Advances in RNA modification in myocardial fibrosis (Review)

Introduction

Myocardial fibrosis is a common end-pathologic phenotype in a variety of cardiomyopathies that is characterized by pathological dilatation of the myocardial interstitium and the abnormal remodeling of extracellular matrix (ECM) components (1). This ECM remodeling involves not only an imbalance in the ratio of type I/III collagen, but an abnormal cross-linking of fibronectin and disruption of the laminin network, thereby leading to an increase in the elastic modulus of myocardial tissue. In the context of acute myocardial injury, fibrosis represents an evolutionarily conserved mechanism of trauma repair. When the investigation of the repair response moved beyond its spatial and temporal constraints, fibrosis research shifted to pathological processes: A sustained activation of the TGF-β/Smad pathway was shown to lead to a reduced mitochondrial oxidative phosphorylation capacity in abnormally proliferating myofibroblasts. These cells were found to depend on glycolysis to maintain a high state of collagen synthesis and an excessive deposition of ECM was shown to result in a marked increase in ventricular stiffness and chronic, deregulated or excessively aggressive fibrotic responses in these tissues were observed to disrupt the structural framework of the heart's structural elements, thereby impeding tissue regeneration and leading to dysfunction (2).

The primary mechanism responsible for cardiac fibrosis is the process of overactivation and proliferation of cardiac fibroblasts (CFs). Activated CFs, also known as myofibroblasts, are the major contributors to ECM production, and are the primary effector cells of cardiac fibrosis (3,4). The abnormal activation and proliferation of CFs underlie the cell biology of cardiac fibrosis. Under physiological conditions, resident CFs exist in a relatively quiescent state and are primarily responsible for maintaining ECM homeostasis, cardiac structure and mediating electrophysiological conduction. However, under conditions of cardiac injury, such as myocardial ischemic injury and myocardial infarction, myofibroblasts secrete more ECM, leading to cardiac fibrosis. The fibrotic process is further exacerbated by sustained activation of the TGF-β/Smad signaling pathway. In previous studies, the binding of TGF-β1 to the TβRII receptor was shown to trigger the phosphorylation of Smad2/3, leading to their subsequent nuclear translocation (5-7). This process directly activates collagen synthesis genes, ultimately leading to a cascade of amplified fibrotic signals. In addition, previous studies have demonstrated that this pathological remodeling inhibits the migration of endogenous cardiac stem cells through the physical barrier effect of migration, ultimately leading to diastolic dysfunction and arrhythmias (8,9). The chronological application of therapeutic interventions, which promote moderate ECM deposition in the acute phase and inhibit pathological remodeling in the chronic phase, has provided novel therapeutic paradigms. Therefore, in order to design further new targets and therapeutic strategies, a deeper understanding of the processes of normal and pathological tissue remodeling is required.

Recent advances in fibrosis research have focused on elucidating the molecular mechanisms underlying cardiac fibrogenesis, with particular emphasis on identifying potential therapeutic targets. Among these, METTL3 has emerged as a key regulatory enzyme in the fibrotic process (10). As was demonstrated in a previously published study (11), METTL3, as the core catalytic component of the methyltransferase complex, performs a crucial role in mediating RNA modifications that influence fibrotic pathways. Despite the advances that have been made in understanding the processes of myocardial fibrosis, however, the therapeutic landscape for myocardial fibrosis remains limited, highlighting the urgent need for an even deeper understanding of its molecular mechanisms. Currently, the most intensively studied RNA modifications are N6-methyladenosine (m6A) and 7-methylguanosine (m7G) modifications, and their specific mechanisms of action in myocardial fibrosis have been partly elucidated. Advancements in this field of research will be crucial for the development of effective therapeutic strategies against this debilitating condition.

Types and functions of RNA modifications

RNA modifications can be categorized into three main groups, namely chemical motif modifications, nucleotide editing and dynamic processing, and other regulatory modifications. The present review focused on the types of RNA modifications in the first two categories. Among the chemical motif modifications, methylation modifications include m6A, methyladenosine (m1A), 5-methylcytosine (m5C) and m7G; acetylation modifications are mainly focused on N4-acetylcytidine (ac4C); and heterodimerization is mainly for pseudouridylation (Ψ), which specifically modifies the U34 position of transfer RNA (tRNA)s. Nucleotide editing and dynamic processing-type modifications include adenosine-to-inosine (A-to-I) base substitutions and uridylation in nucleotide addition (Fig. 1).

Figure 1 A graphical depiction of the chemical modifications occurring in RNA within mammalian cells is provided. (A) The chemical structures of various RNA modifications, such as involving mRNAs and microRNAs, as well as (B) Transfer RNAs, are shown. Specifically, the modification at area 34 of the tRNA regulates swing pairing, encompassing such as m5C, hm5C, and others. A-to-I, adenine-to-inosine; m5C, 5-methylcytosine; m6A, N6-methyladenosine; m1A, 1-methyladenosine, m7G, 7-methylguanosine; ac4C, N4-acetylcytidine.

m6A modifications

Since the discovery of nucleic acid methylation in HeLa cells in the 1970s, m6A and m7G have gradually entered the research landscape as important epitranscriptome markers (12,13). After half a century of intensive research, m6A has now been shown to be the most abundant type of chemical modification in mammalian mRNAs, and its dynamic and reversible modification properties have a key role in the regulation of gene expression. In 2012, researchers achieved the first systematic resolution of m6A modification at the whole transcriptome level in human and mouse through developing m6A-seq, a high-throughput sequencing technology (14). This groundbreaking study revealed that the m6A locus exhibits a remarkable spatial distribution feature, namely, that it is mainly enriched in two functional regions in the transcript: One is the long internal exonic region of the coding region of the gene, whereas the other is the 3'-untranslated region (3'-UTR) close to the translation termination codon. Notably, Meyer et al (15) further revealed, through a finer localization analysis, that ~70% of the m6A sites in the mammalian genome are densely distributed in the 3'-UTR interval 50-200 nucleotides (nt) downstream of the stop codon (16). This remarkable regiospecificity suggests that m6A may be involved in regulating the accessibility of microRNA (miRNA)-binding sites through either spatial site-blocking effects or recruitment of specific reading proteins. This discovery provides important clues towards elucidating the precise molecular mechanism via which m6A modification regulates post-transcriptional processes through RNA-protein interaction networks.

The molecular architecture and functional regulatory mechanisms of the m6A methyltransferase complex have both been gradually elucidated. The complex comprises METTL3 and METTL14 as the core catalytic subunits. Previous studies have shown that METTL3 directly mediates S-adenosylmethionine (SAM)-dependent methyl transfer reactions through its catalytic domain (17,18), whereas METTL14, though it was previously shown not to be catalytically active (16), was responsible for substrate recognition and spatial localization through its superior RNA-binding ability (19-21). The heterodimer formed by the two subunits not only constitutes the catalytic core of the methylation reaction, but its structural complementarity was shown to markedly enhance the thermodynamic stability of the complex (22).

Wilms tumor 1-associated protein, which has been identified as a key regulatory component of the complex, was found to be not enzymatically active (23). Previous studies revealed that it is able to regulate the methylation process through two key mechanisms: First, it was shown to specifically interact with the METTL3/METTL14 heterodimer, thereby promoting the aggregation of the complex in nuclear speckles (19,23); and second, as a splicing factor, it was shown to direct the complexes to target specific sites of pre-mRNA through a phase separation mechanism (19,24). This dynamic regulatory network ensures the spatiotemporal specific deposition of m6A modifications on transcripts.

In addition to the core components, earlier studies revealed that the m6A methylation system comprises multiple classes of cofactors with distinct regulatory roles. For example, Vir-like M6A methyltransferase associated/m6A methyltransferase KIAA1429 was previously shown to mediate co-localization of the complex with RNA polymerase II through acting as a molecular scaffold (25). Similarly, the m6A regulator, RNA binding motif protein 15/15B, has been shown to enhance the complex's affinity for specific RNA motifs through its RNA recognition motifs (26.) In addition, the E3 ubiquitin-protein ligase Hakai and zinc finger CCCH-type containing 13 were reported to stabilize the subcellular localization of the complex via deubiquitination modifications (27,28). These findings collectively demonstrate that the modular assembly of these cofactors enables the methyltransferase complexes to dynamically respond to cellular signals, thereby facilitating the precise regulation of m6A modifications (29).

