Target Sequencing, Cell Experiments, and a Population Study Establish Endothelial Nitric Oxide Synthase (eNOS) Gene as Hypertension Susceptibility GeneNovelty and Significance
A case–control study revealed association between hypertension and rs3918226 in the endothelial nitric oxide synthase (eNOS) gene promoter (minor/major allele, T/C allele). We aimed at substantiating these preliminary findings by target sequencing, cell experiments, and a population study. We sequenced the 140-kb genomic area encompassing the eNOS gene. In HeLa and HEK293T cells transfected with the eNOS promoter carrying either the T or the C allele, we quantified transcription by luciferase assay. In 2722 randomly recruited Europeans (53.0% women; mean age 40.1 years), we studied blood pressure change and incidence of hypertension in relation to rs3918226, using multivariable-adjusted models. Sequencing confirmed rs3918226, a binding site of E-twenty six transcription factors, as the single nucleotide polymorphism most closely associated with hypertension. In T compared with C transfected cells, eNOS promoter activity was from 20% to 40% (P<0.01) lower. In the population, systolic/diastolic blood pressure increased over 7.6 years (median) by 9.7/6.8 mm Hg in 28 TT homozygotes and by 3.8/1.9 mm Hg in 2694 C allele carriers (P≤0.0004). The blood pressure rise was 5.9 mm Hg systolic (confidence interval [CI], 0.6–11.1; P=0.028) and 4.8 mm Hg diastolic (CI, 1.5–8.2; P=0.0046) greater in TT homozygotes, with no differences between the CT and CC genotypes (P≥0.90). Among 2013 participants normotensive at baseline, 692 (34.4%) developed hypertension. The hazard ratio and attributable risk associated with TT homozygosity were 2.04 (CI, 1.24–3.37; P=0.0054) and 51.0%, respectively. In conclusion, rs3918226 in the eNOS promoter tags a hypertension susceptibility locus, TT homozygosity being associated with lesser transcription and higher risk of hypertension.
- blood pressure
- endothelial nitric oxide synthase gene
- population science
- target sequencing
Hypertension is a chronic age-related disease influenced by a large number of genetic and environmental factors, lifestyle, and their interaction.1 Hypertension affects an estimated 25% to 35% of the world’s population and >60% of the elderly.1–3 In 2001, hypertension caused 8 million deaths worldwide, representing 14% of global mortality.4 High blood pressure (BP) is the main driver of ischemic heart disease and stroke3,4 and substantially exceeds the contribution of the 2 other main modifiable risk factors, hypercholesterolemia and smoking, to the global burden of noncommunicable disease.4 Several genome-wide association studies (GWAS) identified a number of single-nucleotide polymorphisms (SNPs), all with a small effect on BP.5,6 In view of the impact of BP as a continuous risk factor, small genetic effects might entail substantial effects on morbidity and mortality.1,3
Using GWAS in a case–control design,7 we recently identified rs3918226 as a new hypertension susceptibility locus. This locus lays in the promoter of the endothelial nitric oxide synthase (eNOS) gene, which encodes the enzyme that produces nitric oxide, a strong vasodilator with a key role in the regulation of systemic vascular resistance. GWAS usually points to genomic regions of interest in relation to a trait, but seldom directly identifies the causal or functional variant. In the present study, we aimed at consolidating the role of eNOS as a hypertension susceptibility gene by fine mapping the DNA sequence tagged by rs3918226, by studying the transcriptional functionality of the rs3918226 alleles in vitro, and by relating the change in BP over time to rs3918226 in a randomly recruited population sample.
From the HYPERGENES study,7 we selected 44 hypertensive patients carrying ≥1 T allele and 48 healthy controls homozygous for the C allele. Analyses of the genetic data confirmed that all patients and controls were of continental Italian descent. We sequenced a 140-kb DNA region of chromosome,7 which, in addition to eNOS, included KCNH2 mapping upstream and 6 genes mapping downstream: ATG9B, ABCB8, ACCN3, CDK5, SLC4A2, and FASTK. Detailed information on the DNA sequencing methods and an accompanying glossary are available in the online-only Data Supplement. We sequenced indexed and multiplexed samples in a paired-end protocol implemented on an Illumina GAIIX platform (Illumina Inc, San Diego, CA).
