Direct RNA Sequencing Reveals Epitranscriptomic Isoform Mechanisms in Heart Failure

Direct RNA sequencing on Oxford Nanopore platforms has mapped the dynamic landscape of N6-methyladenosine RNA modifications at single-base and transcript-isoform resolution in experimental heart failure, uncovering unexpected regulatory mechanisms linking epitranscriptomic changes directly to transcript isoform switching. The investigation, led by Zhihan Xu and colleagues and published in Nature Cardiovascular Research, demonstrates that chemical marks decorating messenger RNA undergo extensive remodeling during cardiac dysfunction. Crucially, their mapping resolved a biological paradox surrounding the structural gene Synpo2l, revealing that an apparent decrease in overall gene methylation does not stem from active demethylation, but rather arises from an alternative promoter switch toward a shorter transcript variant that naturally carries fewer internal chemical modifications.

To model biomechanical stressors underlying cardiac hypertrophy and failure, the team applied transverse aortic constriction to adult male C57BL/6J mice using twenty-seven gauge and thirty gauge needles to induce moderate and severe pressure overload. Mice were randomly assigned to constriction or sham procedures for four weeks, with ketamine and medetomidine anesthesia and postoperative buprenorphine analgesia. Serial echocardiography conducted before surgery and at two and four weeks postoperatively revealed progressive deterioration in cardiac geometry and performance. By week four, both constriction cohorts exhibited marked reductions in left ventricular ejection fraction and fractional shortening, alongside significant expansions in left ventricular internal dimensions at end-diastole. Furthermore, significant elevations in lung weight-to-body weight ratios emerged exclusively in thirty-gauge mice, confirming pulmonary congestion secondary to advanced ventricular decompensation.

Direct transcriptomic profiling confirmed molecular disruptions paralleling this functional decline, capturing substantial induction of hallmark myocardial stress markers including Nppa, Nppb, Acta1, Myh7, and Postn across both operative conditions. While the two surgical models shared one hundred forty-five upregulated and one hundred nineteen downregulated differentially expressed genes, thirty-gauge mice exhibited one thousand seven hundred twenty-one unique differentially expressed genes, compared to ninety-seven unique alterations in twenty-seven-gauge animals. Gene ontology enrichment highlighted distinct programs in the severe cohort, showing that upregulated unique transcripts were predominantly associated with inflammatory signaling, extracellular matrix reorganization, and vascular morphogenesis, whereas downregulated unique transcripts were enriched for fatty acid metabolism, circulatory processes, and metal ion transport.

To establish whether the cardiac epitranscriptome undergoes coordinated alteration alongside transcriptomic disturbances, researchers quantified thirteen distinct messenger RNA modifications using liquid chromatography-coupled triple quadrupole mass spectrometry. Among detectable chemical marks, global N6-methyladenosine levels increased significantly in both surgical models, rising by nine percent in twenty-seven-gauge mice and twenty-two-point-six percent in thirty-gauge animals relative to sham controls, whereas other ribonucleoside modifications remained stable or fluctuated inconsistently. Profiling core epitranscriptomic regulatory enzymes revealed complex shifts across disease stages, marked by prominent discordance between steady-state messenger RNA and translated protein abundance. In twenty-seven-gauge hearts, Fto transcripts were reduced to zero-point-eight-three-fold of baseline, while Western blots showed a non-significant upward trend for METTL3 protein. In contrast, thirty-gauge hearts displayed significant protein upregulation of the core writers METTL3 and METTL14, alongside a one-point-five-four-fold increase in the eraser ALKBH5, illustrating dynamic posttranscriptional compensation during advanced failure.