In terms of substrate profile, the m6A modification catalyzed by METTL3 has a broad RNA-type specificity, covering classical RNA molecules such as mRNAs, tRNAs and ribosomal RNA (rRNA)s, as well as regulatory RNAs, such as miRNA precursors and long-stranded non-coding RNAs (lncRNAs) (30). Notably, among mammalian mRNAs, m6A modifications not only constitute the most abundant post-transcriptional type of modification (31), but its dynamic and reversible nature makes it a key regulatory node in the remodeling of gene expression networks in pathological processes, including those of cardiovascular diseases. This property has enabled METTL3 to gradually become an important research target for the study both of disease mechanisms and of the development of targeted therapies.

The functional realization of m6A modifications relies on the dynamic recognition of their methylation sites via specific reading proteins. Structural and mechanistic analyses have classified m6A reading proteins into three main categories. The first class comprises the canonical YT521-B homology (YTH) domain family (YTHDF1-3 and YTHDC1-2), where conserved aromatic residues within their hydrophobic binding pockets were shown to mediate multiple aspects of mRNA metabolism through the recognition of methylated adenosine residues (32,33). The second category comprises a family of non-classical RNA-binding proteins, including insulin-like growth factor 2 mRNA-binding proteins and heterogeneous nuclear ribonucleoproteins. Although these proteins lack the YTH domain, structural studies have demonstrated that their RNA recognition motifs, or K-homology domains, selectively bind m6A-modified sites to regulate RNA stability and subcellular localization. The third class of m6A reading proteins is defined by translation-regulating factors, such as eukaryotic initiation factor 3 (eIF3). Key studies have shown that eIF3 is able to recognize m6A modifications within mRNA 5'-UTRs through its hydrophobic binding pocket, thereby promoting ribosomal preinitiation complex assembly to exert upstream translational control, while concurrently influencing mRNA stability and localization (32,33). Demethylases, including AlkB homolog 5 (ALKBH5) and fat mass and obesity-associated protein (FTO), were demonstrated to reverse m6A modifications through enzymatic erasure mechanisms, as evidenced by biochemical and structural studies (34) (Fig. 2).

Figure 2 The biological functions of m6A modification are diverse. This reversible process involves 'writers' such as METTL3 and METTL14, and 'erasers' such as FTO and ALKBH5 that remove the modification. The 'readers', such as those with the YTH domain recognize and bind to m6A-modified RNA, thereby influencing RNA processing and its transport to the cytoplasm. Ultimately, m6A modulation controls RNA translation, stability and degradation within the cytoplasm. m6A, N6-methyladenosine; METTL, methyltransferase-like; FTO, fat mass and obesity-associated; ALKBH5, AlkB homolog 5; YTH, YT521-B homology.

m1A modifications

The biological functions of m¹A, a key modification in the epigenetic regulation of RNA, have been gradually unveiled with the advance of research techniques. While earlier studies focused on the distribution of m¹A in fungal tRNAs and rRNAs, a breakthrough discovery based on the methylated RNA immunoprecipitation sequencing (MeRIP-seq) technique showed that m¹A modifications are also present in the coding region (CDS) and the 5'-UTR region of mammalian mRNAs and that the degree of their enrichment is markedly positively associated with the number of variable translation initiation sites (35). However, Grozhik et al (35), by performing chemometric analysis, showed that the abundance of m¹A modifications in mRNAs is markedly lower compared with other RNA types, suggesting that they may exist as precise markers for specific functional sites, rather than as global modifications.

At the methylation regulatory level, members of the tRNA methyltransferase (TRMT) family have been shown to exhibit a subcellular functional division. TRMT61B, which was identified as the first human m1A methyltransferase, primarily catalyzes modifications of cytoplasmic tRNAs and rRNAs (36,37), whereas TRMT10C specifically targets mitochondrial tRNAs and was further demonstrated to maintain mitochondrial homeostasis through the regulation of energy metabolism (38). Previous studies have established that the AlkB homolog (ALKBH) family enzymes precisely coordinate RNA demethylation dynamics; for example, ALKBH3 promotes functional tRNA fragment generation via erasing tRNA methylation, and actively mediates m1A-dependent transcript degradation through its demethylase activity (39,40). Another study (41) showed that ALKBH1 is able to enhance the pool of translationally active tRNAs through selectively removing m¹A modifications at position 58 of tRNAs, thereby increasing their stability. Moreover, ALKBH7 was characterized as a mitochondrial pre-tRNA-specific demethylase that participates in cellular energy metabolism reprogramming (39). Furthermore, FTO not only catalyzes m6A demethylation, but it also recognizes stem-loop structures that enable it to demethylate tRNA m¹A modifications, thereby markedly boosting nucleocytoplasmic protein translation flux (42). In skin scar research, staphylococcal nuclease and tudor domain containing 1 (SND1) was identified as an m¹A reader protein that both enhances RNA stability and drives fibroblast proliferation and collagen deposition in pathological scarring. However, its role in cardiovascular diseases remains poorly understood (43). From the perspective of the functions of SND1 and the existing mechanisms of cardiac fibrosis, future studies should explore the role of SND1 in the progression of cardiac fibrosis by investigating its effect on cardiac fibrosis-associated signaling pathways (such as the TGF-β signaling pathway), gene expression regulation and cellular metabolism.

Notably, m¹A modifications have been shown to exhibit significant environmental stress sensitivity. Under serum starvation or oxidative stress (H2O2-stimulated) conditions, the abundance of m¹A modification sites in mRNAs was found to be markedly increased (39), whereas glucose deprivation led to a decrease in m¹A tRNA levels through the inhibition of TRMT enzyme activity, an effect that could be reversed by knockdown of the ALKBH1 gene (41). This dynamic modification pattern suggests that m¹A may act as a metabolic stress receptor that helps cells to adapt to micro-environmental changes via modulating translational reprogramming. For example, under energetic stress conditions, dynamic changes in mitochondrial m¹A modification would enable rapid regulation of oxidative phosphorylation activity by influencing the translational efficiency of the respiratory chain complex mRNA (Fig. 3).

Figure 3 m1A and m5C modifications. (A) Within the cell nucleus, mRNAs and tRNAs undergo methylation by enzymes including TRMT6 and TRMT61A. Specifically, the methylation of m1A on mt-tRNA, rRNA and mRNA is catalyzed by TRMT10C/TRMT61B. The enzymes FTO and ALKBH1/3/7 are responsible for controlling the demethylation process. Functionally, m1A employs various mechanisms to regulate translation and RNA stability, with YTHDF serving as a potential 'reader' of m1A modifications. (B) Aspartic acid tRNA is a specific target for methylation by the enzyme TRDMT1. The methyltransferases NSUN1-6 function as 'writers' of m5C modifications in RNA, influencing RNA nuclear export, metabolism, and function. Various proteins, including ALYREF, YBX1 and YTHDF2, have been reported to serve as 'readers' of m5C modifications. Additionally, TNT2, ALKBH1 are responsible for catalyzing the oxidation of m5C rather than its elimination. m1A, 1-methyladenosine; m5C, 5-methylcytosine; Mt-mitochondrial; ALKBH1, AlkB homolog 1; YTH, YT521-B homology; FTO, fat mass and obesity-associated; ALKBH5, AlkB homolog 5; YTH, YT521-B homology; rRNA, ribosomal RNA.

m5C modifications

As a key contributor to RNA epigenetic modification, the biological significance of m5C has undergone a cognitive revolution from DNA-specific modification to widespread regulation by RNA. during the early period of its investigation, m5C was considered to exist only in DNA due to limitations in the technologies available for detection, although given the breakthroughs that have been made in high-throughput sequencing and mass spectrometric techniques, researchers have discovered that it is widely distributed in eukaryotic RNAs, and is particularly enriched in high abundance in the anticodon loops of tRNAs and in conserved regions of rRNAs (44). m5C deposition has been shown to be mediated by the NOL1/NOP2/SUN structural domain (NSUN) family, in concert with TRDMT1. In the tRNA modification system, NSUN2 was identified to catalyze the methylation of both miRNA precursors and mRNAs (45,46), whereas NSUN3, NSUN6 and TRDMT1 were shown to form a ternary system targeting tRNA-specific sites, respectively (45,47-49); moreover, functional analyses revealed that, in the ribosomal RNA (rRNA) modification system, NSUN1 and NSUN5 regulate the structural maturation of rRNAs, whereas NSUN4 participates in respiratory chain complex assembly through mitochondrial rRNA/mRNA methylation (50-53).

