Abstract

The electrocardiogram has several advantages in detecting cardiac arrhythmia—it is readily available, noninvasive and cost-efficient. Recent genome-wide association studies have identified single-nucleotide polymorphisms that are associated with electrocardiogram measures. We performed a genome-wide association study using Korea Association Resource data for the discovery phase (Phase 1, n = 6805) and two consecutive replication studies in Japanese populations (Phase 2, n = 2285; Phase 3, n = 5010) for QRS duration and PR interval. Three novel loci were identified: rs2483280 (PRDM16 locus) and rs335206 (PRDM6 locus) were associated with QRS duration, and rs17026156 (SLC8A1 locus) correlated with PR interval. PRDM16 was recently identified as a causative gene of left ventricular non-compaction and dilated cardiomyopathy in 1p36 deletion syndrome, which is characterized by heart failure, arrhythmia and sudden cardiac death. Thus, our finding that a PRDM16 SNP is linked to QRS duration strongly implicates PRDM16 in cardiac function. In addition, C allele of rs17026156 increases PR interval (beta ± SE, 2.39 ± 0.40 ms) and exists far more frequently in East Asians (0.46) than in Europeans and Africans (0.05 and 0.08, respectively).

INTRODUCTION

The electrocardiogram (ECG) has several advantages in detecting cardiac diseases; it is readily available, noninvasive and cost-efficient. QRS complex represents ventricular depolarization, and QRS duration represents the conduction time from atrioventicular node (AV node) to His-Purkinje system and ventricular myocardium (1). PR interval is the time between the onset of atrial depolarization (P-wave) and the onset of ventricular depolarization (R-wave).

QRS duration and PR interval are believed to reflect patient outcomes in several heart diseases (2–4). A diseased ventricular conduction system can lead to life-threatening bradyarrhythmias and tachyarrhythmias (5). Longer QRS duration is a predictor of mortality and sudden death in the general population (6) and in those with hypertension and coronary artery disease (7).

ECG measurements are believed to be complex traits with multiple genetic and environmental determinants (8). The heritability of each ECG measurement ranges from 30 to 50% in several ethnic groups (8–13). Recent genome-wide association studies (GWASs) have identified single-nucleotide polymorphisms (SNPs) that are associated with PR interval (14,15), QRS duration (14,16) and QT interval (17,18). In particular, QT interval has been studied extensively in European descendants by the QTGEN (17) and QTSCD (18) consortiums, and we have also reported a GWAS in East Asians (19). With regard to PR interval, two GWASs reported eight loci in European descendants, five of which (SCN5A-SCN10A, NKK2–5, CAV1/CAV2, SOX5 and TBX5) were linked to atrial fibrillation (AF) (14,15). Twenty-two loci were correlated with QRS duration in two GWASs of European descendants (14,16), and greater number of risk alleles for prolonged QRS duration was also associated with the risk of ventricular conduction defects (16).

No GWAS on PR interval or QRS duration has been performed in the Asian population. To determine the genetic architecture of PR interval and QRS duration in Asians, we conducted a GWAS using Korea Association Resource (KARE) data during the discovery phase (Phase1, n = 6805) and two consecutive replication studies in Japanese populations (Phase 2, n = 2285; Phase 3, n = 5010) (Supplementary Material, Fig. S1).

RESULTS

Discovery GWASs

The clinical characteristics of subjects in the discovery GWAS and two replication studies are described in Table 1. The genomes of discovery subjects were scanned to identify genetic variations that were associated with QRS duration and PR interval. The genotypes in this study consisted of experimentally genotyped SNPs and computationally imputed SNPs. In total, 2.1 million SNPs were examined in the linear regression model as independent variables of ECG traits, controlling for age, sex, recruitment area, body mass index, systolic blood pressure and height as covariates. Q–Q plots of the GWAS in Koreans are shown in Supplementary Material, Figure S2.

Table 1.

Clinical characteristics of subjects in each phase

Variables Phase 1 Phase 2 Phase 3 
KARE Japanese Japanese 
n (% male) 6805 (50.4%) 2285 (31.9%) 5010 (33.3%) 
Age, years 51.6 (8.7) 49.8 (13.9) 56.7 (13.4) 
BMI, kg/m2 24.6 (3.1) 22.2 (3.2) 22.6 (3.2) 
SBP, mm Hg 116.4 (17.9) 120.1 (16.8) 126.1 (19.1) 
Height, cm 160.6 (8.7) 160.4 (8.3) 158.6 (8.7) 
PR interval, ms 163.2 (35.9) 158.1 (21.4) 158.7 (22.0) 
QRS duration, ms 90.3 (10.3) 97.0 (8.2) 95.5 (9.0) 
Variables Phase 1 Phase 2 Phase 3 
KARE Japanese Japanese 
n (% male) 6805 (50.4%) 2285 (31.9%) 5010 (33.3%) 
Age, years 51.6 (8.7) 49.8 (13.9) 56.7 (13.4) 
BMI, kg/m2 24.6 (3.1) 22.2 (3.2) 22.6 (3.2) 
SBP, mm Hg 116.4 (17.9) 120.1 (16.8) 126.1 (19.1) 
Height, cm 160.6 (8.7) 160.4 (8.3) 158.6 (8.7) 
PR interval, ms 163.2 (35.9) 158.1 (21.4) 158.7 (22.0) 
QRS duration, ms 90.3 (10.3) 97.0 (8.2) 95.5 (9.0) 

Data are presented as mean (standard deviation).

BMI, body mass index; SBP, systolic blood pressure.

All P-values are charted in Figure 1A (QRS duration) and B (PR interval), plotting −log10(p) against the chromosomal position on Manhattan plots and are shown in Supplementary Material, Tables S1 and S2. The red line in Figure 1 indicates a genome-wide significance level (P < 5 × 10−8), and the blue line indicates a suggestive level (P < 1 × 10−4). Two loci (CDKN1A and SETBP1) for QRS duration and three loci (SLC8A1, SCN5A/SCN10A and CAV1/CAV2) for PR interval met the genome-wide significance level in the discovery phase. The 323 SNPs for QRS duration and 341 SNPs for PR intervals had P-values that were lower than the suggestive level, encompassed by 21 and 20 loci, respectively. The lead SNP for each locus is listed in Supplementary Material, Table S3.

Figure 1.