Direct RNA sequencing data were processed using the Dorado basecalling environment, where the authors evaluated random-context and DRACH-context neural network architectures to maximize calling precision. Comparative analysis demonstrated that the DRACH-context basecaller substantially reduced low-confidence calls while yielding abundant high-confidence predictions exceeding zero-point-nine probability, leading investigators to enforce a zero-point-eight probability cutoff. The researchers orthogonally validated these computational annotations using glyoxal- and nitrite-mediated deamination of unmethylated adenosines, termed GLORI, coupled with targeted Sanger sequencing on candidate loci within Igf1r and Neat1 transcripts. Categorization of average modification levels across aligned reads revealed that low modification accounted for eighty-seven percent of evaluated sites, medium modification represented eight percent, and high modification occupied five percent. Among sequence motifs, GGACU showed the highest modification probability, followed by GGACA and GAACU. Furthermore, thermodynamic modeling demonstrated that highly modified sites exhibit significantly higher minimum free energy, indicating that cardiac methyltransferases preferentially deposit marks within structurally accessible, open single-stranded RNA architectures.

Differentially methylated region analysis revealed substantial remodeling across both disease severities, identifying six hundred ninety-eight upregulated and six hundred thirty-seven downregulated sites in twenty-seven-gauge mice, compared to one thousand five hundred eighty-seven upregulated and one thousand one hundred sixty-nine downregulated sites in thirty-gauge animals. Baseline methylation characteristics differed markedly between groups: twenty-seven-gauge differential sites possessed median basal methylation of twenty-five percent, whereas thirty-gauge sites exhibited higher baseline levels near forty-five percent. Across the transcriptome, changes in individual methylation sites displayed no broad correlation with corresponding alterations in steady-state gene expression. Even among transcripts harboring at least two consistently regulated methylation sites, the vast majority exhibited no significant change in overall transcript abundance. Furthermore, integrating polyadenylation measurements revealed over two hundred poly(A) tail length alterations in thirty-gauge animals, yet polyadenylation dynamics likewise exhibited no correlation with differential methylation or gene expression, demonstrating that cardiac epitranscriptomic remodeling operates through precise, transcript-specific mechanisms.

Gene ontology assessments of methylated transcripts demonstrated strong enrichment for functional cascades governing RNA metabolism, posttranscriptional processing, and cardiac structural development, prompting the team to investigate evolutionary conservation between murine and human pathology. Comparing murine sequencing datasets with published methylated RNA immunoprecipitation profiles from failing and nonfailing human hearts using LiftOver coordinate conversion, the researchers identified two hundred twenty-seven overlapping transcripts with upregulated methylation in advanced failure and ninety-five overlapping transcripts in nonfailing tissue. Fine-mapping revealed that forty-point-four percent of analyzed murine differential sites mapped directly into human peak boundaries, with an additional five-point-five percent residing within fifty nucleotides of converted peaks. Detailed examination confirmed conserved methylation clusters in noncoding RNA Neat1, receptor transcript Igf1r, and structural gene Synpo2l. Orthogonal GLORI assays performed on human dilated cardiomyopathy biopsy specimens successfully confirmed predicted methylation marks within human NEAT1, affirming substantial cross-species conservation in cardiac epitranscriptomic regulation.

To dissect how epitranscriptomic alterations intersect with transcript diversity, the authors performed differential transcript usage analysis, identifying four hundred sixteen significant isoform-switching events in thirty-gauge failing hearts. Forty-three transcripts exhibited both differential transcript usage and differential methylation, with ten displaying shared regulation in human clinical samples, among which Synaptopodin two-like, or Synpo2l, served as a primary model. In nonfailing murine hearts, the longest transcript, designated isoform one, dominated expression at sixty-three-point-six percent, while isoform three represented twenty-three-point-three percent. In severe failure, this distribution inverted: isoform three surged to seventy-seven-point-eight percent of total expression, while isoform one collapsed to fourteen-point-zero percent. Direct sequencing revealed eighteen-point-four-fold induction of isoform three transcripts, corroborated by quantitative PCR showing fifteen-point-five-fold upregulation. Western blotting confirmed these findings at the protein level, showing a seventeen-fold increase in endogenous seventy-five-kilodalton isoform three protein in failing hearts without significant changes in isoform one abundance.