A previous study also revealed diversified m5C recognition mechanisms; namely, the YTH domain-containing family protein 2 (YTHDF2)-mediated degradation of m5C-methylated mRNAs through its YTH domain, the Y box-binding protein 1 (YBX1)-mediated stabilization of m5C-modified transcripts via cold shock domain binding, and the functioning of Aly/REF export factor (ALYREF) as a nuclear export factor that facilitates the cytoplasmic translocation of m5C-marked mRNAs through specific RNA-protein interactions (54). Notably, the mechanism of dynamic regulation of m5C remains controversial, given that its reversibility has not yet been clarified (unlike that of m6A/m¹A) and this stability implies that m5C modifications have a unique role in long-term episodic memory regulation. Currently, the existing experimental evidence has established that m5C modifications are involved in organ growth, initiation of various pathological conditions, and developmental processes (47,54-57). Taken together, these findings have not only expanded the regulatory dimensions of RNA epigenetic inheritance, but have also provided novel perspectives for the continuing study of disease mechanisms and targeted therapies.

m7G modifications

Since the discovery of m7G modification in the mammalian mRNA 5'-cap structure in the 1970s (58,59), its biological function is now realized to extend from classical mRNA stability regulation to multilevel gene expression networks. Genome-wide mapping has further revealed that, beyond mRNA cap structures, m7G is distributed across non-coding RNAs, including rRNAs, tRNAs and miRNAs, and that this dynamic distribution pattern suggests pervasive roles in post-transcriptional regulation (60).

Another previously published study (61) demonstrated that precise m7G deposition relies on synergistic multi-enzyme complexes; namely, that the mammalian mRNA cap methyltransferase complex, RNMT-RAM, specifically catalyzes m7G modifications in 5'-cap structures by spatially coupling mRNA-capping enzymes into functional post-transcriptional processing modules. In addition, the WBSCR22-TRMT112 complex utilizes the SAM-binding structural domain to catalyze guanosine methylation at position 1,639 of 18S rRNA, a modification that served as a critical checkpoint for the maturation of the small subunit of the ribosome (60) and the METTL1-WDR4 complex has been shown to exhibit multi-functional catalytic properties that are responsible for mRNA, tRNA and miRNA-associated m7G modifications (62-64).

At the level of functional regulation, m7G coordinates RNA metabolism and translation through multiple mechanisms, including prolongation of the mRNA half-life of cap-structured m7G via the inhibition of 5'→3' nucleic acid exonuclease activity (65-67), whereas internal m7G modifications have been shown to promote RNA nucleocytoplasmic translocation by binding to translocation factors. During translation, METTL1-mediated modification of internal m7G mRNA enhances the translation initiation efficiency by reconfiguring the ribosome-binding interface, whereas the m7G modification of tRNA led to an enhancement of the decoding efficiency by optimizing the precision of codon-anti-codon pairing (63,68). Non-coding RNA maturation is also regulated by m7G; for example, the m7G modification of 18S rRNA was identified to be a quality checkpoint for ribosome biogenesis (60), whereas m7G modification of miRNA precursors was shown to regulate the abundance of mature miRNAs by altering the accessibility of the cleavage site of Drosha, a ribonuclease III enzyme that performs a crucial role in the biogenesis of miRNAs.

A cryo-electron microscopy study revealed that the METTL1-WDR4 complex achieves substrate-specific switching through conformational changes: When the complex bound to tRNA, the WD40 structural domain of WDR4 was found to induce active-pocket deformation, which led to a precise recognition of RAGGU motifs (69). Abnormalities in this dynamic catalytic mechanism are closely associated with neurodevelopmental disorders and tumor metastasis; for example, METTL1 deletion led to tRNA fragmentation, with the consequent activation of the p53-dependent apoptotic pathway (63,68). In glioblastoma, METTL1 overexpression was shown to promote tumor cell invasion by enhancing the methylation of oncogenic miRNA (for example, miR-21) precursors (64). Taken together, these findings have not only shed light on the molecular regulatory network of m7G modification, but they also may provide novel ideas for precision therapy targeting RNA methylation.

ac4C modifications

As the only known RNA acetylation modification, the biology of ac4C has undergone a cognitive leap from prokaryotic to eukaryotic systems. Whereas early studies focused on the ac4C modification of bacterial tRNA anticodon stem loops and conserved regions of fungal rRNAs (70), the discovery of N-acetyltransferase 10 (NAT10) revealed the widespread presence of this modification in mammalian RNAs. NAT10 catalyzes, through ATP-dependent acetyltransferase activity, the formation of ac4C in a conserved region of the 18S rRNA site (namely, at position 1,842; ac4C1842). Further structural analysis by cryo-electron microscopy revealed that ac4C1842 serves as a molecular checkpoint for the assembly of the small subunit of the ribosome through stabilizing the helix 69 conformation of the 18S rRNA; its absence led to aberrant processing of the rRNA precursor (70).

In the dynamic regulation of tRNAs, the deletion of NAT10 was shown to lead to a decrease in the overall acetylation level of tRNAs in human colon cancer cells, thereby markedly affecting their structural stability (69). Of particular note, ac4C modification close to the tRNA anticodon swing site led to an increase in ribosome decoding efficiency via the optimization of codon-anticodon geometry matching (71). Subsequent mRNA-level studies identified that ac4C in the CDS region prolonged the half-life of transcripts by resisting 3'→5' exonuclease activity, whereas ac4C in the 5'-UTR region facilitated the translation initiation of upstream open reading frames (ORFs) through a 'leaky scanning' mechanism, via which the efficiency of the main ORFs was enhanced (71,72). Furthermore, in esophageal squamous cell carcinoma, NAT10 was found to specifically catalyze the acetylation modification of the lncRNA CTC-490G23.2, and this modification drove tumor progression (73).

Physiologically, NAT10 promotes the osteogenic differentiation of bone marrow mesenchymal stem cells by stabilizing runt-related transcription factor 2 mRNA, which thereby established its critical role in bone metabolic homeostasis (74). At the pathomechanistic level, NAT10 has been shown to be abnormally highly expressed in malignant tumors such as hepatocellular carcinoma and breast cancer and its stability is enhanced by the acetylation of pro-oncogene mRNAs (for example, Myc and Cyclin D1). Additionally, specific alterations were observed in the serum exosomal ac4C modification profiles of patients with esophageal carcinoma, thereby suggesting that it has potential as a liquid biopsy marker (73,75-79). For therapeutic applications, ac4C-modified synthetic mRNAs have been demonstrated to have unique advantages; specifically, repeated dosing was enabled through reducing Toll-like receptor 7/8-mediated immunogenicity and a 3-fold extension of the shelf-life of mRNA vaccines at ambient temperature was achieved by enhancing their thermal stability (80). These breakthroughs provide new directions for RNA acetylation-based precision medicine (Fig. 4).

