A Rare Noncoding Enhancer Variant in SCN5A Contributes to the High Prevalence of Brugada Syndrome in Thailand

Abstract

BACKGROUND:

Brugada syndrome (BrS) is a cardiac arrhythmia disorder that causes sudden death in young adults. Rare genetic variants in the SCN5A gene encoding the Na_v1.5 sodium channel and common noncoding variants at this locus are robustly associated with the condition. BrS is particularly prevalent in Southeast Asia but the underlying ancestry-specific factors remain largely unknown.

METHODS:

Genome sequencing of BrS probands and population-matched controls from Thailand was performed to identify rare noncoding variants at the SCN5A-SCN10A locus that were enriched in patients with BrS. A likely causal variant was prioritized by computational methods and introduced into human induced pluripotent stem cell (hiPSC) lines using CRISPR-Cas9. The effect of the variant on SCN5A expression and Na_v1.5 sodium channel current was then assessed in hiPSC-derived cardiomyocytes (hiPSC-CMs).

RESULTS:

A rare noncoding variant in an SCN5A intronic enhancer region was highly enriched in patients with BrS (detected in 3.9% of cases with a case-control odds ratio of 45.2). The variant affects a nucleotide conserved across all mammalian species and predicted to disrupt a Mef2 transcription factor binding site. Heterozygous introduction of the enhancer variant in hiPSC-CMs caused significantly reduced SCN5A expression from the variant-containing allele and a 30% reduction in Na_v1.5-mediated sodium current density compared with isogenic controls, confirming its pathogenicity. Patients with the variant had severe phenotypes, with 89% experiencing cardiac arrest.

CONCLUSIONS:

This is the first example of a functionally validated rare noncoding variant at the SCN5A locus and highlights how genome sequencing in understudied populations can identify novel disease mechanisms. The variant partly explains the increased prevalence of BrS in this region and enables the identification of at-risk variant carriers to reduce the burden of sudden cardiac death in Thailand.

Clinical Perspective.

What Is New?

• A rare genetic variant located in an SCN5A noncoding enhancer (RE5) was found to be highly enriched in patients with Brugada syndrome (BrS) from Thailand compared with population-matched controls.

• The variant disrupts a highly conserved nucleotide in a Mef2 transcription factor binding site and caused allele-specific reduction of SCN5A expression and a 30% decrease in sodium channel density in CRISPR-Cas9 edited human induced pluripotent stem cell–derived cardiomyocytes compared with isogenic controls.

• The RE5 variant is found in 4% of patients with BrS from Thailand and helps explain the high prevalence of the condition in Southeast Asia.

What Are the Clinical Implications?

• Although BrS is particularly prevalent in Southeast Asia, the yield of genetic testing in this population is much lower than in other regions, hindering the identification of at-risk individuals.

• By incorporating this variant into clinical genetic testing, the primary causal genetic factor will be detected in more affected families, enabling the diagnosis and clinical management of relatives that could prevent sudden cardiac deaths.

• This study demonstrates how rare noncoding variants, hitherto largely unexplored in cardiac genetic disease, can be discovered and validated and may facilitate the identification of additional variants at the SCN5A locus underlying the increased prevalence of BrS in East Asia.

Brugada syndrome (BrS) is a cardiac arrhythmia disorder associated with increased risk of sudden death in young adults and characterized by ST-segment elevation in the right precordial leads of the ECG, occurring either spontaneously or after challenge with sodium channel blockers (eg, ajmaline).¹ Rare loss-of-function coding variants in SCN5A, encoding the Na_v1.5 sodium channel underlying the cardiac sodium current (I_Na), are found in ≈20% of individuals of European ancestry with BrS.² Several other genes have been implicated in BrS but were considered to have limited evidence of association by ClinGen reevaluation, and SCN5A remains the only clinically actionable gene.^3,4 Genome-wide association studies (GWAS) have demonstrated a strong contribution of common genetic variation to BrS susceptibility (single nucleotide polymorphism [SNP]–based heritability, h2_SNP, between 0.17 and 0.34), indicating the genetic architecture is more complex than a simple monogenic disease model.⁵

Eight of the 21 independent genome-wide significance signals detected by the BrS GWAS were located at the SCN5A–SCN10A locus, further cementing the central role for this genomic region in BrS genetic pathogenesis.⁵ All lead variants are noncoding and located in or near regulatory elements (REs) at the SCN5A–SCN10A locus that have been characterized in detail (Figure 1A), including the SCN5A promoter, a downstream superenhancer encompassing RE6 through RE9 and 2 intragenic SCN5A enhancer elements (RE4/RE5).^6,7 A short transcript of SCN10A, comprising the last 7 exons and controlled by an intronic enhancer (RE1–RE2)–promoter complex, is expressed in the heart.⁸ A common variant in RE1 associated with BrS and ECG traits, rs6801957, leads to reduced expression of SCN10A-short and decreased cardiac I_Na while minimally affecting SCN5A expression.⁸

Figure 1.

