RNA splicing in human cardiomyopathies: shared landscape in diverse pathogenic paths

This editorial refers to ‘Shared splicing dysregulation in heart failure associated with dilated and ischemic cardiomyopathy and spatial specificity across cardiac regions’, by M. Furtado et al., https://doi.org/10.1093/cvr/cvag068. Cardiomyopathies are highly complex and heterogenous diseases, driven by different intrinsic risk factors and external stressors. Global transcriptome reprogramming is a molecular hallmark associated with the onset and progression of cardiomyopathies. The up-and down-regulation of specific cardiac genes has been linked to many of the pathological features, including loss of myocyte contractility and viability, pathological remodelling, and electrophysiological abnormalities.1–3 Importantly, targeted manipulation of cardiac gene expression through transcriptional or epigenetic modulation is being actively pursued as potential future therapies.3–5 However, beyond the mRNA quantities, mature RNA transcripts are produced through multiple post-transcriptional steps, such as alternative splicing (AS), leading to much augmented complexity and specificity in their final products.6–8 While substantial literature has documented aberrant RNA splicing events associated with various forms of cardiomyopathies, a comprehensive chamber-specific RNA splicing profile in human cardiomyopathies remains to be fully elucidated.7 In this issue, Furtado et al. reported an analysis of RNA splicing landscape in two common forms of human cardiomyopathies, the dilated (DCM) and the ischaemic cardiomyopathies (ICM), thus adding important insight to post-transcriptional regulation in human failing hearts.9 In this study, left ventricle, right ventricle and interventricular septum tissues from a substantial cohort of patients without heart failure (control, n = 23) and failing heart explants from patients diagnosed with DCM (n = 54) or ICM (n = 45) were subjected to RNA-seq. A pipeline of three independent RNA splicing analysis platforms (rMATS, vast-tools and MAJIQ) was used to establish the transcriptome profiles at individual exon resolution. Differential splicing activities were collated by pair-wise comparisons among the three cohorts. The functional implications were speculated according to pathway enrichment for the affected genes and a subset of these changes were validated by qRT-PCR method and further profiled across different chambers. The output from this study is a relatively comprehensive chamber-specific landscape of alternative mRNA splicing events in human hearts, across normal and two forms of cardiomyopathies (Figure 1). Considering the important functional impact of post-transcriptional regulation in cardiac pathologies, this dataset is a valuable resource for the field. Illustration of the study flow and RNA splicing patterns identified in human hearts with dilated cardiomyopathy (DCM) vs. ischaemia cardiomyopathy (ICM). (Figure was generated using BioRender). Beyond the transcriptome profiles, the current study was quite preliminary in functional validation or mechanistic investigations. However, several interesting and potentially important observations were made. First, not surprisingly, significant shifts in splicing profiles were found between the control and the cardiomyopathy hearts. The affected genes were identified to be significantly associated with contractility and remodelling. Yet, the scope of RNA splicing changes in the failing hearts was modest relative to the changes at the level of total mRNA expression. Furthermore, there were significant overlap in the splicing affected genes between DCM and ICM, and relatively few RNA splicing events were found to be unique and specific for a particular form of cardiomyopathy. This is unexpected as DCM and ICM are mechanistically distinct and often associated with very different forms of lesions and stress. Secondly, authors performed targeted as well as unbiased motif enrichment analyses in order to identify potential trans-acting RNA-binding proteins involved in the alternative splicing regulation. Although no direct data was provided to validate the candidate factors at functional level, a number of them, including a relatively understudied RNA splicing regulator Quaking, were implicated.10 Interestingly, most of these candidate RNA splicing regulators did not show concordant expression changes between the control and diseased hearts. Finally, among the validated RNA-splicing regulated genes, a number of them showed spatially demarcated splicing patterns among the different chambers of the heart, supporting the overall conclusion that RNA splicing has significant chamber specificity in diseased hearts (Figure 1). Although interesting from fundamental biology perspectives, this insight raises a critical concern for RNA splicing targeted intervention to treat heart failure since most of the current delivery modalities do not have chamber specificity.5–7 Although different forms of cardiomyopathies appear to have similar RNA splicing profiles, this conclusion may be misleading due to a number of underlying limitations. The cardiomyopathy tissues profiled in this study were obtained from the explanted failing hearts, usually at the end stage of disease manifestation. Therefore, the transcriptome profile may reflect a converging end-point of cardiac remodelling and dysfunction, rather than the different pathogenic processes involved. In addition, the heterogeneity of the disease progression and the biopsy samples may have undermined the statistical power needed to identify many of the disease-associated RNA splicing events. While the mRNA expression levels can be counted by the total number of covered reads, the RNA splicing events can only be measured from targeted exon–exon junctions, thus requiring much deeper sequencing datasets to detect all the RNA splicing events with sufficient confidence. In short, this conclusion should be viewed with some caution, considering the limitations in the sequencing depth, sample heterogeneity and the available informatics methods. RNA splicing, as an important step of the post-transcriptional regulatory process, is still vastly understudied relative to the efforts dedicated to transcriptional control in cardiac development and diseases. The current study, although insightful, exposes some key challenges and limitations of the current technologies. Bulk RNA-seq based isoform analysis remains problematic. The fact that very few overlapping splicing events were observed using the three informatics platforms clearly showcased the technical and methodological limitations in RNA splicing studies. The sensitivity and fidelity of the current technologies to uncover the full scope of RNA splicing landscape need to be improved. The bulk RNA-seq may also be limited in isoform profiling as multi-exon spliced products cannot be accurately annotated from short-read sequencing datasets. New technology, such as Long-read Nanopore Sequencing, may overcome some of these limitations, although the high cost barrier and requirement of RNA integrity for the extra-long transcripts (a common feature for extracellular and sarcomere genes) remain to be addressed. Finally, RNA splicing is a highly cell-type specific process, and RNA splicing events need to be profiled in cell-type-specific manner in order to truly uncover their functional implications in cardiac physiology and diseases. The current bulk RNA-seq or long-read RNA-seq may overwhelmingly represent the RNA splicing events in cardiomyocytes, but overlook the important RNA splicing changes in other disease-relevant cell types, such as endothelium, fibroblasts, and immune cells. It is certain that when new advancements in the current RNA analysis platforms can provide us the necessary tools to establish spatially resolved RNA splicing profiling at single-cell resolution, we will be able to peek into RNA splicing, as a new dimension in the cardiac gene expression landscape, where new disease-altering targets are waiting to be explored. This work was supported in part by US National Heart, Lung, and Blood Institute (R01HL176487, R00 HL141626) to CG, Tanoto Foundation in Diabetes Research and NMRC Singapore IRG grant to Y.W. All the data supporting the conclusion of this article has been made available through the original article.