Figure 4 m7G and ac4C modifications. (A) In the nucleus, METTL1-WDR4 serves as the 'reader' for m7G modifications in mRNAs, pre-miRNAs and tRNAs. WBSCR22-TRMT112 is responsible for reading 18S rRNA. The mammalian mRNA cap methyltransferase complex, RNMT-RAM, on the other hand, primarily targets mRNA for m7G modification. m7G has various roles in enhancing translation rates, in accelerating RNA export, in contributing to ribosome synthesis and in facilitating the maturation of miRNAs. (B) Within the nucleus, NAT10 and THUMPD1 primarily function as enzymes that catalyze ac4C modifications on rRNAs, tRNAs and mRNAs. These modifications are associated with translation efficiency, ribosome biosynthesis, and mRNA stability, respectively. METTL, methyltransferase-like; tRNAs m7G, 7-methylguanosine; ac4C, N4-acetylcytidine; miRNA, microRNA; rRNA, ribosomal RNA; NAT10, N-acetyltransferase 10; tRNA, transfer RNA.

Uridylation modifications

Uridylation, a typical representative of RNA 3'-end-tail modification, causes predominantly an enrichment in short poly(A)-tailed or U-tailed RNA molecules (81,82). Its catalytic system consists of the terminal uridylyl transferase (TUT, or TUTase) and terminal nucleotidyl transferase (TENT) families working in concert: TUT4/7 (ZCCHC11/6) was found to specifically recognize the 3'-overhanging structure of mature miRNAs, mediating uridylation modifications of precursor miRNAs, such as pre-let-7 (83,84). In addition, TENT2 was shown to regulate the loading efficiency of Argonaute proteins through selectively catalyzing the addition of uridine to the 3'-end of mature miRNAs (83), and TENT5C was shown to maintain mRNA stability via antagonizing the de-adenylation enzyme CCR4-NOT complex (85).

At the level of dynamic metabolic regulation, uridylation and de-adenylation form a precise balance: When poly(A) binding protein cytoplasmic 1 (PABPC1) bound to long poly(A) tails, it was found to inhibit TUTase activity through spatial site-blocking effects, whereas, in the short poly(A) tail state, PABPC1 was dissociated from the TUTase, thereby initiating the process of uridylation (85-87). This antagonistic mode coexists with a synergistic mode: The miRNA miR-1 was shown to promote de-adenylation through recruiting CCR4, while enhancing TUT4-mediated uridylation, thereby creating a positive feedback loop of mRNA decay (82,85). Mammalian cells have been found to clear uridylated RNA species through a dual pathway: 5'→3' degradation is initiated by XRN1 exonuclease, which specifically recognizes the TUT4/7-extended U-tail (88), whereas 3'→5' degradation is mediated via DIS3-like exonuclease 2 (DIS3L2), which recognizes the uridylation marker through the S1 structural domain. Exosomes remove uridylated mRNAs through the polyadenylate-binding nuclear protein 1 (PABPN1)-DIS3L2 axis, and the exosome also degrades miRNAs (89-91). RNA-binding proteins, such as human-antigen R, antagonize the degradative activity of DIS3L2 by binding to uridine-enriched elements to form protective ribonucleoprotein particles, and this dynamic equilibrium determines the fate of uridylated RNAs (92-95).

In miRNA biosynthesis, uridylation also serves a dual regulatory role: In terms of inhibitory modifications, the RNA-binding protein Lin28 was shown to recruit TUT4 to the GGAG motif of pre-let-7 and other pre-miRNAs, where the addition of an oligo-uridine tail (+UUU) was found to block Dicer cleavage (84). The operation of a similar mechanism was observed in the inhibition of miR-191 processing (96). In terms of mature miRNA functional modulation, uridylated miR-105 was shown to reprogram its target profile by altering seed sequence complementarity (97), whereas, in another study (98), TUT7-mediated pre-miRNA uridylation was found to promote the differential loading of Argonaute-2 (Fig. 5).

Figure 5 Uridylation and A-to-I editing. (A) In the cytoplasm, TUT2, TUT4 and TUT7 are responsible for the addition of uridylates to pre-miRNAs, fully mature miRNAs and mRNAs, whereas TUT1 is responsible for the uridylation of snRNAs in the nucleus. The maturation of miRNAs is regulated by TENT2. Uridylation is important in the maturation of miRNAs, snRNAs, the activity of miRNAs, decay. (B) ADAR1 and ADAR2 function as the 'writer' of A-to-I editing for mRNAs, mature miRNAs and pri-miRNAs. ADAR1/2 affects not only mRNA stability and production, but also miRNA maturation and function. A-to-I, adenine-to-inosine; TUT, terminal uridylyl transferase; miRNA, microRNA; snRNA, small nuclear RNA; ADAR, adenosine deaminase acting on RNA.

A-to-I RNA editing modifications

The adenosine deaminase acting on RNA (ADAR) family has been demonstrated to remodel RNA function through A-to-I editing, and its sites of action are frequently co-localized with Alu elements and long interspersed nuclear elements retrotransposons in the genome (99,100). Although ADAR1 and ADAR2 share catalytic structural domains, they display significant functional heterogeneity: ADAR1 preferentially recognizes the quasi-double-stranded regions formed by Alu elements through its double-stranded RNA-binding domain (101), whereas ADAR2 targets neuron-specific RNA substrates by virtue of its N-terminal nuclear localization signal (102). This difference in substrate selectivity establishes the functional division of labor between the two ADAR family members in the cell.

RNA editing regulates gene expression through a mechanism that operates on multiple levels. At the level of transcript remodeling, ADAR-mediated selective splicing has been shown to alter protein heterodimer composition by retaining specific exons; for example, A-to-I editing of miR-34a led to the alteration of its seed sequence and a loss of inhibitory function for target mRNAs (103), whereas RNA secondary structure remodeling was found to block Staufen double-stranded RNA binding protein 1-mediated RNA nuclear export (101). At the level of protein function regulation, the RNA editing of antizyme inhibitor 1 mRNA was shown to induce its nucleoplasmic shuttling function, thereby regulating polyamine metabolic homeostasis, whereas arginine codon (AGA) to termination codon (UGA) editing triggered nonsense-mediated mRNA degradation (104-108).

During cancer progression, the ADAR family has been shown to exhibit a 'double-edged sword' effect: ADAR1 inhibits melanoma differentiation-associated gene 5 (MDA5) from recognizing endogenous double-stranded RNAs via editing Alu elements, thereby enabling cells to evade immune surveillance (100,109-111), whereas its p150 isoform promotes tumor metastasis through the editing of circular RNAs (112-114). On the other hand, ADAR2 was shown to restore oncogene expression, exerting a tumor-suppressive role (112,113).

Subtype-specific regulation has been demonstrated to further refine the functional network: For example, cytoplasmic ADAR1-p110 regulates mature miRNA function, whereas nuclear ADAR1-p150 maintains embryonic stem cell pluripotency. In cardiac tissues, ADAR2 was shown to protect ADAR1 targets from over-editing through competitive binding (114) and, moreover, a hierarchical regulatory mechanism for editing activity was demonstrated in a different study (115).

In inflammatory regulation, ADAR1 has been shown to maintain immune homeostasis through three mechanisms: First, through editing endogenous double-stranded RNA to block MDA5-mediated intrinsic immune recognition; second, by moderately activating the protein kinase R (PKR)-Z-DNA/RNA binding protein 1 axis to induce adaptive stress responses; and thirdly, by editing viral RNA mimicry via its p150 isoform to prevent type I interferon storms (100,109-111). Disruption of this dynamic equilibrium makes an important contribution to autoimmune diseases and viral infections, thereby providing novel targets for immunomodulatory therapy.

