Introduction

Metabolic syndrome is a common clinical phenotype presenting as concurrent metabolic abnormalities, including central obesity, glucose intolerance, dyslipidemia and hypertension.1 Several different definitions exist2 and there are still debates on the adequacy of its concept; however, metabolic syndrome has attracted considerable interest recently. Although the pathogenesis of metabolic syndrome is not fully understood, the predominant underlying risk factor is considered to be visceral obesity due to an atherogenic diet and physical inactivity in the presence of some genetic background.1, 3 Adipose tissue, especially visceral fat, secretes various adipokines. An increase in adipose tissue mass leads to an alteration in the plasma level of adipokines, resulting in the development of dyslipidemia, hypertension and insulin resistance.3, 4 Studies using twins and families have suggested that genetic and environmental factors contribute to the clustering of metabolic syndrome in various ethnic groups.5, 6, 7, 8, 9, 10 In a study of Japanese-American families, significant genetic influences were noted on all metabolic syndrome components, especially dyslipidemia.7, 10 Several genome-wide linkage studies conducted to identify chromosomal segments that are linked with metabolic syndrome have been reported, but no specific gene or variation has been found.2 More than 20 genetic association studies with metabolic syndrome have been reported.2, 11 The published results at these candidate loci are not consistent and the genetic background of metabolic syndrome remains unclear.

In this study, we screened single-nucleotide polymorphisms (SNPs) in 85 obesity-related genes reported11 for a possible relation to metabolic syndrome. We found that four tag SNPs (rs2294901, rs6133922, rs6077785 and rs6108572) in the McKusick–Kaufman syndrome gene (MKKS) were significantly associated with metabolic syndrome.

Materials and methods

Study subjects

The sample size of the first set of Japanese subjects with metabolic syndrome was 729 (case-1; male to female ratio 337:392; age, 54±14 year). The sample size of the first set of Japanese controls was 441 (control-1; male to female ratio 112:329; age, 47±16 year). The sample size of the second set of Japanese subjects with metabolic syndrome was 351 (case-2; male to female ratio 218:133; age, 54±10 year), whereas that of the second set of controls was 87 (control-2; male to female ratio 57:30; age, 51±7 year). Metabolic syndrome was diagnosed principally according to the new definition criteria released by the Japanese Committee for the Diagnostic Criteria of Metabolic Syndrome in April 2005.12 According to these criteria, metabolic syndrome is defined by the presence of two or more abnormalities in addition to visceral fat obesity (waist circumstance: ⩾85 cm in men, ⩾90 cm in women). These three abnormalities were as follows: (1) triglycerides ⩾150 mg per 100 ml and/or high-density lipoprotein cholesterol <1.03 mmol l−1 (40 mg per 100 ml) or under treatment for this type of dyslipidemia, (2) systolic blood pressure ⩾130 mm Hg and/or diastolic blood pressure blood pressure ⩾85 mm Hg or under treatment for hypertension and (3) fasting glucose ⩾6.1 mmol l−1 (110 mg per 100 ml) or under treatment for diabetes. We used body mass index (>25 kg m−2) instead of waist circumstance, as there are still debates on the criteria for waist circumstance, especially in women. Control was composed of the subjects who were not obese (body mass index <25 kg m−2) and no metabolic abnormalities described above were present. Clinical characteristics of cases and controls are summarized in Table 1.

Table 1 Clinical characteristics of the subjects
Full size table

Subjects with metabolic syndrome were recruited from the outpatient clinics. Control subjects were selected from non-obese Japanese volunteers from the subjects who had undergone a medical examination for common disease screening. Written informed consent was obtained from each subject, and the protocol was approved by the ethics committee of each institution and by that of RIKEN.

DNA preparation and SNP genotyping

Genomic DNA was prepared from each blood sample according to the standard methods. The genes reviewed by Rankinen et al.11 were used for the analysis of metabolic syndrome. The genes located on the X- and Y-chromosomes were not analyzed in this study. The SNPs within such genes were selected from the IMS-JST (Institute of Medical Science-Japan Science and Technology Agency) SNP database.13 A total of 755 SNPs in 122 genes were found in the IMS-JST database. Among the 755 SNPs, we chose for use in this study the 336 SNPs with a minor allele frequency more than 0.2, and the expected allele frequencies were not largely diverged from Hardy–Weinberg equilibrium (P>0.001) (Supplementary Table 1). Invader probes (Third Wave Technologies, Madison, WI, USA) were synthesized for these SNPs, and the SNPs were genotyped in cases and controls by a combination of multiplex PCR and the Invader assay as described earlier.14