Paired-end raw reads were checked for their quality using FastQC, version 0.10.0 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc) and PrinSeq, version 0.19.4 (http://edwards.sdsu.edu/cgi-bin/prinseq/prinseq.cgi). Reads were aligned to the human reference genome (hg19; UCSC assembly; February 2009) using BWA. SAMtools, Picard (http://picard.sourceforge.net), and Genome Analysis Toolkit (GATK) were used to handle the reads and for postalignment quality control checks. Multisamples variant call was performed by GATK, and quality filters were applied to variant call. Only variants with high SNP quality score were evaluated and annotated with Annovar software for type and impact on the gene product (http://www.openbioinformatics.org/annovar).
We imputed missing genotypes based on the HYPERGENES results with MiniMac, a low memory, computationally efficient implementation of the MaCH algorithm,8,9 using as reference panel the 1000 Genome Database, released in June 2011. We tested imputed SNPs with high quality (mean R2, 0.91; SD, ±0.11) for association with hypertension, using logistic regression as implemented in Mach2dat8,9 with adjustments applied for sex and the first 10 significant principal components.
Luciferase Reporter Assays
We obtained the pGL2-eNOS promoter-luciferase plasmid, carrying the C allele of rs3918226, from Addgene (plasmid 19297; http://www.addgene.org) and constructed the T allele by site-directed mutagenesis (see Expanded Methods available in online-only Data Supplement). We transfected HeLa cells in 4 independent experiments and each construct was tested in triplicate. HEK293T cells were transfected in 3 independent experiments and each construct was tested in duplicate. We compared luciferase reporter activities between the C and T alleles by Student t test.
Recruitment for the Flemish Study on Environment, Genes, and Health Outcomes (FLEMENGHO) started in 1985.10,11 From August 1985 to November 1990, a random sample of the households living in a geographically defined area of Northern Belgium was investigated with the goal to recruit an equal number of participants in each of 6 strata by sex and age (20–39, 40–59, and ≥60 years). All household members aged ≥20 years were invited, provided that the quota of their sex–age group had not yet been satisfied. From June 1996 until January 2004, recruitment of families continued using the former participants (1985–1990) as index persons and also including teenagers. The participants were repeatedly followed up. In all study phases, we used the same standardized methods to measure BP and to administer questionnaires.10,11 The European Project on Genes in Hypertension (EPOGH) recruited participants from 1999 to 2001.11,12 The EPOGH investigators received training at the Studies Coordinating Centre in Leuven, Belgium, and applied the same protocol, questionnaires, and follow-up procedures, as used in FLEMENGHO. Questionnaires were translated from Dutch and English into Czech, Italian, Polish, and Russian and back-translated into Dutch and English to ensure that all questions kept the same meaning in all languages. The last follow-up examination took place from 2005 to 2008 in FLEMENGHO11 and from 2006 to 2008 in EPOGH.11 Both studies complied with the Helsinki Declaration for investigation of human subjects.13 Each local institutional review board approved the study protocol. Participants gave written informed consent.
At baseline and follow-up, experienced observers measured each participant’s anthropometric characteristics and BP and administered the standardized questionnaire to collect information on medical history, smoking and drinking habits, and use of medications, including BP-lowering drugs, contraceptive pill intake, and hormonal replacement therapy. At each contact, BP was the average of 5 consecutive auscultatory readings in the sitting position (see Expanded Methods available in online-only Data Supplement). Digit and number preference was checked at 6-month intervals.12 Hypertension was an untreated BP of ≥140 mm Hg systolic, or 90 mm Hg diastolic, or use of antihypertensive drugs. For adolescents (n=33), we used the thresholds specified by the European Society of Hypertension, which are stratified by sex, age, and height percentiles.14 Body mass index was weight in kilograms divided by the square of height in meters.
Datasets for Analysis
From 3785 subjects who initially agreed to participate in FLEMENGHO (n=2593) and EPOGH (n=1192), 53 participants were excluded because the baseline BP measurements were missing, leaving 3732 subjects with a full set of required baseline measurements. Of these, 2981 subjects participated in ≥1 follow-up examination. We additionally excluded 259 participants from analysis because the BP measurements were missing (n=32) or their DNA was of bad quality (n=227). Thus, the blood pressure cohort used to study change in BP included 2722 participants (Figure 1). Changes in BP during follow-up were calculated as the last minus the baseline BP. The hypertension cohort used to study the incidence of hypertension encompassed 2013 participants, who were normotensive at baseline (Figure 1). We censored subjects from further analysis after occurrence of the first diagnosis of hypertension.