Manhattan plot of genome-wide association signals from Phase 1 Study. –log10(p) values are plotted against chromosomal base-pair positions. Green label indicates previously reported loci for QRS duration and PR interval, and red dots indicate previously unreported loci showing associations with QRS and PR in the Phase 1 study and tested for replication. The red line represents the genome-wide significance level (P = 5 × 10−8), and the blue line represents a P-value of 1 × 10−4. (A) QRS duration and (B) PR interval.

Figure 1.

Manhattan plot of genome-wide association signals from Phase 1 Study. –log10(p) values are plotted against chromosomal base-pair positions. Green label indicates previously reported loci for QRS duration and PR interval, and red dots indicate previously unreported loci showing associations with QRS and PR in the Phase 1 study and tested for replication. The red line represents the genome-wide significance level (P = 5 × 10−8), and the blue line represents a P-value of 1 × 10−4. (A) QRS duration and (B) PR interval.

The SETBP1, CDKN1A, SCN5A and HAND1 regions for QRS duration and CAV1, SCN10A and TBX5 regions for PR interval have previously been reported (14–16). The remaining suggestive loci shown in Supplementary Material, Table S3 were examined in replication studies of Japanese populations.

Novel genetic variant of QRS duration and PR interval in East Asians

Seventeen suggestive loci for QRS duration and 17 suggestive loci for PR interval were identified in the discovery GWAS. To confirm these findings, two consecutive replication tests were conducted in two independent Japanese populations.

We carried out Phase 2 follow-up in silico genotyping for 34 SNPs in 2285 Japanese individuals (Supplementary Material, Table S4). In the meta-analysis of Phases 1 and 2, two SNPs (rs2483280 and rs17026156) reached genome-wide significance P-value. These two SNPs and five additional SNPs that became better P-values than Phase 1 in meta-analysis of Phase 1 + 2 were genotyped in a subsequent de novo replication study in 5010 Japanese population (Phase 3). Three SNPs (rs2483280 and rs335206 for QRS duration and rs17026156 for PR interval) reached genome-wide significance in the meta-analyses of Phases 1, 2 and 3 (Table 2).

Table 2.

Replication results of novel SNPs in each phase and meta-analysis

SNP ID CHR BP Genea Coded allele Phase 1 (n = 6805)
 
Phase 2 (n = 2285)
 
Meta-analysis (Phase 1 + 2)
 
AF Beta ± SE P-value AF Beta ± SE P-value Beta P-value Q I2 
QRS 
 rs2483280 3 245 399 PRDM16 0.26 −0.91 ± 0.18 7.47 × 10−7 0.28 −0.61 ± 0.24 0.010 −0.80 3.83 × 10−8 0.33 0.00 
 rs335206 122 532 465 PRDM6 0.33 −0.76 ± 0.17 8.38 × 10−6 0.34 −0.50 ± 0.23 0.026 −0.67 9.63 × 10−7 0.37 0.00 
PR 
 rs17026156 40 614 469 SLC8A1 0.39 2.39 ± 0.40 2.85 × 10−9 0.28 2.00 ± 0.68 0.003 2.29 3.79 × 10−11 0.62 0.00 
 rs7948943 11 97 642 455 Intergenic 0.29 1.96 ± 0.43 5.21 × 10−6 0.28 1.00 ± 0.68 0.139 1.68 3.45 × 10−6 0.23 30.47 
 rs3136189 16 13 942 202 ERCC4 0.24 1.82 ± 0.46 7.60 × 10−5 0.24 0.96 ± 0.71 0.176 1.56 4.97 × 10−5 0.31 3.86 
 rs6499591 16 71 464 505 ZFHX3 0.21 −1.97 ± 0.49 5.58 × 10−5 0.24 −1.12 ± 0.70 0.108 −1.69 2.39 × 10−5 0.32 0.00 
 rs17744182 18 28 286 523 GAREM 0.39 1.84 ± 0.41 7.91 × 10−6 0.39 0.88 ± 0.62 0.157 1.55 6.51 × 10−6 0.20 39.97 
     Phase 3 (n = 5010) Meta-analysis (Phase 1 + 2 + 3)    
AF Beta ± SE P-value Beta P-value Q I2    
QRS    
 rs2483280 3 245 399 PRDM16 0.28 −0.68 ± 0.17 8.19 × 10−5 −0.75 1.51 × 10−11 0.55 0.00    
 rs335206 122 532 465 PRDM6 0.35 −0.64 ± 0.16 8.06 × 10−5 −0.66 3.19 × 10−10 0.66 0.00    
PR    
 rs17026156 40 614 469 SLC8A1 0.29 1.74 ± 0.45 9.44 × 10−5 2.08 2.58 × 10−14 0.55 0.00    
 rs7948943 11 97 642 455 Intergenic 0.29 −1.18 ± 0.45 9.12 × 10−3 0.56 4.16 × 10−2 0.00 92.25    
 rs3136189 16 13 942 202 ERCC4 0.24 −0.82 ± 0.47 8.35 × 10−2 0.61 4.02 × 10−2 0.00 87.72    
 rs6499591 16 71 464 505 ZFHX3 0.24 −0.11 ± 0.47 8.10 × 10−1 −1.03 7.25 × 10−4 0.02 73.24    
 rs17744182 18 28 286 523 GAREM 0.40 1.26 ± 0.42 2.40 × 10−3 1.43 6.25 × 10−8 0.38 0.00    
SNP ID CHR BP Genea Coded allele Phase 1 (n = 6805)
 
Phase 2 (n = 2285)
 
Meta-analysis (Phase 1 + 2)
 
