MicroRNAs in atrial fibrillation target genes in structural remodelling

Abstract

We aim to elucidate how miRNAs regulate the mRNA signature of atrial fibrillation (AF), to gain mechanistic insight and identify candidate targets for future therapies. We present combined miRNA-mRNA sequencing using atrial tissues of patient without AF (n = 22), with paroxysmal AF (n = 22) and with persistent AF (n = 20). mRNA sequencing previously uncovered upregulated epithelial to mesenchymal transition, endothelial cell proliferation and extracellular matrix remodelling involving glycoproteins and proteoglycans in AF. MiRNA co-sequencing discovered miRNAs regulating the mRNA expression changes. Key downregulated miRNAs included miR-135b-5p, miR-138-5p, miR-200a-3p, miR-200b-3p and miR-31-5p and key upregulated miRNAs were miR-144-3p, miR-15b-3p, miR-182-5p miR-18b-5p, miR-4306 and miR-206. MiRNA expression levels were negatively correlated with the expression levels of a multitude of predicted target genes. Downregulated miRNAs associated with increased gene expression are involved in upregulated epithelial and endothelial cell migration and glycosaminoglycan biosynthesis. In vitro inhibition of miR-135b-5p and miR-138-5p validated an effect of miRNAs on multiple predicted targets. Altogether, the discovered miRNAs may be explored in further functional studies as potential targets for anti-fibrotic therapies in AF.

Keywords: Atrial fibrillation; Fibrosis; MicroRNA; Structural remodelling; Transcriptome sequencing.

Fig. 1

Study design. MiRNA and mRNA sequencing were independently performed in atrial tissues from the same patients. Results of the mRNA sequencing were reported previously. Analysed datasets of miRNAs and mRNAs were combined on the basis of target - prediction algorithms (online databases) and Pearson R correlations. Permutation testing and Kolmogorov–Smirnov tests were used to statistically establish a miRNA effect on mRNA signatures. Secondly, correlated and predicted targets were used for gene set enrichment (gProfiler). Inversely, permutation testing was used to determine whether the dysregulated biological processes discovered by the RNA sequencing, were regulated by differentially expressed miRNAs

Fig. 2

MiRNA expression signatures identify nonAF, parAF and persAF patients. a Dimensionality reduction shows a reasonable separation of nonAF, parAF and persAF patients. b Heatmap shows 103 miRNA that were DE between any of the three comparisons. Hierarchical clustering tends to separate nonAF from parAF and persAF patients. Patients show no clustering based on clinical characteristics or comorbidities. Most miRNAs demonstrate a consistent upward or downward trend from nonAF to parAF to persAF. c Log2FCs were plotted for 103 DE miRNAs using nonAF expression as reference. The graph illustrates the continuous expression trend seen in 81 of 103 DE miRNAs

Fig. 3

Top differentially expressed miRNAs in parAF and persAF. a Venndiagram showing in which comparison miRNAs were found to be differentially expressed (FDR<0.05 and |log2FC|> 1). b–i Representative results of miRNA qPCR validation. b–e upregulated miRNAs, f–i downregulated miRNAs. Note that one nonAF sample appears as an outlier. This is not one and the same sample in each figure. To demonstrate that the results were not significantly affected by these apparent outliers, Supplementary Fig. S3 shows the results with exclusion of these outliers. P-values represent the overall variation between the three groups. The miRNAs presented in the graph visually showed an overall ordinal increase or decrease in expression, for which a post-hoc analysis was considered irrelevant and too detailed. P-values were calculated with One-way-ANOVA or Kruskal-Walllis Test when appropriate. Other validated miRNAs can be found in supplementary data

Fig. 4

Top differentially expressed miRNAs that can regulate a multitude of target genes. a–c Correlation graphs show typical examples of miRNA–mRNA sequencing correlations. d Table showing differential expressed miRNAs with more than 10 predicted and correlated targets. e–i Cumulative distribution functions were plotted for predicted and non-targets (CPM > 3) of top differentially expressed miRNAs. P-values were calculated using a Kolmogorov–Smirnov test. e Typical example of a downregulated (miR-138-5p) and f upregulated (miR-182-5p) miRNA that show more negative correlations among predicted targets than among the non-targets. g MiR-208b-3p and h miR-144-5p typify miRNAs with fewer predicted targets. An effect of miRNAs on mRNA expression is suggested visually, but not significant. i MiR-223-3p exemplifies a miRNA with more positive correlations among its predicted targets. j–l qPCR quantification of predicted and correlated targets validated the upregulated expression of ITPRIP and downregulated expression of AJUBA and showed a tended decrease of SULF1. P-values represent the overall variation between the three groups and were calculated with one-way-ANOVA or Kruskal-Walllis test when appropriate

Fig. 5

MiRNAs regulate biological processes involved in atrial fibrillation pathophysiology. a Biological processes enriched in correlated and predicted targets of the top 5 downregulated miRNAs. Only a small number of targets are shown. The figure illustrates that few miRNAs may target a multitude of genes and processes. b Biological processes enriched in correlated and predicted targets of upregulated miRNAs. Only a small number of miRNAs and targets are shown. The figure illustrates how multiple miRNAs may work synergistically to regulate one or few targets and processes

Fig. 6

MiRNA inhibition upregulates multiple target genes. a–e Inhibition of miR-135b-5p demonstrated an increased expression of b Cflar, and a tended increase in the expression of a Dag1 and d Mef2a. P-values were calculated using a T-test or Mann-Whitney U test. f–k Inhibition of miR-138-5p demonstrated an increased expression of f–j Arhgap31, Cspg4, Garre1, Heyl, and Itprip, and a tended increase in the expression of k Pacsin2. P-values were calculated using a non-parametric Mann-Whitney U test. Note that there are three outliers after miRNA inhibition. To demonstrate that the results were not substantially affected by these apparent outliers, Supplementary Fig. S6 shows the results with exclusion of these three outliers. NC, negative control