Introduction

Metabolic syndrome (MetS) is characterized by a cluster of insulin resistance, obesity, dyslipidemia, and hypertension, and is a common cause of cardiovascular disease.1, 2 The prevalence of MetS in Japan has increased over recent decades as a result of changes in diet and physical activity.3 In addition to its role in energy storage, adipose tissue functions as an endocrine organ by secreting various bioactive substances or adipocytokines. These molecules are able to directly or indirectly affect the pathogenesis of metabolic disorders.4, 5

Adiponectin is an adipocytokine whose concentration is reduced in obesity.6 Low plasma adiponectin levels (hypoadiponectinemia) are also associated with type II diabetes,7, 8, 9, 10 with high and low circulating concentrations of triglyceride and high-density lipoprotein (HDL) cholesterol,11 respectively, with hypertension,12 and with coronary artery disease.13, 14, 15 These observations suggest that adiponectin might be a useful biomarker for management of metabolic disorders. Evidence also indicates that adiponectin has beneficial effects on insulin resistance,16 endothelial function, hypertension,17, 18, 19 ischemic heart disease, and atherosclerosis.20, 21 Adiponectin may, therefore, be a key determinant of the development of metabolic disorders.

Adiponectin is encoded by the gene ADIPOQ, which has been mapped to chromosome 3q27.22 Variants of ADIPOQ have been associated with adiponectin level and metabolic traits.23 Case–control studies showed that the Ile164Thr polymorphism of ADIPOQ was associated with hypoadiponectinemia, type II diabetes, hypertension, and coronary artery disease in the Japanese.12, 24, 25 Other such studies found that the rs266729 (–11377C → G) polymorphism of ADIPOQ was associated with type II diabetes in Caucasians.26, 27 The rs1501299 (+276G → T) polymorphism was also shown to be associated with adiponectin level and type II diabetes in Caucasians as well as in obese Japanese.28, 29 However, the clinical relevance of ADIPOQ variants to metabolic traits has remained to be fully explored in a population-based cross-sectional study of the Japanese population. We have now examined the relation of ADIPOQ variants to serum adiponectin level and metabolic traits in such a study. Our data indicate that a common variant of ADIPOQ, the minor allele of rs1656930, was associated with low adiponectin levels and hypertension. The association of this polymorphism with hypoadiponectinemia was replicated in an independent population.

Subjects and methods

Study subjects

The Kita-Nagoya Genomic Epidemiology (KING) study (ClinicalTrials.gov identifier: NCT00262691) is a population-based cross-sectional study based in Kita-Nagoya, Japan, for which a total of 3975 volunteers were recruited from participants of annual health checkups and which has been described in detail previously.30 Of these volunteers, 3297 individuals (1456 men, 1841 women; age, 63.6±6.5 years) whose serum adiponectin concentration was measured were enrolled in this study. A replication study included 750 men (57.8±3.1 years) recruited from Japanese company employee volunteers who were considered to be engaged mostly in light labor (described in detail under ClinicalTrials.gov identifier NCT00408824). A physical examination was performed and a medical history was obtained for each subject at an annual health checkup. Body mass index (BMI), waist circumference (at the midpoint between the lowest rib and the iliac crest after normal expiration), and blood pressure were measured for all subjects. The study protocol was approved by the ethics committees of Aichi-Gakuin University School of Dentistry and Nagoya University School of Medicine, and all participants provided written informed consent.

Definitions of risk factors

The overweight condition was defined as a BMI of ≥25 kg/m.2 Hypertension was defined as systolic blood pressure of ≥130 mm Hg or diastolic blood pressure of ≥85 mm Hg on repeated measurement, or as treatment with antihypertensive medication. Hyperglycemia was defined as a fasting serum glucose concentration of ≥100 mg/dl or a hemoglobin (Hb) A1c content of ≥5.9%, or as treatment with antidiabetes agents. Dyslipidemia was defined as a serum concentration of total cholesterol of ≥220 mg/dl, a serum triglyceride concentration of ≥150 mg/dl, or a serum HDL-cholesterol concentration of <40 mg/dl in men or <50 mg/dl in women, or as treatment for dyslipidemia. MetS was defined according to the recent International Diabetes Federation consensus criteria,1 which stipulate central obesity as being essential for the diagnosis.

