Identification of microRNA-mRNA dysregulations in paroxysmal atrial fibrillation

Abstract

Background: The molecular mechanisms underlying the early development of atrial fibrillation (AF) remain poorly understood. Emerging evidence suggests that abnormal epigenetic modulation via microRNAs (miRNAs) might be involved in the pathogenesis of paroxysmal AF (pAF).

Objective: To identify key molecular changes associated with pAF, we conducted state-of-the-art transcriptomic studies to identify the abnormal miRNA-mRNA interactions potentially driving AF development.

Methods: High-quality total RNA including miRNA was isolated from atrial biopsies of age-matched and sex-matched pAF patients and control patients in sinus rhythm (SR; n=4 per group) and used for RNA-sequencing and miRNA microarray. Results were analyzed bioinformatically and validated using quantitative real-time (qRT)-PCR and 3'UTR luciferase reporter assays.

Results: 113 genes and 49 miRNAs were differentially expressed (DE) in pAF versus SR patients. Gene ontology analysis revealed that most of the DE genes were involved in the "gonadotropin releasing hormone receptor pathway" and "p53 pathway". Of these DE genes, bioinformatic analyses identified 23 pairs of putative miRNA-mRNA interactions that were altered in pAF (involving 15 miRNAs and 17 mRNAs). Using qRT-PCR and 3'UTR luciferase reporter assays, the interaction between upregulation of miR-199a-5p and downregulation of FKBP5 was confirmed in samples from pAF patients.

Conclusion: Our combined transcriptomic analysis and miRNA microarray study of atrial samples from pAF patients revealed novel pathways and miRNA-mRNA regulations that may be relevant in the development of pAF. Future studies are required to investigate the potential involvement of the gonadotropin releasing hormone receptor and p53 pathways in AF pathogenesis.

Keywords: Paroxysmal atrial fibrillation; RNA sequencing; Transcriptomic study; microRNA.

Figure 1. Columnwise side-by-side boxplots of normalized samples showing comparable count distribution across samples in the RNA-Seq experiment

SR=control patients in sinus rhythm; pAF=paroxysmal atrial fibrillation.

Figure 2. RNA-Seq of atrial samples from sinus rhythm (SR) and paroxysmal AF (pAF) patients

A. Volcano plot of all the transcripts detected by RNA-Seq highlighting the 113 transcripts that are differentially expressed between SR and pAF samples (red circles). B. Heat-map of the 113 differentially expressed transcripts. Gene ontology pie charts of the differentially expressed transcripts based on the “Molecular Function” (C) and “Biological Process” (D) categories.

Figure 3. miRNA microarray of atrial samples from sinus rhythm (SR) and paroxysmal AF (pAF) patients

A. Volcano plot of all the miRNAs detected by microarray highlighting the 49 miRNAs that are both differentially expressed between SR and pAF samples and have a high signal on the microarray (red circles). B. Heat-map of the 49 differentially expressed miRNAs. C. Pie chart showing known roles of the differentially expressed miRNAs. “Novel” means no known function has been established for the miRNAs.

Figure 4. miRNA-mRNA interactions

A. Schematic of the workflow to identify putative miRNA-mRNA interactions relevant for AF pathogenesis. DE=differentially expressed. B. qRT-PCR validation of 8 mRNAs selected from the 23 putative miRNA-mRNA interaction pairs. N=6-8 per group. C. qRT-PCR validation of 5 miRNAs selected from the 23 putative miRNA-mRNA interaction pairs in pAF vs. SR samples. N=4 per group. *P<0.05, **P<0.01, vs. SR.

Figure 5. Interaction between hsa-miR-199a-5p and FKBP5 3’fUTR

A. Schematic of bioinformatically predicted interaction site between hsa-miR-199a-5p and Homo sapiens FKBP5 3’UTR based on microRNA.org.[23] Top panel: wild-type (WT) 3’UTR of FKBP5. Bottom panel: mutant (Mut) 3’UTR of FKBP5 where the seed site for hsa-miR-199a-5p is partially ablated. B. Conservation of the hsa-miR-199a-5p seed site based on TargetScanHuman 6.2.[26] C. 3’UTR luciferase assay showing a relative decrease in the normalized luciferase signal from the WT FKBP5 3’UTR luciferase construct in the presence of hsa-miR-199a-5p but not from the mutant (Mut) construct. N=4 replicates per group. **P<0.01 vs. Scramble.