Loss of microRNA-106b-25 Cluster Promotes Atrial Fibrillation by Enhancing Ryanodine Receptor Type-2 Expression and Calcium Release

Abstract

Background

Enhanced sarcoplasmic reticulum (SR) Ca²⁺-leak via ryanodine receptor type-2 (RyR2) contributes to the pathogenesis of atrial fibrillation (AF). Recent studies have shown that the level of RyR2 protein is elevated in atria of paroxysmal AF (pAF) patients, suggesting that microRNA-mediated post-transcriptional regulation of RyR2 might be an underlying mechanism. Bioinformatic analysis suggests that miR-106b and miR-93, members of the miR-106b-25 cluster, could bind to RYR2-3′UTR and suppress its translation. Thus, we tested the hypothesis that loss of the miR-106b-25 cluster promotes AF via enhanced RyR2-mediated SR Ca²⁺-leak.

Methods and Results

Quantitative real-time PCR showed that the levels of mature miR-106b, miR-93 and miR-25 were lower in atria of pAF patients compared with patients with sinus rhythm. In vitro assay showed that miR-93 reduced RYR2-3′UTR luciferase activity. Total RyR2 protein in atrial tissue of miR-106b-25^−/− mice was increased by 42% compared to wild-type (WT) littermates, but still maintained a normal subcellular distribution. Ca²⁺-spark frequency and total SR Ca²⁺-leak were increased in atrial myocytes of miR-106b-25^−/− mice. Telemetry ECG recordings revealed that miR-106b-25^−/− mice exhibited more frequent atrial ectopy and were also more susceptible to pacing-induced AF than WT littermates. Increased SR Ca²⁺-release and AF susceptibility in miR-106b-25^−/− mice were abolished by the RyR2-blocker K201.

Conclusions

These results suggest that miR-106b-25 cluster mediated post-transcriptional regulation of RyR2 is a potential molecular mechanism involved in pAF pathogenesis. As such, the miR-106b-25 cluster could be a novel gene-therapy target in AF associated with enhanced RyR2 expression.

Introduction

Atrial fibrillation (AF) is the most commonly encountered cardiac arrhythmia in the clinical setting and the leading cause of stroke among other morbidities and mortalities. Currently, pharmacologic treatments of AF are not optimal due to an incomplete understanding of AF pathogenesis.¹ Emerging evidence suggests a central role for ryanodine receptor type-2 (RyR2) in AF development including both initiation and maintenance of AF.^{2, 3} Specifically, inappropriate diastolic Ca²⁺-release from the sarcoplasmic reticulum (SR) through RyR2 has been causally linked to both atrial ectopic activity as well as atrial remodeling critical for AF maintenance and progression.^{2, 4} In chronic AF (cAF), hyperactivity of RyR2 has been attributed to RyR2 hyperphosphorylation caused by overactive kinases.⁵ In contrast, recent studies have found that RyR2 is not hyperphosphorylated in paroxysmal AF (pAF), the earlier stage of AF before extensive structural remodeling occurs.^{4, 6} Instead, these studies showed that RyR2 protein levels are upregulated in pAF and that there is an increase in SR Ca²⁺-leak in these patients.^{4, 6} While a number of biological processes could contribute to protein upregulation, we hypothesized that altered epigenetic regulation of RyR2 in pAF due to a downregulation of microRNA (miRNA)-mediated translational repression leads to the increased SR Ca²⁺-leak and atrial arrhythmogenesis.

miRNAs are a class of 21-24 nucleotide non-coding RNAs that plays a key role in post-transcriptional regulation of mRNAs by either suppressing their translation or promoting their degradation.⁷ A number of studies have uncovered the role of miRNAs in cardiac diseases such as myocardial infarction,⁸ cardiac hypertrophy,⁹ as well as AF.^10–12 However, these studies did not address the changes in pAF nor investigate the regulation of RyR2 by miRNAs despite its central role in AF pathogenesis. In this study, we identified miRNAs that are bioinformatically predicted to regulate the RYR2-3′UTR and confirmed using luciferase reporter assay that miR-93, a member of the miR-106b-25 cluster, regulates RYR2-3′UTR. We also showed that expression of the miR-106b-25 cluster is reduced in pAF patients, pointing to a potential causal relationship between the downregulation of this cluster and the upregulation of RyR2 protein. We directly tested this hypothesis in mice and demonstrated that miR-106b-25 ablation led to 1) increased RyR2 protein levels, 2) enhanced Ca²⁺-sparks frequency and SR Ca²⁺-leak in atrial myocytes, and 3) increased frequency of atrial ectopy and susceptibility to pacing-induced AF.