For database management and statistical analysis, we used SAS software, version 9.3 (SAS Institute Inc, Cary, NC). Between-group comparisons of means, medians, proportions, and Kaplan–Meier survival functions relied on the standard normal z test or ANOVA, Kruskal–Wallis ANOVA or Fisher exact test, and the log-rank test, respectively. We applied McNemar test to evaluate changes over time in categorical variables. We computed 95% confidence intervals (CIs) of rates as R±1.96×√(R/T), where R and T are the rate and the denominator used to calculate the rate.
We studied the association between change in BP and the rs39188226 genotypes, using mixed models with indicator variables (0,1). Multivariable-adjusted models with BP changes as dependent variables accounted for cohort, sex, age, baseline BP, follow-up duration, baseline and follow-up body mass index, intake of female sex hormones or nonsteroidal antiinflammatory drugs at baseline and follow-up, and 3 indicator variables coding for antihypertensive drug intake (starting or stopping treatment between baseline and follow-up and remaining on treatment). A sensitivity analysis additionally accounted for family clusters modeled as a random effect.
To study the incidence of hypertension, we applied Cox regression adjusted for the same covariables as in the continuous analyses. We checked the proportional hazards assumption by the Kolmogorov supremum test. To account for family clusters, we used the PROC SURVIVAL procedure of the SUDAAN 10.01 software (Research Triangle Institute, NC). We computed the positive predictive value of TT homozygosity as (R×D)/([G/100]×[R–1]+1), where R is the multivariable-adjusted hazard ratio, D is the incidence of hypertension in the whole population (34.5%), and G is the prevalence of TT homozygosity (1.09%).15 The attributable risk is given by ([R–1]×100)/R and the population-attributable risk by ([G/100]×[R–1]×100)/([G/100]×[R–1]+1).15
In 44 hypertensive patients and 48 healthy controls, we captured 91% (SD, 6) of the targeted genomic region with a 63-fold amplification above the genomic background. In each sample, the region of interest was covered on average 20 times with mapping and base quality scores of ≥20 and 17, respectively.
We identified 338 variants, of which 15 were nonsynonymous; 23 synonymous; 23 and 8 in the 3′ UTR and 5′ UTR, respectively; 198 intronic; and 71 intergenic. Among the 338 variants, 277 were already annotated in dbSNP135 (Table S1 in the online-only Data Supplement) and 61 were novel variants (Table S2). We had genotyped 76 of 277 annotated SNPs in the HYPERGENES GWAS, using the Illumina 1M array.7 The genotype concordance rate between the 2 technologies was high (r2=≈0.988). Of the 61 novel SNPs, 55 heterozygous variants were rare, only present in a single subject. Table S3 lists the 6 other novel variants. Of these, 3 were intronic in KCNH2 (1 homozygous in 1 subject and 2 other heterozygous in 2 and 4 individuals). One variant was intergenic, mapping ≈3 kb from KCNH2 and 9.5 kb from eNOS (3 heterozygotes). One was intronic in eNOS (2 heterozygotes). One was intronic in ABCB8 (9 heterozygotes and 1 homozygote).
The haplotype analysis appears in page S5 and Figure S1. Five novel variants were located in a region of linkage disequilibrium upstream of rs3918226. Their annotation did not suggest any functional role (Table S3). We, therefore, did not consider them in further analyses. Variants imputed in the entire HYPERGENES study population with a probability value of ≤10–3 appear in Table S4. rs3918226 remained the SNP most closely associated with hypertension.
Luciferase Reporter Assays
SNP rs3918226 is located in the promoter region of eNOS. Compared with the C allele, the risk-carrying T allele was associated with lower transcriptional activity of the eNOS gene ranging from ≈20% (P<0.0001) when tested in HeLa cells (Figure 2A) to ≈40% (P<0.01) in HEK293T cells (Figure 2B).
