Skip Navigation


Human Molecular Genetics Advance Access originally published online on November 14, 2007
Human Molecular Genetics 2008 17(4):617-627; doi:10.1093/hmg/ddm335
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
17/4/617    most recent
ddm335v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Kato, N.
Right arrow Articles by Tomoike, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kato, N.
Right arrow Articles by Tomoike, H.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

High-density association study and nomination of susceptibility genes for hypertension in the Japanese National Project

Norihiro Kato1,*, Toshiyuki Miyata3, Yasuharu Tabara4, Tomohiro Katsuya6, Kazuyuki Yanai1, Hironori Hanada3, Kei Kamide3, Jun Nakura5, Katsuhiko Kohara5, Fumihiko Takeuchi2, Hiroyuki Mano7, Michio Yasunami8,9, Akinori Kimura8,9, Yoshikuni Kita10, Hirotsugu Ueshima10, Tomohiro Nakayama11, Masayoshi Soma12, Akira Hata13, Akihiro Fujioka14, Yuhei Kawano3, Kazuwa Nakao15, Akihiro Sekine16, Teruhiko Yoshida17, Yusuke Nakamura16,18, Takao Saruta19, Toshio Ogihara6, Sumio Sugano20, Tetsuro Miki5 and Hitonobu Tomoike3

1 Department of Gene Diagnostics and Therapeutics 2 Department of Medical Ecology and Informatics, Research Institute, International Medical Center of Japan, Tokyo, Japan 3 National Cardiovascular Center, Suita, Osaka, Japan 4 Department of Basic Medical Research and Education 5 Department of Geriatric Medicine, Ehime University Graduate School of Medicine, Toon, Ehime, Japan 6 Department of Geriatric Medicine, Osaka University Graduate School of Medicine, Suita, Osaka, Japan 7 Division of Functional Genomics, Jichi Medical University, Shimotsuke-shi, Tochigi, Japan 8 Department of Molecular Pathogenesis, Medical Research Institute 9 Laboratory of Genome Diversity, School of Biomedical Science, Tokyo Medical and Dental University, Tokyo, Japan 10 Department of Health Science, Shiga University of Medical Science, Otsu, Shiga, Japan 11 Division of Molecular Diagnostics, Advanced Medical Research Center 12 Division of Nephrology and Endocrinology, Department of Internal Medicine, Nihon University School of Medicine, Tokyo, Japan 13 Department of Public Health, Graduate School of Medicine, Chiba University, Chiba, Japan 14 Amagasaki Health Medical Foundation, Amagasaki, Hyogo, Japan 15 Department of Medicine and Clinical Science, Kyoto University Graduate School of Medicine, Kyoto, Japan 16 SNP Research Center, Institute of Physical and Chemical Research, Tokyo, Japan 17 Genetics Division, National Cancer Center Research Institute, Tokyo, Japan 18 Human Genome Center, Institute of Medical Science, University of Tokyo, Tokyo, Japan 19 Department of Internal Medicine, School of Medicine, Keio University, Tokyo, Japan 20 Division of Bioscience, Graduate School of Frontier Sciences, University of Tokyo, Tokyo, Japan

* To whom correspondence should be addressed at: Department of Gene Diagnostics and Therapeutics, Research Institute International Medical Center of Japan, 1-21-1 Toyama, Shinjuku-ku, Tokyo 162-8655, Japan. Tel: +81 332027181; Fax: +81 332027364; Email: nokato{at}ri.imcj.go.jp

Received August 23, 2007; Accepted November 12, 2007


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 FUNDING
 REFERENCES
 
Essential hypertension is one of the most common, complex diseases, of which considerable efforts have been made to unravel the pathophysiological mechanisms. Over the last decade, multiple genome-wide linkage analyses have been conducted using 300–900 microsatellite markers but no single study has yielded definitive evidence for ‘principal’ hypertension susceptibility gene(s). Here, we performed a three-tiered, high-density association study of hypertension, which has been recently made possible. For tier 1, we genotyped 80 795 SNPs distributed throughout the genome in 188 male hypertensive subjects and two general population control groups (752 subjects per group). For tier 2 (752 hypertensive and 752 normotensive subjects), we genotyped a panel of 2676 SNPs selected (odds ratio ≥ 1.4 and P ≤ 0.015 in tier 1) and identified 75 SNPs that showed similar tendency of association in tier 1 and tier 2 samples (P ≤ 0.05 for allele frequency and P ≤ 0.01 for genotype distribution tests). For tier 3 (619 hypertensive and 1406 normotensive subjects), we genotyped the 75 SNPs and found nine SNPs from seven genomic loci to be associated with hypertension (P ≤ 0.05). In three of these loci, the lowest P-values were observed for rs3755351 (P = 1.7 x 10–5) in ADD2, rs3794260 (P = 0.0001) in KIAA0789 and rs1805762 (P = 0.0003) in M6PR when case–control comparison was made in the combined data. An SNP (rs3755351) within ADD2 had the lowest P-value and its experiment-wide significance level is 0.13. Thus, these results have nominated several susceptibility genes for hypertension, and independent replication will clarify their etiological relevance.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 FUNDING
 REFERENCES
 
Essential hypertension (MIM 145500 [OMIM] ) is a multifactorial trait, in which interactions among genetic, environmental and demographic factors are involved. Substantial contribution of genetic factors to the overall disease etiology has been documented by a number of epidemiological studies. For example, family studies controlling for a common environment indicate that blood pressure heritability is in the range of 15–35% (1–3). Accordingly, considerable efforts have been made in the study of molecular genetics of hypertension, but the inherently complex nature has hampered progress in the elucidation of the genes involved (4). Over the last decade, multiple genome-wide linkage analyses have been conducted by using microsatellite markers to localize genes influencing hypertension status and/or blood pressure levels in a number of populations derived from various ethnic groups. Although no single study has so far yielded definitive evidence for ‘principal’ hypertension susceptibility gene(s), some of these studies provide consistency of linkage results in a few chromosomal regions (5–7). It is therefore assumed that multiple genes contribute to the etiology of hypertension independently or synergistically, with each gene exerting small effects under a certain environmental condition.

