A common variant alters SCN5A-miR-24 interaction and associates with heart failure mortality

Abstract

SCN5A encodes the voltage-gated Na+ channel NaV1.5 that is responsible for depolarization of the cardiac action potential and rapid intercellular conduction. Mutations disrupting the SCN5A coding sequence cause inherited arrhythmias and cardiomyopathy, and single-nucleotide polymorphisms (SNPs) linked to SCN5A splicing, localization, and function associate with heart failure-related sudden cardiac death. However, the clinical relevance of SNPs that modulate SCN5A expression levels remains understudied. We recently generated a transcriptome-wide map of microRNA (miR) binding sites in human heart, evaluated their overlap with common SNPs, and identified a synonymous SNP (rs1805126) adjacent to a miR-24 site within the SCN5A coding sequence. This SNP was previously shown to reproducibly associate with cardiac electrophysiological parameters, but was not considered to be causal. Here, we show that miR-24 potently suppresses SCN5A expression and that rs1805126 modulates this regulation. We found that the rs1805126 minor allele associates with decreased cardiac SCN5A expression and that heart failure subjects homozygous for the minor allele have decreased ejection fraction and increased mortality, but not increased ventricular tachyarrhythmias. In mice, we identified a potential basis for this in discovering that decreased Scn5a expression leads to accumulation of myocardial reactive oxygen species. Together, these data reiterate the importance of considering the mechanistic significance of synonymous SNPs as they relate to miRs and disease, and highlight a surprising link between SCN5A expression and nonarrhythmic death in heart failure.

Keywords: Cardiology; Genetic variation; Genetics; Ion channels; Noncoding RNAs.

Figure 1. The synonymous coding SCN5A SNP, rs1805126, modulates the miR-24–SCN5A interaction.

(A) Schematic of the conserved miR-24–SCN5A interaction identified by Ago2 HITS-CLIP (read coverage shown above). The site resides within the coding region of the SCN5A terminal exon and is adjacent to the synonymous SNP rs1805126 (*) that has previously been linked to heart rhythm abnormalities by GWAS. (B–D) The effect of miR-24 on SCN5A expression, and the potential impact of rs1805126 on this interaction, was tested in cell culture experiments. Mouse N2a cells were cotransfected in triplicate with synthetic pre-miRs (4 nM) and human full-length SCN5A expression plasmids harboring either the C or T allele for rs1805126, or synonymous mutations that disrupt the miR-24 seed site (mut*, base mutations are indicated in bold in panel A). At 48 hours after transfection, Western blot and QPCR analyses were used to measure protein and mRNA expression. (B) Representative Western blots show that miR-24 strongly suppresses Na_V1.5 expression, having a more robust effect on the C allele, relative to T. Densitometry analysis of Western blot data (n = 9 biological replicates from 3 separate studies for C versus T, n = 4 for mutated) supports a significant allele-specific difference in miR-24 suppression of Na_V1.5 levels (C), as does QPCR analyses measuring SCN5A mRNA levels (D; n = 6 biological replicates from 2 studies). Data are represented as the mean ± SEM, and P values were obtained using 2-tailed unpaired t test comparing the indicated groups.

Figure 2. Effect of miR-24 overexpression on SCN5A function and expression in neonatal rat cardiomyocytes.

NRCMs were cotransfected with synthetic pre-miRs (25 nM) and plasmids coexpressing miRs/siRNAs and EGFP. Na⁺ current in EGFP-positive cells was measured 2–4 days later using patch clamp. Shown are representative traces of cardiac Na⁺ currents and pulse protocol (A) and the current-voltage relationship of Na⁺ current (B) in NRCMs transfected with miR-Neg1 (miR-N, negative control), miR-24, or siSCN5A (positive control). (C) Sodium current density at –10 mV is also plotted. These patch-clamp data support the hypothesis that miR-24 potently suppresses sodium current density in NRCMs, to a similar degree as Scn5a-targeted siRNA. Data are represented as the mean ± SEM. Sample numbers combined from 3 separate experiments are denoted in the figure, and P values were calculated using 1-way ANOVA with Dunnett’s post hoc test (versus miR-Neg1). (D) NRCMs were transfected with 25 nM synthetic pre-miR mimics and endogenous Na_V1.5 protein levels were measured by Western blot 48 hours later; a representative Western blot image is shown. (E) Combined Western blot (WB) densitometry analysis from 2 separate experiments (n = 6 total biological replicates per treatment; mean ± SEM) is plotted. Pre–miR-29 and -Neg1 serve as negative controls, and β-actin or GAPDH levels were used for normalization. *P value derived using 1-way ANOVA with Dunnett’s post hoc test (versus miR-Neg1).