Blood collection

Blood samples were obtained from subjects in the fasted condition for measurement of serum biomarkers and extraction of DNA. After the subject had rested for 10 min in the sitting position, vital signs were recorded and 14 ml of blood were collected from the antecubital vein into tubes containing EDTA. The blood was centrifuged at 3000 rpm for 15 min at 4°C, and the resulting supernatant was stored at −80°C until analysis. Serum adiponectin level was determined with the use of a latex turbidometric immunoassay (Otsuka Pharmaceutical Corporation, Osaka, Japan). Glucose, HbA1c, and lipids were measured by standard methods.

Genotyping of polymorphisms

Genomic DNA was isolated with the use of a FlexGene DNA kit (Qiagen, Chatsworth, CA, USA). Genotyping of 30 single nucleotide polymorphisms (SNPs) of ADIPOQ was performed with a fluorescence- or colorimetry-based allele-specific DNA primer–probe assay system (Toyobo Gene Analysis, Tsuruga, Japan) as described previously.31 To confirm the accuracy of genotyping, we randomly selected 100 DNA samples and subjected them to polymerase chain reaction (PCR) and restriction fragment length polymorphism analysis or to direct DNA sequencing of the PCR products. In each instance, the genotype determined by the allele-specific DNA primer–probe assay system was identical to that determined by the confirmatory methods. Furthermore, we genotyped all samples with a positive control (90 samples with a positive control at a time), and we confirmed that the positive control was genotyped correctly. The genotyping success rate was >99%, and all the SNPs were found to be in Hardy–Weinberg equilibrium (P>0.05 for all). Primer and probe sequences as well as PCR conditions for genotyping are shown in Supplementary Table 1.

Selection of SNPs

In an initial screening study, we genotyped 30 SNPs spanning the ADIPOQ locus in 500 subjects randomly selected from the total study population. The relation of these SNPs to serum adiponectin level was assessed, and those that showed such an association were genotyped in the remaining 2797 subjects. The 30 SNPs examined encompassed the genomic region (25 kbp) containing ADIPOQ exons and introns plus 8 kbp upstream and 1 kbp downstream (chromosome 3: genome positions 188034376–188059821) and were selected from those listed in the National Center for Biotechnology Information database dbSNP build 129 or those described in published studies. Several of the SNPs have previously been associated with adiponectin level32 and metabolic disorders including obesity,33, 34 type II diabetes mellitus,24, 28 or cardiovascular disease.25 In addition, we only selected SNPs for which the minor allele frequency (MAF) was shown to be >1% in Japanese subjects. The average distance between analyzed SNPs was 0.85 kb.

Statistical analysis

Allele frequencies were calculated by allele counting. The χ2 test was used to identify departures from Hardy–Weinberg equilibrium. The relation between serum adiponectin level and various parameters or ADIPOQ variants was evaluated by simple linear regression analysis or multivariate regression analysis. The relation of ADIPOQ genotype to continuous clinical parameters was evaluated by analysis of variance or analysis of covariance, with Scheffe's test used for post hoc analysis. Categorical variables were compared among genotypes with the use of the χ2 test. Student's t-test was used for dominant model analysis. Statistical analysis was performed with SPSS version 14.0 software (SPSS, Chicago, IL, USA). For continuous variables, results are presented as means±SD. Given that the distributions of serum adiponectin and triglyceride concentrations were skewed, values were logarithmically transformed for statistical analysis. In the screening study, the cutoff P-value was defined as <0.1 for simple linear regression analysis in order to avoid false negative results. Otherwise, a P-value of <0.05 was considered statistically significant. Haplotype analysis to detect linkage disequilibrium (LD) was performed with Haploview 4.0 software (Broad Institute, Cambridge, MA, USA).