In parallel with family-based linkage analyses across the entire genome, population-based association studies have been performed, particularly focusing on individual candidate genes to search for genetic influences on hypertension. Association studies for mapping disease-related genes have recently gained popularity over traditional family-based linkage analyses mainly because of their far greater statistical power to detect the presence of genes with relatively ‘minor’ effects (8,9). Some researchers criticize the liability to false-positive or non-replicable claims. Nevertheless, population-based association studies have become an alternative and complementary approach to family-based linkage analyses in practice.

Given the limitation of statistical power that can be achieved by family-based linkage analyses with sample size practically collectable, population-based association studies are now underway in a genome-wide scale for a number of multifactorial diseases (10). Here, we performed a high-density association study of hypertension with a three-tiered genotyping approach in the Japanese population (Fig. 1).


Figure 1
View larger version (34K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 1. Schematic presentation of a three-tiered screening strategy in the present study. Gene-centered genome-wide exploratory test was performed in tier 1, followed by case–control study of disease-associated SNPs in tier 2 and tier 3 samples. A panel of 188 male hypertensive patients were compared with each of two population control panels in tier 1. Subsequently, ‘case versus unaffected control’ study was repeated twice to identify the best candidate SNPs. In transitions from tier 1 to tier 2, and from tier 2 to tier 3, the number of SNPs was reduced according to the selection criteria that we arbitrarily defined. See details in the Materials and Methods section.

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 FUNDING
 REFERENCES
 
Multi-tiered case–control study
We performed a large-scale case–control association study of hypertension using SNP markers selected from the Japanese SNP (JSNP) database (11,12). These SNP markers were distributed throughout the genome (Table 1). Only male hypertensive individuals were tested in tier 1, and a total of 80 795 SNPs distributed on 22 autosomes were used for the association study. Details of the high-throughput genotyping were same as previously described (13,14), and technical evaluation of our genotyping assay (e.g. overall success rate and accuracy of the genotyping assay) is shown in the supplementary material (Supplementary Explanation). JSNP had been developed as a database for the SNP discovery project with particular focus on common gene variations in the Japanese population. Although SNP marker resources used in the current study showed a certain degree of diversity in terms of the number of typed SNPs per gene locus, this partially reflected the variable size of re-sequenced fragments depending on the individual gene structure (12).


View this table:
[in this window]
[in a new window]

  		Table 1. Summary of SNPs genotyped in tier-1 screening and genome coverage estimated by HapMap data

 
The gene-centered genome-wide exploratory test in tier 1 identified 2676 SNPs with odds ratio (OR) ≥ 1.4 and P ≤ 0.015 in at least one test comparing allele frequency and/or genotype distribution (dominant or recessive models) between 188 hypertensive patients and 752 population control subjects in either of two panels (see Materials and Methods). In this exploratory test, the SNPs showing inverted tendency of OR between two pairs of case–control comparisons and significant deviations from Hardy–Weinberg equilibrium (HWE) in any panel (P ≤ 0.01) were excluded. Subsequently, we performed a screening of these 2676 SNPs with 752 hypertensive patients and 752 normotensive controls in tier 2, which constituted the first ‘case versus unaffected control’ study panel, i.e. comparison between 940 cases and 752 controls, together with the 188 cases in tier 1. On the basis of relatively stringent criteria, we identified 75 SNPs that showed P-values of ≤0.01 for genotype distribution and P-values of ≤0.05 for allele frequency in the {chi}2-test statistic. To further examine the association signals, we performed a replication study of these 75 SNPs with another panel of 619 hypertensive subjects and 1406 normotensive controls in tier 3. Cases and unaffected controls collected in tiers 2 and 3 were enrolled according to the identical criteria and their baseline characteristics are shown in Table 2. There were some trait differences in cases between tiers 2 and 3, such as blood pressure measurements and percentages of the subjects taking anti-hypertensive medication. This could be largely attributed to differences in sample enrollment settings between tiers 2 and 3; that is, cases in tier 3 were enrolled from either the annual medical checkup of a medical institution or the clinic practices of general practitioners, whereas a major part of cases in tier 2 were from the clinic practices of university hospitals. Among the 75 SNPs showing P-values between 0.05 and 4.4 x 10–5 in the first ‘case versus unaffected control’ study, only nine SNPs showed borderline association (at the level of P ≤ 0.05) in the second ‘case versus unaffected control’ study (Fig. 2). Of these, we found six SNPs that showed P-values of ≤0.05 for both genotype distribution and allele frequency in the {chi}2-test statistic (Table 3). rs3755351 and rs3771426 were located within the assumed intron 1 of ADD2, rs3787240 and rs3787241 were located in the same intron of EYA2, and the remaining two SNPs—rs3794260 and rs1805762—each located in KIAA0789 and M6PR. To adjust for three covariates—age, gender and body mass index (BMI), we also performed logistic regression analysis for the significant SNPs (Supplementary Material, Table S1). With consideration of genetic model consistency, an SNP (rs3755351) showed the strongest association in the identical model (an additive model by logistic regression analysis) among three tiers. Further details of the association results are described in the Discussion.


