3	53
m5C	NSUN1	NSUN1 regulates the formation of a distinct active chromatin structure at nascent RNAs via its interaction with RNA polymerase II and BRD4	52
A-to-I	ADAR	A-to-I editing of double stranded RNAs via ADAR leads to gene silencing of heterochromatin through a mechanism involving Vigilin and its high affinity for inosine-containing RNAs	85
5′ phosphomethylation	MEPCE	MePCE preferentially binds to the tail of histone H4, functioning as a P-TEFb activator at specific genes important for cellular identity, while this histone binding interrupts MePCE’s RNA methyltransferase activity toward 7SK	89

N⁶-Methyladenosine (m⁶A)

m⁶A, discovered back in 1974, is the most abundant eukaryotic mRNA modification, with the majority of mRNAs containing approximately three modified residues^58–60. Catalysis of m⁶A is mediated by a methyltransferase complex consisting of METTL3 and METTL14, together with their cofactors WTAP, RBM15, RBM15B, ZC3H13, VIRMA and HAKAI^61–66. Moreover, the formation of the METTL3-METTL14 complex has been shown to be negatively affected by the interaction between the transcription factor ZFP217 and METTL3⁶⁷. Very recently, METTL16 was proposed as another m⁶A RNA methyltransferase, modifying the 3′-UTR of specific mRNAs such as MAT2A. METTL16 was also suggested to target important noncoding RNAs including the spliceosomal component U6 small nuclear RNA, which pairs with the 5’ splice sites of pre-mRNAs during splicing^68,69. m⁶A has been strongly linked to alterations of mRNA translation, mRNA degradation, nuclear-cytoplasmic mRNA export and splicing^34,70–74. The publication of the crystal structure of the METTL3/METTL14 complex revealed that METTL3 exclusively acts as the catalytic subunit of the complex, while METTL14 mostly functions as a structural scaffold with roles in complex maintenance and substrate binding^75–77. Analysis of the METTL3/METTL14 binding sites suggested that this protein complex targets m⁶A predominantly in coding sequences and 3’ UTRs, with a considerable distribution bias to the latter. In addition, a substantial portion of m⁶A is present on intronic sequences, indicating possible co-transcriptional functions of the modification⁷⁸.

Indeed, there is compelling evidence of co-transcriptional involvement of key members of the m⁶A methyltransferase complex. Initially, m⁶A was shown to act as a link between transcription and translation through the interaction of RNAPII and METTL3⁷⁹. The authors demonstrated that RNAPII and its dynamics effect the deposition of m⁶A, with suboptimal transcription leading to enhanced appearance of m⁶A on mRNAs and reduced translation. METTL3 was further reported to have a pivotal role in the UV-mediated DNA damage response⁸⁰. Upon UV-mediated DNA damage, DNA polymerase κ (Pol κ) and METTL3 colocalise to UV-induced DNA damage sites in order to facilitate repair and cell survival, a function strictly dependent on METTL3 enzymatic activity. The mechanism of METTL3 recruitment to DNA, however, remained unknown. Further connections of m⁶A to chromatin were shown using chromatin immunoprecipitation assays and sequencing (ChIP-seq). One of the studies demonstrated that METTL3 has a preferential enrichment at the 3’ end of protein-coding genes, reflecting the canonical m⁶A distribution on mRNAs⁷¹. The authors further showed that following acute temperature stress, METTL3 together with DGCR8 localises to selected chromatin regions, where they co-transcriptionally mark mRNAs for RNA degradation. Another study reported that in acute myeloid leukemia (AML), METTL3, independently from METTL14, localizes to transcriptional start sites (TSSs) of cancer-related genes and induces m⁶A modification within the coding region of the associated mRNAs, thereby boosting their translation by alleviating ribosomal stalling⁵⁷. Recruitment of METTL3 to TSSs is facilitated by the CAATT-box binding protein CEBPZ and mediates the translational efficiency of oncogenic factors like SP1 and SP2, creating a novel cancer pathway with appealing druggable potential. Remarkably, analysis of METTL3 and METTL14 ChIP-seq datasets in human AML cells revealed minimal genomic overlap suggesting separate roles for the two METTLs on chromatin.