Identification of a noncoding enhancer variant in SCN5A associated with BrS. A, Overview of the coding sequence and regulatory elements at the SCN10A–SCN5A locus, highlighting single nucleotide polymorphisms (SNPs) from Brugada syndrome (BrS) genome-wide association study (GWAS; lead SNP and SNPs with r²>0.5 in European populations), the EMERGE track of cardiac-specific epigenetic markers, the location of the 4 rare and low-frequency noncoding variants associated with BrS in Thai patients, and odds ratios (ORs) and P values for each haplotype. B, Principal component analysis for patients with BrS and controls from Thailand. C, BrS case and control frequencies for SCN5A coding variants: ultrarare variants (gnomAD filtering allele frequency <0.00001), the low-frequency variant p.Arg965Cys, and other low-frequency variants (gnomAD filtering allele frequency <0.001). D, Sequence alignment of 6 mammalian species, including human and mouse, of the hs2177 enhancer core that encompasses the GRCh38:3-38580380-A-C variant (indicated by box). The variant is located in and disrupts a predicted and conserved Mef2 motif site (indicated by blue shading) and is flanked by T-box (red shading) and Gata (green shading) recognition sites as predicted by Homer. E, UCSC browser views of the hs2177-SCN5A RE5 enhancer in human (hg38) and mouse (mm10). Relevant cardiac transcription factor chromatin immunoprecipitation sequencing and assay for transposase-accessible chromatin with sequencing traces from the literature were plotted (see Table S1 for data set references). ACM indicates atrial cardiomyocytes; CM, cardiomyocytes; and VCM, ventricular cardiomyocytes.

BrS is several times more prevalent in East (especially Southeast) Asia compared with regions of predominantly European ancestry. After its initial description in 1992,⁹ BrS was recognized as the major cause of sudden unexplained death syndrome in Thailand, a condition causing sudden nocturnal death in young men and traditionally known locally as Lai Tai.^10,11 The high rate of BrS in Southeast Asia is not attributable to an increased prevalence of ultrarare loss-of-function SCN5A coding variants; indeed, a relatively lower diagnostic yield is observed compared with European ancestry cases (≈5% versus 20%).^2,12 We previously demonstrated enrichment of a low-frequency SCN5A coding variant of intermediate effect size, p.Arg965Cys, in Thai patients with BrS compared with population-matched controls.¹² However, other genetic factors underlying the higher prevalence of BrS in Southeast Asia are still largely unknown. Here, we identify and functionally validate a novel rare noncoding variant in an SCN5A enhancer that is highly enriched in patients with BrS from Thailand and partly explains the increased prevalence of BrS in this region.

METHODS

Data Availability

The RNA sequencing data set is available in the Gene Expression Omnibus database (accession record GSE264359). All other supporting data are available within the article and in the Supplemental Material.

BrS and Control Cohorts

The patients with BrS were recruited from 7 participating institutions in Thailand. All patients were diagnosed with BrS according to the 2013 Heart Rhythm Society/European Heart Rhythm Association/Asia Pacific Heart Rhythm Society Expert consensus statement.¹³ Blood samples were collected for DNA extraction and genotyping. The control samples were collected from volunteers who had no Brugada pattern (type 1) on standard 12-lead ECG. All patients and controls were of Thai ethnic origin by self-report. The study was approved by the institutional review board of the faculty of medicine, Chulalongkorn University, Bangkok, Thailand (institutional review board no. 431/58). Informed consent was obtained from all participants. All methods were performed in accordance with relevant guidelines or regulations.

Genome Sequencing and Analysis

Data Generation and Processing

The samples were sequenced by the Illumina FastTrack service on HiSeqX with a PCR-free library. Data processing, alignment, and variant calling were performed using standardized approaches as described in the Methods in the Supplemental Material. Samples were excluded on the basis of a missing genotype rate of >0.03, relatedness (PI_HAT>0.2; 1 of related sample group retained with priority for cases), high inbreeding rate (F>0.1), or sex check inconsistencies. Sample ancestry was determined using principal component analysis (PCA) with samples from the 1000 Genomes (1KG) data set. All samples clustered with East Asian 1KG samples. A PCA of cases and controls was plotted to demonstrate ancestry matching.

Genetic Variant Analysis

All SCN5A coding variants were described with respect to the ENST00000333535 transcript. Coding variants were classified using the guidelines of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology¹⁴ as previously described.² Noncoding variants at the SCN5A–SCN10A locus were assessed for enrichment in cases versus controls using Fisher exact test (for samples with or without variants). The following regulatory regions at the locus were queried for variants (all coordinates based on GRCh38): RE1/SCN10A enhancer (chr3:38725264-38728203), RE2/SCN10A enhancer (chr3:38721868-38725264), SCN10A-short promoter (chr3:38716810-38719605), SCN5A promoter (chr3:38642253-38652554), RE4/SCN5A enhancer (chr3:38613169-38618856), RE5/SCN5A enhancer (chr3:38575835-38584663), and RE6-9/SCN5A superenhancer (chr3:38526650-38546621), as well as the intermediate sequence between these regions (Figure 1A).⁶ To assess for linkage disequilibrium between variants, r² values were calculated for each variant pair in this region (with a combined minimum case and control allele count of 3). Variants of interest were further annotated with CADD and FINSURF scores,¹⁵ frequency in gnomADv3 East Asian genome sequences (n=2604), and conservation of affected base across species using Ensembl version 109. The frequency of the putative causal variant was also assessed in other genomic databases with individuals of Asian ancestry: the Genomics Thailand database of 10 043 individuals of Thai ancestry; the Singaporean SG10K_Health v5.3 database of individuals of Chinese, Indian, or Malay ancestry¹⁶; the jMORP/54KJPN database of individuals of Japanese ancestry¹⁷; and the KOVA database of individuals of Korean ancestry.¹⁸

Transcription Factor Motif Analysis

Transcription factor (TF) motif scanning analysis was performed using Homer¹⁹ to detect the presence of 1847 motifs in the Homer¹⁹ and Jaspar²⁰ databases for the wild-type and variant sequences in the enhancer region of interest. The enhancer region was further characterized using human and mouse chromatin immunoprecipitation sequencing data sets (Table S1).