AF Beta ± SE P-value AF Beta ± SE P-value Beta P-value Q I2 
QRS 
 rs2483280 3 245 399 PRDM16 0.26 −0.91 ± 0.18 7.47 × 10−7 0.28 −0.61 ± 0.24 0.010 −0.80 3.83 × 10−8 0.33 0.00 
 rs335206 122 532 465 PRDM6 0.33 −0.76 ± 0.17 8.38 × 10−6 0.34 −0.50 ± 0.23 0.026 −0.67 9.63 × 10−7 0.37 0.00 
PR 
 rs17026156 40 614 469 SLC8A1 0.39 2.39 ± 0.40 2.85 × 10−9 0.28 2.00 ± 0.68 0.003 2.29 3.79 × 10−11 0.62 0.00 
 rs7948943 11 97 642 455 Intergenic 0.29 1.96 ± 0.43 5.21 × 10−6 0.28 1.00 ± 0.68 0.139 1.68 3.45 × 10−6 0.23 30.47 
 rs3136189 16 13 942 202 ERCC4 0.24 1.82 ± 0.46 7.60 × 10−5 0.24 0.96 ± 0.71 0.176 1.56 4.97 × 10−5 0.31 3.86 
 rs6499591 16 71 464 505 ZFHX3 0.21 −1.97 ± 0.49 5.58 × 10−5 0.24 −1.12 ± 0.70 0.108 −1.69 2.39 × 10−5 0.32 0.00 
 rs17744182 18 28 286 523 GAREM 0.39 1.84 ± 0.41 7.91 × 10−6 0.39 0.88 ± 0.62 0.157 1.55 6.51 × 10−6 0.20 39.97 
     Phase 3 (n = 5010) Meta-analysis (Phase 1 + 2 + 3)    
AF Beta ± SE P-value Beta P-value Q I2    
QRS    
 rs2483280 3 245 399 PRDM16 0.28 −0.68 ± 0.17 8.19 × 10−5 −0.75 1.51 × 10−11 0.55 0.00    
 rs335206 122 532 465 PRDM6 0.35 −0.64 ± 0.16 8.06 × 10−5 −0.66 3.19 × 10−10 0.66 0.00    
PR    
 rs17026156 40 614 469 SLC8A1 0.29 1.74 ± 0.45 9.44 × 10−5 2.08 2.58 × 10−14 0.55 0.00    
 rs7948943 11 97 642 455 Intergenic 0.29 −1.18 ± 0.45 9.12 × 10−3 0.56 4.16 × 10−2 0.00 92.25    
 rs3136189 16 13 942 202 ERCC4 0.24 −0.82 ± 0.47 8.35 × 10−2 0.61 4.02 × 10−2 0.00 87.72    
 rs6499591 16 71 464 505 ZFHX3 0.24 −0.11 ± 0.47 8.10 × 10−1 −1.03 7.25 × 10−4 0.02 73.24    
 rs17744182 18 28 286 523 GAREM 0.40 1.26 ± 0.42 2.40 × 10−3 1.43 6.25 × 10−8 0.38 0.00    

aGenes are defined as the gene closest to the SNP within a 200-kb window (HaploReg v2). Bold indicates genome-wide significant P-values (5 × 10−8).

CHR, chromosome; BP, base pair; AF, coded allele frequency; Q, P-value for Cochrane's Q statistic; I2, heterogeneity index.

The genetic regions of these three SNPs and their association results are depicted as signal plots in Figure 2. rs2483280 lies in the third intron of PRDM16 (based on the NM_022114.3 transcript), rs335206 resides in the fifth intron of PRDM6 (based on the NM_001136239.1 transcript) and rs17026156 is located 21 kb upstream of SLC8A1.

Figure 2.

Signal plots for three novel loci across a 1-Mb window. Association of individual SNPs in the Phase 1 study plotted as −log10(p) against chromosomal base-pair position. The y-axis on the right shows the recombination rate, estimated from the HapMap CHB and JPT populations. All P-values are from the discovery phase. The purple diamond represents the meta-analysis results of the Phase 1, 2 and 3 studies. (A) rs2483280 of PRDM16, (B) rs335206 of PRDM6 and (C) rs17026156 of SLC8A1.

Figure 2.

Signal plots for three novel loci across a 1-Mb window. Association of individual SNPs in the Phase 1 study plotted as −log10(p) against chromosomal base-pair position. The y-axis on the right shows the recombination rate, estimated from the HapMap CHB and JPT populations. All P-values are from the discovery phase. The purple diamond represents the meta-analysis results of the Phase 1, 2 and 3 studies. (A) rs2483280 of PRDM16, (B) rs335206 of PRDM6 and (C) rs17026156 of SLC8A1.

Extension of variants identified in European descendants to Koreans

To compare the genetic architecture of QRS duration and PR interval between Europeans and Koreans, the SNPs that were previously identified in European descendants were examined in discovery GWAS (Table 3 and Supplementary Material, Table S5). Three GWASs reported 21 SNPs for QRS duration and 10 SNPs for PR intervals (14–16). In the KARE genotype data of discovery GWAS, there were only five SNPs that matched the reported SNPs. Thus, we added SNPs with linkage disequilibrium (LD) (r2 > 0.8 and D' > 0.9) of the lead SNPs in the European studies. A total of 22 SNPs were examined for their association in Koreans (Supplementary Material, Table S5).

Table 3.

Extension of variants identified in European descendants to Koreans

Reported gene Ref. European GWAS
 
Korean GWAS
 
SNP Coded allele AF  Beta ± SE P-value SNP Coded allele AF Beta ± SE P-value 
QRS 
NFIA 16 rs9436640 0.46  −0.59 ± 0.07 4.57 × 10−18 rs2103883 0.35 −0.57 ± 0.17 6.85 × 10−4 
CRIM1 16 rs7562790 0.40  0.39 ± 0.07 8.22 × 10−9 rs7562790 0.65 0.64 ± 0.17 1.52 × 10−4 
HEATR5B-STRN 16 rs17020136 0.21  0.51 ± 0.08 1.90 × 10−9 rs2160411 0.55 0.55 ± 0.16 6.76 × 10−4 
HAND1-SAP30L 16 rs13165478 0.36  −0.55 ± 0.07 7.36 × 10−14 rs6580083 0.30 −0.85 ± 0.18 1.29 × 10−6 
CDKN1A 14,16 rs9470361 0.25  0.87 ± 0.08 3.00 × 10−27 rs9470366 0.15 1.16 ± 0.23 4.02 × 10−7 
VTI1A 16 rs7342028 0.27  0.48 ± 0.08 4.95 × 10−10 rs10885378 0.47 0.34 ± 0.16 3.55 × 10−2 
SETBP1 16 rs991014 0.42  0.42 ± 0.07 6.2 × 10−10 rs4890489 0.32 0.97 ± 0.17 2.32 × 10−8 
PR 
MEIS1 15 rs11897119 0.39  1.36 ± 0.21 4.62 × 10−11 rs4430933 0.76 −1.54 ± 0.47 1.02 × 10−3 
ARHGAP24 14,15 rs7692808 0.31  −2.01 ± 0.22 5.99 × 10−20 rs10012090 0.92 −1.87 ± 0.76 1.33 × 10−2 
SOX5 15 rs11047543 0.15  −2.09 ± 0.29 3.34 × 10−13 rs4246224 0.14 −1.89 ± 0.58 1.16 × 10−3 
QRS, PR 
EXOG-SCN5A-SCN10A 14–16 rs9851724 0.33 QRS −0.66 ± 0.07 1.91 × 10−20 rs7633988 0.29 −0.86 ± 0.18 1.50 × 10−6 
rs6800541 0.40 PR 3.77 ± 0.21 2.10 × 10−74 rs7433306 0.15 4.33 ± 0.56 1.45 × 10−14 
CAV1-CAV2 14,15 rs3807989 0.40 QRS 3.3 1.10 × 10−4 rs11773845 0.34 0.60 ± 0.17 5.36 × 10−4 
PR 2.30 ± 0.21 3.66 × 10−28 3.21 ± 0.42 3.33 × 10−14 
TBX5-TBX3 15,16 rs10850409 0.27 QRS −0.49 ± 0.08 3.06 × 10−10 rs3914956 0.49 −0.41 ± 0.16 1.11 × 10−2 
rs1896312 0.28 PR 1.95 ± 0.23 3.13 × 10−17 rs10744836 0.52 1.29 ± 0.40 1.41 × 10−3 
Reported gene Ref. European GWAS
 