Methods

Detailed methods are provided in Supplemental Material.

Human atrial samples

Human right atrial appendages were collected with written informed consent under protocols approved by the ethics committee of the Medical Faculty Essen, University Duisburg-Essen (12-5268-BO). Detailed characteristics of patients are provided in Supplemental Table S1.

Study animals

Animal protocols were approved by the Institutional Animal Care and Use Committee of Baylor College of Medicine. miR-106b-25 homozygous knockout mice (miR-106b-25^−/−) were purchased from The Jackson Laboratory.

Bioinformatic identification of candidate RYR2-regulating miRNAs

We used three established miRNA target prediction algorithms: TargetScan6.2, PITA, and microrna.org, to identify potential RYR2-regulating miRNAs. Only thirteen miRNAs were predicted to target RYR2 gene by all three algorithms (Supplemental Table S2). The ranking of these thirteen miRNAs from each algorithm were combined and sorted. The top five miRNAs based on the combined ranking were considered for further analysis (Supplemental Table S2).

Quantitative real-time PCR

Total RNA including miRNAs was isolated using Direct-zol^™ RNA MiniPrep (Zymo Research, Irvine, CA). For detection of mRNAs, reverse transcription was performed using iScript^™ cDNA Synthesis Kit (Bio-Rad, Hercules, CA). For detection of miRNAs, reverse transcription and quantitative real-time (qRT)-PCR were carried out using a modified protocol as previously described.¹³ All qRT-PCR reactions were performed using PerfeCTa® SYBR® Green FastMix® (Quanta Biosciences, Gaithersburg, MD) in triplicates in 96-well plates in Mastercycler ep realplex (Eppendorf, Hamburg, Germany). Expression levels were compared using the relative CT (cycle number) method after normalization to L7. The primers used for mRNA and miRNA detection are listed in Supplemental Tables S4 and S5, respectively.¹⁴

Luciferase assay

The 1437-bp 3′UTR of Homo sapiens RYR2 and 1418-bp 3′UTR of Mus musculus Ryr2 were amplified respectively from human and mouse cardiac cDNA libraries and subcloned into pRL-CMV 6x CXCR4 Renilla luciferase vector using forward primer 5′-CTGATGTCTAGAGCAGGAATCTTATGTCTGGAAGATG -3′ and reverse primer 5′-CTGATGGCGGCCGCGTTAAATAGTTTAATCAGAGCGGTTTTACAG-3′ for the human RYR2 and forward primer 5′-GTCGTCTCTAGATAGCATCATACCTTGATTGTCTCTG-3′ and reverse primer 5′-GTCGTCGCGGCCGCGTTAGTTAAATAGTTTAATTAGAGCCGTTTTAC-3′ for the mouse Ryr2. The putative interaction sites between miR-93 and the 3′UTRs were searched using RNAhybrid.¹⁵ Mutagenesis was carried out using QuikChangeII (Agilent Technologies, Santa Clara, CA) to disrupt these interactions using forward primer 5′-TCGACATAAAGAAAATAAAGATGTTTTTACCGGTACGTATCTTCTGCATAGCTCAACCCACA-3′ and reverse primer 5′-TGTGGGTTGAGCTATGCAGAAGATACGTACCGGTAAAAACATCTTTATTTTCTTTATGTCGA-3′ for the human RYR2 3′UTR and forward primer 5′-GATGTAAAGAAAATAAAGGTGTTTGACCGGTACATTATCTTCTGCATAGCTCAACCCACTTATA-3′ and reverse primer 5′-TATAAGTGGGTTGAGCTATGCAGAAGATAATGTACCGGTCAAACACCTTTATTTTCTTTACATC-3′ for the mouse Ryr2 3′UTR.

HEK293 cells were seeded in 24-well plates using antibiotics-free DMEM medium (Invitrogen, Carlabad, CA) supplemented with 10% fetal bovine serum and 2 mM L-glutamine. When the cells reached 50–90% confluency, they were co-transfected using LipoD293 (SignaGen Laboratories, Rockville, MD) and Opti-MEM (Life Technologies) with each of the 3′UTR Renilla luciferase constructs, the pGL3 firefly luciferase construct, and each of the following miRNA mimics: miR-25, miR-93, miR-106a, miR-106b, and non-targeting scramble (Life Technologies, Carlsbad, CA) used at a final concentration of 100 nM. Eighteen to 24 hours after transfection, the cells were harvested and assayed using the Dual-Luciferase® Reporter Assay System (Promega, Madison, WI) according to the manufacturer’s instructions. Renilla luciferase signals were normalized to the firefly luciferase signals.