Ψ modifications

Ψ, a very abundant post-transcriptional modification in RNA, occurs via the action of pseudouridine synthetase (PUS), catalyzing the isomerization of uridine. Its unique C-C glycosidic bond (compared with the C-N bond of uridine) results in both a higher thermodynamic stability of, and irreversible changes to, RNA molecules, also serving as a consistent marker of the epitranscriptome (116-118). The catalytic pathways may be divided into two categories, first, the RNA-independent pathway, in which PUS family members (for example, PUS1) directly recognize RNA-specific structures (such as the T-arm deletion conformation of tRNAs) to accomplish site-specific modifications through conserved catalytic structural domains (119,120); and second, the RNA-dependent pathway, which is mediated via the box H/ACA-type small nucleolar ribonucleoprotein complexes, in which the protein Dyskerin acts as the core catalytic subunit to direct the Ψ-modification of rRNA and small nuclear RNA (snRNA) (121-123).

Mammalian PUS family members exhibit a fine subcellular division of labor: In the cytoplasmic system, PUS10 has been shown to specifically catalyze Ψ-modification at tRNA positions 54/55 (124,125), whereas PUS1 regulates transcript stability by recognizing mRNA stem-loop structures (119,126). In the mitochondrion, TruB pseudouridine synthase family member 1 (TRUB1) maintains the Ψ-modification at tRNA position 55, and its absence was found to lead to tRNA conformational disorder, which resulted in triggering the defective assembly of the respiratory chain complex (127). In the nucleoplasmic system, PUS3 targets tRNA modifications at positions 38/39 (128,129), and PUS7 was found to recognize the UGUAR motif to regulate both small nucleolar RNA (snoRNA) processing and pre-mRNA splicing (120,130). Ψ also has an important role in the control of translation. Pseudouridylation of tRNAs has been reported to regulate translation; in addition, the pseudouridylation of rRNAs also affects mRNA translation. In 293T cells, the presence of Ψ in mRNA codons was shown to increase protein production and to accelerate the rate of recognition of associated tRNAs (131).

The Ψ-modification has been shown to regulate the translation process via a mechanism that operates on multiple levels. Ψ in the mRNA CDS region forms a 'ribosomal speed bump' that enhances the translation efficiency of rare codons through prolonging the ribosomal retention time, whereas the Ψ-modification of termination codons (namely, changing UGA to ΨGA) results in an inhibition of the recognition of transcripts by the eukaryotic release factor eRF1, resulting in an ≤15% prolongation of readthrough translation (132,133). Under stress conditions, Ψ-modification has also been shown to maintain the eIF2A non-phosphorylated state by inhibiting PKR activation, resulting in an increased efficiency of translation initiation (134). In addition, Ψ has been shown to be involved in the global regulation of RNA metabolism: The RNA pseudouridine synthase D4 (RPUSD4)/PUS1/PUS7-mediated Ψ-modification of the pre-mRNA splice site was demonstrated to alter the efficiency of splicing processing, and it was proposed that this could be explained by its facilitation of RNA-binding protein binding (135) (Fig. 6).

Figure 6 Pseudouridylation and U34 modification. Pseudouridylation and U34 modification are important processes that occur on tRNAs. In mammals, the RNA pseudouridylation 'writers' primarily include PUS1/3/7/10, PRUSD2-4, TRUB1/2 and Dyskerin. The catalysis of pseudouridylation by Dyskerin relies on H/ACA snoRNA. From a functional standpoint, pseudouridylation contributes to RNA processing, stability and functionality. The translation of tRNA is markedly affected by alterations at the U34 position, such as cm5U, τm5U, MCM5U, MCM5s2U and MCM5Um. MCM5U and its derivatives are formed when ALKBH8 methylates cm5U, which is catalyzed by the ELP1-6 complex. The τm5U modification in mt-tRNA is catalyzed by GTP binding protein 3 or mitochondrial translation optimization 1 homolog, whereas SHMT2 provides the starting material for methyl synthesis. PUS, pseudouridine synthetase; tRNA, transfer RNA; TRUB1/2, TruB pseudouridine synthase family member 1/2; snoRNA, small nucleolar RNA; SHMT2, serine hydroxymethyltransferase 2.

At the pathological level, oxidative stress has been shown to induce an increase in Ψ sites within the mRNA CDS region of HeLa cells, a dynamic modification that assists tumor cells in adapting to the microenvironment through stabilizing key transcripts such as hypoxia-inducible factor-1α (HIF-1α) (132). For example, the Ψ-modification of HIF-1α mRNA promotes tumor cell survival under hypoxic conditions via inhibiting nuclease degradation and enhancing the glycolytic pathway activity mediated by HIF-1α. The discovery of this epitranscriptomic buffering mechanism has provided novel ideas for targeting RNA modifications in cancer therapy.

Modifications of U34 on tRNA

The chemical modification of U34 is a central regulatory element for codon-decoding precision. In a couple of previously published studies, U34 is able to extend the tRNA recognition of synonymous codons through non-classical base pairing (for example, the G-U wobble), a dynamic pairing mechanism that fulfills a critical role in maintaining a balance between translation rate and fidelity that is especially indispensable in decoding NNN-AA-type codons, and an absence of U34 modifications led to the accumulation of misfolded proteins, disruptions in unfolded protein response processes and impaired translation elongation (136,137) (Fig. 6).

Role of RNA modification in myocardial fibrosis

Methylation in myocardial fibrosis

m6A in myocardial fibrosis

Research has shown that METTL3 is a key driver of myocardial fibrosis, which regulates pro-fibrotic gene expression through m6A modification to drive fibroblast activation and the excessive deposition of ECM. Overexpression of METTL3 markedly promotes the transdifferentiation of CFs into myofibroblasts, both through upregulating the expression of type I/III collagen and mesenchymal fibrosis markers and through activating the TGF-β1-Mad2/3 signaling pathway (10) (Fig. 7). In addition, m6A modifications weaken double-stranded RNA stability by triggering a significant conformational shift of the RNA double strand to a hairpin structure (for example, in the case of oligonucleotide 19/21), which directly affects the ability to interact with recognition proteins, such as FTO, ALKBH5 and YTHDF2. FTO and ALKBH5 were found to be able to recognize the RNA double strand through m6A-induced RNA conformational changes, thereby enabling them to distinguish among substrates of highly similar sequences with dynamic selectivity. Contrary to the currently held consensus view, the GG(m6A)CU conserved motif was found not to be essential for FTO/ALKBH5, whereas METTL3/YTHDF2 was strictly dependent on it. This demonstrated that the m6A modification itself could regulate demethylase substrate selectivity through dynamic changes in RNA structure, challenging the traditional 'sequence motif determinism' hypothesis, and providing a novel perspective for our understanding of the functional complexity of the m6A epitranscriptome (34).

Figure 7 When METTL3 is overexpressed in cardiac fibroblasts, there is a notable increase in the proliferation of both type I and type III collagen, the activation of cardiac fibroblasts and the expression of α-SMA. METTL, methyltransferase-like; α-SMA, α-smooth muscle actin.

The extracellular matrix glycoprotein tenascin C (TNC) has been identified as a key downstream target of METTL3 in recent years. A previous study showed that METTL3 is able to enhance the mRNA stability of TNC by increasing its m6A modification level (namely, through increasing the number of methylated sites), leading to an elevated level of TNC protein expression, which, in turn, exacerbates cardiomyocyte apoptosis and fibrosis through the integrin αvβ3 signaling pathway (138). Overexpression of METTL3 has also been shown to promote the m6A modification of fibrogenic genes [for example, collagen, type I, α 1 (COL1A1) and collagen, type III, α 1 (COL3A1)] through YTHDF2-dependent mechanisms, which, in turn, promotes cardiomyocyte apoptosis and fibrosis, leading to increased collagen deposition (139). In addition, it has been shown that the RNA-methylated reading protein YTHDF2 inhibits excessive mitochondrial autophagy via recognizing the m6A modification site on BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (BNIP3) mRNA and promoting its degradation. In myocardial ischemia-reperfusion (I/R) injury, due to an absence of YTHDF2 led to an accumulation of BNIP3 protein, which resulted in the induction of excessive mitochondrial autophagy and myocardial cell death; on the other hand, overexpression of YTHDF2 was shown to reduce the level of BNIP3, leading to a decrease in the area of myocardial infarction (140). In exploring pathway synergies, the mRNA stability of METTL3 was found to be upregulated through m6A modification (with the consequent prolongation of its half-life), leading to increased insulin-like growth factor binding protein 3 protein expression. This molecule formed a synergy with TGF-β signaling to promote α-smooth muscle actin (α-SMA) expression, thereby markedly increasing the ability of CFs to migrate and to synthesize collagen (141). By contrast with previous findings on METTL3 overexpression, however, inhibition of this protein was found to be associated with the downregulation of fibrotic markers and decreased activity of CFs. In myocardial fibrosis, m6A regulates programmed cell death, and cardiomyocyte apoptosis is mediated via an upregulation of METTL3. Elevated m6A levels could promote the transcriptional degradation of autophagy-associated genes, which ultimately exacerbate cardiac injury and fibrosis (142).