Korean GWAS
 
SNP Coded allele AF  Beta ± SE P-value SNP Coded allele AF Beta ± SE P-value 
QRS 
NFIA 16 rs9436640 0.46  −0.59 ± 0.07 4.57 × 10−18 rs2103883 0.35 −0.57 ± 0.17 6.85 × 10−4 
CRIM1 16 rs7562790 0.40  0.39 ± 0.07 8.22 × 10−9 rs7562790 0.65 0.64 ± 0.17 1.52 × 10−4 
HEATR5B-STRN 16 rs17020136 0.21  0.51 ± 0.08 1.90 × 10−9 rs2160411 0.55 0.55 ± 0.16 6.76 × 10−4 
HAND1-SAP30L 16 rs13165478 0.36  −0.55 ± 0.07 7.36 × 10−14 rs6580083 0.30 −0.85 ± 0.18 1.29 × 10−6 
CDKN1A 14,16 rs9470361 0.25  0.87 ± 0.08 3.00 × 10−27 rs9470366 0.15 1.16 ± 0.23 4.02 × 10−7 
VTI1A 16 rs7342028 0.27  0.48 ± 0.08 4.95 × 10−10 rs10885378 0.47 0.34 ± 0.16 3.55 × 10−2 
SETBP1 16 rs991014 0.42  0.42 ± 0.07 6.2 × 10−10 rs4890489 0.32 0.97 ± 0.17 2.32 × 10−8 
PR 
MEIS1 15 rs11897119 0.39  1.36 ± 0.21 4.62 × 10−11 rs4430933 0.76 −1.54 ± 0.47 1.02 × 10−3 
ARHGAP24 14,15 rs7692808 0.31  −2.01 ± 0.22 5.99 × 10−20 rs10012090 0.92 −1.87 ± 0.76 1.33 × 10−2 
SOX5 15 rs11047543 0.15  −2.09 ± 0.29 3.34 × 10−13 rs4246224 0.14 −1.89 ± 0.58 1.16 × 10−3 
QRS, PR 
EXOG-SCN5A-SCN10A 14–16 rs9851724 0.33 QRS −0.66 ± 0.07 1.91 × 10−20 rs7633988 0.29 −0.86 ± 0.18 1.50 × 10−6 
rs6800541 0.40 PR 3.77 ± 0.21 2.10 × 10−74 rs7433306 0.15 4.33 ± 0.56 1.45 × 10−14 
CAV1-CAV2 14,15 rs3807989 0.40 QRS 3.3 1.10 × 10−4 rs11773845 0.34 0.60 ± 0.17 5.36 × 10−4 
PR 2.30 ± 0.21 3.66 × 10−28 3.21 ± 0.42 3.33 × 10−14 
TBX5-TBX3 15,16 rs10850409 0.27 QRS −0.49 ± 0.08 3.06 × 10−10 rs3914956 0.49 −0.41 ± 0.16 1.11 × 10−2 
rs1896312 0.28 PR 1.95 ± 0.23 3.13 × 10−17 rs10744836 0.52 1.29 ± 0.40 1.41 × 10−3 

Ref, References; AF, coded allele frequency.

Seven of 14 QRS-related loci, 3 of 5 PR-related loci and all 3 loci for both traits were associated with Koreans, based on P < 0.05 (Fig. 3). Of the seven QRS duration-associated loci, three (HAND1-SAP30L, CDKN1A and SETBP1) had P-values of <1 × 10−5. The three loci (EXOG-SCN5A-SCN10A, CAV1-CAV2 and TBX5-TBX3) that were linked to both traits were also significantly associated with Koreans. Further, the EXOG-SCN5A-SCN10A and CAV1-CAV2 regions had large effect sizes compared with other loci (beta ± SE = 4.33 ± 0.56, P = 1.45 × 10−14 and beta ± SE = 3.21 ± 0.42, P = 3.33 × 10−14, respectively).

Figure 3.

Comparison of effect size (β) of SNPs previously identified in Europeans with those in East Asians. Each dot refers to an association signal, with colors (red, P < 5 × 10−8; orange, 5 × 10−8P < 10−5; green, 10−5P < 0.05; white, P ≥ 0.05). CAV1-CAV2 locus was not included in QRS plot because it did not reach genome-wide significance in European GWAS. Effect size was presented beta per copy of the coded allele. (A) QRS duration and (B) PR interval.

Figure 3.

Comparison of effect size (β) of SNPs previously identified in Europeans with those in East Asians. Each dot refers to an association signal, with colors (red, P < 5 × 10−8; orange, 5 × 10−8P < 10−5; green, 10−5P < 0.05; white, P ≥ 0.05). CAV1-CAV2 locus was not included in QRS plot because it did not reach genome-wide significance in European GWAS. Effect size was presented beta per copy of the coded allele. (A) QRS duration and (B) PR interval.

