Introduction

In 2000, a number of national cooperative projects were started in Japan under the banner of ‘Millennium Projects’. These projects aimed to achieve bold technological innovation in three areas of vital importance to Japan: information utilization, aging society and the environment. Among them, genome research in five fields of genetics, namely disease genes, human genome variation, the rice genome, bioinformatics, and development/differentiation/regeneration, was collectively termed the ‘Millennium Genome Project (MGP)’. This project was launched to shed light on some of the unsolved challenges of the aging society by gaining an understanding of disease pathogenesis and by establishing genome-based personalized medical care (Figure 1).1 In the Disease Genes division, the five most frequent diseases in developed countries that have a harmful impact on lifelong health and quality-of-life were selected, namely essential hypertension, diabetes, cancer, asthma and Alzheimer's disease. Among the study groups for these five target diseases, the group for hypertension consists of 16 researchers whose focus is the genetic aspects of essential hypertension (Figure 2). These researchers were selected from among scientists with established achievements in genetics and/or epidemiology of hypertension. The aim of this review is to summarize the results of gene hunting conducted by the study group for hypertension.

Figure 1
figure 1

Overview of the MGP.

Full size image
Figure 2
figure 2

Investigators in the MGP for hypertension.

Full size image

Study overview

Two antithetical but complementary approaches were adopted to identify common variants associated with individual blood pressure susceptibility, a genetic statistics-based genome-wide approach and a knowledge-based candidate gene approach (Figure 3). A multi-stage genome-wide association study (GWAS) using approximately 80 000 gene-associated single-nucleotide polymorphisms (SNPs) was performed as a five study group initiative as the flagship study of the Disease Genes division. The Japanese-oriented SNPs (JSNPs) analyzed in this GWAS2, 3 and the high-throughput low-cost genotyping system based on the Invader technology were both developed by the Human Genome Variation division.

Figure 3
figure 3

Overall study design of the MGP for hypertension.

Full size image

The study group for hypertension undertook a second GWAS using approximately 20 000 microsatellite (MS) markers with an average spacing of 146 kb. Although many studies have used commercially available MS marker sets consisting of <1000 markers to narrow down candidate gene loci, a Japanese research group has successfully identified 30 000 MS marker sets by systemic whole-genome screening and developed a high-throughput genotyping system based on the DNA-pooling method.4

The candidate gene approach was concomitantly designed and performed by each of the study groups. The study group for hypertension performed a systemic multiple candidate gene analysis, with candidate gene selection made from among genes related to signal transduction pathways, including receptors, soluble carrier proteins, binding proteins, channels, enzymes and G-proteins that were possibly related to blood pressure regulation.

GWAS using SNP markers

To overcome financial and technological limitations, mostly owing to a lack of DNA chip technology, the Disease Genes division launched a coordinated two-stage-screening approach (Figure 4).5 The theoretical basis of this exploratory test scheme has been reported elsewhere.6 In the first stage, a total of 940 samples from 188 patients with one of the five target diseases were simultaneously genotyped, and the association analysis was performed using the case samples of the other four study groups as a pseudo control. Allele frequency data for 752 individuals from a general Japanese population established by the JSNP project were also used as a second control. Both this exploratory analysis as well as the second-stage analysis used cases of men with very severe hypertension gathered by the nationwide MGP collaboration. Among the 80 795 SNPs initially genotyped, the 2676 top-hit SNPs that showed an odds ratio (OR) >1.4 with a P-value <0.015 in at least one test comparison of allele frequency or genotype distribution were analyzed in the next stage analysis. A total of 75 SNPs that showed positive associations in the second-stage analysis, namely a P-value <0.01 for genotype distribution and <0.05 for allele frequency on comparison of 940 cases (188 first-stage cases and 752 second-stage cases) and 752 controls, were further analyzed in a second panel of 619 hypertensive subjects and 1406 normotensive controls. After the third-stage analysis, which was the study group’s own extensive analysis, six SNPs located in four genes, ADD2, EYA2, KIAA0789 and M6PR, were finally explored.

