Introduction

Type 2 diabetes is a prevalent chronic disease caused by the complex interplay of multiple genetic variants and environmental factors. In the past two decades, a total of 18 loci have been confirmed as type 2 diabetes risk loci with modest effect sizes in populations of European descent.1, 2 Although most of these same loci have been confirmed to be associated with type 2 diabetes in Chinese,3, 4, 5, 6 differences in allele frequency, genetic background, lifestyle and environmental exposures may lead to inconsistent results or different effect size of some genetic variants on type 2 diabetes risk between Chinese and European (for example, TCF7L2, HHEX-IDE, JAZF1, CDC123/CAMK1D, TSPAN8/LGR5).3, 6

The association between the five loci (JAZF1, CDC123/CAMK1D, TSPAN8/LGR5, VEGFA and ADAMTS9) and type 2 diabetes were first identified by Zeggini et al.7 with meta-analysis of three genome-wide association studies. Subsequently, several replication studies indicated that most of the novel type 2 diabetes genes (JAZF1, TSPAN8/LGR5, CDC123/CAMK1D and so on) appeared to affect various specific aspects of β-cell function.8, 9, 10 However, no significant association was detected for single-nucleotide polymorphisms (SNPs) from JAZF1, TSPAN8-LGR5, THADA, ADAMTS9, NOTCH2 in the previous Han Chinese study.3 Replication studies in other Asian populations just identified marginal significance of type 2 diabetes risk associated with some of these novel loci.11, 12, 13 We thought that the possible reasons for no or marginal significance were the limited sample size of these Asian population studies. To identify the effects of these genes on type 2 diabetes, we conducted a meta-analysis by pooling our data with those reported in the previous studies of East Asians.

HHEX-IDE locus was located at 10q region, which has previously been linked to fasting glucose, HbA1c levels and type 2 diabetes.14 Both HHEX and IDE genes have strong biological claims for a role in type 2 diabetes pathogenesis. However, all significant variants of this locus identified by genome-wide association studies of different ethnic groups were all located in 3′ region of HHEX gene,15, 16, 17, 18 therefore, some studies suggested that HHEX was the most likely causal candidate gene for type 2 diabetes in chromosome 10q23–q2519 and paid much attention to the HHEX gene. Here, eight common tag-SNPs, which covered both HHEX and IDE genes, were genotyped to analyze their effects on type 2 diabetes. Moreover, we also conducted a meta-analysis only in Chinese population to examine rs1111875 type 2 diabetes risk for the inconsistent results of previous studies.

Materials and methods

Participants

A total of 3953 individuals, Han Chinese, were enrolled in the current analysis from Shanghai, China.20 An interview was conducted by trained physicians or public health workers from the Pudong and Baoshan Centers for Disease Control and Prevention. The study individuals need to meet the following two criteria: (1) they were stable residents in the areas and (2) they were free from severe psychological disorders, physical disabilities and diagnosed with cancer, tuberculosis, AIDS and other communicable diseases within 6 months. Written informed consent was obtained from all the participants.

A total of 1912 type 2 diabetes patients (785 men, 1127 women; age 63.9±9.5 years) were defined in accordance with World Health Organization criteria (fasting plasma glucose (FPG) 7.0 mmol l−1 and/or 2-hour plasma glucose 11.1 mmol l−1). Moreover, more than 50% of type 2 diabetes patients were diagnosed at the age of 45–65 years and have 5–15 years duration of disease. In all, 2041 controls (635 men, 1406 women; age 58.1±9.4 years),4, 20, 21 with a FPG concentration <6.1 mmol l−1, were enrolled from the same geographical area with cases. The overall age and sex distribution of controls were matched with that of case group.

Height, weight, hip and waist circumference, and blood pressure were measured by trained medical professionals using a standardized protocol. Body mass index (BMI) was calculated as weight (kg) per (height (m))2. Total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, HbA1c and FPG were measured enzymatically according to standard methods using Roche modular P800 autoanalyzer (Roche, Mannheim, Germany) with reagents (Roche Diagnostics CmbH). Data are showed as medians (25–75% range) or means±s.d. (Supplementary Table 1). High molecular weight genomic DNA was prepared from venous blood using the QuickGene 610L Automatic DNA/RNA Extraction System (Fujifilm, Tokyo, Japan).

SNPs selection

A total of 13 SNPs were genotyped, providing at least one representative SNP for each of the seven genes previously reported to be of some genome-wide significance in Europeans. These included five SNPs (rs864745 in JAZF1, rs7961581 in TSPAN8/LGR5, rs12779790 in CDC123/CAMK1D, rs4607103 in ADAMTS9 and rs9472138 in VEGFA) with minor allele frequency >5% (7). For the HHEX-IDE locus, only representative SNPs with r2<0.80 based on HapMap Han Chinese were selected for genotyping, and eight variants (rs1111875, rs2275729, rs5015480, rs7923837, rs1887922, rs2209772, rs1999763 and rs11187007) which had positive association with type 2 diabetes or diabetes-related traits in previous studies were selected.15, 16, 17, 18, 22, 23, 24, 25, 26

Genotyping

All 13 variants were genotyped using TaqMan genotyping assay on an ABI7900 system (Applied Biosystems, Foster City, CA, USA). Genotyping success rates were above 95% except rs12779790 and all mismatch rates were below 1% in duplicate samples.