Characteristics of the Participants
The Table lists the characteristics of the participants by cohort and rs3918226 genotype. The blood pressure cohort (n=2722) included 1442 (53.0%) women and 55 (2.0%) diabetic patients. All participants were white Europeans. Age averaged 40.7 years (range 19.5–83.5). At baseline, 709 (26.1%) participants had hypertension, of whom 322 (45.4%) were on antihypertensive drug treatment. The hypertension cohort consisted of 2013 participants, who were normotensive at baseline (Table). In 3 groups of unrelated participants (n=717), randomly selected from the blood pressure cohort, the frequencies of the rs3918226 genotypes did not deviate from Hardy–Weinberg equilibrium (0.19<P<0.99). The genotype and allele frequencies were equally distributed across cohorts (P=0.67; Table S6).
Cross-sectional Analyses of the Blood Pressure Cohort
At baseline (Table), systolic BP was similar across the rs3918226 genotypes (P=0.42), but TT homozygotes had higher (P=0.030) diastolic BP than C allele carriers. At follow-up (Table), systolic (P=0.012) and diastolic (P=0.0002) BPs were higher in TT homozygotes than in C allele carriers. The differences in baseline and follow-up systolic and diastolic BPs between the CT and CC genotypes were not statistically significant (P≥0.23; Table). With adjustments applied for cohort, sex, age, body mass index, and antihypertensive drug intake, baseline (P=0.035) and follow-up (P=0.0007) diastolic BPs and follow-up systolic BP (P=0.038), but not baseline systolic BP (P=0.48), remained higher in TT homozygotes than in C allele carriers.
Blood Pressure Cohort
Follow-up data were available at 1, 2, or ≥3 occasions in 1269, 459, and 994 participants, respectively. Median follow-up was slightly longer (P=0.053) in TT homozygotes (10.4 years; interquartile range [IQR], 7.2–14.0) than in C allele carriers (7.6 years; IQR, 6.1–12.3; Table). In multivariable-adjusted analyses (see Methods) of the BP changes over follow-up (Figure 3), systolic BP increased 9.7 mm Hg (CI, 4.2–15.1; P=0.0005) in TT homozygotes and by 3.9 mm Hg (CI, 1.8–5.9; P=0.0003), 3.8 mm Hg (CI, 2.1–5.4; P<0.0001), and 3.8 mm Hg (CI, 2.1–5.4; P<0.0001) in CT heterozygotes, CC homozygotes, and C allele carriers, respectively. Diastolic BP increased by 6.8 mm Hg (CI, 3.3–10.3; P<0.0001) in TT homozygotes and by 2.0 mm Hg (CI, 0.6–3.3; P=0.0036), 1.9 mm Hg (CI, 0.9–3.0; P=0.0004), and 1.9 mm Hg (CI, 0.9–3.0; P=0.0004) in CT heterozygotes, CC homozygotes, and C allele carriers, respectively. Systolic BP increased 5.8 mm Hg (CI, 0.4–11.2; P=0.034) and 5.9 mm Hg (CI, 0.7–11.1; P=0.028) more in TT homozygotes than in CT heterozygotes and CC homozygotes. Similarly, diastolic BP increased 4.8 mm Hg (CI, 1.4–8.3; P=0.0062) and 4.8 mm Hg (CI, 1.5–8.2; P=0.0046) more in TT homozygotes than in CT and CC genotype carriers. Furthermore, the BP increases were 5.9 mm Hg systolic (CI, 0.6–11.1; P=0.028) and 4.8 mm Hg diastolic (CI, 1.5–8.2; P=0.0046) greater in TT homozygotes than in C allele carriers. In models additionally accounting for family clusters, the effect sizes were 5.3 mm Hg systolic (CI, –0.5 to 11.2; P=0.071) and 3.1 mm Hg diastolic (CI, –0.6 to 6.8; P=0.097). The interactions between the TT genotype and sex or age were not significant (P≥0.66).