Immunoblotting

Protein extraction and Western blotting were performed as previously described.² Briefly, mouse or human tissue lysates or SR fractions were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinyl difluoride (PVDF) membranes. The membranes were then incubated with antibodies in 5% milk against the following proteins: RyR2 (1:5,000; MA3-916, Thermo Fisher Scientific, Houston, TX), Na⁺/Ca²⁺-exchanger type-1 (NCX1; 1:1,000; R3F1, Swant, Bellinzona, Switzerland), calsequestrin type-2 (CSQ2; 1:1,000; PA1-913, Thermo Fisher Scientific), Ca²⁺/calmodulin protein kinase II phosphorylated RyR2 at Serine2814 (pS2814; 1:1,000; custom-made),¹⁶ and GAPDH (1:10,000; MAB-374, Millipore, Billerica, MA). Membranes were then washed in tris-buffered saline and incubated with secondary anti-mouse and anti-rabbit antibodies (both 1:10,000) conjugated respectively to Alexa-Fluor 680 (Life Technologies, Carlsbad, CA) and IR800Dye (Rockland Immunochemicals, Gilbertsville, PA). The fluorescence was visualized on an Odyssey infrared scanner (Li-Cor, Lincoln, NE) and the bands were quantified using ImageJ (National Institute of Health, Bethesda, MD).

Telemetry ECG recording

Mice were implanted with telemeters (Data Sciences International, MN, USA) as previously described.¹⁷ 24-hour ambulatory ECG was monitored in conscious mice at ages of 3 months. Telemetry ECG recordings were analyzed using ECG-Auto software (Emka Technologies).

Intracardiac electrophysiology

In vivo electrophysiology studies were performed in mice at the age of 4–5 months, as previously described.¹⁶ Additional details are provided in the Supplemental Material.

Co-immunoprecipitation

Atrial lysates was prepared and co-immunoprecipitation was performed as previously described.¹⁸ 1000μg of lysates was suspended with lysis buffer: 1% CHAPS in 1000μl RIPA buffer (pH 7.4) containing 150 mM NaCl, 10 mM Tis-HCl, 20 mM NaF, 1 mM NaVO₃ and protease and phosphatase inhibitor cocktails (cOmplete, Mini and PhosSTOP from Roche Applied Science, Indianapolis, IN). The samples were incubated at 4 °C with 3μl of RyR2 antibody (MA3-916; Thermo Scientific) or mouse IgG (M5284; Sigma) and rotated overnight. The next day, 50μl of protein G beads slurry was added to each sample and further rotated at 4 °C for 1 hour. The pellet was collected and washed 2 times with the same RIPA buffer (except without CHAPS). Afterwards, the pellet was incubated with 20μl of loading buffer at 70°C for 10 min before the supernatant was used for SDS-PAGE.

Statistical analysis

Data are presented as mean ± SEM. Statistical analysis was performed using SPSS Exact Tests (IBM, USA). To compare continuous variables with a skewed distribution, the Mann-Whitney test was applied. Fisher’s exact test was used to compare categorical data. Ca²⁺ sparks and leak data were analyzed using R and the R packages lme4 and languageR, by using linear mixed effects models where the sparks or leak values are the dependent variable while genotype/treatment group is set as the fixed effect and the mouse and cell (or spark for the spark parameters) are set as random effects. The Markov-chain Monte Carlo sampling or MCMC-estimated P-values of these models were reported. A P-value of 0.05 or less was considered statistically significant.