Figure 2
View larger version (18K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 2. Statistical significance of {chi}2-test of disease association in the two-staged ‘case versus unaffected control’ study. –Log10 P-values were used instead of raw P-values in each scatter plot. The dashed lines indicate P = 0.05. (A) As for SNPs genotyped in the second ‘case versus unaffected control’ panel, P-values for allele frequency in the second ‘case versus unaffected control’ panel were plotted against those in the first ‘case versus unaffected control’ panel, where SNPs located in three gene loci—ADD2, KIAA0789 and M6PR—are depicted with solid circles to which the following symbols are attached: A, ADD2; K, KIAA0789; M, M6PR. (B) As for SNPs genotyped in the second ‘case versus unaffected control’ panel, P-values for genotype distribution ([2 x 3] contingency table) in the second ‘case versus unaffected control’ panel were plotted against those in the first ‘case versus unaffected control’ panel, where SNPs located in three gene loci—ADD2, KIAA0789 and M6PR—are depicted with solid circles as mentioned earlier.

 


View this table:
[in this window]
[in a new window]

  		Table 2. Clinical characteristics of participants

 


View this table:
[in this window]
[in a new window]

  		Table 3. Summary of genomic SNPs associated with hypertension status in two-staged ‘case versus unaffected control’ study

 
SNP discovery and further test of association in three selected genes
Because a group of SNPs from three genes, ADD2, KIAA0789 and M6PR, were particularly noted for their significant association with hypertension (Table 3), we searched for potentially functional SNPs by re-sequencing the 5'- and 3'-untranslated regions, all exons and exon–intron borders of the individual loci, on the basis of the gene structure deposited in the human genome database (http://www.ncbi.nlm.nih.gov/). We detected a total of 74 SNPs—25 SNPs in ADD2, 40 SNPs in KIAA0789 and 9 SNPs in M6PR—and thereby selected 25 tag SNPs for genotyping 2025 subjects in tier 3 (see Supplementary Material, Table S2A, B and C). Apart from four SNPs which had been already included in the JSNP screening marker set, we found four additional SNPs, two in ADD2 (rs2024453 and rs10084293) and one each in KIAA0789 (rs9739493) and M6PR (rs1805725), to be significantly associated with hypertension (Table 3). Thus, in each gene, we identified at least two SNPs showing modest evidence of association with hypertension (P ≤ 0.05 level in tier 3) but these SNPs did not necessarily belong to the same linkage disequilibrium (LD) block (Fig. 3 and LD group in Supplementary Material, Table S2A, B and C). The analysis of haplotypes inferable from tag SNPs did not show more significant disease association than the analysis of individual SNPs in any of three genes tested (data not shown).


Figure 3
View larger version (35K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 3. LD relations between SNPs in the ADD2, KIAA0789 and M6PR genes (top) and disease association of markers from the corresponding genomic regions (bottom). In the top, the LD between a pair of markers is indicated by the color of the block above and to the left of the intersection of the markers. For the sake of readability, only the names of SNPs showing significant association are shown to the right of the vertical axis of the LD plot. The rest of the SNP information is described in Supplementary Material, Table S2A, B and C. In the upper bottom, the location of genetic markers studied in the corresponding genomic region is shown with relation to gene structure. Here, green and red circles indicate the SNPs with low (MAF < 0.05) and high (MAF ≥ 0.05) allele frequencies in the Japanese, respectively. In the lower bottom, –log10 P-values of the differences in allele frequencies between case and control subjects in tier 3 (i.e. the second ‘case versus unaffected control’ study) were plotted against the location of individual genetic markers genotyped.

 
Consideration of study power and multiple testing
We first estimated a type I error probability for the three-tiered screening to be 6.8 x 10–5: 0.036 for tier 1, 0.0009 for tiers 1 and 2 combined and 0.076 for tier 3 screening. Then, we estimated overall sensitivities (which could represent the statistical power) to be 0.10–0.45, 0.04–0.23 and 0.01–0.08 for a disease-associated SNP of OR = 1.4, 1.3 and 1.2, respectively, assuming the disease allele frequency within 0.1–0.9, the disease prevalence of 0.25 and the multiplicative genotype model. Since we had adopted relatively generous criteria for screening association signals, we evaluated the false discovery rate (FDR) to account for multiple testing (15). FDR for the nine SNPs found as significant was 0.69. A multi-staged screening in the current study could be largely categorized into two steps: tiers 1 and 2 (which constitute the first ‘case versus unaffected control’ study) and tier 3 (which constitutes the second ‘case versus unaffected control’ study). We therefore assessed experiment-wise type I errors with particular focus on the last-stage screening in tier 3. By permutation, the chance of observing a P-value of 0.0019 (for allele frequency test at rs3755351 in ADD2) in tier 3 was estimated to be 0.13.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 FUNDING
 REFERENCES
 
With the recent advent of high-throughput genotyping technologies and high-resolution maps of SNP markers, it is expected that genome-wide association studies allow us to identify systematically the contributions of common genetic variations to human multifactorial diseases (16–18). In this line, our study has attempted to discover common hypertension susceptibility gene variants via a gene-centered genome-wide association design for the first time. Despite the modest genetic impacts assumed for hypertension, e.g. the {lambda}-values (the relative risk for siblings of the affected probands) have been reported to be approximately 4 (19), we have nominated several susceptibility genes for hypertension (Table 3). Among these genes, findings for ADD2 and KIAA0789 are particularly noteworthy, because the former has been known to be a physiological candidate gene for hypertension and the latter is a novel gene with as-yet unknown physiological function.