Generation of Human Induced Pluripotent Stem Cell–Derived Cardiomyocytes

hiPSC Culture, Genome Editing, and Characterization of Clones

The Personal Genome Project 1 human induced pluripotent stem cell (hiPSC) line, PGP1-SV1, referred to as PGP1 in this article (detailed information is available at https://arep.med.harvard.edu/gmc/PGP_cells.html), as well as the derived edited clones, were maintained on growth factor–reduced Matrigel Matrix (Corning)–coated plates in presence of mTeSR1 medium (Stemcell Technologies). Cells were passaged every 3 to 4 days using 0.5 mM EDTA (Life Technologies) in PBS without CaCl₂ and MgCl₂ and replated in presence of Rho kinase inhibitor (Fasudil HCl; Selleck Chemicals) for the first 24 hours. Cells were maintained at 37 °C, with 5% CO₂ and 20% O₂. The RE5 GRCh38:chr3-38580380-A-C variant was introduced in heterozygous state in the PGP1 hiPSC line by CRISPR/Cas9-mediated editing (Synthego). The Cas9 and gRNA, as ribonucleoprotein, and the single strand oligodeoxynucleotide carrying the single nucleotide variant for homology-directed repair (sequences in Table S2) were introduced by nucleofection. Two heterozygous, edited clones (A7 and C8) were selected after Sanger sequencing of their genomic DNA (primers in Table S2 and Figure S2A). Quality controls were performed on these edited clones, which confirmed karyotype integrity by KaryoStat Assay (Figure S2B) and pluripotency by PluriTest Assay (Figure S2C; Synthego). One predicted off-target genomic locus, exhibiting 2 mismatches with the guideRNA spacer sequence, was identified by the Wellcome Sanger Institute Genome Editing off-target finding tool (https://wge.stemcell.sanger.ac.uk/find_off_targets_by_seq) at position GRCh38:chr8-15043569-15043591. An ≈200-bp window of genomic DNA from the A7 and C8 clones, centered around this predicted potential off-target site, was sequenced by Sanger sequencing to detect any sequence alterations (primer sequences in Table S2). No sequence alterations were detected in in the 2 edited clones (A7 and C8).

hiPSC–Derived Cardiomyocytes Generation

The parental PGP1 and edited A7 and C8 hiPSC lines were maintained in undifferentiated state in presence of mTeRS1 medium on Matrigel Matrix-coated plates until they reached 65% to 80% confluence. Differentiation into cardiomyocytes was then performed as described by Maas et al²¹ without cardiomyocytes expansion and with modifications described in the Methods in the Supplemental Material. The efficacy of the cardiomyocyte differentiation, and the consistency between the wild-type and edited lines, was demonstrated through RNA sequencing data for cardiomyocyte differentiation and maturation-relevant genes, as well as the expression ratio of the “fetal” and “adult” SCN5A exon 6 (Figure S3).

RNA Sequencing

RNA Isolation

Total RNA was isolated from hiPSC–derived cardiomyocytes (hiPSC-CMs) derived from 4 independent batches of differentiation of the parental PGP1 and heterozygous RE5 variant–carrying A7 hiPSC isogenic lines. Cardiomyocyte cultures were lysed by TRIzol Reagent (Ambion/Life Technologies) followed by chloroform addition (200 μL/mL of TRIzol Reagent) and hydrophilic phase isolation after centrifugation at 4 °C. RNA was then precipitated in presence of isopropanol, pelleted by centrifugation at 4 °C, and washed twice with 75% ethanol. The RNA pellet was then briefly air-dried and resuspended in RNase-free water.

Determination of Phase Between the RE5 Variant and an Exonic SNP

To determine the SCN5A transcript output from each allele of the A7 and PGP1 lines using bulk messenger RNA sequencing analysis, we took advantage of the presence of a heterozygous SNP (GRCh38:3-38580976-T-C; rs7430407) in SCN5A exon 17. The phase between the SNP and the RE5 variant in the A7 line was determined by amplifying an 802 base pair region encompassing the 596-bp interval between the rs7430407 SNP and the RE5 variant (primers in Table S2), followed by cloning and Sanger sequencing. This demonstrated that the C-allele of the RE5 variant was in cis with the C-allele of rs7430407 (Figure 2B).

Figure 2.

Experimental characterization of the RE5 variant. A, Overview of experimental validation: generation of human induced pluripotent stem cell–derived cardiomyocytes (hiPSC-CMs) from PGP1 cell line, isogenic control (red), and RE5 variant line (blue), where the variant was introduced in the heterozygous state by CRISPR-Cas9 editing (graphic produced with BioRender). B, Allelic balance of SCN5A expression in the 2 lines was assessed using the heterozygous synonymous coding variant c.3183A>G present in the PGP1 line (596 bases upstream of the RE5 variant in exon 17), with the C exonic allele demonstrated to be in phase with the RE5 variant. C, Allelic balance of SCN5A expression in the control and RE5 variant lines as determined by RNA sequencing (4 biological replicates each), where P value refers to the Z test differences in allelic expression ratios. D, Sodium current (I_Na) characterization in control hiPSC-CMs (PGP1; 26 cells from 6 differentiations) and hiPSC-CMs of the 2 RE5 variant lines (A7, 21 cells from 6 differentiations; and C8, 21 cells from 5 differentiations), showing representative traces of I_Na activated during the first depolarizing pulses of the voltage clamp protocol shown in the inset. E, Average current–voltage (I-V) relationships of peak I_Na (left) and boxplots depicting I_Na densities (median and boxes represent interquartile range), determined at −20 mV (right). *Significance compared with controls (P<0.05; Kruskal-Wallis test, followed by Dunn comparisons).