Figure 4
figure 4

Schematic presentation of the three-stageSNP-based GWAS strategy.

Full size image

Identification of the ADD2 gene as a candidate locus for BP regulation is particularly noteworthy, because β-adducin, which is encoded by the ADD2 gene, has been suggested to have a role in a physiological pathway of BP regulation.7 Adducin is a heterodimeric cytoskeleton protein composed of α-, β- and γ-subunits, and is highly conserved across species. Point mutations on the Add1 and Add2 genes account for the 50% BP difference between Milan hypertensive and normotensive strains via the modulation of Na+–K+ activity.8, 9 In humans, association of the ADD1 G460W polymorphism (rs4961) with hypertension has been widely investigated in various populations7 under the hypothesis that the ADD1 W allele accelerates Na+–K+ co-transport,10 which in turn results in increased renal sodium reabsorption. As for the ADD2 gene, several articles have reported that this may confer susceptibility in association analyses.11, 12 A linkage analysis also reported that the 2p13 region in ADD2 was also a hypertension-related locus.13 However, a recent large-scale GWAS in the European population,14 as well as our study in East Asians,15 found no SNPs of genome-wide significance near the ADD2 gene, or the other three identified genes, namely EYA2, KIAA0789 (renamed to WSCD2) and M6PR. Statistical power of previous GWAS was not enough to detect SNPs with smaller effect sizes. As larger GWAS samples can detect larger numbers of common susceptibility variants with smaller effects, increasing sample size may provide more convincing evidence.

GWAS using MS markers

A total of 385 hypertensive patients and 385 normotensive control subjects were analyzed in the three-stage screening using 18 977 MS markers (Figure 3).16 A portionof the hypertensive cases analyzed in the first and second screenings were shared with the SNP-based GWAS. To bring down the cost and technical burden of genotyping 1000 MS markers, individual DNA samples of each stage were pooled within cases and controls. The pooled DNA was then amplified using the 18 977 MS markers, and the PCR products were genotyped by a standard protocol using capillary DNA sequencers. This DNA-pooling method enabled us to obtain the allele frequencies of MS markers in pooled Japanese individuals by measuring the height of multiple peaks. An association analysis was performed by comparing the height of peaks between the case and controls. P-values <0.05 were considered significant in all stage analyses to avoid potential false negatives.

After the three-stage analysis, 54 markers, which showed P-values <0.05 in either of the 2 × 2 (allelic) or 2 × m (genotype, m=number of alleles) analysis, were initially identified as potential susceptible loci for hypertension. These markers were again genotyped in all 770 individual samples (not pooled) used in the first- to third-stage analyses (385 cases and 385 controls), and finally 19 loci were identified. Among these 19 loci, three chromosomal locations, 6q27, 2p25.1 and 2p11.1-q12.3, overlapped with a region identified in other linkage17, 18 or admixture mapping studies.19 Although these previous studies did not permit sufficient narrowing down of the region to allow a candidate gene to be identified, our dense mapping did allow the former two markers to be dropped into the SMOC2 and LPIN1 gene regions, respectively. Although the SMOC2 gene encodes a matricellular protein that promotes matrix assembly and can stimulate endothelial cell proliferation and migration, an association between this gene and hypertension has not been suggested. Further, LPIN1 is recognized as a candidate gene for human lipodystrophy, which is characterized by a loss of body fat, fatty liver, hypertriglyceridemia and insulin resistance. Although one study reported replication data which suggested that one of the 12 tagged SNPs in the LPIN1 gene region was a susceptibility locus for hypertension in Chinese men,20 another study failed to replicate the associations.21