Follow-up data were available at 1, 2, or ≥3 contacts in 874, 351, and 788 participants, respectively. The median duration of follow-up (7.1 years; IQR, 5.5–10.4) was similar (P=0.18) among the rs3918226 genotypes (Table). In the entire cohort, 692 participants developed hypertension, of whom 216 (31.2%) were on antihypertensive drug treatment at the least follow-up visit. In 476 untreated patients (68.8%), the diagnosis of hypertension relied on thresholds being exceeded for systolic or diastolic BP or both in 171 (35.9%), 166 (34.9%), and 139 (29.2%) patients, respectively. The TT genotype conferred a higher risk of hypertension compared with the CT and CC genotypes (P≤0.011) or C allele carriers (P=0.003), with no difference between the C allele–carrying genotypes (P=0.55; Figure 4). Expressed per 1000 person-years of follow-up, incidence rates of hypertension were 40.6 cases (CI, 37.6–43.6) in the entire hypertension cohort, 86.8 (CI, 44.3–129.3) in TT homozygotes, 43.7 (CI, 35.8–51.6) in CT heterozygotes, 39.4 (CI, 36.1–42.7) in CC homozygotes, and 40.1 (CI, 37.1–43.1) in C allele carriers. In multivariable-adjusted Cox regression (Figure 5), the hazard ratios associated with the TT homozygosity were 2.04 (CI, 1.23–3.37; P=0.0056), 2.06 (CI, 1.21–3.50; P=0.0075), and 2.04 (CI, 1.24–3.37; P=0.0054) compared with CT heterozygotes, CC homozygotes, and C allele carriers. In an analysis adjusted for family clusters, the hazard ratio expressing the risk of hypertension in TT homozygotes versus C allele carriers was 2.4 (CI, 1.20–3.46; P=0.0082). All Cox models complied with the proportional hazards assumption. The positive predictive values, attributable risk, and population-attributable risk associated with TT homozygosity were 69.3%, 51.0%, and 1.1%, respectively.
The key finding of the present study was that rs3918226 in the eNOS promoter tags a hypertension susceptibility locus, TT homozygosity being associated with lesser transcription of the gene product and a 2-fold higher risk of hypertension. Our current findings confirm the previously reported HYPERGENES case–control study.7 The discovery phase of this project7 involved 1865 hypertensive patients and 1750 controls, who were genotyped with a Illumina 1M array. The validation study included 1385 cases and 1246 controls, who were genotyped with a 14-K Illumina Infinium custom array. HYPERGENES showed that rs3918226 in the eNOS gene promoter (–665 C>T) tags a hypertension susceptibility locus.7 The odds ratio associated with the T allele was 1.54 (CI, 1.37–1.73; P=2.58×10–13). In a meta-analysis, using both in silico and de novo genotyping data in 21 714 subjects, the odds ratio was 1.34 (CI, 1.25–1.44; P=1.03×10–14).7 In the current study, using BP as a continuous phenotype in a randomly recruited European population sample followed up for 7.6 years (median), systolic and diastolic BPs increased 5.9 and 4.8 mm Hg more in TT homozygotes than in C allele carriers.
Fine mapping the eNOS genomic region in hypertensive patients and healthy controls and imputing approaches in the whole HYPERGENES cohort using 1000 Genome Database released in 2011 as reference further validated rs3918226 as the SNP most closely associated with hypertension. Of 61 novel variants discovered in the genomic area of interest, 55 were singletons with very low minor allele frequency (<0.5%), and 5 were located in a region of linkage disequilibrium and were not functional. The remaining newly discovered variant in the ABCB8 gene located downstream of rs3918226 and was not genotyped in the HYPERGENES sample, because it could not have been tagged by rs3918226 in view of the high recombination rate in this genomic region.
In our initial report,7 we tested whether rs3918226 falls into a regulatory binding site. Using the PATCH algorithm of the TRANSFAC database16 and the TFSEARCH software17 (score 87.37), we characterized a putative binding site for transcription factors of the ETS (E-twenty six) family only 1 nucleotide away from rs3918226. The members of the ETS family, ETS-1 and ELF-1, are present in endothelial cells and are essential for the activation of the eNOS promoter.18 We then hypothesized that rs3918226 maps in an open chromatin region. Indeed, DNaseI and FAIRE (Formaldehyde-Assisted Isolation of Regulatory Elements) experiments in HUVEC (Human Umbilical Vein Endothelial Cells) cells from ENCODE show significant signals around rs3918226 either for DNaseI (P=1.9e−10) or FAIRE (P=5e−7). Moreover, methylation and acetylation histone marks (H3K4ME1 and H3K27Ac) provide signals above the 98th percentile in the same region.