Bulk RNA Sequencing and Analysis

Quality and integrity of the isolated RNA was determined by 4200 TapeStation (Agilent) analysis, which indicated a RIN score >9 for all samples. Cardiomyocytes from the parental PGP1 and isogenic A7 hiPSC lines were used for bulk RNA sequencing. Library preparations were made after ribosomal RNA depletion by KAPA Total RNA HyperPrep with RiboErase kit (Roche). Sequencing was performed on an Illumina platform NovaSeq S4.300 in pair-end mode with read length of 150 bp (PE150) and a sequencing depth of 60 million reads per sample. We used STAR²² to map reads to the hg19 build of the human genome/transcriptome. We used VarScan 2²³ to detect variants and their allele-specific occurrence within the samples. We used a minimum base quality Phred score of 15, a minimum read depth of 10, and a minimum variant allele frequency threshold of 0.20 in VarScan 2 for variant detection. To determine whether a variant was specifically enriched or depleted in either cell line, indicative for allele-specific enhancer activity, a Z test was performed on the observed allele frequencies (P<0.05; Z test).²⁴

Electrophysiology Patch Clamp Analysis

Single-Cell Electrophysiology Data Acquisition

The hiPSC-CM cultures were dissociated to obtain single cardiomyocytes as described in the Methods in the Supplemental Material. I_Na was recorded from hiPSC-CMs using an Axopatch 200B amplifier (Molecular Devices). Voltage control, data acquisition, and analysis were realized with custom software.²⁵ Signals were low-pass–filtered with a cutoff of 5 kHz and digitized at 20 kHz. Cell membrane capacitance (C_m) was calculated by dividing the time constant of the decay of the capacitive transient after a −5 mV voltage step from −40 mV by the series resistance. Series resistance was compensated for by at least 80%. Following the procedure of Veerman et al,²⁶ we selected single, spontaneously beating hiPSC-CMs showing regular, synchronous contractions in modified Tyrode solution, after which this extracellular solution was switched to a solution suitable for specific I_Na measurements. Modified Tyrode solution contained the following (in mmol/L): NaCl 140, KCl 5.4, CaCl₂ 1.8, MgCl₂ 1.0, glucose 5.5, HEPES 5.0, with pH 7.4 (NaOH).

Sodium Current Measurements

I_Na was recorded at room temperature (≈20 °C) using the ruptured patch-clamp technique. Extracellular solution contained the following (in mmol/L): NaCl 20, CsCl 120, CaCl₂ 1.8, MgCl₂ 1.0, glucose 5.5, HEPES 5.0, with pH 7.4 (CsOH). Nifedipine (10 µmol/L; Sigma) was added to block the L-type Ca²⁺ current in hiPSC-CMs.²⁷ Patch pipettes were pulled from borosilicate glass (GC100F-10; Harvard Apparatus) and had resistances of ≈2.5 MΩ after filling with the solution consisting of (in mmol/L) NaCl 3.0, CsCl 133, MgCl₂ 2.0, Na₂ATP 2.0, TEACl 2.0, EGTA 10, and HEPES 5.0, with pH 7.2 (CsOH). I_Na was measured using a double pulse protocol from a holding potential of −120 mV (Figure 2D, inset; cycle length of 5 seconds). The first pulse was used to determine the I–V relationship and voltage dependency of activation; the second pulse was used to analyze the voltage dependency of inactivation. I_Na was defined as the difference between peak and steady-state current and current densities were calculated by dividing current amplitude by C_m. Voltage dependence of activation and inactivation curves were fitted with Boltzmann function (I/I_max=A/{1.0+exp[(V_1/2−V)/k]}), where V_1/2 is the half-maximal voltage of (in)activation and k the slope factor (in mV). The time course of current inactivation was fitted by a double-exponential equation: I/I_max=A_f ×exp(−t/τ_f)+A_s×exp(−t/τ_s), where A_f and A_s are the fractions of the fast and slow inactivation components, and τ_f and τ_s are the time constants of the fast and slow inactivating components, respectively.

Statistical Analysis

Clinical Characteristics Statistical Analysis

Clinical characteristics including demographic data, history of cardiac events, and ECG measures were compared among the 4 genotype groups (RE5 variant, ultrarare SCN5A coding variants, SCN5A:p.Arg965Cys, and no low-frequency SCN5A coding variants). Statistical analyses were performed using χ² or Fisher exact tests for categorical variables and ANOVA for continuous variables.

Electrophysiology Patch Clamp Statistical Analysis

Data are expressed as mean±SEM, unless stated otherwise. Statistical analysis was carried out with SigmaStat 3.5 software (Systat Software, Inc.). Normality and equal variance assumptions were tested with the Kolmogorov-Smirnov and the Levene median test, respectively. Significance between measures was tested using 1-way repeated-measures ANOVA followed by pairwise comparison using the Student-Newman-Keuls test or by the Kruskal-Wallis test followed by pairwise comparisons with Dunn method. P<0.05 defines statistical significance.

RESULTS

Identification of Rare Noncoding Variants at SCN5A–SCN10A Locus

Genome sequencing data were generated for BrS probands and controls from Thailand (see Methods for details of data generation, read mapping, variant calling, and quality control checks). After sample-level quality control, a total of 231 unrelated probands and 500 ancestry-matched controls were included, with principal component analysis demonstrating the cohorts were of comparable ancestral background (Figure 1B). Of 231 BrS probands (mean age, 47.6±12.8 years; 95% male), 88% had a documented spontaneous BrS ECG pattern and 76% were symptomatic (Table 1). A total of 5.6% of cases had an ultrarare (gnomAD filtering allele frequency <0.00001) SCN5A coding variant (versus 1.0% in controls; P=4.0E-04), 6.5% carried the SCN5A:p.Arg965Cys low-frequency variant (versus 0.4% in controls; P=1.6E-06), and 3.9% had other low-frequency (gnomAD filtering allele frequency <0.001) SCN5A coding variants, although the latter were only marginally enriched over controls (3.0%; P=0.51; Figure 1C; Table S3).