Systematic candidate gene screening

Candidate gene analysis is the classic approach to find susceptible genotypes. Previous studies selected several candidate genes on the basis of known biochemical or physiologic components related to BP regulation. The results of previous studies which investigated a single or only a few arbitrarily selected genes have not been consistent, and positive findings have rarely been replicated. As hypertension is a polygenic disease, alleles at many uninvestigated loci would have contributed to the ultimate disease trait. We therefore conducted a systematic candidate gene screening by evaluating 307 genes selected from those related to signal transduction, binding proteins, channels, enzymes and G-proteins that were possibly related to BP regulation.22, 23

The first screening was conducted using a case–control panel with 758 severe young-onset hypertensive patients and 726 strictly normotensive elderly persons enrolled from four independent regions throughout Japan (Table 1). One SNP per one gene, preferably located in the promoter region or exons with the highest minor allele frequency in the Japanese population, was selected based on the information published in the JSNP database (http://snp.ims.u-tokyo.ac.jp). Of the 307 SNPs analyzed, 38 showed P-values <0.05 in either analysis comparing allele frequency or genotype distribution, including a dominant or recessive model. To further identify susceptible SNPs, replication genotyping was performed using an independent nested population of 1929 cases and 1993 controls chosen from the approximately 14 000 cohort samples (Table 1). This cohort panel consist of six independent population- or company employee-based genetic epidemiological cohorts established by several of the members of the study group for hypertension. Results repeatedly showed the association of ATP2B1 rs2070759 with hypertension (OR=1.18, P=4.0 × 10−4). The subsequent dense SNP analysis around the 150 kb region of the ATP2B1 gene identified another significant SNP at the promoter region (rs11105378: OR=1.31, P=4.1 × 10−11), and the BP susceptibility of this SNP was replicated in the whole cohort sample analysis. Further, a trans-ethnic analysis using the Global BPgen data set,24 a large-scale GWAS samples of European descent, confirmed that the ATP2B1 gene was a susceptibility locus for hypertension (OR=1.13, P=5.9 × 10−4).

Table 1 Study samples used in candidate gene screening
Full size table

The per-allele effect size of rs11105378 for systolic BP trait was 1.33 mm Hg (P=1.5 × 10−8) in the Japanese cohort sample. Meta-analysis with samples of the Global BPgen (0.59 mm Hg, P=0.001) and the CHARGE25 (1.31 mm Hg, P=9.1 × 10−11) consortia allowed more precise estimation, at 1.02 mm Hg for systolic BP (P=1.4 × 10−18) and 0.54 mm Hg for diastolic BP (P=1.6 × 10−13). Further, combined analysis with other susceptible SNPs identified by recent GWAS in samples of European origin, namely FGF5 rs1458038, CYP17A1 rs1004467 and CSK rs1378942, showed a stepwise association with BP traits and hypertension risk. Individuals carrying seven or eight risk genotypes had higher systolic/diastolic BP (136.2/81.5 mm Hg), in contrast to the lower BP of individuals with only two or less-risk genotypes (125.3/57.5 mm Hg) (Figure 5). The OR of the high-risk group (OR=1.43, P=1.0 × 10−4) was 2.27 compared with the lowest risk group(OR=0.63, P=0.020).

Figure 5
figure 5

Adjusted OR for hypertension and systolic blood pressure by the number of risk genotypes. Number of risk genotypes was calculated by the following four SNPs: ATP2B1 rs1105378, FGF5 rs1458038, CYP17A1, rs1004467 and CSK rs1378942. Age, age,2 sex, BMI and cohort variable-adjusted OR and systolic BP (line graph) is shown in the panel. Number of subjects in each group is represented as a bar graph.

Full size image

In the SNP-based MGP-GWAS, two SNPs located in ATP2B1 gene region, namely rs2070758 (TT/TG/GG=153/34/1) and rs2070759 (AA/AC/CC=49/102/35), were genotyped. However, in the association analysis with752 pseudo control samples, the P-value did not reach statistical significance (allele frequency; P=0.460 and P=0.341, respectively). MS markers located 2.4 kb upstream of the ATP2B1 gene also did not show a positive association with hypertension. The most plausible reason for the failure to identify ATP2B1 in the MGP-GWAS may be insufficient statistical power owing to a limited number of subjects. The susceptibility conferred by the ATP2B1 gene is therefore clear, given the negative results of the MGP-GWAS.