In our current study, we consolidated our previous results7 in transfected HeLa and HEK293T cells showing that the T allele, which is the risk factor for hypertension, is associated with a significant reduction of eNOS transcription compared with the C allele. We hypothesize that this can impair endothelial NO production in vivo. Luizon et al19 reported that the rs3918226 polymorphism does not affect plasma nitrite levels in 181 healthy self-reported blacks. However, Luizon et al’s results are difficult to interpret, because rs3918226, according to HapMap data, is not polymorphic in blacks. We, therefore, presume a substantial admixture with whites in this black study population. The T allele frequency was only 0.04.19 Thus, the low frequency of the risk-conferring T allele and the small sample size probably render a correct estimation of T allele effect on plasma nitrite levels impossible.20
In mammals, NO can be generated by 3 different isoforms of the enzyme NO synthase, referred to as neuronal nNOS (NOS1), inducible iNOS (NOS2) produced by macrophages, and endothelial eNOS (NOS3).21 The human eNOS gene spans 21 kb with 26 exons on chromosome 7q35–q36. Blockade of NO synthesis with inhibitory l-arginine analogues leads to peripheral vasoconstriction and a rise in BP.22–24 Genetically engineered mice with disrupted eNOS are hypertensive and have no endothelium-derived relaxant activity.25 The physiologically most important determinants for the continuous generation of NO and thus the regulation of local blood flow are fluid shear stress and pulsatile stretch.26 NO dilates all types of blood vessels by stimulating soluble guanyl cyclase and increasing the cGMP concentration in smooth muscle cells.22 eNOS is not only a physiological vasodilator but also conveys vascular protection in several ways.21,22 NO released toward the vascular lumen is a potent inhibitor of platelet aggregation and adhesion to the vascular wall and prevents the release of platelet-derived growth factors that stimulate smooth muscle proliferation. NO decreases the expression of chemoattractant protein MCP-1 (Monocyte chemoattractant protein) and of a number of surface adhesion molecules and inhibits leukocyte adhesion to vascular endothelium and leukocyte migration into the vascular wall. This offers protection against the early phases of atherosclerosis.22 The decreased endothelial permeability, the reduced influx of lipoproteins into the vascular wall, and the inhibition of low-density lipoprotein oxidation contribute to the antiatherosclerotic properties of eNOS-derived NO. Finally, NO inhibits DNA synthesis and proliferation of vascular smooth muscle cells as well as smooth muscle cell migration, thereby protecting against the later stages of atherogenesis.22
Given the central role of eNOS in cardiovascular regulation, several previous studies addressed the association between hypertension or cardiovascular disease and genetic variation in eNOS. Niu and Qi27 published a meta-analysis of 3 widely investigated polymorphisms, G894T (rs1799983) in exon 7, 4b/a in intron 4, and T–786C (rs2070744) in the promoter in relation to hypertension, published in English and Chinese. Overall comparison between allele 894T and 894G across all studies (cases/controls: 19 284/26 003) yielded an increased risk of hypertension, amounting to 16% overall, 32% in Asians, and 40% in Chinese. The risk associated with the 4a versus 4b allele was 29% overall and 42% in Asians. For T–786C, ethnicity-stratified analyses suggested that in whites the risk of hypertension was 25% and 69% higher in carriers of the –786C allele and the –786CC genotype, respectively.27 Moreover, the T allele of T–786C polymorphism is a predisposing factor to coronary spasm and reduces the eNOS promoter activity.28
The candidate gene and GWAS studies published so far on hypertension6 identified 47 distinct genetic variants robustly associated with BP, but collectively these variants explained only a few percent of the heritability of BP. HYPERGENES was the first GWAS to identify rs3918226 (C–685T) as a hypertension susceptibility locus.7 An international consortium applied a gene-centric assay in an independent discovery sample of 25 118 individuals that combined hypertensive case–control and general population samples and followed up 10 suggestive SNPs in a further 59 349 individuals.6 An analysis of combined discovery and follow-up data identified rs3918226 (T/C: 0.08/0.92) as being significantly (2.2×10–9) associated with diastolic BP. The effect size per –690T allele