Through a multi-tiered screening, nine SNPs derived from seven distinct gene loci have remained to show some evidence of association out of the 80 795 SNPs initially screened. Although the selection criteria were arbitrarily defined in the present study, a small percentage of the SNPs have passed the criteria in transitions from tier 1 to tier 2 (3.3%) and from tier 2 to tier 3 (2.8%). In the ADD2 gene, for example, the minor allele frequency (MAF) of rs3755351 is lower in case groups (0.14–0.19) than that in control groups (0.21–0.22) throughout three tiers. A P-value of 1.7 x 10–5 and an OR of 1.30 (95% CI 1.15–1.46) are attained for allele frequency comparison of rs3755351 when the subjects studied in different tiers are combined and finally categorized into the case (tiers 1–3) and unaffected control (tiers 2 and 3) groups. None of our results appears to be significant with the use of a strict Bonferroni correction, a very conservative evaluation of significance, and further replication in an independent population is indispensable.

The candidacy of ADD2 as a hypertension susceptibility gene has been supported by several physiological and biochemical findings (20–22), together with some evidence from the studies of molecular genetics (23–27). Adducin is a ubiquitously expressed membrane-skeleton heteromeric protein composed of different subunits, {alpha}-, β- and {gamma}-subunits. It is known to play a substantial role in the regulation of membrane ion transport. Point mutations of the {alpha}- and β-adducins account for up to 50% of the blood pressure difference between Milan hypertensive and normotensive rat strains, probably via the modulation of the Na+–K+ ATPase activity (one of major Na+-channels) in the kidney (23,24). In this line, of note is the fact that β-adducin-deficient mice show significant increases in systolic and diastolic blood pressures and pulse pressure (21). The human homolog of β-adducin spans over 100 kb on chromosome 2p13 and comprises 17 exons. It has been reported that a common SNP (rs4984) identified at position 1797 in exon15 is associated with an increased risk of hypertension under certain pathological conditions in European populations (25–27), whereas this SNP itself is not polymorphic in Asian populations (http://www.ncbi.nlm.nih.gov/SNP/). Also, it has to be noted that one previous study (28) showed significant evidence for hypertension linkage in the 2p13 region (a peak of 2.84 LOD at 93 cM), where the ADD2 locus is exactly located among several positional candidate genes. Despite our investigation in the ADD2 locus, we could not find either a clear LD block-like structure or potentially functional SNPs in the vicinity of three disease-associated SNPs (rs2024453, rs3755351 and rs3771426), which are located in the putative promoter region and intron 1, apart from rs10084293 located within an LD block of ADD2 (Fig. 3). We have assessed the independence of multiple associated SNPs in ADD2 by logistic regression analysis and have found that the observed association in this gene could be explained principally by the most significant SNP (rs3755351) (see Supplementary Explanation). Once these associations are validated in an independent study panel, further extensive searches of functional SNPs in the ADD2 locus are warranted.

Our high-density association study has also highlighted the KIAA0789 gene located on chromosome 12q23.3. This gene encodes a hypothetical protein, LOC9671, which is expressed principally in the central nervous system and modestly in the pancreas (unpublished data). The predicted gene structure of KIAA0789 involves 9 exons, spanning ~120 kb. There is a clear LD block in the 5' region of the putative exon 1 (~3.8 kb in size), whereas we have found two other LD block-like structures within the KIAA0789 gene (Fig. 3). Two disease-associated SNPs (rs3794260 and rs9739493) have turned out to reside in different LD blocks, and the construction of their haplotypes does not seem to provide much additional information on disease association. Although the precise gene structure and gene function remain unknown, KIAA0789 appears to contain a carboxy-binding WSC domain, and its homologs are likely to exist in mice and rats according to the database information (http://www.ncbi.nlm.nih.gov/). Again, detailed investigation including independent replication of disease association will lead us to clarify the etiological relevance of KIAA0789 to hypertension.

Another, potential disease association, though modest statistical significance, has been found for M6PR. The M6PR gene encodes a cation-dependent receptor for mannose-6-phosphate groups on lysosomal enzymes and plays a critical role in the segregation and targeting of lysosomal enzymes to lysosomes. Thus far, no functional relation between M6PR and hypertension has been reported. Similar to KIAA0789, this gene could also allow us to identify a novel, as-yet unnoticed blood pressure regulatory mechanism.