Although METTL3 has been shown to regulate pre-METTL3 pre-fibrotic pathways through m6A modifications, such as the TGF-β/HIF-1α/reactive oxygen species (ROS) signaling pathway, identifying its therapeutic targets still presents a major challenge: First, there is a need to clarify the METTL3 pre-repair (early stage of injury) and pre-fibrotic (chronic) processes, and second, there is the need to analyze the synergistic reading proteins, such as those involved in the YTHDF2 and IGF2BP1 regulatory network. Small-molecule inhibitors targeting METTL3 have entered preclinical studies and are able to markedly reduce fibrotic areas by selectively inhibiting methyltransferase activity. These advances will provide novel ideas for enabling us to reverse pathological cardiac remodeling, although researchers need to be wary of the risk of the genomic instability that may be caused by overall m6A inhibition.

m7G in myocardial fibrosis

The dysregulation of m7G modification is associated with a variety of diseases; however, the role of METTL1 in cardiac fibrosis remains poorly understood. One study (143) identified the critical regulatory role of METTL1-mediated RNA m7G methylation in myocardial fibrosis, and the underlying molecular mechanism. Multidimensional experimental validation studies have shown that, in fibrotic myocardial tissues from patients with myocardial infarction and also on the basis of TGF-β1-induced in vitro models, myocardial fibroblasts exhibit upregulated METTL1 expression levels (which are elevated compared with controls) and a significant increase in the abundance of transcriptome-wide m7G methylation (143). In a tamoxifen-induced, fibroblast-specific METTL1-knockout mouse model (Col1a2-CreERT; METTL1flox constructed using the Cre-loxP system), cardiac function indices were found to be markedly improved following myocardial infarction and histopathological analyses further revealed a decreased interstitial collagen volume fraction and reduced transformation rates of α-SMA-positive myofibroblasts. Moreover, mechanistic studies demonstrated that METTL1, as a key component of the m7G methylation system, could regulate the TGF-β/Smad3 signaling pathway (including Smad7 and Smurf2) without affecting fibroblast gene stability but by impacting mRNA translation efficiency, thereby promoting an excessive deposition of ECM proteins (namely, Col1a1, Col3a1 and fibronectin). Single-cell sequencing analysis revealed that the deletion of METTL1 specifically inhibited the differentiation of a fibroblast subpopulation towards a pro-fibrotic phenotype, whereas no significant effects were exerted on cardiomyocytes or on other cardiac resident cells. This conclusion not only established for the first time a direct link between RNA epigenetic modifications and cardiac fibrosis, but it also provided a theoretical and experimental basis for the development of novel anti-fibrotic therapies targeting the METTL1-m7G axis.

Acetylation in myocardial fibrosis

Recent studies (144,145) have shown that the RNA acetyltransferase NAT10 dynamically regulates the cardiac fibrotic process through ac4C modification. In neonatal CFs, NAT10, through ac4C modification, increased the mRNA stability and translational efficiency of angiomotin-like protein 1 (Amotl1), facilitating the interaction of the Amotl1 protein with Yes-associated protein 1 (YAP), and driving YAP nuclear translocation. This process led to a significant promotion of CF proliferation (increased rate) and transdifferentiation of the CFs into myofibroblasts (upregulation of α-SMA expression) through activation of the Hippo/YAP signaling pathway. Consequently, experiments wherein Amotl1 gene silencing was combined with the YAP-selective inhibitor verticibufungin revealed an effective blockade of the fibrotic phenotype induced by NAT10 overexpression, which consequently led to reduced collagen deposition.

EGR3, an early growth response factor, promotes myocardial fibrosis by activating the TGF-β/Smad3 signaling pathway, which thereby promotes ECM remodeling and exacerbates myocardial fibrosis (as denoted by an increased fibrotic area). The lncRNA tsr007330 was shown to mediate ac4C modification of EGR3 mRNA through recruiting NAT10, leading to an increased expression of EGR3 protein, which, in turn, exacerbated the condition. These findings confirmed the centrality of NAT10 in post-myocardial infarction fibrosis, also revealing the pathological significance of the non-coding RNA-ac4C-EGR3 axis as a novel regulatory cascade reaction (146).

ac4C modifications enhance fibrotic gene expression through transcriptome reprogramming, additionally fulfilling a key role in chromatin accessibility remodeling to activate key signaling pathways (for example, the Hippo/YAP and TGF-β/Smad signaling pathways). Small-molecule inhibitors targeting NAT10 (for example, remodeling proteins) have been shown to have preclinical potential, given their ability to markedly reduce ac4C modification levels and to improve cardiac function, although aspects of their tissue-specific delivery and long-term safety have yet to be evaluated in depth.

Role of RNA modification in other cardiovascular diseases

Hypertension

The pathogenesis of hypertension, as the leading preventable risk factor for cardiovascular disease worldwide, involves a complex interaction of environmental and genetic factors and is multiply regulated by the immune response, oxidative stress, sympathetic nerves and the renin-angiotensin system. Studies have revealed that RNA epigenetic modifications have a key role in blood pressure homeostasis through regulating the function of vascular smooth muscle cells (147,148).

ADAR2-mediated RNA editing presents an important mechanism for maintaining vascular function. This enzyme regulates the contractile properties of vascular smooth muscle cells by specifically editing the fine filament protein A (FLNA) precursor mRNA. When ADAR2 activity becomes aberrant, conformational changes of the FLNA protein lead to an imbalance in the regulation of vascular tone, which is closely associated with elevated diastolic blood pressure. This finding revealed the precise regulatory role of RNA editing in vascular dynamics (149).

Genetic polymorphisms in the FTO gene, an m6A demethylase, have been found to be associated with susceptibility to hypertension. Although the exact mechanism has not yet been fully elucidated, this regulation at the epigenetic level has provided novel perspectives to explain the genetic predisposition to hypertension (150).

The RNA-binding protein HuR also acts as a molecular hub of multiple modification sites, and participates in the blood pressure regulatory network through binding m6A, uridylated and A-to-I-modified transcripts. Reduced levels of its expression in aortic tissues are associated with the progression of hypertension and the creation of animal models has confirmed that HuR deficiency triggers vasodilatory dysfunction and compensatory myocardial hypertrophy. Mechanistic studies have shown that HuR maintains the homeostasis of the NO/cGMP signaling pathway by stabilizing soluble guanylate cyclase and foveolar protein 1 mRNA (151,152).

It should be noted that these regulatory mechanisms do not exist in isolation. For example, HuR may indirectly affect the editing efficiency of FLNA by regulating the mRNA stability of ADAR2, whereas FTO-mediated m6A modification may interfere with the ability of HuR to bind to target RNAs, forming a multilevel epistatic regulatory network. This cross-talk property highlights the systems biology significance of RNA modifications in blood pressure regulation, and also provides a theoretical basis for the development of novel antihypertensive strategies targeting the epitranscriptome.