was +0.78 mm Hg (SE, 0.21; P=9.5×10–5).6 The recent GWAS of systolic and diastolic BP performed by the International Consortium for Genome-Wide Association Studies (ICBP) used a multi-stage design in 200 000 individuals of European descent.29 Overall, the eNOS region was poorly covered in this study. Genotyping had been performed in most of the cohorts with arrays older that the Illumina 1M, which do not include rs3918226. Moreover, imputation was done using the HapMap panel as reference that does not include rs3918226. In the ICBP data set,29 a SNP mapping 779 bp from rs3918226, rs1800783 (position 150689397), shows a high D′ (1.000) but a low R-sq (0.141) with rs3918226. Because of the low R-sq, allele frequencies are different, and rs1800783 cannot be considered a proxy of rs3918226. In GenHAT,30 the hazard ratio for the primary end point, fatal coronary heart disease, and nonfatal myocardial infarction in T allele versus CC genotype carriers was 1.12 (CI, 1.00–1.26; P=0.048). Conen et al31 analyzed 3 SNPs in the eNOS gene (rs3918226, rs1800779, and rs1799983) in 18 436 white women enrolled in the Women’s Health Study. The participants were all health professionals and normotensive at baseline. BP was self-reported. Over 9.8 years, 29.6% of the women developed hypertension. The hazard ratios for the eNOS polymorphisms were 1.01 (CI, 0.97–1.06), 1.06 (CI, 0.99–1.14), and 1.05 (CI, 1.01–1.09), respectively.31 Progression of BP across 3 increasing categories was not associated with the eNOS polymorphisms, but follow-up for this soft end point was only 4 years. Seidlerová et al32 reported a pilot study examining the association between arterial properties and the rs3918226 polymorphism in 101 untreated volunteers. Among 31 smokers, carriers of the mutated T allele (n=8) had a marginally higher aortic pulse wave velocity (10.0 versus 8.7 m/s; P=0.051) and a higher aortic augmentation index (172 versus 153%; P=0.024). Seidlerová et al32 hypothesized that pending confirmation in a larger study genetic modulation of intermediate arterial phenotypes might lead to higher BP.
Taking into account our current findings and the literature, the C to T substitution at position –690 in the eNOS promoter strongly complies with the Bradford Hill criteria33 as a cause of hypertension. The association is strong,6,7,27 consistent across studies,6,7,27 and specific for hypertension- or hypertension-related complications.6,7,27,30 The current study established temporality and provided a possible mechanism adding to the plausibility. Some studies, but not ours, suggested a dose effect based on the number of T alleles.6,27
We demonstrated that TT homozygosity at the rs3918226 locus in the eNOS gene promoter enhances the age-related increase in BP and increases the risk of hypertension, probably by reducing the transcriptional activity of the eNOS gene. The implications of our current findings span both the prevention and treatment of hypertension and its associated cardiovascular complications. The prevalence of TT homozygosity is low, explaining why the population-attributable risk for hypertension is only 1.1%. However, the attributable risk in TT homozygotes is 51.0%. Combined with other genetic markers, the rs3918226 polymorphism might, therefore, contribute to the stratification of cardiovascular risk. Our findings also support pharmacological interference with the NO signaling pathway. In GenHAT,30 amlodipine, compared with lisinopril, was more effective in the prevention of stroke in minor allele carriers (hazard ratios CT+TT versus CC: 0.49 versus 0.85; P=0.04). Overall, amlodipine reduced systolic BP 1.2 mm Hg more than lisinopril.34 Stroke is the complication of hypertension that is most closely linked to the BP level.1 Moreover, amlodipine enhances endothelial NO availability via stimulation of NO formation35 and by prolonging the NO half-life through antioxidative properties.35,36 Recently developed compounds act downstream in the NO signaling pathway by inhibition of cGMP-specific phosphodiesterase type 537 or by stimulation (haem-dependent) or activation (NO- and haem-independent) of soluble guanylate cyclase activity.38 Further clinical research should establish whether eNOS might be a target for preventive or therapeutic intervention.
Dr Charles J. Lowenstein (University of Rochester Medical Center) provided the pGL2-eNOS promoter-luciferase plasmid. Dr Cédric Howald (University of Lausanne) and ENCODE researchers helped with the open chromatin analysis. Sandra Covens provided expert clerical assistance.