We should bear in mind several limitations inherent in the present study. First, the level of genome coverage is an issue of heated debate (10,18). Some people may argue that our a priori marker selection strategy is gene-centric without utilizing LD information and hence it is not sufficient to pick up as many modest associations as possible in genome-wide searches of hypertension susceptibility genes. A comprehensive framework of common variations throughout the human genome has been made available by the recent completion of the International HapMap Project (29). On the basis of our assessment, the JSNP screening markers in this study cover 20.6% of the HapMap SNPs, whereas a substantial proportion (~30%) of SNPs appear to be unique to JSNP (Table 1 and Supplementary Material, Fig. S1). Under these circumstances, an ideal set of SNPs for our study would encompass deliberately selected tag SNPs (principally common genetic variants) and additional ‘singleton’ SNPs (sometimes rare genetic variants). Besides this argument of tag SNPs, there are two points of weakness regarding genome coverage as follows: (i) sex chromosome markers have been excluded from the analysis because of the pre-determined policy of multi-disease collaborative study in the Japanese Millennium Genome Project, and (ii) a substantial part of the expressed human genes is not covered by the JSNP database (11), in which the fundamental SNP data were almost fixed in the middle of 2003. Second, the statistical power attainable by our study panel needs to be taken into consideration. For the last few years, genotype costs have fallen dramatically, yet present economic and experimental conditions make it necessary, in practice, to reduce the number of genotyped samples down to a moderately sized case group (188 subjects in our study) at the initial screening with approximately 80 000 SNPs. We arbitrarily set the selection criteria of OR ≥ 1.4 and P ≤ 0.015 in transition from tier 1 to tier 2, where the overall statistical power is estimated to be 10–45% for a disease-associated SNP of OR = 1.4 and 1–8% for that of OR = 1.2, assuming the disease allele frequency within 0.1–0.9 and the disease prevalence of 0.25. Thus, it is likely that our study design allows for capturing less than half of the true disease associations particularly with regard to modest genetic susceptibility. Third, ethnic diversity has not been tested within the scope of the present study. Instead of using commercially available SNP sets aimed at full genomic coverage, we have attempted to focus on potentially functional variants and also relatively common SNPs (MAF ≥ 0.1) in the Japanese population. Accordingly, some of disease-associated SNPs listed in Table 3 may be rare or not polymorphic in the other ethnic groups. To clarify allele frequency representation of individual loci and etiological impacts attributable to them, further examination is required in the context of ethnic diversity.

During our preparation of this report, two genome-wide association studies for hypertension and/or blood pressure have been performed in Caucasians (30,31). When our results are compared with public data sets for these association statistics, a few SNPs in the regions of interest appear to show a tendency of association with hypertension or blood pressure; for example, an SNP (rs17006246) in ADD2, which is in strong LD (r2 = 0.806 and D'= 1 in the HapMap JPT population) with rs3755351, the most significant SNP in our study, is modestly associated with hypertension status (P = 0.029) in the Diabetes Genetics Initiative (DGI) study but the direction of effect is opposite between rs3755351 typed in this study and rs17006246 typed in the DGI study. On the other hand, rs1805740, in strong LD with an SNP (rs1805762) in M6PR, is modestly associated with hypertension status (P = 0.036) in the Wellcome Trust Case Control Consortium study with the same direction of effect as in this study (see Supplementary Material, Table S3).

In summary, our high-density association study provides a list of gene loci potentially predisposing people to hypertension, which awaits replication across populations. With the available samples, we have observed an association of SNPs including three SNPs clusters (or gene loci) in the Japanese populations. In face of the complex nature of disease etiology, it seems to be a formidable task but worth challenging that we eventually apply the SNPs information to improved prevention, diagnosis and treatment of hypertension.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
SUPPLEMENTARY MATERIAL
 FUNDING
 REFERENCES
 
Study design
We performed a large-scale association study for genes susceptible to hypertension by using a three-tiered genotyping approach (tiers 1, 2 and 3) as depicted in Figure 1. All methods of the study were approved by the review committees of the individual institutions involved in the present study. All subjects provided written informed consent for participation.

In the gene-centered genome-wide exploratory test in tier 1, we carried out genotyping of 83 802 SNPs (3007 of which were excluded from the analysis because they are on sex chromosomes or in the unknown locations) using genomic DNAs from 188 Japanese male hypertensive patients and 752 unrelated Japanese individuals (referred to as general population controls) and another panel of 752 Japanese subjects (referred to as arbitrarily defined controls) who were affected with any of the other four common diseases including gastric cancer, diabetes mellitus, bronchial asthma and Alzheimer’s disease; each of these was investigated as the ‘Japanese Millennium Genome Project’ (Fig. 1). The theoretical basis of adopting this exploratory test scheme was previously reported elsewhere (32). Cases were enrolled from the clinical practice or the annual medical checkup of university hospitals and medical institutions according to the uniformly defined criteria. These included (i) systolic blood pressure ≥160 mmHg, diastolic blood pressure ≥95 mmHg, or both on two consecutive visits for untreated subjects; (ii) patients receiving long-term antihypertensive treatments; (iii) no secondary form of hypertension as evaluated by an extensive workup; (iv) family history of hypertension, i.e. at least one hypertensive subjects detectable among parents and siblings of the participants; (v) an age of onset known to be between 30 and 59 years. Moreover, only male subjects with BMI < 25 kg/m2 were selected in tier 1. We compared allele frequencies and/or genotype distributions in hypertensive patients and two population control panels and evaluated deviation from HWE at each of the genotyped loci. For the subsequent screening in tier 2, we selected SNPs (i) with OR ≥ 1.4 and P ≤ 0.015 against either of two population control panels and with concordant OR tendency against two control panels; (ii) with MAF ≥ 0.1 and (iii) not showing significant deviations (P= 0.01 level) from Hardy–Weinberg expectations in the patient or control panels.

In tier 2 (which comprised 752 hypertensive patients and 752 normotensive controls), we further tested the SNPs thus screened in tier 1, which effectively constituted the first ‘case (tiers 1 and 2) versus unaffected control (tier 2)’ study. Here, cases in tier 2 were selected according to the criteria (i)–(v) mentioned earlier for tier 1. Normotensive controls, on the other hand, were defined as follows: (i) systolic blood pressure ≤130 mmHg and diastolic blood pressure ≤85 mmHg without receiving antihypertensive treatments; (ii) age ≥50 years and (iii) no family history of hypertension. Both males and females were included in tier 2 without reference to BMI. We selected SNPs (i) with P-value ≤0.05 when comparing allele frequency; and (ii) with P-value ≤0.01 when comparing genotype distribution between (tiers 1 and 2) cases and (tier 2) controls by {chi}2 test statistics.