Atherosclerosis and coronary heart disease

Atherosclerosis, as an inflammation-driven vascular lesion, is characterized by lipid deposition and fibrous plaque formation as its core pathology, which ultimately leads to coronary artery stenosis and thrombotic events (153). Studies have shown that m6A modification is heavily involved in the disease process through regulating vascular inflammation and endothelial function. In patients with coronary artery disease and in models of atherosclerosis, the expression levels of the methyltransferases METTL3 and METTL14 are markedly upregulated (154-156) and their mediated m6A modifications promote plaque development through a dual mechanism: First, by enhancing monocyte-endothelial cell adhesion and activating the NF-κB/IL-6 signaling pathway, which drives the inflammatory cascade; and second, by inhibiting the mRNA stability of MyD88 in macrophages, which hinders M2-type polarization and exacerbates the inflammatory microenvironment within the plaque. Notably, knockdown of METTL14 leads to a marked reduction in the macrophage inflammatory phenotype via inhibiting the NF-κB pathway (156), whereas the METTL3-m6A-YTHDF2 signaling axis further amplifies monocyte/macrophage inflammatory responses by regulating the metabolism of oxidized low-density lipoprotein (157). In addition, the synergistic interaction of METTL3/14 with ADAR1 regulates vascular calcification and neovascularization, suggesting their involvement in a complex mechanism underlying plaque stability and ischemic compensation (158,159).

In the progression of coronary heart disease, the dynamic balance of RNA modifications exerts a key regulatory role in cardiac function. A previous study specifically on cardiomyocytes showed that the downregulation of FTO expression under hypoxic conditions leads to an increase in m6A levels and that FTO overexpression modulates both intracellular calcium homeostasis and improved myocardial contractile function through stabilizing ATP2A2 mRNA (encoding the SERCA2a protein) (160). By contrast, ALKBH5 overexpression was found to inhibit cardiomyocyte proliferation and to reduce cardiac function through the m6A-YTHDF1-YAP signaling pathway, revealing a dual role for demethylases in myocardial repair. In an acute myocardial infarction model, endothelial cell-derived extracellular vesicles were shown to carry the METTL3-m6A-HNRNPA2B1 complex, thereby exacerbating cardiomyocyte apoptosis and dysfunction through the upregulation of miR-503 (160).

The mechanism of I/R injury, a serious complication of coronary heart disease, has been demonstrated to be closely associated with the m6A modification network. In a myocardial I/R model, METTL3 overexpression was found to promote the binding of heterogeneous nuclear ribonucleoprotein D to m6A-modified TFEB mRNA, to inhibit autophagy, and to induce apoptosis in cardiomyocytes. TFEB inhibits METTL3 stability, and activates ALKBH5 transcription through a negative feedback loop, forming a dynamic regulatory balance (160). Gene intervention experiments confirmed that the myocardium-specific knockdown of METTL14 led to a reduction in the expression of atrial natriuretic peptide (ANP), a marker of heart failure, and also a significant improvement in cardiac function following I/R injury (160). These findings not only revealed the multilevel regulatory role of RNA modification in coronary artery disease, but also provided a theoretical basis for therapeutic strategies that are aimed at targeting the m6A network (for example, METTL3 inhibitors or FTO activators). Furthermore, m7G RNA modification exerts dual roles through dynamic regulation via the methyltransferase METTL1/WDR4 complex. A previous study (161) showed that, during the early reperfusion phase (1-3 h), AKT-dependent phosphorylation activates METTL1, thereby transiently enhancing the translation efficiency of protective genes (for example, HIF-1α) to maintain metabolic energy homeostasis. However, in the middle-to-late reperfusion phase (>6 h), the sustained overexpression of METTL1 drives excessive m7G modification. This serves to stabilize the mRNAs encoding the pro-apoptotic factor caspase-3, and the pro-fibrotic factor TGF-β1, thereby triggering oxidative stress bursts and mitochondrial dysfunction. Mechanistic analyses have demonstrated the core process involving the METTL1-mediated serine and arginine rich splicing factor 9 (SRSF9)/nuclear factor of activated T-cells, cytoplasmic 4 (NFATc4) signaling axis, where m7G modification enhances the RNA stability of splicing factor SRSF9, facilitating the alternative splicing of NFATc4 and nuclear translocation, thereby activating the calcineurin pathway. This cascade ultimately leads to pathological cardiomyocyte hypertrophy, collagen deposition and cardiac functional deterioration. Targeting METTL1 has also been shown to be a strategy to markedly mitigate cardiac injury: Myocardial-specific knockout models exhibited a 40% reduction in infarct size, with improved cardiac function (ejection fraction and fractional shortening), whereas delivery strategies (for example, ROS-responsive carriers) that block METTL1 activity may offer novel translational avenues (162).

Cardiomyopathy

Primary cardiomyopathies encompass various types of hypertrophic cardiomyopathy (HCM) and dilated cardiomyopathy, and their pathogeneses are closely associated with genetic variation. In HCM ~70% of cases were shown to be driven by mutations in the genes of myosin-binding protein C and β-myosin heavy chain (MYH7) (163), whereas mutations in other key genes, such as α-myosin heavy chain, troponin T and myosin ganglionic protein, may also lead to structural disorders of the cardiac myocardium (164,165). Abnormal mitochondrial function, specifically mitochondrial dysfunction in the myocardium, fulfills an important role in the progression of cardiomyopathy, and mutations in the GTP-binding protein 3 or mitochondrial translation optimization 1 homolog genes were shown to elicit an impaired assembly of respiratory chain complexes through impairing the 5-taurinomethyluridine (τm5U) modification of mitochondrial tRNAs, ultimately leading to HCM (166). In a dilated cardiomyopathy (DCM) model, specific knockdown of the pentatricopeptide repeat domain 1 gene disrupted ribosomal large subunit assembly through interfering with the Ψ-modification at position 2509 of 16S rRNA, which had the effect of triggering defects in mitochondrial energy metabolism with dysregulated mTOR signaling (167). In addition, conditional knockdown experiments of YTHDC1 and ADAR1 confirmed that the deletion of RNA-binding proteins could induce DCM through aberrant m6A modification (168) and dysregulated RNA editing (110), respectively.

Unlike genetic lesions, secondary cardiomyopathies (for example, diabetic cardiomyopathy and chemotherapy cardiotoxicity) are primarily driven by acquired factors. In diabetic cardiomyopathy, METTL14-induced hypermethylation of the lncRNA TINCR was found to enhance the degradation of TINCR via YTHDF2, which consequently reduced the inhibitory effect of TINCR on NLRP3 inflammasome targets. Furthermore, the lncRNA Airn was shown to stabilize p53 mRNA through m6A modification, also inhibiting the ubiquitin-mediated disruption of eIF2C, thereby attenuating myocardial fibrosis in diabetic cardiomyopathy (169,170). In adriamycin-induced cardiotoxicity, METTL14 was shown to exacerbate myocardial injury by promoting iron death (specifically, through increasing the levels of lipid peroxidation) through modulation of the m6A modification of genes involved in glutathione metabolism (171).

The aforementioned studies have revealed the centrality of RNA epitope modification in cardiomyopathy: First, METTL14 inhibitors may inhibit inflammatory vesicle activation through restoring TINCR expression; subsequently, the m6A-reading function, targeting YTHDC1, may correct metabolic derangements in the cardiomyocytes; and lastly, ADAR1 activators may both repair RNA-editing aberrations and improve myocardial electrophysiological stability. However, resolution of the tissue-specific delivery systems with dynamic modification profiles remains a key challenge for clinical translation.

Cardiac hypertrophy

Cardiac hypertrophy, as a compensatory response of the myocardium to stress load, can be categorized into two types, namely physiological and pathological, which exhibit significant differences in their epitranscriptome regulatory mechanisms. Physiological hypertrophy (for example, that induced by exercise) was shown to be accompanied by a decrease in the overall m6A levels of myocardial mRNA, and this was potentially associated with the downregulation of METTL14 expression (160,172). By contrast, the molecular mechanism of pathological hypertrophy (for example, that induced by pressure overload), a major trigger of heart failure, involves a complex dysregulation of the dynamic balance of m6A modification. It is caused by a variety of cardiac diseases, including both heart failure with maintained ejection fraction and heart failure with reduced ejection fraction. In the aortic arch constriction model, the myocardium-specific knockdown of FTO was shown to exacerbate ejection fraction-reduced heart failure (173). On the other band, FTO overexpression was demonstrated to inhibit the expression of pathologic hypertrophy-associated genes, such as ANP and B-type natriuretic peptide, through demethylation, resulting in a reduction in the cardiomyocyte cross-sectional area (174). This bidirectional regulatory effect suggests that FTO fulfills a key role in maintaining the fine balance between myocardial adaptation and pathological remodeling.