DISCUSSION

In silico annotation of novel SNP sites

Our discovery GWAS in Koreans and two replication studies in Japanese identified three novel loci for QRS duration and PR interval: rs2483280 (PRDM16 locus) and rs335206 (PRDM6 locus) for QRS duration and rs17026156 (SLC8A1 locus) for PR interval. Because the three SNPs lay in noncoding regions, we examined their function in regulating gene expression using ENCODE data and the web-based program RegulomeDB. Further, their evolutional conservation was studied by comparing the allelic sequences with primate genome sequences using Ensembl data.

The DNA sequence that encompassed the rs2483280 SNP was predicted to be a ZBTB3 transcription factor-binding motif and an open chromatin region (DNase I-hypersensitive region). However, the SNP sequence was not conserved in primate DNA (Fig. 4A). Thus, we searched for high-LD SNPs near the lead SNP and identified rs2255212 1.5 kb away from the lead SNP (LD score r2 = 0.98 and D’ = 1.00) (Fig. 4A). rs2255212 was predicted to be a TCF4 transcription factor-binding site and an open chromatin region, and the SNP was highly conserved in all primates. However, the association P-value of rs2255212 (1.16 × 10−6) was not better than that of the lead SNP (rs2483280, P = 7.47 × 10−7). rs335206 was predicted to be a ZNF263-binding site and conserved in all primates, but not an open chromatin region (Fig. 4B).

Figure 4.

Evolutionary conservation of three novel genetic variants. The human and primate sequences of the SNP ± 10 bp were obtained from Ensembl website. (A) rs2483280 of PRDM16, (B) rs335206 of PRDM6 and (C) rs17026156 of SLC8A1.

Figure 4.

Evolutionary conservation of three novel genetic variants. The human and primate sequences of the SNP ± 10 bp were obtained from Ensembl website. (A) rs2483280 of PRDM16, (B) rs335206 of PRDM6 and (C) rs17026156 of SLC8A1.

rs17026156 did not match any functionally conserved sequence, although it was conserved in all primates. Thus, we searched for LD SNPs near the lead SNP and identified a high-LD (r2 = 0.93 and D’ = 1.00) SNP (rs13017846) that was predicted to be a PIT-1-binding site (Fig. 4C). However, the association P-value of rs13017846 (4.47 × 10−9) was also not better than that of the lead SNP (rs17026156, P = 2.85 × 10−9).

Based on simulation study, the functional variant in a GWAS locus may not have the most significance owing to random sampling. Therefore, functional validation is required to implicate or dismiss rs13017846 and rs2255212, although they did not exceed the significance P-values of the lead SNPs.

Notably, the allele frequency of rs17026156 varies widely between ethnicities. The allele frequencies of the ancestral C allele are as low as 0.05 in Europeans (HapMap-CEU) and 0.08 in Africans (HapMap-YRI), whereas they reach as high as 0.57 in Chinese (HapMap-HCB) and 0.35 in Japanese (HapMap-JPT). The rs17026156 was not identified in previous GWASs in European descendants, possibly due to its low allelic frequency in Europeans. Individuals with the C allele of rs17026156 increased PR intervals (beta ± SE, 2.39 ± 0.40 ms) in East Asian population.

Annotation of proximal genes

rs2483280, associated with QRS duration, lies in the third intron of PRDM16 (PR domain-containing 16), which encodes a protein with a zinc finger DNA-binding domain and PR domain. PRDM16 regulates brown adipose tissue differentiation (20), and its genetic region translocates frequently chromosome 3q21, causing acute myeloid leukemia and myelodysplastic syndrome (21). Recently, fine mapping analysis of 1p36 deletion syndrome implicated a mutation in PRDM16 as a cause of cardiomyopathy with left ventricular non-compaction and dilated cardiomyopathy, both of which are characterized by progressive cardiac dysfunction, resulting in heart failure, arrhythmia and sudden cardiac death (22).

PRDM16, expressed in the nuclei of cardiomyocytes, potentiates cardiomyocyte proliferation. Haploinsufficiency of PRDM16 in zebrafish results contractile dysfunction and the reduction of ventricular conduction velocity (22), supporting our finding that rs2483280 in PRDM16 is associated with QRS duration.

Another novel SNP was associated with QRS duration—rs335206, in the fourth intron of PRDM6 (PR domain-containing 6), a transcriptional repressor in smooth muscle cells. PRDM6 regulates the development, differentiation and proliferation of blood vessels (23), and rs335206 was recently linked to systolic blood pressure (24). It is recently reported in mice that Prdm6 knockout embryos die during development, displaying signs of cardiac insufficiency including a thinning of the myocardial walls (25).

rs17026156, associated with PR interval, lies ∼21 kb upstream of SLC8A1 (sodium/calcium exchanger 1 precursor). SLC8A1 extrudes calcium from cardiac myocytes during relaxation and returns the myocardium to its resting state after excitation (26). Targeted disruption of SLC8A1 causes defects in heartbeat—SLC8A1-/- mouse embryos experience slow and arrhythmic heart contractions (27). We have also reported this locus to correlate with QT interval traits in East Asians (19). Based on the previous reports, SLC8A1 appears to mediate electrophysiological conductivity during heart.

In conclusion, we have identified three novel loci for QRS duration and PR interval and confirmed 13 previously reported loci. These data will increase our understanding in the genetic architecture that underlies the mechanisms of electrocardiographic traits, QRS duration and PR interval.

MATERIALS AND METHODS

Subjects

The study subjects have been described in a QT interval GWAS (19). Briefly, 6805 subjects from KARE were selected from an ongoing population-based cohort, as part of the Korean Genome and Epidemiology Study (KoGES). Subjects without a self-reported history of cardiac disease, concurrent use of medications that interfere with the ECG measurements and abnormal electrolyte values at the ECG were included. Written informed consent was obtained from all participants, and this project was approved by the institutional review board of the Korea National Institute of Health.

The Phase 2 Japanese subjects were part of the Nagahama Prospective Genome Cohort for Comprehensive Human Bioscience (The Nagahama Study). The Nagahama Study cohort was recruited from 2008 to 2010 from the general population in Nagahama City, a largely rural city of 125 000 inhabitants in Shiga Prefecture that lies in the center of Japan. Of the 9804 participants, persons whose genome-wide SNP was analyzed (n = 3710) and who were free of symptomatic cardiovascular disease and abnormal ECG readings (n = 2285) were used in the second GWAS panel. All clinical measurements and sampling of blood were performed on enrollment. Genomic DNA was extracted from peripheral blood samples with phenol–chloroform.