In tier 3 (which comprised 619 hypertensive patients and 1406 normotensive controls), we performed the second ‘case versus unaffected control’ study to examine significant associations observed in tiers 1 and 2. The diagnostic criteria in tier 3 were identical to those in tier 2. For the assessment of assumptions when using statistical models in the present study, quantile–quantile plots of P-values were depicted for each stage of association test described in Supplementary Explanation.

No significant population stratification was observed for samples in tier 1 when it was assessed with the methods reported by Patterson et al. (33). However, the presence of population stratification was indicated for samples in the first stage ‘case (tiers 1 and 2) versus unaffected control’ study. We observed moderate bias in genotype frequency of some SNPs between the two tiers, which may have resulted from technical/experimental artifacts between genotyping of cases in tiers 1 and 2. Therefore, the trend test statistic at this analytical stage was corrected according to the significant eigenvector (see Supplementary Explanation). Stratification in tier 3 was not detected but could not be ruled out because of the relatively small number of SNPs (n = 75) genotyped in tier 3. As for the nine SNPs that showed significant disease association after multi-stage screening, they were not correlated with the significant eigenvector detected in tiers 1 and 2 cases and tier 2 controls. The P-values for nine SNPs were similar between the nominal and the EIGENSTRAT-corrected ones; for example, the nominal P-value was 0.0029 and the EIGENSTRAT-corrected P-value was 0.0069 at rs3755351 in ADD2.

SNP marker resource and genotyping
Most of the SNP markers used in the present study were same as the markers used in the previous reports (14) and derived from the JSNP database. The samples in tiers 1 and 2 were genotyped by PCR amplification of multiple genomic fragments with 20 ng of genomic DNA followed by characterization with the invader assay. Genotyping of the samples in tier 3 was undertaken using the TaqMan® SNP Genotyping Assays (Applied Biosystems). To secure the accuracy and completeness of genotyping, which is critical for large-scale studies (34), we attached a set of ‘flags’ to individual SNP data mainly dependent on the data completeness, after two independent investigators had checked the raw data robustness by looking at the scatter plot of the assay.

SNP discovery in the selected genes
Approximately 38 kb of genomic sequence spanning the exons and the 5'- and 3'-untranslated regions of three genes, ADD2, KIAA0789 and M6PR, was re-sequenced in 48 Japanese control individuals to identify potentially functional SNPs. Since KIAA0789 had not been fully annotated, the arbitrary positions of translation initiation sites were estimated according to the human genome database. From the SNPs thus identified, tag SNPs were selected for the three genes with the algorithm that we previously reported (35). These tag SNPs were then used for the case–control analysis in tier 3 to further examine association signals seen throughout the multi-staged screening. We deposited the identified SNP information in the NCBI’s SNP database and also in our own database, JMDBase (Japan Metabolic Disease Database).

Statistical analysis
The SNPs were tested individually for the statistical significance of disease association with the {chi}2-test statistic, which evaluated three inheritance models—[2 x 3] contingency table, dominant and recessive models—for genotype distributions and independence on [2 x 2] contingency table for allele frequencies. Here, the most significant P-values among three inheritance models were adopted for genotype distributions when we selected SNPs for screening in tier 3. The criteria for declaring suggestive evidence of disease association were arbitrarily set at each analytical stage as summarized in Figure 1, and they are described in the Results section. SNPs' genotype departures from HWE were tested using the {chi}2-test with 1 degree of freedom.

In the three genes showing significant association signals, the extent of LD was measured in terms of an LD coefficient r2 before the analysis of haplotype structure. Within each LD block, haplotypes were inferred from genotype data by the SNPHAP software for the case and control groups, respectively.

We randomly permutated the genotype of individuals across different panels, 100 times per SNP, and counted the ratio of permutations that fulfill the screening criteria. This ratio indicates the specificity of the study. According to the P-value distribution of the permutations, we evaluated the probability of observing an SNP with P-value no larger than the actual minimum. This probability indicates the experiment-wise P-value. For the specific prevalence and penetrance, we calculated genotype frequency and randomly generated genotypes according to their frequency. We generated genotypes for 1000 simulations of each panel and computed the ratio of simulations that could pass the screening. This ratio is considered the sensitivity of the study.

Values were expressed as means ± SD unless otherwise indicated.

Uniform resource locators
The JSNP database is available at http://snp.ims.u-tokyo.ac.jp/index.html. The National Center for Biotechnology Information’s SNP database is available at http://www.ncbi.nlm.nih.gov/SNP/. The JMDBase is available at http://www.jmdbase.jp. SNPHAP is available at http://www.gene.cimr.cam.ac.uk/clayton/software/


   SUPPLEMENTARY MATERIAL
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
FUNDING
 REFERENCES
 
Supplementary Material is available at HMG Online.


   FUNDING
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 FUNDING
REFERENCES
 
This work was supported by the Grant-in-Aid for Scientific Research on Priority Areas (Millennium Genome Project) from the Ministry of Education, Science, Sports, and Culture of Japan, the grants from the Program for Promotion of Fundamental Studies in Health Sciences of Pharmaceuticals and Medical Devices Agency (PMDA) and a grant for the Core Research for the Evolutional Science and Technology, from the Japan Science Technology Agency (JST).