Table 1.

Clinical Characteristics of Participants Grouped According to SCN5A Genotype

Rare and low-frequency noncoding variants at the SCN5A-SCN10A locus encompassing the defined promoter and enhancer regions were assessed for enrichment in BrS probands versus controls. Four variants in linkage disequilibrium (r²>0.2) were identified (Table 2): chr3-38580380-A-C in the SCN5A enhancer RE5, chr3-38719550-C-T in the SCN10A-short transcript promoter, chr3-38724980-T-C in the SCN10A-short enhancer RE2, and chr3-38683338-C-T in the intergenic region between SCN5A and SCN10A (Figure 1A; GRCh38 coordinates).

Table 2.

Details of the Noncoding Variants at the SCN5A-SCN10A Locus Associated With BrS in Patients and Controls From Thailand

For these 4 variants, the largest case-control odds ratios (ORs) were observed for the RE5 and SCN10A-short promoter variants (in complete linkage disequilibrium in this data set). Haplotypes containing only one or both of the RE2 and intergenic variants had no association with BrS, confirming they are merely tagging the causal variant (Figure 1A). The RE5 variant alters a nucleotide fully conserved across 35 mammalian species with available orthologous sequence data, in contrast to the other 3 variants (Figure S1), and has higher computational deleteriousness scores from the CADD and FINSURF¹⁵ algorithms (Table 2). The RE5 variant is also absent in the gnomADv3 database (n=76 165 genomes, including n=2604 of East Asian ancestry), whereas the SCN10A-short promoter variant has a gnomAD–East Asian frequency of 0.001, indicating it also exists on haplotypes distinct from the RE5 variant. Based on this cumulative evidence, the RE5 variant was deemed to be the putative causal variant for this risk haplotype.

The RE5 variant was detected in 3.9% (n=9) of cases versus 0.2% (n=1) of controls (P=2.0E-04; OR, 20.2 [2.5–160.6]). We searched the Genomics Thailand database, comprising 10 043 individuals with a range of clinical and pharmacogenetics indications (https://genomicsthailand.com/Genomic/home), to attain a more precise estimate of the variant frequency in Thailand. It was detected in 6 individuals (0.06%), yielding an OR of 45.2 (14.5–141.3) compared with the BrS cohort (P=5.6E-12). In addition to gnomADv3–East Asian, the RE5 variant is also not detected in other genomic data sets of Asian ancestry (comprising 87 867 individuals; Table S4), suggesting it is restricted to Thai or Southeast Asian populations.

Clinical Characterization of RE5 Variant Carriers

None of the 9 patients with BrS with the RE5 variant had rare or low-frequency coding variants in SCN5A. When grouping cases by genotype (ie, RE5 variant, ultrarare SCN5A coding variants, SCN5A:p.Arg965Cys, no low-frequency [gnomAD filtering allele frequency <0.001] SCN5A coding variants), no statistically significant differences in clinical characteristics were observed (Table 1), although a significant leftward shift of ECG QRS axis was observed in cases with the RE5 variant (7.88±52° versus 46.4±51°; P<0.05), with 2 cases displaying left QRS axis deviation (−30° to −150°).

Patients with the RE5 variant were particularly symptomatic: 89% (8/9) had a history of sudden cardiac arrest, 7 of whom had documented ventricular tachycardia or ventricular fibrillation, compared with only 58% of cases without any SCN5A variant (Table S5). For one patient, analysis of their family pedigree revealed the variant largely segregated with phenotype, although one genotype-negative individual displayed a drug-induced BrS ECG pattern (Figure 3A). However, this should be interpreted in the context of an appreciable false-positive rate for this test,²⁸ and the fact that it is not atypical in BrS families for individuals lacking the familial pathogenic SCN5A variant to be phenotype-positive due to the complex genetic pathogenesis of the disease.²⁹

Figure 3.

Clinical aspects of carriers of the RE5 variant. A, Family pedigree of one of the patients with Brugada syndrome (BrS) carrying the RE5 variant with ages indicated (the proband is II-7, as indicated). Three family members with the variant tested positive for BrS after ajmaline challenge (II-1, III-1, III-3; black fill). Two other family members with the variant did not consent to ajmaline challenge: II-4 displayed a borderline baseline Brugada type I pattern and III-6 displayed a baseline Brugada type III pattern. One individual also tested positive for BrS after ajmaline challenge despite not carrying the RE5 variant (III-5), which is not atypical for SCN5A pathogenic variants in BrS families and reflects the complex genetic pathogenesis of the disease. No other noncarriers display the Brugada marker on ECG. B, Epicardial substrate ablation areas (after ajmaline challenge) for patients with BrS carrying the RE5 variant (n=3; substrate sizes 40.0, 29.2, and 17.2 cm²; mean±SD, 28.7±11.5 cm²) compared with data from Ciconte et al³¹ for carriers of pathogenic coding variants in SCN5A (SCN5A-positive; n=49; mean±SD, 18.8±5.7 cm²) and patients with BrS without SCN5A variants (SCN5A-negative; n=146; mean±SD, 11.9±4.8 cm²; 1-way ANOVA, P<0.0001; Tukey honestly significant difference post hoc for RE5 vs SCN5A-negative, P<0.0001).³¹ C, Baseline (above) and postajmaline (below) epicardial mapping for the proband of the pedigree in A, with the color bar representing local activation time from early (red) to late (blue) activation (scale in milliseconds). The latest activation was localized at the right ventricular outflow tract (RVOT), the target for radiofrequency ablation, before and after ajmaline administration. However, the latest activation at the RVOT was markedly prolonged after ajmaline, increasing from 130 ms before to 200 ms after. AP indicates anterior-posterior; INF, inferior; LL, left lateral; LV, left ventricle; and RV, right ventricle.