Results

Increased RyR2 protein expression in pAF patients

Previous studies demonstrated an increased level of RyR2 protein in atrial samples from pAF patients.^{4, 6} Using the same or similar samples, RyR2 protein and mRNA levels were assessed to pinpoint the underlying mechanism of dysregulation. Increased protein levels of RyR2 were found in SR fractions of pAF patients compared with control patients in sinus rhythm (SNR) (P<0.05; Figure 1A), consistent with previous reports.^{4, 6} In contrast, however, RYR2 mRNA levels were unchanged in pAF compared with SNR patients (P>0.05; Figure 1B). This suggests a putative role for miRNA-mediated post-transcriptional regulation that normally would repress the translation of RYR2 mRNA. A bioinformatic analysis to identify candidate miRNAs was performed using three established miRNA target prediction algorithms (TargetScan6.2, PITA, and microrna.org). Thirteen miRNAs were predicted by all three algorithms to target the RYR2 3′UTR (Supplemental Table S2). Following sorting by a score (“Rank”) that takes into account of all 3 algorithms, the top five miRNAs (miR-93, miR-129, miR-506, miR-124, and miR-106a) were chosen for further studies (Supplemental Table S2).

Figure 1.

Upregulation of RyR2 and downregulation of the miR-106b-25 cluster in pAF patients. A) WB analysis revealed increased RyR2 protein levels in sarcoplasmic reticulum preparation of atrial biopsies from patients with pAF versus control patients in SNR. B) qRT-PCR showing unchanged levels of RYR2 mRNA in atrial samples of pAF patients. C) qRT-PCR showing downregulation of the miR-106b-25 cluster and miR-106a in pAF patients. Numbers in bars indicate number of patients. *P<0.05, **P<0.01 vs. SNR.

The miR-106b-25 cluster is downregulated in pAF patients

Since miRNAs negatively regulate mRNAs, we hypothesized that RYR2-regulating miRNAs would be downregulated in pAF patients. The levels of the top five predicted miRNAs were therefore measured using qRT-PCR in atrial samples from pAF and SNR patients. Only miR-93 and miR-106a were significantly downregulated in pAF patients by 44% (P<0.01) and 53% (P<0.05), respectively, compared to patients in SNR (Figure 1C). Because miR-106b and miR-25 are co-members of miR-93 in the miR-106b-25 cluster,¹⁹ we also measured their levels in pAF patients and found that they were also significantly downregulated in pAF samples by 38% (P<0.05) and 39% (P<0.05), respectively (Figure 1C). These data demonstrate that the entire miR-106b-25 cluster is downregulated in pAF patients, which is most likely due to the downregulation of its host gene, minichromosome maintenance complex component 7 (MCM7) (Supplemental Figure S1).

Since the expression level of miR-93 is about 2.5-fold higher than that of miR-106a in all four chambers of the human heart²⁰ and miR-93 is the top-ranked candidate that regulates RyR2 (Supplemental Table S2), we focused our investigation on miR-93 and the miR-106b-25 cluster in the regulation of RyR2 in pAF.

miR-93 targets the RYR2 3′UTR

Computational analysis of the Homo sapiens and Mus musculus RYR2-3′UTR revealed conserved binding sites for miR-93 (Figure 2A–C). To determine whether miR-93 could bind to the 3′UTR of RYR2, dual luciferase activity of RyR2-3′UTR construct was measured in the presence of miR-106b, miR-93, miR-25, miR-106a, or scramble (negative control) mimics. miR-93 mimic significantly reduced the luciferase activity of both the human and mouse RYR2-3′UTR (P<0.01), compared with the scramble mimic (Figure 2D). Mutation of the predicted binding site on both 3′UTRs abolished this inhibitory effect of the miR-93 mimic. In addition, miR-106b (P<0.05) and miR-106a (P<0.01) mimics also reduced the luciferase activity of the mouse Ryr2-3′UTR but not the human RYR2-3′UTR, which effect was also abolished by the mutation on the predicted binding site on mouse Ryr2-3′UTR. In contrast, miR-25 mimic had no effect on either human or mouse RYR2-3′UTR luciferase activity. Together, these results demonstrate that miR-93 negatively regulates RYR2-3′UTR, which could lead to decreased RyR2 protein expression.