   ACKNOWLEDGEMENTS
 
We acknowledge the outstanding contributions of the SNP Research Center, the Institute of Physical and Chemical Research (RIKEN), the National Cancer Center (NCC) and the International Medical Center of Japan (IMCJ) employees who provided technical and infrastructural support for this work. Above all, we thank the patients and study subjects who made this work possible and who give it value. We thank all the people who continuously support the Suita Study. We also thank Drs Yukio Yamori, Yoshihiro Kokubo, Masakuni Serizawa, Ken Sugimoto, Hiroshi Akasaka, Chikanori Makibayashi and many physicians in Amagasaki Medical Association, Sadahiko Iwamoto, Yoichi Ohno, Mikano Hishiki, Kotoko Kosuge, Noriko Aoi, Akira Haketa, Yoichi Izumi, Toru Izumi, Taishi Sasaoka, Shigeru Hohda and Takeharu Hayashi, and Ms Kaoru Sugama for their contribution to collecting DNA samples and clinical information accompanied; Drs Hiromi Rakugi and Masafumi Kuremura and Ms Akemi Muramoto and Kazuko Iwasa for their advice and assistance.

Conflict of Interest statement. None declared.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 FUNDING
 REFERENCES
 

  1. Mitchell G.F., DeStefano A.L., Larson M.G., Benjamin E.J., Chen M.H., Vasan R.S., Vita J.A., Levy D. Heritability and a genome-wide linkage scan for arterial stiffness, wave reflection, and mean arterial pressure: the Framingham Heart Study. Circulation (2005) 112:194–199.[Abstract/Free Full Text]

  2. Williams R.R., Hunt S.C., Hasstedt S.J., Hopkins P.N., Wu L.L., Berry T.D., Stults B.M., Barlow G.K., Schumacher M.C., Lifton R.P., et al. Are there interactions and relations between genetic and environmental factors predisposing to high blood pressure? Hypertension (1991) 18(Suppl_3):I29–I37.[Medline]

  3. Biron P., Mongeau J.G., Bertrand D. Familial aggregation of blood pressure in 558 adopted children. Can. Med. Assoc. J. (1976) 115:773–774.[Abstract]

  4. Gong M., Hubner N. Molecular genetics of human hypertension. Clin. Sci. (2006) 110:315–326.[CrossRef][Web of Science][Medline]

  5. Rice T., Cooper R.S., Wu X., Bouchard C., Rankinen T., Rao D.C., Jaquish C.E., Fabsitz R.R., Province M.A. Meta-analysis of genome-wide scans for blood pressure in African American and Nigerian samples. The National Heart, Lung, and Blood Institute GeneLink Project. Am. J. Hypertens. (2006) 19:270–274.[CrossRef][Web of Science][Medline]

  6. Liu W., Zhao W., Chase G.A. Genome scan meta-analysis for hypertension. Am. J. Hypertens. (2004) 17:1100–1106.[CrossRef][Web of Science][Medline]

  7. Province M.A., Kardia S.L., Ranade K., Rao D.C., Thiel B.A., Cooper R.S., Risch N., Turner S.T., Cox D.R., Hunt S.C., et al. National Heart, Lung and Blood Institute Family Blood Pressure Program. Am. J. Hypertens. (2003) 16:144–147.[CrossRef][Web of Science][Medline]

  8. Jones H.B. The relative power of linkage and association studies for the detection of genes involved in hypertension. Kidney. Int. (1998) 53:1446–1448.[CrossRef][Web of Science][Medline]

  9. Risch N., Merikangas K. The future of genetic studies of complex human diseases. Science (1996) 273:1516–1517.[Abstract/Free Full Text]

  10. Barrett J.C., Cardon L.R. Evaluating coverage of genome-wide association studies. Nat. Genet. (2006) 38:659–662.[CrossRef][Web of Science][Medline]

  11. Hirakawa M., Tanaka T., Hashimoto Y., Kuroda M., Takagi T., Nakamura Y. JSNP: a database of common gene variations in the Japanese population. Nucleic Acids Res. (2002) 30:158–162.[Abstract/Free Full Text]

  12. Haga H., Yamada R., Ohnishi Y., Nakamura Y., Tanaka T. Gene-based SNP discovery as part of the Japanese Millennium Genome Project: identification of 190,562 genetic variations in the human genome. Single-nucleotide polymorphism. J. Hum. Genet. (2002) 47:605–610.[CrossRef][Web of Science][Medline]

  13. Ohnishi Y., Tanaka T., Ozaki K., Yamada R., Suzuki H., Nakamura Y. A high-throughput SNP typing system for genome-wide association studies. J. Hum. Genet. (2001) 46:471–477.[CrossRef][Web of Science][Medline]

  14. Yamazaki K., McGovern D., Ragoussis J., Paolucci M., Butler H., Jewell D., Cardon L., Takazoe M., Tanaka T., Ichimori T., et al. Single nucleotide polymorphisms in TNFSF15 confer susceptibility to Crohn’s disease. Hum. Mol. Genet. (2005) 14:3499–3506.[Abstract/Free Full Text]

  15. Benjamini Y., Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Statist. Soc. B. (1995) 57:289–300.

  16. Wang W.Y., Barratt B.J., Clayton D.G., Todd J.A. Genome-wide association studies: theoretical and practical concerns. Nat. Rev. Genet. (2005) 6:109–118.[CrossRef][Web of Science][Medline]

  17. Hirschhorn J.N., Daly M.J. Genome-wide association studies for common diseases and complex traits. Nat. Rev. Genet. (2005) 6:95–108.[Web of Science][Medline]

  18. Palmer L.J., Cardon L.R. Shaking the tree: mapping complex disease genes with linkage disequilibrium. Lancet (2005) 366:1223–1234.[CrossRef][Web of Science][Medline]