Three patients with the RE5 variant underwent epicardial substrate ablation using the sodium channel blocker ajmaline to define the substrate areas of the right ventricular outflow tract and right ventricular inferior epicardium.³⁰ These procedures revealed large substrate areas more similar to SCN5A-positive than SCN5A-negative cases,³¹ suggesting a substantial loss-of-function effect of the RE5 variant (Figure 3B and 3C).

Functional Validation of the RE5 Variant

To assess the potential consequences of the RE5 variant, we performed a transcription factor (TF) motif scanning analysis using Homer¹⁹ on a 1286 base pair stretch of the hs2177 enhancer sequence³² (which is fully encompassed by the RE5 region and includes the variant site), scanning for the presence of 1847 motifs in the Homer¹⁹ and Jaspar²⁰ databases. Nine variant-dependent motif recognition sites were detected: 8 exclusive to the wild-type sequence and 1 to the RE5 variant sequence (Table S6). Of the 8 sites lost by the RE5 variant, 6 are different representations of the Mef2 motif (CTAWWWWTAG; −5 to +5). The variant substitutes the A nucleotide at position +4 of the motif, previously shown to cause a strong reduction in binding affinity for Mef2.³³ This Mef2 binding sequence is conserved in evolution and its functionality is supported by Mef2 chromatin immunoprecipitation sequencing data from mouse hearts³⁴ (Figure 1D and 1E; Figure S4A). The motif gained by the RE5 variant represents a binding site for the Gfi1B transcriptional repressor.³⁵

The effect of the RE5 variant on expression, and specifically through loss of Mef2 binding, was assessed using luciferase assays in hiPSC-CMs (which endogenously express cardiac-specific TFs including Mef2) and HEK-293 cells (using cotransfection with Mef2C and Myocardin, a Mef2 coactivator; see the Methods in the Supplemental Material). The wild-type RE5 fragment drove robust luciferase activity in both the hiPSC-CMs and HEK-293 cells, whereas the RE5 variant allele showed significantly lower reporter gene activity (Figure S4B andS4C).

To evaluate the functional consequences of the presence of the RE5 variant in the context of a human cardiomyocyte, we introduced the heterozygous variant into a hiPSC line. The PGP1 hiPSC line (details in Methods) was edited by the CRISPR/Cas9-mediated homology-directed repair method and 2 isogenic clones (A7 and C8) carrying the single nucleotide variant in heterozygous form were isolated and characterized (Figure S2). hiPSC-CMs from the parental and heterozygous mutant lines were generated by adaptation of a protocol based on small molecules–mediated modulation of the cWnt pathway²¹ (details in Methods).

We evaluated the potential cis-regulatory effect of the RE5 variant on SCN5A allelic expression through RNA sequencing analysis of hiPSC-CMs from the A7 variant line and the isogenic control (PGP1). To distinguish transcripts originating from the 2 alleles, we took advantage of the presence of a heterozygous SNP located in an exon adjacent to the RE5 intronic variant (rs7430407, GRCh38:3-38580976-T-C, located 596 bases from the RE5 variant) and determined their phase in the A7 line (details in Methods; Figure 2B). The allele-specific transcript output ratios in the hiPSC-CMs of the 2 isogenic lines (PGP1 and A7) were quantified and compared, with A7 hiPSC-CMs containing the RE5 variant displaying a 34% reduction of the SCN5A transcript allelic ratio compared with PGP1 controls (Z test, P=0.0004; Figure 2C; Table S7). This method of comparing the allelic balance between the wild-type and variant alleles, within each cell line and RNA sequencing sample, controls for the expected variability between samples to more accurately gauge the effect of the variant on SCN5A expression. However, the overall normalized SCN5A expression also shows a 28% reduction in the RE5 variant line. In mice, the 2 Scn5a alleles are independently regulated.⁶ These data further indicate allele-specific reduced RE5 enhancer activity and SCN5A expression due to loss of the Mef2 motif as a consequence of the variant.

I_Na measurements were performed in control hiPSC-CMs (PGP1) and in the hiPSC-CMs of the 2 isogenic lines (A7 and C8) with the RE5 variant. Figure 2D shows typical I_Na recordings (−80 to 0 mV) from all groups and Figure 2E (left) shows the average current–voltage (I–V) relationships of I_Na. The I_Na density was significantly lower in hiPSC-CMs of both RE5 variant lines, with a 30% reduction at −20 mV (Figure 2E, right). Neither the voltage dependency of activation (Figure S5A) nor the voltage dependency of inactivation (Figure S5B) were affected by the RE5 variant. The time constants of current inactivation, determined at a test potential of −20 mV, were not different in hiPSC-CMs of the 2 RE5 variant lines compared with the control (data not shown). Action potential measures were not assessed, but because the action potential upstroke velocity in cardiomyocytes (see Berecki et al³⁶ and primary references cited therein), including hiPSC-CMs (Figure S6), is strongly regulated by I_Na density (or availability), a 30% reduction in I_Na density would result in a 24% reduced upstroke velocity. In summary, the RE5 variant results in a decrease in I_Na density without changes in gating properties, concordant with the reduced SCN5A expression from the allele harboring the RE5 variant.