The conceptual role of m⁶A in a chromatin context was further highlighted by a study demonstrating an interaction between the transcription factor SMAD2/3 and the METTL3-METTL14-WTAP complex in embryonic stem cells⁸¹. This interaction promotes co-transcriptional binding of the m⁶A methyltransferase complex to selected transcripts mainly involved in pathways regulating early cell fate decisions, thus marking them for fast-track downregulation upon differentiation. Intriguingly, histone H3 trimethylated at Lys36 (H3K36me3), a marker for transcription elongation, was found to directly promote global deposition of m⁶A⁵⁶. Specifically, METTL14 was shown to preferential bind H3K36me3, thus recruiting the METTL3-METTL14-WTAP complex to adjacent RNAPII, triggering subsequent m⁶A modification of the nascent RNA.

Collectively, there is considerable evidence supporting a chromatin-associated function of m⁶A, regulating the fate of nascent mRNA substrates in a context-dependent manner (Table 1). We anticipate that further studies will not only provide evidence for new roles of m⁶A but also dissect how these novel functions are related to chromatin regulation.

Other RNA modifications

Despite the progress in understanding the functions of m⁶A, there is relatively little known about other RNA modifications, although this gap narrowing, also revealing connections to chromatin (Table 1). 5-methylcytosine (m⁵C) modification of RNA was identified several decades ago^82,83. m⁵C is an abundant modification found in both DNA and RNA, and several RNA modifying enzymes have been identified. The majority of the known m⁵C methyltransferases belong to the highly conserved NOL1/NOP2/Sun (NSUN) domain containing family^84,85. NSUN2 had originally been identified as a tRNA m⁵C methyltransferase, but it is also responsible for m⁵C methylation in several mRNAs and lncRNAs^24,86,87. Furthermore, the export factor Aly/REF was shown to recognize m⁵C and control the export of bound mRNA transcripts in an NSUN2-dependent mechanism⁸⁷. m⁵C methylation of tRNAs by NSUN2 has been linked to the generation of small tRNA fragments, regulation of protein synthesis and the stress response, as well as normal development and carcinogenesis^41,88. It was also reported that m⁵C on RNA appears to be increased in circulating tumour cells of patients with lung cancer further indicating a potential role in carcinogenesis⁸⁸. However, a co-transcriptional impact of m⁵C was not suggested until recently, despite the established presence of the modification on mRNAs. Pioneering data now demonstrated that the catalytic activity of NSUN1 promotes formation of a distinct active chromatin structure at nascent sites of transcription, through interactions with RNA polymerase II and BRD4⁵⁵. Future efforts will likely uncover additional co-transcriptional mechanisms behind the catalytic function of m⁵C methyltransferases.

The conversion of adenosine nucleosides to inosine (A-to-I), orchestrated by the ADAR family of enzymes, is one of the most prevailing form of post-transcriptional RNA modification. Recent efforts have identified important functions of ADAR enzymes, but there is still uncertainty regarding the functional and regulatory role of A-to-I editing events per se^89–91. It is also unclear how exactly ADARs recognise a specific A-to-I RNA editing site and whether recognition occurs co-transcriptionally. Importantly, however, A-to-I editing of double stranded RNAs leads to heterochromatic gene silencing through a mechanism involving Vigilin and its high affinity for inosine-containing RNAs⁹².

The catalytic activity of MePCE has been tightly linked to O-methylation of the 5′ triphosphate on the gamma position of RNA molecules (5′-Pγ)^93,94. MePCE inhibits the degradation of the small nuclear RNA 7SK either through direct binding or by 5′-Pγ-methyl capping⁹⁵. Reduction of MePCE negatively affects migration and invasion of human breast cancer cells suggesting that this methyltransferase plays a role in tumorigenesis⁹⁶. Besides, MePCE loss modulates the expression of oncogenic pathways associated with breast cancer via a mechanism involving RNAPII, indicating a possible chromatin association. Interestingly, results from in vitro assays have illustrated that MePCE preferentially binds to the tail of histone H4, an interaction mediated by the C terminal of its S-adenosyl-methionine-binding domain (SAM-BD)⁹⁶. BCDIN3D, a homolog of MePCE has been previously connected to dimethylation of the 5′-phosphate on specific microRNA precursors, which blocks Dicerdependent processing³⁹. To date, there is however no evidence suggesting a co-transcriptional function of BCDIN3D.