Results

Clinical characteristics

The clinical characteristics of the KING study and replication study participants are shown in Table 1. The 3297 subjects of the KING study had a mean age of 63.6±6.5 years, with 55.8% of them being female. The average serum adiponectin level was 10.9 μg/ml (95% confidence interval (CI), 10.7–11.1 μg/ml), with that in women being significantly higher than that in men (Table 1; Supplementary Table 2). The 750 male subjects of the replication study had a mean age of 57.8±3.1 years and an average serum adiponectin level of 8.5 μg/ml (95% CI, 8.2–8.8 μg/ml).

Table 1 Characteristics of the study subjects
Full size table

Relation of adiponectin level to clinical parameters

We first examined the relation between serum adiponectin concentration and clinical parameters in all KING study subjects (Supplementary Table 2). Serum adiponectin level was positively associated with serum HDL-cholesterol concentration and negatively associated with BMI, blood pressure, serum concentrations of low-density lipoprotein cholesterol and triglyceride, fasting serum glucose concentration, and HbA1c level by simple linear regression analysis and by multivariate regression analysis with adjustment for age and sex. Consistent with previous observations,11 it also showed a significant association with hypertension, hyperglycemia, dyslipidemia, and the overweight condition.

Selection of SNPs in ADIPOQ and analysis of LD

We performed a screening study for 30 polymorphisms in and around ADIPOQ with 500 subjects randomly selected from the 3297 participants of the KING study (Table 1; Figure 1). Consistent with previous observations,32 we detected two LD blocks, and several SNPs also showed strong pairwise LD (Figure 2). SNPs rs4632532 (–19148C → T), rs1648707 (–19119C → A), and rs822388 (–8564T → G) were in strong LD (r2>0.95), as were rs16861194 (−11426A → G) and rs60806105 (–11156del/insCA); rs822391 (–7045T → C), rs822394 (–4120C → A), and rs822396 −3971A → G); rs2241766 (+45T → G), rs2241767 (+349A → G), and rs2082940 (+3317T → C); and rs3774261 (+712G → A), +2019del/insA, and rs1063538 (+3336T → C). Minor alleles of four SNPs (rs17300539 (−11391G → A), Gly90Ser, rs17366743 (Tyr111His), and Arg112Cys) were not found (MAF=0). Among the 30 SNPs analyzed, rs1656930 (−17991G → A), rs9882205 (–450A → G), and Ile164Thr were significantly associated with serum adiponectin level (P=0.034, 0.054, and 9.0 × 10–4, respectively) (Table 2).

Figure 1
figure 1

Schematic representation of the genomic structure and polymorphic variants of ADIPOQ. Open and closed boxes represent noncoding and coding regions of exons, respectively. Nucleotide positions are relative to the translation initiation codon (ATG). Asterisks indicate rare polymorphisms that result in amino-acid substitutions.

Full size image
Figure 2
figure 2

Haploview plot of the 30 SNPs in and around ADIPOQ. Dark red blocks, D′=1 with logarithm of the odds ratio for linkage (LOD) >2.0; blue blocks, D′=1 with LOD <2; light red and pink blocks, D′<1.0 with LOD >2.0; white blocks, D′<1 with LOD <2.0. Numbers in blocks are D′ values × 102.

Full size image
Table 2 Simple linear regression analysis of the relation between 30 SNPs of ADIPOQ and serum adiponectin level in an initial screening study
Full size table

Relation of ADIPOQ SNPs to serum adiponectin concentration and metabolic traits

The three SNPs associated with serum adiponectin level in the screening study as well as rs266729 (−11377C → G), the minor allele of which was previously associated with hypoadiponectinemia and type II diabetes,26 were genotyped in the full complement of 3297 subjects and examined for their relation to serum adiponectin concentration. The minor alleles of rs1656930, rs9882205, and Ile164Thr, but not that of rs266729, were associated with low serum adiponectin levels by simple linear regression analysis. These associations remained significant in multiple regression analysis after adjustment for age and sex either alone or together with other clinical parameters including BMI (Table 3). The minor allele of rs1656930, which was strongly associated with hypoadiponectinemia (Figure 3), was relatively common (MAF=6.2%). In contrast, the minor allele of Ile164Thr, which showed the strongest association with low serum adiponectin levels, was rare (MAF=0.9%). A 70-year-old man homozygous for the minor allele of Ile164Thr had the lowest adiponectin level (0.6 μg/ml) among the study population and was diagnosed with MetS, including hypertension and dyslipidemia. The minor allele of rs9882205 was common (MAF=38.5%) and showed a weak but significant association with hypoadiponectinemia. Conditional regression analysis with adjustment for each SNP indicated that rs1656930, Ile164Thr, and rs9882205 were independently associated with serum adiponectin level, although Ile164Thr and rs9882205 showed evidence of possible interdependence (Supplementary Table 3).