Sources of Funding
The European Union (grant FP7-HEALTH-2007-A-201550), HYPERGENES, and InterOmics (PB05 MIUR-CNR Italian Flagship Project) provided financial support for the genotyping and experimental studies. The European Union (grants IC15-CT98-0329-EPOGH, LSHM-CT-2006-037093-InGenious HyperCare, FP7-HEALTH-2007-A-201550- HYPERGENES, HEALTH-2011.2.4.2-2-EU-MASCARA, HEALTH-F7-305507 HOMAGE, and the European Research Council Advanced Researcher Grant-2011-294713-EPLORE) gave support to the Studies Coordinating Centre, Leuven, Belgium, and the FLEMENGHO and EPOGH population studies. The Fonds voor Wetenschappelijk Onderzoek Vlaanderen, Ministry of the Flemish Community, Brussels, Belgium (G.0734.09, G.0881.13 and G.0880.13), also supported the FLEMENGHO study.
The European Working Party on eNOS is an ad hoc collaboration between investigators involved in the Flemish Study on Environment, Genes, and Health Outcomes (FLEMENGHO), the European Project on Genes in Hypertension (EPOGH), and the HYPERGENES project.
↵* These authors are joint first authors.
↵† These authors are joint senior authors.
This paper was sent to Morris Brown, Guest editor, for review by expert referees, editorial decision, and final disposition.
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.113.01428/-/DC1.
- Received March 31, 2013.
- Revision received May 10, 2013.
- Accepted August 12, 2013.
- © 2013 American Heart Association, Inc.
- Salvi E,
- Kutalik Z,
- Glorioso N,
- et al
- Stolarz-Skrzypek K,
- Kuznetsova T,
- Thijs L,
- Tikhonoff V,
- Seidlerová J,
- Richart T,
- Jin Y,
- Olszanecka A,
- Malyutina S,
- Casiglia E,
- Filipovský J,
- Kawecka-Jaszcz K,
- Nikitin Y,
- Staessen JA
- Matys V,
- Fricke E,
- Geffers R,
- et al
- Akiyama T
- Karantzoulis-Fegaras F,
- Antoniou H,
- Lai SL,
- Kulkarni G,
- D’Abreo C,
- Wong GK,
- Miller TL,
- Chan Y,
- Atkins J,
- Wang Y,
- Marsden PA
- Luizon MR,
- Metzger IF,
- Lacchini R,
- Tanus-Santos JE
- Salvi E,
- Cusi D
- Förstermann U,
- Sessa WC
- Rees DD,
- Palmer RM,
- Moncada S
- Fleming I,
- Busse R
- Nakayama M,
- Yasue H,
- Yoshimura M,
- Shimasaki Y,
- Kugiyama K,
- Ogawa H,
- Motoyama T,
- Saito Y,
- Ogawa Y,
- Miyamoto Y,
- Nakao K
- Zhang X,
- Lynch AI,
- Davis BR,
- Ford CE,
- Boerwinkle E,
- Eckfeldt JH,
- Leiendecker-Foster C,
- Arnett DK
- Bradford-Hill A
- 34.↵The ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs diuretic. The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA. 2003;288:2981–2997.
- Ganafa AA,
- Walton M,
- Eatman D,
- Abukhalaf IK,
- Bayorh MA
- Reffelmann T,
- Kloner RA
- Stasch JP,
- Pacher P,
- Evgenov OV
Novelty and Significance
What Is New?
In the general population, TT homozygosity at the rs3918226 locus (minor/major allele, T/C allele) in the eNOS gene promoter enhances the age-related increase in blood pressure and increases the risk of hypertension. Sequencing confirmed rs3918226, a binding site of E-twenty six transcription factors, as the SNP most closely associated with hypertension. In luciferase reporter assays, the risk-carrying T allele was associated with a 20% to 40% lower transcriptional activity than the C allele.
What Is Relevant?
The prevalence of TT homozygosity is low, explaining why the population-attributable risk for hypertension is only 1.1%. However, the attributable risk in TT homozygotes is 51.0%. Combined with other genetic markers, the rs3918226 polymorphism might, therefore, contribute to the stratification of cardiovascular risk. Further clinical research should establish whether eNOS might be a target for preventive or therapeutic intervention.
rs3918226 in the eNOS promoter tags a hypertension susceptibility locus, TT homozygosity being associated with lesser transcription and higher risk of hypertension.