  19. Altmuller J., Palmer L.J., Fischer G., Scherb H., Wjst M. Genomewide scans of complex human diseases: true linkage is hard to find. Am. J. Hum. Genet. (2001) 69:936–950.[CrossRef][Web of Science][Medline]

  20. Gilligan D.M., Lozovatsky L., Silberfein A. Organization of the human beta-adducin gene (ADD2). Genomics (1997) 43:141–148.[CrossRef][Web of Science][Medline]

  21. Marro M.L., Scremin O.U., Jordan M.C., Huynh L., Porro F., Roos K.P., Gajovic S., Baralle F.E., Muro A.F. Hypertension in beta-adducin-deficient mice. Hypertension (2000) 36:449–453.[Abstract/Free Full Text]

  22. Muro A.F., Marro M.L., Gajovic S., Porro F., Luzzatto L., Baralle F.E. Mild spherocytic hereditary elliptocytosis and altered levels of alpha- and gamma-adducins in beta-adducin-deficient mice. Blood (2000) 95:3978–3985.[Abstract/Free Full Text]

  23. Tripodi G., Valtorta F., Torielli L., Chieregatti E., Salardi S., Trusolino L., Menegon A., Ferrari P., Marchisio P.C., Bianchi G. Hypertension-associated point mutations in the adducin alpha and beta subunits affect actin cytoskeleton and ion transport. J. Clin. Invest. (1996) 97:2815–2822.[Web of Science][Medline]

  24. Ferrandi M., Salardi S., Tripodi G., Barassi P., Rivera R., Manunta P., Goldshleger R., Ferrari P., Bianchi G., Karlish S.J. Evidence for an interaction between adducin and Na(+)–K(+)-ATPase: relation to genetic hypertension. Am. J. Physiol. (1999) 277:H1338–H1349.[Web of Science][Medline]

  25. Tikhonoff V., Kuznetsova T., Stolarz K., Bianchi G., Casiglia E., Kawecka-Jaszcz K., Nikitin Y., Tizzone L., Wang J.G., Staessen J.A. beta-Adducin polymorphisms, blood pressure, and sodium excretion in three European populations. Am. J. Hypertens. (2003) 16:840–846.[CrossRef][Web of Science][Medline]

  26. Tikhonoff V., Kuznetsova T., Stolarz K., Bianchi G., Casiglia E., Kawecka-Jaszcz K., Nikitin Y., Tizzone L., Wang J.G., Staessen J.A., European Project on Genes in Hypertension Investigators. Blood pressure phenotypes in relation to the beta-adducin C1797T polymorphism in the European Project on Genes in Hypertension (EPOGH). Blood. Press. Monit. (2003) 8:151–154.[CrossRef][Web of Science][Medline]

  27. Wang J.G., Staessen J.A., Barlassina C., Fagard R., Kuznetsova T., Struijker-Boudier H.A., Zagato L., Citterio L., Messaggio E., Bianchi G. Association between hypertension and variation in the alpha- and beta-adducin genes in a white population. Kidney Int. (2002) 62:2152–2159.[CrossRef][Web of Science][Medline]

  28. Barkley R.A., Chakravarti A., Cooper R.S., Ellison R.C., Hunt S.C., Province M.A., Turner S.T., Weder A.B., Boerwinkle E., Family Blood Pressure Program. Positional identification of hypertension susceptibility genes on chromosome 2. Hypertension (2004) 43:477–482.[Abstract/Free Full Text]

  29. International HapMap Consortium. A haplotype map of the human genome. Nature (2005) 437:1299–1320.[CrossRef][Medline]

  30. Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature (2007) 447:661–678.[CrossRef][Web of Science][Medline]

  31. Saxena R., Voight B.F., Lyssenko V., Burtt N.P., de Bakker P.I., Chen H., Roix J.J., Kathiresan S., Hirschhorn J.N., et al, Diabetes Genetics Initiative of Broad Institute of Harvard MIT, Lund University, and Novartis Institutes of BioMedical Research. Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science (2007) 316:1331–1336.[Abstract/Free Full Text]

  32. Sato Y., Suganami H., Hamada C., Yoshimura I., Yoshida T., Yoshimura K. Designing a multistage, SNP-based, genome screen for common diseases. J. Hum. Genet. (2004) 49:669–676.[CrossRef][Web of Science][Medline]

  33. Patterson N., Price A.L., Reich D. Population structure and eigenanalysis. PLoS Genet. (2006) 2:e190.[CrossRef][Medline]

  34. Clayton D.G., Walker N.M., Smyth D.J., Pask R., Cooper J.D., Maier L.M., Smink L.J., Lam A.C., Ovington N.R., Stevens H.E., et al. Population structure, differential bias and genomic control in a large-scale, case–control association study. Nat. Genet. (2005) 37:1243–1246.[CrossRef][Web of Science][Medline]

  35. Takeuchi F., Yanai K., Morii T., Ishinaga Y., Taniguchi-Yanai K., Nagano S., Kato N. Linkage disequilibrium grouping of single nucleotide polymorphisms (SNPs) reflecting haplotype phylogeny for efficient selection of tag SNPs. Genetics (2005) 170:291–304.[Abstract/Free Full Text]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 GeneticsHome page
K. M. Weiss
Tilting at Quixotic Trait Loci (QTL): An Evolutionary Perspective on Genetic Causation
Genetics, August 1, 2008; 179(4): 1741 - 1756.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
17/4/617    most recent
ddm335v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Kato, N.
Right arrow Articles by Tomoike, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kato, N.
Right arrow Articles by Tomoike, H.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?