Figure 2.

miR-93 target RYR2-3′UTR. A) Sequences of mature miR-106b, miR-93, miR-25, and miR-106a. B) Predicted interaction between human (h) Ryr2-3′UTR and miR-93 (i) and mutation of the binding site (ii). C) Predicted interaction between mouse (m) Ryr2-3′UTR and miR-93 (i) and mutation of the binding site (ii). D) Dual luciferase assay of 3′UTR constructs in the presence of scramble, miR-106b, miR-93, miR-25, and miR-106a mimics. n=4 or 8 per group. **P<0.01.

miR-106b-25 deficiency causes upregulation of RyR2 protein

To determine whether loss of the miR-106b-25 cluster could lead to RyR2 upregulation, both RyR2 mRNA and protein levels were measured in atrial tissue of miR-106b-25^−/− mice. Consistent with pAF patients, RyR2 mRNA levels were comparable in atria of miR-106b-25^−/− and WT littermates (Figure 3A), while RyR2 protein was increased by 42% in miR-106b-25^−/− atrial lysates (P<0.05; Figure 3B). Protein levels of other major Ca²⁺-handling proteins including Na⁺/Ca²⁺ exchanger type-1 (NCX1), calsequestrin type-2 (CSQ2), sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase 2a (SERCA2a), and phospholamban (PLN), as well as the phosphorylation levels of RyR2 at serine (S)2814 were unchanged in miR-106b-25^−/− atria (Figure 3C). Interestingly, there was no change in RyR2 protein levels in the ventricles of miR-106b-25^−/− mice (Supplemental Figure S2).

Figure 3.

Increased RyR2 protein levels in miR-106b-25^−/− atria. A) qRT-PCR showing unchanged levels of Ryr2 mRNA in miR-106b-25^−/− atria. B) Representative western blots and dot plot showing upregulated RyR2 protein levels in miR-106b-25^−/− atrial lysates. *P<0.05. C) Representative western blots and bar graph showing unchanged expression of other major Ca²⁺-handling proteins and phosphorylation of RyR2 at serine (S)2814 in miR-106b-25^−/− atria. N=3–4 animals per group.

Immunocytochemistry experiment in atrial myocytes isolated from miR-106b-25^−/− mice demonstrated that RyR2 exhibited normal distribution patterns within the junctional SR membrane where it co-localizes with junctophilin-2 (JPH2; Supplemental Figure S3). This suggests that the “extra” RyR2 channels are proportionally distributed within the cells, most likely participating in SR Ca²⁺-release.

miR-106b-25 deficiency increases Ca²⁺-release via RyR2 clusters

To evaluate whether the “extra” RyR2 channels can increase Ca²⁺-release, we measured Ca²⁺-sparks frequency (CaSF) in isolated atrial myocytes. CaSF normalized to SR Ca²⁺-load was increased in miR-106b-25^−/− (2.2 ± 0.5 a.u.) compare to WT littermates (1.0 ± 0.2 a.u., P<0.05; Figure 4A, B). While the amplitude (Figure 4C), duration and decay of Ca²⁺-sparks (Supplemental Table S3) were unaltered in miR-106b-25^−/−, the full-width at half maximum (FWHM) was increased in miR-106b-25^−/− (1.46±0.03μm) compared to WT littermates (1.36±0.03μm; P<0.05; Figure 4D). Moreover, incubation with the RyR2-inhibitor K201²¹ reduced the CaSF (0.7 ± 0.2 a.u., P<0.05) and the FWHM (1.11 ± 0.03μm, P<0.001) in atrial myocytes of miR-106b-25^−/− (Figure 4B, D). Together these findings support the notion that the increased amount of RyR2 in miR-106b-25^−/− mice resulted in larger functional RyR2 clusters on the SR.

Figure 4.

Enhanced RyR2 Ca²⁺-release activity in miR-106b-25^−/− atrial myocytes. A) Representative line-scan images of Ca²⁺-spark recordings in isolated atrial myocytes. B) Bar graph showing increased CaSF normalized to SR load in miR-106b-25^−/− atrial myocytes, normalized by K201. C) Bar graph showing unchanged Ca²⁺-spark amplitude comparing miR-106b-25^−/− and WT atrial myocytes. D) Bar graph showing an increase in the FWHM of Ca²⁺-sparks in miR-106b-25^−/− atrial myocytes, normalized by RyR2 blocker K201. Numbers inside bars indicate number of cells studied from 3 mice per group. *P<0.05, **P<0.01, ***P<0.001.

We also analyzed the kinetics of the Ca²⁺-induced-Ca²⁺-release (CICR). We found that the Ca²⁺-transient amplitude was increased in atrial myocytes isolated from the miR-106b-25^−/− mice while the decay constant (tau) of Ca²⁺-transient was unchanged (Supplemental Figure S4). The enhanced CICR could be attributed to more SR Ca²⁺-release via either a larger RyR2 cluster or increased number of RyR2 clusters, both of which facilitates the synchronization between RyR2 cluster and lead to a greater Ca²⁺-transient amplitude. This also suggests that not only is diastolic SR Ca²⁺-leak increased but CICR is also increased. On the other hand, the unchanged tau suggests that SERCA2a activity is unchanged with miR-106b-25 deficiency, consistent with the unchanged SERCA2a protein level in the miR-106b-25^−/− atria (Figure 3C).