Table 3 Relation of serum adiponectin level to variation in ADIPOQ
Full size table
Figure 3
figure 3

Association of genotype (dominant model) for rs1656930 with serum adiponectin level in the KING study participants. Data are means with 95% CI. *P=5.2 × 10–6, †P=4.4 × 10−3, ‡P=2.1 × 10−5 (Student's t-test).

Full size image

A replication study confirmed the association of the minor alleles of rs1656930 and Ile164Thr with serum adiponectin level in an independent population, although this population did not include any subjects who were homozygous for the minor alleles of these SNPs (Table 3). No significant association of rs9882205 and serum adiponectin level was detected in the replication study.

Finally, we assessed the relation of the four genotyped SNPs to clinical parameters in the KING study participants. The minor allele of rs1656930 was positively associated with systolic blood pressure and the prevalence of hypertension (Table 4). Indeed, all subjects homozygous for the minor allele of rs1656930 had hypertension. This association with systolic blood pressure remained significant in multiple regression analysis after adjustment for adiponectin level (Supplementary Table 4). The minor allele of Ile164Thr was associated with the prevalence of MetS (P<0.03), but not with that of the other metabolic disorders including hypertension. The minor alleles of rs9882205 and rs266729 were not associated with any clinical parameters (data not shown).

Table 4 Relation between clinical characteristics and genotype for rs1656930 in the KING study participants
Full size table

Discussion

We have performed a fine-mapping analysis of ADIPOQ and assessed the relation of gene variants to serum adiponectin concentration in a population-based study. As far as we are aware, this study provides the first demonstration of an association between a common variant of ADIPOQ (the minor allele of rs1656930) and hypoadiponectinemia.

ADIPOQ SNPs and adiponectin level

The SNP rs1656930 (−17991G → A) is located 6.6 kbp upstream of the transcription start site of ADIPOQ. The adiponectin gene contains two regions of high LD, one encompassing the promoter region and the other including intron 2 as well as the coding and 3′ untranslated regions of exon 3.32 We confirmed that the LD block in the promoter region extended to and included rs1656930 (Figure 2). It is thus possible that rs1656930 affects transcriptional activity or is in high LD with an unknown SNP in this region that affects such activity. The rs1656930 and rs864265 SNPs are in perfect pairwise LD (r2=1) in the HapMap database, but we did not genotype this latter SNP in this study. The combined group of subjects with the GA or AA genotype of rs1656930 had a significantly lower adiponectin level than did those with the GG genotype. This difference was also apparent when men and women were analyzed separately (Figure 3).

We found that the minor allele of the Ile164Thr polymorphism of ADIPOQ was also significantly associated with low adiponectin levels, consistent with previous observations.12, 25 Case–control studies have shown that heterozygosity for the Ile164Thr polymorphism was associated with metabolic abnormalities including hypertension and type II diabetes in Japanese.12, 24 We have now found that this polymorphism was associated with the prevalence of MetS, but not with that of hypertension or hyperglycemia. One subject homozygous for the Thr allele of this polymorphism had a markedly low adiponectin level (0.6 μg/ml) and features of MetS. As far as we are aware, a subject with the Thr/Thr genotype for this SNP has not been previously described. The discrepancy between our current data and the previous results may be due to differences in age or the frequency of hypertension or diabetes among study populations.

Whereas rs1501299 (+276G → T) was previously found to be associated with adiponectin level in obese Japanese subjects,28 it was not associated with serum adiponectin concentration in our initial screening study. In the Framingham Offspring study,35 two SNPs in the promoter region of ADIPOQ (rs822387, rs17300539) were shown to be strongly associated with plasma adiponectin level, whereas rs17366743 (Tyr111His) was associated with the prevalence of type II diabetes. However, in this study, none of these three SNPs was associated with adiponectin level or hyperglycemia in the screening population.