DISCUSSION

In this study, we have identified and validated a rare noncoding SCN5A enhancer variant associated with BrS in Thailand, one of the first examples of a rare regulatory variant of proven pathogenicity in a cardiac disease gene. The variant is highly enriched in patients with BrS, with an OR between 20 and 45 based on our study controls and the Genomics Thailand database, respectively, and is located within a characterized enhancer region for SCN5A (RE5). Computational evidence was highly supportive of pathogenicity, based on deleteriousness scores of the CADD and FINSURF algorithms, full conservation of the affected nucleotide base, and the predicted disruption of a Mef2 TF binding site. The causality of the variant was subsequently confirmed by the demonstration of a significant reduction in both allele-specific SCN5A expression and Na_v1.5 sodium channel function in CRISPR-Cas9–edited hiPSC-CMs compared with isogenic controls. These effects are consistent with the loss-of-function mechanism of action of known pathogenic BrS coding variants in SCN5A (either missense or protein-truncating variants). The frequency of the variant in cases, its apparent specificity to the Thai population, alongside its demonstrated functional effect, suggest that it is an important contributor to the high prevalence of the disorder in Thailand.

The diagnostic yield for clinical genetic testing in patients with BrS from Thailand (ie, ultrarare loss-of-function coding variants in SCN5A) is several fold lower than in patients with BrS of European ancestry (5% versus 20%). This likely reflects the increased prevalence of BrS in this region (ie, whereas the relative rate is lower, the absolute rate of such variants is expected to be broadly similar across populations in the absence of specific founder variants; Figure 4). We previously described a 9% yield in Japanese patients with BrS, which reflects a disease prevalence intermediate between Europe and Southeast Asia.² Because the RE5 variant is found in ≈1 in every 25 patients with BrS from Thailand, it may partly explain the increased prevalence of this condition in this region, in conjunction with the SCN5A:p.Arg965Cys intermediate effect variant we previously described (present in 6.5% of cases; Figure 4). Indeed, the RE5 variant accounts for almost as many Thai patients with BrS as all of the ultrarare SCN5A coding variants combined and a relatively equivalent proportion of patients with BrS of European ancestry with causal SCN5A coding variants.

Figure 4.

Overview of BrS genetic architecture in Thailand and European-ancestry populations. Top, Graphical representation of the genetic contribution of rare and low-frequency variants to Brugada syndrome (BrS) in Thailand (left) and European ancestry populations (right), using a likely conservative estimate of a 4-fold higher prevalence of BrS in Thailand.^37,38 The absolute frequency of ultrarare loss-of-function coding variants in SCN5A is likely to be broadly similar in both ancestries (represented by the dark blue squares), but the diagnostic yield of such variants in BrS cohorts from Thailand is relatively lower (≈5% vs ≈20%) because of the different disease prevalence between ancestries (highlighted by the overall square sizes). A proportion of the additional susceptibility in Thai patients with BrS is explained by the low-frequency SCN5A coding variant p.Arg965Cys (≈6.5% of cases) and the noncoding RE5 enhancer variant described in this study (≈4% of cases). Bottom, Population frequencies of SCN5A p.Arg965Cys and RE5 variants in South and East Asian countries (see Table S8 for details of data sets used and exact variant frequencies; gray = data not available).

The strong observed effect size of this variant (based on both case–control OR and functional analysis) is reflected in the severe phenotype observed in BrS carriers and may be explained by the predicted concurrent loss of a Mef2 motif (a critical family of cardiac TFs) and possible gain of a Gfi1 transcriptional repressor, although further studies will be needed to probe the exact transcriptional mechanisms underlying SCN5A downregulation. The 30% reduction in I_Na caused by the RE5 variant is a remarkably strong effect for a single nucleotide variant in a noncoding regulatory region, given the complex regulatory machinery at the SCN5A gene, which encompasses numerous promoter and enhancer regions. By comparison, direct disruption of the SCN5A transcript by heterozygous nonsense or frameshift coding variants has been shown in numerous studies to lead to a maximum of 40% to 50% reduction in I_Na (using coexpression of wild-type and mutated proteins in electrophysiology assays to reflect the heterozygosity observed in patients): p.Glu473Ter,³⁹ p.Arg535Ter,⁴⁰ p.Trp822Ter,^41,42 p.Trp1191Ter,⁴³ and p.Leu1393Ter.⁴⁴ SCN5A variants affecting gene dosage (whether coding or noncoding) often have less deleterious effects than pathogenic SCN5A missense variants, which can exhibit dominant negative effects.⁴²

The effect of rare regulatory variants at the SCN5A locus in individual patients is likely to be modified by the presence of risk or protective common variant haplotypes that have been identified in BrS GWAS. The complex genetic architecture of BrS at this and other genomic loci will at least partially explain the variable phenotypes observed in carriers of the RE5 variant, ranging from the highly symptomatic probands to the milder (or nonsymptomatic) phenotypes in relatives and the general Thai population database. The cumulative effect of these rare and common regulatory variants will ultimately determine the overall dosage of SCN5A across both alleles and, in conjunction with polygenic risk at other genomic loci, act to influence disease penetrance in individuals. Future precision medicine approaches will require not just a comprehensive assessment of overall genetic risk but an understanding of the allele-specific interaction between common and rare variation (eg, through the construction of haplotypes using long read genomic sequencing).