It was previously shown that JPH2 binding to RyR2 stabilizes RyR2 and that reduced JPH2/RyR2 ratio in pAF patients causes RyR2 hyperactivity associated with AF pathogenesis.⁶ Thus, we evaluated the interaction between RyR2 and JPH2 in atrial samples of miR-106b-25^−/− mice. Co-immunoprecipitation experiment revealed that more RyR2 was pulled-down from the miR-106b-25^−/− atrial lysates (Supplemental Figure S5A). Despite the similar amount of JPH2 that was co-immunoprecipitated with RyR2, the ratio of JPH2/RyR2 was significantly reduced in miR-106b-25^−/− atria (P<0.01; Supplemental Figure S5). This finding suggests that the extra RyR2 molecules are not stabilized by JPH2 and therefore may be hyperactive, contributing to the increased SR Ca²⁺-leak in miR-106b-25^−/− atrial myocytes.

miR-106b-25 deficiency increases spontaneous Ca²⁺-waves

To test whether the increased amount of RyR2 in miR-106b-25^−/− directly affects SR Ca²⁺-release, we measured SR Ca²⁺-leak.² SR Ca²⁺-leak normalized to SR Ca²⁺-load was significantly higher in miR-106b-25^−/− (20.1±3.8%) compared to WT littermates (10.5±1.4%, P<0.05; Figure 5A, B). In addition, pre-incubation of atrial myocytes with the RyR2 inhibitor K201 (1μmol/L) for 30 minutes reduced the SR Ca²⁺-leak to 9.4±1.7% in the miR-106b-25^−/− (P<0.05 vs untreated cells; Figure 5B).

Figure 5.

Enhanced SR Ca²⁺-leak and arrhythmogenic Ca²⁺-release in miR-106b-25^−/− atrial myocytes. A) Representative SR Ca²⁺-leak measurements in atrial myocytes. NT=normal Tyrode solution; TTC=tetracaine. B) Bar graph showing increased relative levels of SR Ca²⁺-leak in miR-106b-25^−/− atrial myocytes. Representative tracing (C) and summary bar graph (D) showing increased spontaneous Ca²⁺-wave frequency in high Ca²⁺-buffer with isoproterenol in atrial myocytes from miR-106b-25^−/− versus WT littermates (n=3–4 mice per group). Red arrow marks a spontaneous Ca²⁺-wave. Numbers inside bars indicate number of cells studied. *P<0.05, **P<0.01.

To assess whether enhanced SR Ca²⁺-leak could evoke potentially arrhythmogenic Ca²⁺-waves, the extracellular Ca²⁺-concentration was increased to 2.5 mM and isoproterenol (1 μmol/L) was added to the perfusate. Under these conditions 67% of myocytes from miR-106b-25^−/− mice (9 of 16) developed spontaneous Ca²⁺-waves, whereas this behavior was observed in only 13% of myocytes from WT littermates (2 of 16, P<0.01; Figure 5C, D). These data indicate that miR-106b-25 deficiency increases the incidence of proarrhythmic Ca²⁺-release events due to upregulation of RyR2 levels.

miR-106b-25 deficiency promotes ectopic activity and an AF substrate

To examine whether miR-106b-25^−/− develop spontaneous atrial arrhythmias, 24-hour ECG telemetry recordings were performed. Although no episodes of spontaneous AF were observed, an increase in premature atrial contractions (PACs) were seen in the miR-106b-25^−/− (6.0 ± 2.0 beats/hr) versus WT littermates (1.0 ± 0.4 beats/hr, P<0.05; Figure 6 A, B). This suggests that loss of the miR-106b-25 cluster promotes atrial ectopy in mice.

Figure 6.

miR-106b-25 deficiency leads to increased atrial ectopy and atrial weight. A) Representative ECG telemetry recordings and B) dot plot showing premature atrial contractions (PACs) per hour in miR-106b-25^−/− mice. C) Representative whole mount photographs and atrial weight normalized to tibial length (AW/TL; D) showing enlarged atria in miR-106b-25^−/− mice. **P<0.01 vs WT.