The Phase 3 replication panel comprised Japanese from three independent subcohorts. First, the Anti-Aging Center (AAC) cohort included consecutive participants in the medical check-up program at Ehime University Hospital, which was designed specifically to evaluate age-related disorders, including atherosclerosis, cardiovascular disease, physical function and mild cognitive impairment. All clinical data in this study were obtained through the check-up process.

With regard to the second subcohort, the Takashima Study is an ongoing longitudinal study, based on community residents in Takashima City. Takashima City is a semiurban area in Shiga Prefecture, with a population of ∼54 000. Study subjects were recruited between 2002 and 2003 from participants of the annual medical check-up program, held by Takashima City. The basic clinical parameters in this study were obtained from the personal medical check-up records of the subjects. The third subcohort of the replication analysis comprised the remaining sample of the Nagahama Study.

All study procedures in Japan were approved by the ethics committee of Ehime University Graduate School of Medicine, Shige University of Medical Science and Kyoto University Graduate School of Medicine.

ECG measurements

PR interval and QRS duration values were obtained from a supine 12-lead ECG using digital electrocardiographic recorders—Phase 1, MAC5000 (GE Medical System, CT, USA); Phase 2, FCP-7411 and FCP-7431 (Fukuda Denshi, Tokyo, Japan); Phase 3, AAC, ECG-1500 (Nihon Kohden, Tokyo, Japan) and Takashima, FCP-4720 (Fukuda Denshi, Tokyo, Japan). ECGs with insufficient quality (e.g., owing to baseline drift or missing leads) and those with rhythms other than sinus rhythm or AF were excluded. PR interval was measured from the onset of the P-wave to the onset of ventricular depolarization. QRS duration was measured from the onset of ventricular depolarization to the J point.

Genotyping

The genotyping data were obtained from KARE, which used the Affymetrix Genomewide Human SNP Array 5.0. The genotype quality control criteria have been reported in a previous GWAS study (28). Briefly, the criteria for the inclusion of SNPs were genotype call rate of >0.98, minor allele frequency (MAF) of >0.01 and Hardy–Weinberg equilibrium (HWE) (P > 1 × 10−6). The related individuals were excluded from the KARE genotype dataset, whose computed average pairwise identity-by-state value was higher than that estimated from first-degree relatives of Korean sib-pair samples (>0.80, n = 601). Ultimately, 352 228 SNPs passed the quality control process and were subsequently used in the GWASs for PR interval and QRS duration. SNP imputation was performed with IMPUTE (29) using the JPT and CHB sets of HapMap Phase 2 as references. After removing SNPs with MAF of <0.01 and SNP missing rate of >0.05, we combined the remaining 1.8 million imputed SNPs with the SNPs that were typed directly in KARE for the association analysis.

Genome-wide SNP genotyping of the Nagahama sample was performed using a series of BeadChip DNA arrays (Illumina, San Diego, CA, USA). Genotyping quality was controlled by excluding SNPs with call rates of <99%, with an MAF of <0.1%, and deviating significantly from HWE (P < 1 × 10−7). Individuals who met the following criteria were excluded from analysis: average genotype call rate <95%, high degree of kinship (Pi-hat >0.35 [PLINK version 1.07 (30)]), and identified as an ancestry outlier by principal component analysis with the HapMap Phase 2, release 28 JPT dataset as the reference [EIGENSTRAT version 2.0 (31)]. Genotype imputation was performed using MACH, version 1.0.16 (32). Imputed SNPs for which the MAF was <0.01 or R-square value was <0.5 were excluded from the association analysis.

Replication genotyping of the Phase 3 sample was performed using a TaqMan probe assay and commercially available primer and probe sets (Life Technologies Corporation, Carlsbad, CA, USA). The fluorescence level of the PCR products was measured on a 7900HT Fast Real-Time PCR System (Applied Biosystems, Foster, CA, USA).

Statistical analysis

The effect of a genotype was analyzed by linear regression. The effect size (beta) and standard error (SE) of coded alleles were calculated on PR interval and QRS duration. All analyses were adjusted for age, sex, recruitment area, BMI, systolic blood pressure and height. PLINK (30) was used for all statistical tests. All tests were based on an additive model, and Phase 1 SNPs for replication test were selected, based on P < 1 × 10−4. We combined Phase 1 and Phase 2 data by inverse-variance meta-analysis under the assumption of fixed effects using Cochran's Q test to determine between-study heterogeneity (33). Phase 1 + Phase 2 SNPs were selected, based on meta-analyses P-values that were more significant than Phase 1 P-values. Finally, Phase 1 + Phase 2 + Phase 3 meta-analyses were conducted, and through which we identified significant genome-wide-level variants. All meta-analysis calculations were implemented in PLINK (30) (version 1.07).

In silico functional analysis of novel SNPs

Proximal SNP and LD were computed using SNAP, a web-based software program (http://www.broadinstitute.org/mpg/snap/ldsearchpw.php) (34). Evolutional conservation was confirmed using the Ensembl Genome browser (http://www.ensembl.org/index.html), comparing the SNP ± 10 bp in primates. The functional elements that were linked to the associated SNPs were analyzed using the RegulomeDB (http://regulome.stanford.edu/), which was developed by the ENCODE project (35).

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

Conflict of Interest statement. None declared.

FUNDING

The genotype and epidemiological data were provided by the Korean Genome Analysis Project (4845-301) and the Korean Genome and Epidemiology Study (4851-302), funded by the Ministry for Health and Welfare, Republic of Korea. This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education, Science and Technology (NRF-2013R1A1A2012069). This work was supported by a National Research Foundation of Korea (NRF) grant, funded by the Korean government (MSIP)(NRF-2011-0030072).

APPENDIX

Principal investigators of the Japanese study cohorts are as follows:

Nagahama Study: Fumihiko Matsuda (chairperson), Yasuharu Tabara, Takahisa Kawaguchi, Yoshimitsu Takahashi, Kazuya Setoh, Chikashi Terao, Ryo Yamada, Akihiro Sekine, Shinji Kosugi and Takeo Nakayama (Kyoto University Graduate School of Medicine, and School of Public Health); the AAC study: Yasuharu Tabara (chairperson), Katsuhiko Kohara, Michiya Igase and Tetsuro Miki (Ehime University Graduate School of Medicine)

Takashima study: Yoshikuni Kita (chairperson), Hirotsugu Ueshima and Naoyuki Takashima (Shiga University of Medical Science).