Figure 3. Association of rs1805126 genotype with heart failure outcomes.

(A–C) Patient samples from GRADE (a multicenter prospective study of heart failure outcomes) were genotyped for the rs1805126 SNP and Kaplan-Meier analyses were performed to compare overall survival rates from all-cause death (A), and freedom from appropriate ICD shocks (B), a surrogate for sudden cardiac death. Homozygous CC individuals had significantly poorer survival outcomes relative to the other genotypes (i.e., T allele carriers). No differences in shock outcomes were observed between the groups. P values were obtained using log-rank test. Cohort sizes are as follows: TT = 532, TC = 628, and CC = 358. (C) CC homozygous individuals had lower left ventricular ejection fractions (LVEF; mean ± SEM; n = 541 TT, 673 TC, and 385 CC). **P < 0.01 using 1-way ANOVA with Dunnett’s post hoc test, CC versus TT genotypes. (D) A replication study was performed in the GRAHF cohort (based on DNA sample availability); consistent with GRADE, rs1805126-CC homozygous GRAHF patients showed a near-significant trend towards worse heart failure outcomes (i.e., lower A-HeFT composite scores). Group mean (dark line) and 95% confidence intervals (light shading) are plotted; P value determined by 2-tailed unpaired t test. Cohort sizes are as follows: TT = 32, TC = 99, and CC = 139.

Figure 4. Association of rs1805126 genotype with cardiac SCN5A expression.

(A and B) Western blot (WB) using antibodies against both N- and C-terminal ends of SCN5A and accompanying densitometry analysis shows that Na_V1.5 protein expression is decreased in nonfailing CC homozygous human cardiac tissue samples (n = 7), relative to TT samples (n = 6). β-Catenin serves as loading control. **P < 0.01 by 2-tailed unpaired t test. (C) Reanalysis of available genome-wide human cardiac eQTL data from nonfailing hearts reveals that CC samples express less SCN5A mRNA, compared with TT samples; cohorts from the University of Pennsylvania (left) and The Netherlands (right). Normalized microarray probe intensity (log2) is plotted; further analysis indicates approximately 10%–20% lower SCN5A mRNA in CC versus TT. Sample numbers are denoted in the figure. *P ≤ 0.05 by 2-tailed unpaired t test. (D) Allele-specific expression analysis in heterozygous individuals was done using RNA-seq data obtained from nonfailing and failing human hearts (n = 29), offering further support that the C allele is expressed at lower levels, relative to T. ***P < 0.001 by 1-sample t test.

Figure 5. Evaluation of oxidative stress in Scn5a heterozygous knockout mouse hearts.

Oxidation of dihydroethidium (DHE, a surrogate marker of steady-state levels of superoxide) was measured in fresh-frozen cardiac tissue sections collected from 7- to 8-month-old male Scn5a^+/– mice and wild-type littermates (n = 3 each). Positive controls included sections treated with antimycin A (AntA, a mitochondrial electron transport chain blocker that is known increase ROS generation). Images were captured at ×40 magnification. Scale bars: 100 μm. (A) Representative photomicrographs of DHE-stained cardiac sections are shown. (B) Signal intensity quantified in 90 cells per group (i.e., 30 per mouse) is plotted (mean ± SEM). ****P < 0.0001 by 2-tailed t test.