Replication results of the BP susceptibility of ATP2B1

GWAS in people of European descent25 and Koreans26 also reported that the ATP2B1 gene variants are the most significant determinants of BP traits and hypertension. The same results were also observed in another Japanese population.27 A meta-analysis of the East-Asian GWAS data28 further supported that the association was strongest for ATP2B1, except for the ALDH2 rs671 genotype, which determines alcohol intolerance in East-Asians. A very recently published world-wide study for BP genetics29 based on 270 000 persons in various populations, including our MGP data sets, provided conclusive evidence on this issue.

Mechanisms of the association between ATP2B1 and BP

The ATP2B1 (so-called PMCA1) gene encodes the plasma membrane calcium ATPase isoform 1, which removes bivalent calcium ions from eukaryotic cells against very large concentration gradients and has a critical role in intracellular calcium homeostasis. Although the pathophysiological implications of ATP2B1 gene products in the development of hypertension are uncertain, we obtained an important clue for the mechanism from an ex vivo expression analysis of ATP2B1 mRNA. ATP2B1 mRNA expression in umbilical artery smooth muscle cells differed among rs11105738 genotypes, and cells carrying the risk allele showed significantly lower levels of mRNA.23

In mammals, calcium ATPase isoforms are encoded by at least four separate genes (ATP2B1–ATP2B4).30 Although we examined the possible association of ATP2B4 gene polymorphisms with hypertension, the most promising of the four isoforms, we found no significant SNPs around the region, suggesting that the association of the plasma membrane Ca2+ pump with BP regulation is isoform-specific.

We recently reported31 that the trimetric intracellular cation A (TRIC-A) channel knockout mouse showed markedly higher BP, and that several SNPs locating around the TRIC-A gene were significantly associated with hypertension in humans. TRIC channels are intracellular monovalent cation channels, which are postulated to mediate counter-ion movements facilitating physiological Ca2+ release from internal stores. It is particularly interesting that the two genes, which were identified by the independent two antithetical approaches of non-hypothesis statistic-based analysis (ATP2B1) and hypothesis-based analysis (TRIC-A), are molecules that play important roles in Ca2+ homeostasis (Figure 6). Dense SNP analysis or re-sequencing of other molecules relating to Ca2+ transport may further elucidate genetic factors for BP regulation.

Figure 6
figure 6

Molecules associated with Ca2+ movement at vascular smooth muscle cells. Cav, L-type Ca2+ channel; BK Ch, BK channel; IP3R, inositol trisphosphosphate receptor; RyR, ryanodine receptor.

Full size image

Perspectives

Hypertension is one of the most prevalent complex genetic disorders, for which the genetic heritability has been estimated at up to 60%. Owing to its large impact on a number of cardiovascular diseases, hypertension is a major contributor to the global health burden. Identification of genetic variants associated with BP regulation is important in two aspects, recognition of genetically high-risk populations for effective prevention and early diagnosis and treatment, and elucidation of the molecular pathogenesis to identify target molecules for diagnosis, treatment and prevention. A series of genetic studies in the MGP for hypertension and recent GWAS in various populations identified several genes responsible for BP regulation. However, the effect size of each implicated SNP was small (<1 mm Hg), and the combination of all identified loci accounted for <10% of overall BP variation. As GWAS are firstly designed to identify common genetic variants for multifactorial diseases and have limited ability to detect less common or rare variants, missing heritability may be traced by analyzing the uncommon variants using a million chip based on the 1000G project data (www.1000genomes.org) or a next generation sequencer. BP is a phenotype that widely fluctuates as a result of non-genetic factors. Accurate BP measurement by ambulatory monitoring or home-measurement may also be indispensable for unlocking the genetic basis of hypertension.