Resolving individual methylation sites across Synpo2l isoforms provided a definitive mechanistic explanation for its apparent hypomethylation during heart failure. Although gene-level analysis initially classified Synpo2l as hypomethylated based on three downregulated differential sites in its shared terminal exon, isoform-resolved profiling of eleven common exon sites revealed that ten remained completely stable across disease states. Rather than undergoing enzymatic demethylation, the gene shifted from isoform one, which naturally carries the highest density of methylated sites per transcript, to isoform three, which intrinsically harbors the lowest methylation levels. Specifically, five sites clustered near the beginning of the terminal exon exhibited wide variation among isoforms, consistently displaying maximal methylation on isoform one and minimal methylation on isoform three. Consequently, selective accumulation of undermethylated isoform three diluted overall gene-level methylation measurements, establishing that transcript isoform switching can directly generate apparent epitranscriptomic remodeling without active mark removal.

Investigation of human clinical specimens confirmed identical isoform dynamics, showing significant upregulation of SYNPO2L isoform three in human dilated cardiomyopathy apical biopsies and in cultured human stem cell-derived cardiomyocytes exposed to endothelin-one. To determine how distinct transcription start sites are controlled, the researchers cloned isoform-specific promoter segments into luciferase reporter vectors. At baseline, promoters for isoforms one and two exhibited robust activity exceeding four hundred-fold above minimal controls, whereas shorter promoter fragments for isoform three displayed negligible baseline induction. However, extending the isoform three promoter to one thousand five hundred nucleotides, which overlaps the adjacent isoform two promoter, restored substantial reporter expression and conferred responsiveness to endothelin-one stimulation. Furthermore, incorporating an upstream distal regulatory element linked to a splicing quantitative trait locus selectively enhanced isoform three transcription while suppressing isoform one activity, demonstrating that distal enhancers and alternative promoters coordinate stress-induced isoform switching during cardiac remodeling.

To establish whether methylation machinery exerts direct posttranscriptional control over SYNPO2L transcripts, investigators perturbed methyltransferase activity in human cardiomyocytes using genetic overexpression and pharmacological inhibition. Ectopic expression of catalytic wild-type METTL3 induced cellular stress pathways and promoted isoform three transcription indirectly by activating the stress-responsive promoter region. Conversely, applying selective METTL3 catalytic inhibitor STM2457 with endothelin-one diminished overall methylation enrichment while augmenting SYNPO2L transcript levels. To evaluate transcript clearance directly, the authors performed actinomycin D chase experiments monitoring decay kinetics over eight hours. Pharmacological METTL3 inhibition significantly prolonged the functional half-lives of both isoform two and isoform three without altering the degradation rate of isoform one, which exhibited the highest intrinsic stability. These findings indicate that while methyltransferase activity indirectly spurs isoform three transcription through cardiac stress signaling, direct methyl mark deposition accelerates transcript clearance, illustrating layered epitranscriptomic control over cardiac gene expression.

Subject of Research: Transcriptome-wide m6A RNA modification mapping and isoform regulation in heart failure

Article Title: Comprehensive m6A landscape from direct RNA sequencing reveals SYNPO2L RNA isoform regulation mechanisms in heart failure

Article References: Xu, Z., Tian, C., Chan, J. W., Tew, W., Tang, G. L. Q., Qi, R., Tao, Z., Yim, D. G. R., Cui, L., Luu, D. A. T., Amrute, J. M., Tergaonkar, V., Lee, C. J. M., & Foo, R. S. Y. (2026). Comprehensive m6A landscape from direct RNA sequencing reveals SYNPO2L RNA isoform regulation mechanisms in heart failure. Nature Cardiovascular Research. https://doi.org/10.1038/s44161-026-00898-0

Image Credits: AI Generated

DOI: 10.1038/s44161-026-00898-0

Keywords: direct RNA sequencing, m6A RNA modification, heart failure, SYNPO2L, epitranscriptomics, isoform switching, cardiomyocytes, METTL3, Comprehensive, landscape, direct, sequencing