In addition to rs2294901, three SNPs (rs221667, rs6133922 and rs6077785) were selected as tag SNPs using a Haploview 4.0,15 according to the haplotype map of the human genome (HapMap). In addition, SNP rs6108572 that was reported by Benzinou et al.16 in the MKKS gene was also genotyped. Four SNPs (rs221667, rs6133922, rs6077785 and rs6108572) were genotyped using TaqMan probes (C_2230457_10, C_25623400_10 and C_2482540_10; Applied Biosystems, Foster City, CA, USA). TaqMan probe for rs6108572 was designed and synthesized by Applied Biosystems.

Statistical analysis

For each case–control study, χ2-test (additive model) was performed according to Sladek et al.17 We coded genotypes as 0, 1 and 2, depending on the number of copies of the risk alleles. Odds ratio (OR) adjusted for age and gender was calculated using multiple logistic regression with genotypes, age and gender as independent variables.

We used the Q-value as an approach to the problem of multiple testing.18 Statistical analysis was performed using the software R (http://www.r-project.org/), and the Q-value was calculated using Q-VALUE (http://faculty.washington.edu/~jstorey/qvalue/).

The Hardy–Weinberg equilibrium was assessed using the χ2-test.19 We used the linkage disequilibrium (LD) measures D′ and r2, calculated as reported earlier.20 Haplotype phasing was estimated using the EM algorithm.21

Results

The cutoff points for the visceral fat obesity (waist circumstance: 85 cm in men, 90 cm in women) are based on the cutoff point for the visceral fat area (100 cm2) determined by computerized tomography scan.12, 22 Neither body mass index nor waist circumstance could estimate the visceral fat mass as precisely as computerized tomography scan. Therefore, we performed the case–control association study using case-1 and control-1 subjects who had no risk of the metabolic syndrome. We found that three SNPs (rs1545, rs1547 and rs2294901) in the MKKS gene were significantly related to metabolic syndrome even when the conservative Bonferroni’s correction was applied (Table 2). The log-quantile-quantile P-value plot indicates that P-values of these three SNPs were significantly lower than the expected P-values (Figure 1). Q-values of these SNPs were also significantly small (Table 2), suggesting that MKKS gene would be related to the metabolic syndrome. To confirm the associations of these SNPs, we performed case–control association using the second set of subjects. Although the minor allele frequency of case-2 were higher than control-2, it was not significant because of the small number of control-2 (Table 2). When the case-2 was compared with control-1, significant associations were observed (P=0.017 in rs1547, P=0.022 in rs1545 and P=0.019 in rs2294901).

Table 2 Genotype frequencies and association tests of significant SNPs and replication study
Full size table
Figure 1
figure 1

The log-quantile-quantile P-value plot for the analysis of case-1 vs control-1. The horizontal line, y=−log10 (0.05/336), indicates the threshold of significance using Bonfferoni correction. The three single-nucleotide polymorphisms (SNPs) reached the threshold of significance. Black circles represent significant SNPs.

Full size image

We examined the power of the test for this association study. The test was χ2-test (additive model) with the significance level 0.05/336=1.49 × 10−4 (the Bonferroni's correction). Prevalence of metabolic syndrome was estimated as 12.1 and 1.7% among male and female groups, respectively.12 Thus, we assumed that the penetrances of associated SNPs were different among males and females. We also assumed two conditions: one was that the effect of increasing one risk allele for disease was equal among males and females; the other was that allele frequencies of the SNPs were the same between male and female groups. Under these assumptions, we examined the power of the test for 729 cases (case-1) and 441 controls (control-1) using Bayes’ theorem. We considered the cases of four different levels of minor allele frequencies (0.1–0.4) and eight different levels of the effects (1.1–1.8), and performed 100 000 times simulation studies for each condition. Obtained power was indicated in Supplementary Table 2. We also give obtained powers in the 1080 cases and 528 controls namely combined two case and control groups (Supplementary Table 3).