We recently showed that RyR2-mediated SR Ca²⁺-leak promotes the development of an AF substrate including enlarged atria and slowed atrial conduction velocity permissive of AF progression.² Here, we found that miR-106b-25^−/− also exhibited an increased atrial weight normalized to tibia length (AW/TL; 0.47 ± 0.02 mg/mm) compared to WT littermates (0.33 ± 0.03 mg/mm, P<0.01; Figure 6 C, D). This suggests that the loss of the miR-106b-25 cluster may also lead to the development of an AF-maintaining substrate, which is not secondary to the impaired ventricular function (supplemental Figure S6). Echocardiography revealed comparable ejection fraction (EF), left ventricular posterior wall thickness during systole (LVPWs) and end-diastolic diameter (EDD) in both miR-106b-25^−/− and WT littermates. This could be due to unaltered RyR2 protein level in the miR-106b-25^−/− ventricles (supplemental Figure S2).

miR-106b-25 deficiency increases the susceptibility to AF

To determine whether miR-106b-25^−/− mice are more prone to AF development, programmed intracardiac stimulation was performed. Baseline electrophysiological parameters including heart rate, QRS, QTc, sinus node recovery time, and atrial refractory period were similar in miR-106b-25^−/− and WT. Upon rapid atrial pacing, 75% of miR-106b-25^−/− mice developed reproducible AF (6 of 8), compared to only 12.5% of WT (1 of 8, P<0.05; Figure 7A, B, D). To test whether the increased AF inducibility was due to enhanced SR Ca²⁺-release via RyR2, we repeated AF induction in miR-106b-25^−/− 15–20 minutes after injected with the RyR2-inhibitor K201 (10 mg/kg, i.p.).²¹ K201 indeed reduced the incidence of pacing-induced AF to 14.3% (1 of 7, P<0.05 vs. before K201 injection; Figure 7 C, D). Altogether, our data support the hypothesis that miR-106b-25 deficiency leads to atrial arrhythmogenesis via enhanced RyR2-mediated SR Ca²⁺-release.

Figure 7.

miR-106b-25 deficiency leads to enhanced AF susceptibility. Representative recordings of surface ECG and intracardiac electrograms in WT (A), miR-106b-25^−/− (B), and miR-106b-25^−/− mice treated with RyR2 blocker K201 (C). D) Bar graph summarizing the incidence of pacing-induced AF showing an increase in miR-106b-25^−/− mice, suppressed by K201. Numbers inside bars indicate numbers of mice studied. *P<0.05.

Discussion

Our study suggests a novel molecular mechanism underlying AF development, namely defective epigenetic regulation of RyR2 expression as a result of reduced miR-106b-25 cluster activity. Our data revealed that the loss of the miR-106b-25 cluster lead to upregulation of RyR2 protein levels and proarrhythmic SR Ca²⁺-release, which were associated with an increased incidence of atrial ectopy. The impaired miRNA-mediated RyR2 dysregulation increased AF susceptibility in mice and may contribute to the pathogenesis of paroxysmal AF in humans.

Emerging role of miRNA-mediated epigenetic regulations in AF development

Epigenetic regulations are emerging as important mechanisms underlying the development of cardiac arrhythmias.²² Recent advancement in DNA and RNA sequencing technologies has enabled epigenome mapping and provided novel insights into the abnormal expressions of key proteins implicated in arrhythmogenesis.^{23, 24} A number of studies have demonstrated the role of miRNAs in AF pathogenesis.^{11, 12, 25} For example, miR-328,¹¹ miR-499,²⁶ and miR-26¹² have been associated with AF development via regulating L-type Ca²⁺-channel, small-conductance Ca²⁺-activated potassium channel 3, and inward-rectifier potassium current, respectively. In addition, miR-133, miR-590, and miR-21 have been shown to promote atrial fibrosis associated with AF.^{25, 27}

Our study revealed that an entire miR cluster, the miR-106b-25 cluster, is downregulated in pAF patients and that genetic ablation of this cluster promoted arrhythmogenesis through increased RyR2-mediated Ca²⁺-release. Interestingly, a recent study by Wahlquist et al.²⁸ showed that enhanced miR-25 mediated regulation of sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase 2a (SERCA2a) contributes to the reduced cardiac contractility in heart failure (HF). While the suppressed SERCA2a-mediated Ca²⁺-reuptake is closely related to the development of HF,^{29, 30} its role in the pathology of AF has not been clearly define. Moreover, in our study, we found that only miR-93, but not miR-25, regulates RYR2-3′UTR, suggesting that miR-93 is the main player within the miR-106b-25 cluster that regulates RyR2 Ca²⁺-release. Finally, in the study by Wahlquist et al.,²⁸ miR-25 was found to be upregulated in patients with HF whereas the entire miR-106b-25 cluster was found to be downregulated in patients with pAF in the present study (Figure 1C). Despite the differences, these findings highlight the importance of this cluster in regulating Ca²⁺-homeostasis and suggest potential co-regulation, the mechanism of which would require further studies.