References

1
Saksena
S.
Camm
J.A.
2011
 
Electrophysiological Disorders of the Heart Elsevier Saunders
2
Algra
A.
Tijssen
J.G.
Roelandt
J.R.
Pool
J.
Lubsen
J.
QTc prolongation measured by standard 12-lead electrocardiography is an independent risk factor for sudden death due to cardiac arrest
Circulation
 , 
1991
, vol. 
83
 (pg. 
1888
-
1894
)
3
Desai
A.D.
Yaw
T.S.
Yamazaki
T.
Kaykha
A.
Chun
S.
Froelicher
V.F.
Prognostic significance of quantitative QRS duration
Am. J. Med.
 , 
2006
, vol. 
119
 (pg. 
600
-
606
)
4
Benjamin
E.J.
Chen
P.S.
Bild
D.E.
Mascette
A.M.
Albert
C.M.
Alonso
A.
Calkins
H.
Connolly
S.J.
Curtis
A.B.
Darbar
D.
, et al.  . 
Prevention of atrial fibrillation: report from a national heart, lung, and blood institute workshop
Circulation
 , 
2009
, vol. 
119
 (pg. 
606
-
618
)
5
Fang
F.
Sanderson
J.E.
Yu
C.M.
Potential role of biventricular pacing beyond advanced systolic heart failure
Circ. J.
 , 
2013
, vol. 
77
 (pg. 
1364
-
1369
)
6
Aro
A.L.
Anttonen
O.
Tikkanen
J.T.
Junttila
M.J.
Kerola
T.
Rissanen
H.A.
Reunanen
A.
Huikuri
H.V.
Intraventricular conduction delay in a standard 12-lead electrocardiogram as a predictor of mortality in the general population
Circ. Arrhythm. Electrophysiol.
 , 
2011
, vol. 
4
 (pg. 
704
-
710
)
7
Marijon
E.
Trinquart
L.
Otmani
A.
Waintraub
X.
Kacet
S.
Clementy
J.
Chatellier
G.
Le Heuzey
J.Y.
Competing risk analysis of cause-specific mortality in patients with an implantable cardioverter-defibrillator: the EVADEF cohort study
Am. Heart J.
 , 
2009
, vol. 
157
 (pg. 
391
-
397 e391
)
8
Li
J.
Huo
Y.
Zhang
Y.
Fang
Z.
Yang
J.
Zang
T.
Xu
X.
Familial aggregation and heritability of electrocardiographic intervals and heart rate in a rural Chinese population
Ann. Noninvasive Electrocardiol.
 , 
2009
, vol. 
14
 (pg. 
147
-
152
)
9
Havlik
R.J.
Garrison
R.J.
Fabsitz
R.
Feinleib
M.
Variability of heart rate, P-R, QRS and Q-T durations in twins
J. Electrocardiol.
 , 
1980
, vol. 
13
 (pg. 
45
-
48
)
10
Mathers
J.A.
Osborne
R.H.
Degeorge
F.V.
Studies of blood pressure, heart rate, and the electrocardiogram in adult twins
Am. Heart J.
 , 
1961
, vol. 
62
 (pg. 
634
-
642
)
11
Moller
P.
Heiberg
A.
Berg
K.
The atrioventricular conduction time - a heritable trait? III. Twin studies
Clin. Genet.
 , 
1982
, vol. 
21
 (pg. 
181
-
183
)
12
Russell
M.W.
Law
I.
Sholinsky
P.
Fabsitz
R.R.
Heritability of ECG measurements in adult male twins
J. Electrocardiol.
 , 
1998
, vol. 
30
 
Suppl
(pg. 
64
-
68
)
13
Singh
J.P.
Larson
M.G.
O'Donnell
C.J.
Tsuji
H.
Evans
J.C.
Levy
D.
Heritability of heart rate variability: the Framingham Heart Study
Circulation
 , 
1999
, vol. 
99
 (pg. 
2251
-
2254
)
14
Holm
H.
Gudbjartsson
D.F.
Arnar
D.O.
Thorleifsson
G.
Thorgeirsson
G.
Stefansdottir
H.
Gudjonsson
S.A.
Jonasdottir
A.
Mathiesen
E.B.
Njolstad
I.
, et al.  . 
Several common variants modulate heart rate, PR interval and QRS duration
Nat. Genet.
 , 
2010
, vol. 
42
 (pg. 
117
-
122
)
15
Pfeufer
A.
van Noord
C.
Marciante
K.D.
Arking
D.E.
Larson
M.G.
Smith
A.V.
Tarasov
K.V.
Muller
M.
Sotoodehnia
N.
Sinner
M.F.
, et al.  . 
Genome-wide association study of PR interval
Nat. Genet.
 , 
2010
, vol. 
42
 (pg. 
153
-
159
)
16
Sotoodehnia
N.
Isaacs
A.
de Bakker
P.I.
Dorr
M.
Newton-Cheh
C.
Nolte
I.M.
van der Harst
P.
Muller
M.
Eijgelsheim
M.
Alonso
A.
, et al.  . 
Common variants in 22 loci are associated with QRS duration and cardiac ventricular conduction
Nat. Genet.
 , 
2010
, vol. 
42
 (pg. 
1068
-
1076
)
17
Newton-Cheh
C.
Eijgelsheim
M.
Rice
K.M.
de Bakker
P.I.
Yin
X.
Estrada
K.
Bis
J.C.
Marciante
K.
Rivadeneira
F.
Noseworthy
P.A.
, et al.  . 
Common variants at ten loci influence QT interval duration in the QTGEN Study
Nat. Genet.
 , 
2009
, vol. 
41
 (pg. 
399
-
406
)
18
Pfeufer
A.
Sanna
S.
Arking
D.E.
Muller
M.
Gateva
V.
Fuchsberger
C.
Ehret
G.B.
Orru
M.
Pattaro
C.
Kottgen
A.
, et al.  . 
Common variants at ten loci modulate the QT interval duration in the QTSCD Study
Nat. Genet.
 , 
2009
, vol. 
41
 (pg. 
407
-
414
)
19
Kim
J.W.
Hong
K.W.
Go
M.J.
Kim
S.S.
Tabara
Y.
Kita
Y.
Tanigawa
T.
Cho
Y.S.
Han
B.G.
Oh
B.
A common variant in SLC8A1 is associated with the duration of the electrocardiographic QT interval
Am. J. Hum. Genet.
 , 
2012
, vol. 
91
 (pg. 
180
-
184
)
20
Borensztein
M.
Viengchareun
S.
Montarras
D.
Journot
L.
Binart
N.
Lombes
M.
Dandolo
L.
Double Myod and Igf2 inactivation promotes brown adipose tissue development by increasing Prdm16 expression
FASEB J.
 , 
2012
, vol. 
26
 (pg. 
4584
-
4591
)
21
Mochizuki
N.
Shimizu
S.
Nagasawa
T.
Tanaka
H.
Taniwaki
M.
Yokota
J.
Morishita
K.
A novel gene, MEL1, mapped to 1p36.3 is highly homologous to the MDS1/EVI1 gene and is transcriptionally activated in t(1;3)(p36;q21)-positive leukemia cells
Blood
 , 
2000
, vol. 
96
 (pg. 
3209
-
3214
)
22
Arndt
A.K.
Schafer
S.
Drenckhahn
J.D.
Sabeh
M.K.
Plovie
E.R.
Caliebe
A.
Klopocki
E.
Musso
G.
Werdich
A.A.
Kalwa
H.
, et al.  . 
Fine mapping of the 1p36 deletion syndrome identifies mutation of PRDM16 as a cause of cardiomyopathy
Am. J. Hum. Genet.
 , 
2013
, vol. 
93
 (pg. 
67
-
77
)
23
Davis
C.A.
Haberland
M.
Arnold
M.A.
Sutherland
L.B.
McDonald
O.G.
Richardson
J.A.
Childs
G.
Harris
S.
Owens
G.K.
Olson
E.N.
PRISM/PRDM6, a transcriptional repressor that promotes the proliferative gene program in smooth muscle cells
Mol. Cell. Biol.
 , 
2006
, vol. 
26
 (pg. 
2626
-
2636
)
24
Gaal
E.I.
Salo
P.
Kristiansson
K.
Rehnstrom
K.
Kettunen
J.
Sarin
A.P.
Niemela
M.
Jula
A.
Raitakari
O.T.
Lehtimaki
T.
, et al.  . 
Intracranial aneurysm risk locus 5q23.2 is associated with elevated systolic blood pressure
PLoS Genet.
 , 
2012
, vol. 
8
 pg. 
e1002563
 