According to the Hapmap database, 38 SNPs (minor allele frequency >5%) exist in the MKKS gene and they are in one LD block (Supplementary Figure 1). To investigate whether MKKS gene would be associated with the metabolic syndrome, we selected four tag SNPs (rs2294901, rs221667, rs6133922 and rs6077785). These SNPs captured 38 of 38 (100%) alleles at r2>0.95. As Benzinou et al.16 reported that rs6108572 was associated with metabolic disorder in French obese children and there was no allele frequency data in the Japanese, we also genotyped rs6108572 in our sample. The four SNPs (rs2294901, rs6133922, rs6077785 and rs6108572) showed strong association with metabolic syndrome, whereas rs221667 showed weak association (Table 3). Minor alleles of four SNPs (rs2294901, rs6133922, rs6077785 and rs6108572) and major allele of rs221667 were risk allele for the metabolic syndrome. All SNPs were in Hardy–Weinberg equilibrium (P>0.10).

Table 3 Genotype frequencies and association tests of tag SNPs in the MKKS gene
Full size table

Linkage disequilibrium analysis of SNPs in the MKKS gene showed that they were in one block (Table 4) and rs6108572 was not captured by other tag SNPs. Thus, we added rs6108572 as a tag SNP in the MKKS gene. Haplotype analysis showed that TGAAA haplotype was protective against the metabolic syndrome (P=0.0074), and CCGTT haplotype was susceptible (P=0.00070) to the metabolic syndrome (Table 5).

Table 4 Linkage disequilibrium (LD) coefficients between three single-nucleotide polymorphisms (SNPs) and other tag SNPs of the MKKS gene
Full size table
Table 5 Haplotype structure and frequencies in the MKKS gene
Full size table

The prevalence of metabolic syndrome increased with age and was approximately 6–7 times higher in men than in women in Japan.12 Thus, interactions of age and gender as well as genetic factors play important roles in the development of metabolic syndrome. When the case–control association study was performed in men and women separately, P-values were less than 0.05, except for rs6133922 in men and rs221667 in women (Table 6). When the case and control were divided by age (over 50 years and under 50 years), significant associations were observed both in younger and elder groups, except for rs221667 in elder group (Table 6). This finding is most likely due to the decrease in the number of each genotype. To investigate the effect of age and gender, logistic regression analysis was performed. The objective variable was the binary variable for the presence or absence of metabolic syndrome. The explanatory variables were age, gender and SNP. The results are shown in Table 7. These tag SNPs (rs2294901, rs221667, rs6133922, rs6077785 and rs6108572) showed a significant association with metabolic syndrome after correction for the confounding factors (age and gender).

Table 6 The association tests of tag SNPs in the MKKS gene stratified by gender and age
Full size table
Table 7 The results of logistic regression analysis of tag single-nucleotide polymorphisms (SNPs) in the MKKS gene
Full size table

Discussion

We found that SNPs in MKKS were significantly associated with the metabolic syndrome. Mutations in MKKS cause Bardet–Biedl syndrome, which is an autosomal recessive mendelian disorder characterized by progressive obesity, retinal dystrophy, learning disability, polydactyly, renal and cardiac malformations and hypogenitalism.23, 24 In some cases of Bardet–Biedl syndrome, the patients also had metabolic syndrome,25 and MKKS-null mice developed obesity and hypertension in addition to the phenotype resembling that of the Bardet–Biedl syndrome.26 Recently, T variant of rs6108572 was reported to be associated with quantitative components of the metabolic syndrome (dyslipidemia) as well as obesity and in French Caucasian children.16 T variant of rs6108572 was also associated with obesity in French Caucasian adults, and TT carriers of obese patients increased prevalence of arterial hypertension. Therefore, MKKS would be a good candidate gene for the development of metabolic syndrome, as the variant of MKKS was indicated to be related to metabolic syndrome in two populations, the Japanese and French Caucasians. SNPs rs1547 and rs1545 are the nonsynonymous SNPs Arg517Cys and Gly532Val, respectively. SNP rs2294901 is located in the 3′-flanking region of MKKS. MKKS is a chaperonin-like protein and a novel centrosomal component required for cytokinesis.27 The apical domain region (amino acids 198–370) is important for centrosomal localization. Thus, Arg517Cys (rs1547) and Gly532Val (rs1545) may not affect its function as severely as Bardet–Biedl syndrome; however, these nonsynonymous changes may alter the function of the MKKS protein and contribute to the development of metabolic syndrome. SNPs rs6108572 and rs6077785 are located in intron 1 and there are several SNPs in almost absolute LD (r2=1.0) with rs6077785 in intron 1. Thus, these SNPs would affect the transcriptional activity of MKKS gene, although further investigation would be necessary.

In summary, we have successfully identified SNPs in the MKKS gene that is related to metabolic syndrome. Our approach would be effective, although further studies to elucidate the mechanisms by which these SNPs affect the development of metabolic syndrome are needed.