We observed a strong male predominance in the patients with BrS analyzed, both for the overall cohort (95%) and the RE5 variant carriers (100%; Table 1). The reflects a well-established sex bias for heritable cardiac diseases in general, and BrS in particular, which remains an incompletely understood phenomenon. Further research is required to elucidate the factors underlying the large difference in sex-based penetrance in carriers of BrS pathogenic variants, and the implications this has for cascade screening and risk stratification in affected families.

The strong polygenic and common variant contributions to BrS highlighted by recent GWAS have revealed an increasingly complex genetic architecture for this condition.⁵ This can lead to complicated inheritance patterns in affected families, as demonstrated in Figure 3A, where one of the individuals had a positive ajmaline test despite not carrying the RE5 variant. Such “nonsegregation” is not atypical in BrS family pedigrees; a previous study identified 8 noncarriers of familial rare SCN5A variants in 13 families who tested positive after ajmaline challenge, including in 2 families with truncating SCN5A variants of indisputable pathogenicity.²⁹ This complexity necessitates a nuanced approach to evaluating the pathogenicity of novel BrS-associated variants that encompasses genetic, computational, and functional lines of evidence.

This study highlights the potential of rare noncoding variation to account for some of the missing heritability in diseases such as BrS. The integrated approaches we describe here demonstrate the type of methods that can be applied to address the challenging task of characterizing noncoding pathogenic variants as we enter the era of routine genome sequencing. In particular, detailed epigenetic knowledge across disease-relevant genomic loci and the judicious use of computational evidence can help reduce the search space of the vast noncoding genome and identify putative variants of interest for functional validation. The identification of such novel pathogenic variant classes also benefits from expanding research to historically understudied populations, where founder variants can be readily identified in case–control studies, especially where diseases under investigation are more prevalent (such as BrS in Southeast Asia).

The persistence of at least 2 recurrent SCN5A loss-of-function variants of intermediate to large effect size in the Thai population (the RE5 variant and the SCN5A:p.Arg965Cys intermediate effect variant) is intriguing given their association with sudden death as BrS risk factors. Whereas selective pressure against pathogenic BrS variants may be relatively modest, as sudden death usually occurs in fourth or fifth decade, these variants (and likely other such risk factors in East Asian populations) appear to be restricted to this region. Several examples of disease risk factors that are protective against infectious diseases are known, such as the relationship between malaria and sickle cell trait or APOL1 polymorphisms that protect against African sleeping sickness but increase the risk for chronic kidney disease.⁴⁵ Whether genetic variants associated with partially reduced sodium channel function confer similar protection is unknown, but would offer an intriguing hypothesis for the highly increased prevalence of BrS across Southeast Asia.

There are some limitations to this study. Whereas the genetic and functional data validate the pathogenicity of the RE5 variant and its role as the primary causal variant in these patients, the genetic architecture of BrS is complex. Additional common (and possibly rare) variant risk factors are likely to be contributing to the phenotype in the described cases, explaining the severity of the observed phenotypes. At the SCN5A/SCN10A locus, interaction between rare and common risk factors may determine the extent of reduced SCN5A expression and I_Na, which will require complex analysis and experimentation to resolve. Larger population genomics data sets will also be required to more precisely define the OR for the variant and determine whether it is also a risk factor in other Southeast Asian countries.

We have identified a rare, noncoding regulatory variant of large effect size at the SCN5A locus that is a major contributor to BrS risk in the Thai population. Improved knowledge of the genetic basis of arrhythmogenic diseases such as BrS offers the opportunity for proactively identifying and monitoring at-risk individuals to reduce the burden of sudden cardiac death, especially in Southeast Asia, where BrS is particularly prevalent. The inclusion of the RE5 variant in clinical genetic testing assays for BrS in Thailand may increase the diagnostic yield by one third (Figure 4). This will enable the identification and diagnosis of at-risk individuals in affected families allowing for targeted clinical monitoring and potentially treatment.⁴⁶ Based on the variant frequency in the Genomics Thailand database, we estimate that there are ≈43 000 carriers of the RE5 variant alone in Thailand. As genome sequencing becomes integrated into routine medical practice, the identification of variant carriers will enable more accurate assessment of its penetrance in the population and the initiation of clinical intervention, where appropriate, to prevent sudden cardiac death.

ARTICLE INFORMATION

Acknowledgments

The authors thank the National Biobank of Thailand, National Science and Technology Development Agency, for their assistance in gathering variant population frequency from the Genomics Thailand preliminary data.

Sources of Funding

Dr Mauleekoonphairoj received support by the Second Century Fund (C2F), Chulalongkorn University. Drs Poovorawan, Nademanee, and Khongphatthanayothin received support from the National Research Council of Thailand. Dr Wu received support from the Yin Shu-Tien Foundation, Taipei Veterans General Hospital–National Yang-Ming University Excellent Physician Scientists Cultivation Program (grant 108-V-A-013). Dr Behr received support from the Robert Lancaster Memorial Fund and UKRI. Dr Christoffels received support from EIC Pathfinder Challenges (grant Nav1.5-CARED) and the Netherlands Organization for Scientific Research (grant OCENW.GROOT.2019.029). Dr Bezzina received support from the Dutch Heart Foundation (grant CVON PREDICT2), the Netherlands Organization for Scientific Research (VICI fellowship 016.150.610), Fondation Leducq (grant 17CVD02), and EIC Pathfinder Challenges (grant Nav1.5-CARED).

Disclosures

None.

Supplemental Material

Methods

Tables S1–S8

Figures S1–S6

References 47–61

Supplementary Material

cir-151-31-s002.pdf ^{(1.5MB, pdf)}

cir-151-31-s003.xlsx ^{(82.9KB, xlsx)}

cir-151-31-s004.docx ^{(14.2KB, docx)}