One of the most evolutionarily conserved modifications on eukaryotic mRNA is the 5′ m⁷G cap. The m⁷G RNA modification operates as a distinctive molecular module that facilitates functions such as pre-mRNA splicing and nuclear export as well as cap-dependent protein synthesis. Cap-related m⁷G methylation is mediated by the methyltransferase RNMT⁹⁷. Conversely, deposition of m⁷G on tRNAs and rRNAs is mediated by the methyltransferase METTL1, in association with its co-factor WDR4⁹⁸. METTL1 has been shown to be critical for the growth of several cancer types^43,47,57 and it is well-known that mutations in WDR4 are causing microcephalic primordial dwarfism in humans⁹⁹. Recent evidence suggested that reduction of m⁷G on tRNAs due to loss of METTL1 leads to global effects on tRNA function as well as defects in embryonic stem cell self-renewal and differentiation¹⁰⁰. Moreover, two studies recently revealed that METTL1-mediated m⁷G also occurs internally in miRNAs and mRNAs, despite the general dogma that the modification in mRNAs is limited to the cap^101,102. m⁷ G modification of miRNAs was suggested to increase miRNA maturation and processing by disruption of the inhibitory secondary structure of G-quadruplexes present in numerous miRNA precursors¹⁰¹. It was additionally demonstrated that the catalytic activity of METTL1 was associated with suppression of lung cancer cell migration, indicating a possible tumour suppressive role¹⁰¹. Furthermore, loss of internal m⁷G on mRNAs by METTL1 knockdown was shown to affect mRNA translation of m⁷G-modified transcripts¹⁰². The identification of m⁷G on miRNA and mRNA is expanding the functional repertoire of the gene regulation control by the epitranscriptome. It will be interesting now to investigate whether there are links between m⁷G modification of target RNAs and chromatin.

Concluding remarks and future perspectives

Recent discoveries in epitranscriptomics have revealed novel functions for RNA modifications in nearly every class of cellular RNA, some with strong links to disease. To date, there are however only few studies illustrating epigenetic–epitranscriptomic regulatory feedback loops and synchronization between these two regulatory layers. Work in this area is evidently at an early stage. Nevertheless, the observed interplay between specific RNA modifications and chromatin suggests promising therapeutic opportunities. Indeed, in various cancer types, there is remarkable overlap between druggable targets or pathways regulated by both chromatin and RNA modifications. For instance, the control of oncogenic proteins like MYC, BCL2 and BRD4 has been linked to m⁶A-mediated mechanisms, supporting the therapeutic promise of targeting the catalytic function of METTL3 as an anticancer approach^42,57,103. Intriguingly, BRD4 is functionally involved in an NSUN1 and m⁵C-dependent co-transcriptional mechanism that regulates the response of cancer cells to 5-Azacitidine⁵⁵. These kinds of studies draw a translational path for future efforts in identifying effective combinatorial therapies using clinically approved (or under clinical development) “epigenetic” drugs coupled with small-molecules inhibitors of selected RNA modifying enzymes.

Pharmacological targeting of key RNA modifying enzymes is anticipated and we envisage that it will be beneficial in several diseases. While RNA methyltransferase inhibitors have not yet been described, inhibitors for the m⁶A demethylase FTO have recently shown encouraging anticancer effects in a preclinical study¹⁰⁴. However, given that individual RNA modifications have been shown to mediate contrasting phenotypes in different tissue types^57,101, it will also be crucial to assess the potential caveats and general effects of such therapies. A strong paradigm is m⁶A, which was shown to either promote or disrupt the development and/or maintenance of malignant phenotypes dependent on the tumor type^42,105–108. Similarly, loss of METTL1-mediated m⁷G is promoting migration and invasion in lung cancer¹⁰¹ while it appears to be essential for acute myeloid leukemia cells⁵⁷.

An important avenue to probe therapeutic and functional interactions between druggable chromatin targets and modifiers of RNA are CRISPR sensitisation screening platforms in clinically relevant disease models. Such powerful genetic approaches have the potential to unearth synthetic lethal interactions that could provide patient stratification routes and link RNA modifications with an improved response to existing therapies or evasion of drug resistance. In that regard, a recent publication presented compelling evidence that RNA modifying enzymes responsible for the mcm⁵s²U modification on the wobble U position in tRNA anticodons are key mediators of resistance to BRAF inhibitors observed in BRAF^V600E-melanomas¹⁰⁹.

As the field of epitranscriptomics develops, we anticipate novel chromatin associations. New and improved detection technologies could pave the way for the exploration of other, currently uncharacterised RNA modification. In addition, the development of specific pharmacological inhibitors of RNA modifying enzymes will revolutionise research in this field and open exciting new approaches for the treatment of cancer and other diseases.