Novel epigenetic mechanism underling RyR2 hyperactivity in AF

Many studies have demonstrated that spontaneous SR Ca²⁺-release via RyR2 can cause delayed afterdepolarizations (DADs),¹⁶ which may constitute the cellular correlate of focal atrial activity in vivo. We have shown that mice with dysfunctional RyR2 channels are more susceptible to pacing-induced AF, which is associated with DAD-mediated triggered activity.⁵ Recently, we discovered that RyR2-mediated SR Ca²⁺-leak drives atrial remodeling leading to the development of a substrate permissive of AF development.² Thus, RyR2-mediated Ca²⁺-release plays an essential role in both the initiation and perpetuation of AF.

Evidence suggests that altered post-translational modifications (e.g. phosphorylation⁵ and nitrosylation³¹) along with aberrant protein-protein interactions mediated by the binding partners (e.g. FKBP12.6³² and JPH2⁶) are responsible for the hyperactivity of RyR2 channels in cardiac diseases. Our recent studies and current findings revealed a prominent upregulation of RyR2 in pAF patients,^{4, 6} and identified a novel mechanism contributing to RyR2 hyperactivity in AF, suggesting that altered miRNA-mediated regulations of RyR2 could be an important epigenetic mechanism involved in AF pathogenesis.

The mechanisms underlying the arrhythmogenic Ca²⁺-release due to the miR-106b-25 deficiency could include: 1) the “extra” RyR2 molecules are not stabilized by its binding partners, such as JPH2, leading to a hyperactive RyR2 channel itself; 2) the increased amount of RyR2 channels could form larger cluster to release more Ca²⁺ during each opening; 3) the synchronization between RyR2-clusters may be improved due to a possibly reduced distance between clusters; 4) the propensity of diastolic Ca²⁺ propagate to activate the NCX channel is ultimately increased. One or all of the mechanisms mentioned above could increase the arrhythmogenic SR Ca²⁺-release and thereby increase the AF susceptibility.

Therapeutic implications

The most important challenge in AF management is finding a way to forestall the progression of the AF substrate, i.e. to break the “AF begets AF” downward spiral. Our recent study demonstrated that inhibition of the RyR2-mediated SR Ca²⁺-leak could prevent the progression from pAF to persistent long-lasting AF.² Efforts are underway to develop effective compounds to correct RyR2 hyperactivity in various cardiac diseases.³³ On the other hand, our study provides a novel therapeutic approach aimed at normalizing the protein homeostasis using miRNAs as opposed to ion channel blockers, which are suboptimal and sometimes paradoxically pro-arrhythmogenic.¹ Future studies are needed to specifically evaluate whether miR-93 mimic can reduce AF susceptibility by downregulating RyR2 expression and thereby correct SR Ca²⁺-leak in mouse models of AF and ultimately in AF patients.

Study limitations

Although we have found that the host gene of miR-106b-25 cluster, MCM7, is downregulated in pAF patients, future studies are needed to establish the mechanistic regulation causing this dysregulation. Moreover, functional alterations in miR-106b-25^−/− are confined to atria not ventricles, despite that the miR-106b-25 cluster is missing in entire cardiac tissue. In line with these current observations, a few recent studies also have shown the similar findings.^{2, 3, 6} Further studies are needed to delineate the mechanisms underlying the differential responses in atria and ventricle.

Conclusion

Our study suggests that abnormal miR-106b-25^−/− regulation of RYR2 expression is one potential molecular mechanism involved in the arrhythmogenesis in both mice and patients with paroxysmal AF. Targeting miRNAs may constitute a novel therapeutic strategy to correct the potentially arrhythmogenic SR Ca²⁺-leak in pAF patients. miR-93 mimic may therefore constitute a novel therapeutic strategy in treating pAF associated with RyR2 hyperactivity.