METTL3 functions as an m6A methyltransferase in an environment-dependent manner: Its overexpression in the basal state enhances cardiac compensatory function, although the myocardium-specific knockdown of METTL3 was found to accelerate the progression of heart failure under pressure overload conditions. Mechanistic studies have shown that the Piwi-interacting RNA CHAPIR promotes pathological ventricular hypertrophy by binding to and inhibiting METTL3 methylation activity, thereby leading to hypomethylation of the mRNA of the ADP-ribosyltransferase PARP10, and blocking the YTHDF2-mediated degradation pathway (175,176). Notably, the m6A-reading protein YTHDF2, whose expression is upregulated in heart failure, was shown to promote the cardiac hypertrophic phenotype through stabilizing myosin-7 (MYH7) mRNA (namely, prolonging its half-life), thereby creating a vicious circle (176).

Taken together, these findings have revealed a dual role for the m6A modification network: Its first role is to regulate myocardial compensation and dystrophic transition through the balance of FTO-METTL3, and its second role is to drive sustained progression of the hypertrophic phenotype through the YTHDF2-MYH7 axis. Targeting this regulatory network (for example, through the development of METTL3 mutagenic activators or YTHDF2 inhibitors) may provide possible novel strategies, although for this to occur, technical challenges presented by matters such as tissue-specific delivery and dynamic modification monitoring will need to be addressed.

Summary and future outlook

The RNA epitranscriptional regulatory network of myocardial fibrosis is associated with both breakthrough therapeutic opportunities and complex biological challenges. Methylation reactions (for example, m6A and m7G) and ac4C serve as core modification mechanisms that drive fibrotic progression through precisely regulating fibroblast activation and extracellular matrix deposition. METTL3 silencing markedly reduces the levels of fibrosis markers such as collagen, revealing its potential as a drug target; however, the potential toxicity of the molecule to cardiomyocytes presents an obstacle to its clinical application: A paradox that highlights the critical importance of cell-specific effects. By way of contrast, the METTL1-m7G axis has unique properties: In fibrotic tissues, from patients with myocardial infarction and in TGF-β1-induced in vitro models, METTL1 is specifically overexpressed in cardiac fibroblasts and is accompanied by elevated levels of transcriptome-wide m7G methylation; knockdown of METTL1 not only improves cardiac function, but it also reduces collagen deposition and α-SMA-positive myofibroblast transformation, more critically, without affecting cardiomyocyte function. This selectivity stems from the fact that METTL1 specifically promotes the excessive deposition of ECM proteins by finely regulating mRNA translation efficiency, rather than the stability of key genes of the TGF-β pathway (or example, Smad7/Smurf2) through m7G modification.

Acetylation modifications amplify fibrotic signals through mechanisms at multiple levels in concert with methylation networks. NAT10, as an RNA acetyltransferase, dynamically regulates cardiac fibrosis through ac4C epigenetic modifications: It enhances Amotl1 mRNA stability and translational efficiency, promotes the binding of Amotl1 protein to YAP, and activates the Hippo/YAP signaling pathway, ultimately accelerating fibroblast proliferation and transdifferentiation. Notably, silencing of the Amotl1 gene combined with YAP-specific inhibitors has been shown to effectively block the fibrotic phenotype induced by NAT10 overexpression, and the synergistic effect of EGR3 with the TGF-β/Smad3 pathway, and the multilayered regulation of fibrosis by NAT10, further established its centrality to post-infarction fibrosis. These findings reveal a hierarchical modification network architecture: The METTL3-m6A modification initiates the TGF-β transcriptional cascade, the NAT10-ac4C modification stabilizes key effectors (for example, EGR3 mRNA), and the METTL1-m7G modification enhances the translational efficiency of downstream genes. Altogether, the three types of modification form a self-amplifying pathogenic cycle (Fig. 8).

Figure 8 The three-axis regulatory model is shown: (A) Methylation-acetylation synergy, where METTL3-mediated m6A modifications and NAT10-driven ac4C modifications co-operatively target the Hippo/YAP and TGF-β/Smad signaling pathways, exacerbating fibroblast activation and excessive collagen production. (B) Cell-type-specific programming revealed by single-cell analysis, as exemplified by METTL1-catalyzed m7G modifications selectively promoting fibroblast differentiation into pro-fibrotic phenotypes while protecting cardiomyocytes. (C) Cross-modification crosstalk, illustrated by the HuR protein integrating m6A, uridylation, and A-to-I editing signals to regulate extracellular matrix dynamics, while the METTL3/FTO balance influences stress-responsive RNA stability. Collectively, these three axes drive the pathological phenotype of myocardial fibrosis, characterized by excessive collagen deposition, increased ventricular stiffness and diastolic dysfunction. METTL1/3, methyltransferase-like 1/3; m6A, N6-methyladenosine; NAT10, N-acetyltransferase 10; YAP, Yes-associated protein; A-to-I, adenine-to-inosine; FTO, fat mass and obesity-associated.

These post-transcriptional modifications have been demonstrated to affect cell function and RNA metabolism, to regulate gene expression, and to influence both organ development and cell differentiation, ultimately determining the cell's fate. Through these chemical changes, cells have an improved ability to adapt to internal and external stimuli (39,41,160). Innovative solutions are required to circumvent the core bottlenecks faced by translational medicine: The dual role of METTL3 in both the early repair of injury and the chronic fibrosis stage requires the development of spatiotemporal regulation strategies, such as ROS-responsive nanocarriers for precise drug delivery of myocardial infarction. The fibroblast specificity of METTL1 offers opportunities for spatial targeting, which can be accomplished using the Col1a2 promoter-driven CRISPR-Cas9 system for cell-type specific interventions. Future research should focus on three frontiers: First, resolving the precise modification sites of METTL1 on Smad7/Smurf2 mRNA and elucidating the underlying mechanism of interaction with the translation machinery; second, exploring the network interactions of m7G dynamic modifications with other nodes of the TGF-β pathway; and third, developing METTL1-specific small-molecule inhibitors and a serum exosome modification fingerprinting diagnostic system. Current antifibrotic treatments are still in the developmental stages; although RNA modification networks have opened new therapeutic windows, existing strategies are mostly limited to symptom control. Fundamental breakthroughs are required for the integration of epitranscriptomics, non-coding RNA and signaling pathway research; for example, the development of CRISPR-Cas13d-mediated editing of the Amotl1 mRNA ac4C locus, or patient stratification (high m7G/ac4C ratio populations) to achieve precision intervention. Ultimately, successful translation of primary research results to the clinic will depend on our ability to harness both cell-specific targeting (for example, fibroblast-restricted METTL1 inhibition) and critical node breakthroughs to move epitranscriptional regulation from the mechanistic discovery mode to finding clinical solutions.

Availability of data and materials

Not applicable.

Authors' contributions

KW conceived the present study and provided suggestions for the revision of the manuscript. XW, RW and XC wrote the manuscript and made the figures. Data authentication is not applicable. All authors read and approved the final manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Use of artificial intelligence tools

During the preparation of this work, a generative AI/AI-assisted technology, namely DeepSeek, was used to improve the readability and language of the manuscript.

Acknowledgements

Not applicable.

Funding

The present study was supported by Qingdao Science and Technology Benefiting the People Demonstration Project (grant no. 24-1-8-smjk-7-nsh), Major Basic Research Projects in Shandong Province (grant no. ZR2024ZD46) and Taishan Scholar Distinguished Expert.

References