A genome-wide association study recently showed that rs17366568 and eight other SNPs within ADIPOQ were associated with plasma adiponectin level in Europeans.36 Although we did not genotype rs17366568, rs1656930 is independent of this SNP (r2=0) in the HapMap database. Of the other eight SNPs, only rs822394 showed possible pairwise LD (r2=0.585) with rs1656930 in the HapMap database. However, in this study, unlike rs1656930, rs822394 did not have a significant effect on serum adiponectin level (Table 2). Moreover, rs864265, which is in perfect LD with rs1656930, showed no significant association with adiponectin level in the European study.

The discrepancies in findings among these various studies may result from underlying differences in the study populations including the frequencies of ADIPOQ polymorphisms.

ADIPOQ SNPs and metabolic traits

ADIPOQ has been studied previously as a candidate gene for metabolic disorders,22 and the relation of gene variants to circulating adiponectin levels has long been recognized.23, 37 Despite the strong association between adiponectin concentration and metabolic disorders described in various studies, other analyses have failed to detect a significant association between ADIPOQ variants and metabolic phenotype.36, 38 We also detected few associations between metabolic abnormalities and the selected variants of ADIPOQ that are present within genomic regions previously shown to be associated with metabolic phenotype. We had a power of 100% in the KING study to detect associations of ADIPOQ variants with metabolic abnormalities similar to those described by Iwashima et al12 (assuming α=0.05), but the previously observed association of Ile164Thr with hypertension was not replicated in our study. Such discrepancies may reflect the multiple genetic or environmental factors that contribute to metabolic disorders.

Adiponectin and hypertension

We found that only the minor allele of rs1656930 was positively associated with systolic blood pressure and the prevalence of hypertension. Of note, the increase in systolic blood pressure associated with the minor allele of this SNP was apparent only for homozygotes (Table 4). Although this allele was also associated with a low adiponectin level, the association with systolic blood pressure remained significant after adjustment for adiponectin level. Hypertension is a complex disorder that results from the interplay of genetic and environmental influences. Hypoadiponectinemia has been implicated as an important risk factor for hypertension.12, 39 Hypertensive individuals with insulin resistance were also found to have a low adiponectin level.18, 40 Plasma adiponectin concentrations were significantly lower in men with hypertension than in normotensive men and were negatively correlated with blood pressure in subjects without hypertension.12 In addition, adiponectin levels were negatively correlated with blood pressure in normotensive subjects without abnormal glucose tolerance.12 Consistent with these clinical observations, adiponectin knockout mice manifest impaired endothelium-dependent vasodilation on an atherogenic diet17 as well as salt-sensitive hypertension.41 Adiponectin supplementation ameliorated elevated blood pressure in both obese KKAy mice and salt-fed adiponectin knockout mice.41 These results suggest that hypoadiponectinemia contributes to the development of hypertension.

Study limitations

In this study, there was no subject homozygous for the minor allele of rs1656930 in the replication population, reflecting the relatively small number of subjects in this population (n=750) compared with the total number of subjects in the KING study (n=3297). We were thus not able to confirm the association of the minor allele of rs1656930 with hypertension in the replication study. We had a power of 22.6% in the replication study to detect effects similar to those observed in the KING study (assuming α=0.05). Verification that rs1656930 is associated with hypertension will thus require independent confirmation in adequately powered samples. Similarly, we could not confirm the association of rs9882205 with serum adiponectin level in the replication study.

Conclusion

Our data demonstrate an association of the minor allele of the rs1656930 polymorphism of ADIPOQ both with a low serum adiponectin level and with the prevalence of hypertension in the Japanese population. Hypoadiponectinemia resulting from physical inactivity and excessive calorie intake is thought to contribute to the development of metabolic disorders and cardiovascular disease. Screening for a common genetic background of hypoadiponectinemia may, therefore, provide important information for the assessment and management of metabolic disorders.