25
Gewies
A.
Castineiras-Vilarino
M.
Ferch
U.
Jahrling
N.
Heinrich
K.
Hoeckendorf
U.
Przemeck
G.K.
Munding
M.
Gross
O.
Schroeder
T.
, et al.  . 
Prdm6 is essential for cardiovascular development in vivo
PLoS One
 , 
2013
, vol. 
8
 pg. 
e81833
 
26
Bers
D.M.
Despa
S.
Na+ transport in cardiac myocytes; implications for excitation-contraction coupling
IUBMB Life
 , 
2009
, vol. 
61
 (pg. 
215
-
221
)
27
Wakimoto
K.
Kobayashi
K.
Kuro
O.M.
Yao
A.
Iwamoto
T.
Yanaka
N.
Kita
S.
Nishida
A.
Azuma
S.
Toyoda
Y.
, et al.  . 
Targeted disruption of Na+/Ca2+ exchanger gene leads to cardiomyocyte apoptosis and defects in heartbeat
J. Biol. Chem.
 , 
2000
, vol. 
275
 (pg. 
36991
-
36998
)
28
Cho
Y.S.
Go
M.J.
Kim
Y.J.
Heo
J.Y.
Oh
J.H.
Ban
H.J.
Yoon
D.
Lee
M.H.
Kim
D.J.
Park
M.
, et al.  . 
A large-scale genome-wide association study of Asian populations uncovers genetic factors influencing eight quantitative traits
Nat. Genet.
 , 
2009
, vol. 
41
 (pg. 
527
-
534
)
29
Marchini
J.
Howie
B.
Myers
S.
McVean
G.
Donnelly
P.
A new multipoint method for genome-wide association studies by imputation of genotypes
Nat. Genet.
 , 
2007
, vol. 
39
 (pg. 
906
-
913
)
30
Purcell
S.
Neale
B.
Todd-Brown
K.
Thomas
L.
Ferreira
M.A.
Bender
D.
Maller
J.
Sklar
P.
de Bakker
P.I.
Daly
M.J.
, et al.  . 
PLINK: a tool set for whole-genome association and population-based linkage analyses
Am. J. Hum. Genet.
 , 
2007
, vol. 
81
 (pg. 
559
-
575
)
31
Price
A.L.
Patterson
N.J.
Plenge
R.M.
Weinblatt
M.E.
Shadick
N.A.
Reich
D.
Principal components analysis corrects for stratification in genome-wide association studies
Nat. Genet.
 , 
2006
, vol. 
38
 (pg. 
904
-
909
)
32
Li
Y.
Willer
C.J.
Ding
J.
Scheet
P.
Abecasis
G.R.
MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes
Genet. Epidemiol.
 , 
2010
, vol. 
34
 (pg. 
816
-
834
)
33
Ioannidis
J.P.
Patsopoulos
N.A.
Evangelou
E.
Heterogeneity in meta-analyses of genome-wide association investigations
PLoS One
 , 
2007
, vol. 
2
 pg. 
e841
 
34
Johnson
A.D.
Handsaker
R.E.
Pulit
S.L.
Nizzari
M.M.
O'Donnell
C.J.
de Bakker
P.I.
SNAP: a web-based tool for identification and annotation of proxy SNPs using HapMap
Bioinformatics
 , 
2008
, vol. 
24
 (pg. 
2938
-
2939
)
35
Boyle
A.P.
Hong
E.L.
Hariharan
M.
Cheng
Y.
Schaub
M.A.
Kasowski
M.
Karczewski
K.J.
Park
J.
Hitz
B.C.
Weng
S.
, et al.  . 
Annotation of functional variation in personal genomes using RegulomeDB
Genome Res.
 , 
2012
, vol. 
22
 (pg. 
1790
-
1797
)

Author notes

Full list of the Japanese study group is given in Appendix.
K.-W.H. and J.E.L. contributed equally to this work.

Supplementary data