Statistical analysis

We first evaluated the Hardy–Weinberg equilibrium for specific allele-genotype frequencies between cases and controls. We then calculated linkage disequilibrium (LD) coefficients (D′ and r2) in the control group.27 The association of all SNPs with type 2 diabetes was then examined by logistic regression adjusting for sex, age and loge-transformed BMI. We conducted a meta-analysis using comprehensive Meta Analysis software (Version 2.0, BIOSTAT, and Englewood, NJ, USA). Cochran's Q-test was performed to assess heterogeneity across different studies. For quantitative traits in relation to genotypes, a general linear statistical methodology was used, applying additive, dominant and recessive models while adjusting for the effect of age and sex (BMI, waist-to-hip ratio), or age, sex and loge-transformed BMI (all other traits). The statistical analyses were performed by using the SPSS program (SPSS, Chicago, IL, USA). A P-value of <0.05 was considered significant.

Results

The genotype distributions of each SNP selected for the present study (see Methods) in the case and control groups were shown in Table 1. All 13 SNPs genotyped were in Hardy–Weinberg equilibrium among controls (Table 1). We further compared frequencies of the reported risk allele in the Chinese populations with those in the Japanese, Indian and European populations (Table 2 and Supplementary Table 2). The allele frequencies for most of the SNPs examined in the present study were considerably different from the populations of Indian and European descent, but close to the Japanese population.

Table 1 Association of candidate SNPs with type 2 diabetes in case and control individuals
Table 2 Comparison of risk-allele frequencies in the Chinese populations with those in other ethnic populations in previous studies

As shown in Table 1, rs7961581 in TSPAN8/LGR5 showed allelic difference between case and control groups (odds ratio (OR)=1.13, 95% confidence interval (CI) 1.01–1.26, P=0.031). After adjusting with age, gender and loge-transformed BMI, rs864745 and rs7961581 tend to have the same direction of the association with type 2 diabetes (P=0.085 and 0.055, respectively), as previously reported in European genome-wide association studies.7 However, we could not observe any association between other SNP loci (rs12779790, rs4607103, rs9472138) and type 2 diabetes (nominal P>0.05). Similarly, subsequent meta-analyses (including six populations of >10 000 individuals)3, 11, 13 only revealed that rs864745 (OR=1.09, 95% CI 1.03–1.16, P=3.49 × 10−3) and rs7961581 (OR=1.11, 95% CI 1.05–1.17, P=1.89 × 10−4) were significantly associated with type 2 diabetes risk in East Asian populations (Table 3).

Table 3 Meta-analysis of the association of candidate SNP loci with type 2 diabetes

For the HHEX-IDE locus, the LD coefficients (D′ and r2) (Supplementary Table 3) suggested low pair-wise LD among most of the eight SNPs. However, rs11187007 in IDE and rs1111875 in HHEX, although which were about 250 kb apart from each other, were in a high LD (D′=0.90, r2=0.77). Meanwhile, both rs11187007 and rs1111875 were identified to be associated with type 2 diabetes risk after further adjustment for age, gender and BMI (for both variants: OR=1.15, 95% CI 1.04–1.28, P=0.009 in an additive model; Table 1). In addition, meta-analysis of six populations, total individuals of 15 467 Chinese (7207 cases and 8260 controls)3, 5, 6, 28 confirmed the association of the HHEX rs1111875 with type 2 diabetes (OR=1.15, 95% CI 1.10–1.21, P=1.93 × 10−8) (Table 3).

In analysis of quantitative traits, rs7923837, rs2275729, rs1111875 and rs11187007 appeared to affect plasma triglyceride levels in our Chinese participants, whereas rs7923837 showed effects on FPG. However, none of these associations still reach conventional statistical significance after adjusting for multiple comparisons (Supplementary Table 4).

Discussion

In the present study, we identified a variant (rs11187007) in the IDE gene with increased risk of type 2 diabetes and confirmed the effects of HHEX-IDE locus on type 2 diabetes in Chinese population. Our meta-analysis pooling data emanated from East Asians indicated that the JAZF1 and TSPAN8 loci may confer modest susceptibility to type 2 diabetes risk, as identified in the European decent populations.

Although Florez et al.29 previously conducted a well-designed large-scale study and reported that genetic variants in the human IDE gene did not contribute to susceptibility of type 2 diabetes, available evidence does support that IDE has strong functional roles in type 2 diabetes’ pathogenesis. The IDE gene encodes a metallopeptidase, which can degrade insulin, initiate cellular insulin processing and terminate its action. Homozygous mice with IDE gene deletion resulted in hyperinsulinemia and glucose intolerance, which are classical hallmarks in type 2 diabetes development.30 Here, we found a novel association of rs11187007 with type 2 diabetes and triglycerides, in line with the findings that variants and haplotypes in IDE were associated with FPG, HbA1c and type 2 diabetes.22, 23, 30, 31

On the other hand, the association study and meta-analysis further confirmed the effect of HHEX rs1111875 on type 2 diabetes in Chinese population, consistent with previous genome-wide association studies and replication studies in other ethnic populations.15, 16, 17, 18, 19, 25, 26, 32, 33, 34 The HHEX gene encodes a transcription factor that is also involved in the Wnt signaling pathway and that has been shown to be essential for pancreas development. van Vliet-Ostaptchouk et al.19 suggested that HHEX was the most likely causal candidate in chromosome 10q. However, in our study, both IDE rs11187007 and HHEX rs1111875 were detected to be associated with type 2 diabetes. It also shows that a 270-kb LD block covers the whole HHEX-IDE locus in the HapMap date of Chinese subjects. Moreover, the SNPs do not reside within the coding or putative regulatory regions of any gene. Therefore, it is difficult to explain which gene may be the susceptibility gene. Given the consistency of the association between the HHEX-IDE locus and type 2 diabetes risk observed in multiple populations, it does seem warranted to conduct deep sequencing of this region to identify potential causal variants in these two genes that may directly affect type 2 diabetes risk.

The JAZF1 gene encodes a transcriptional repressor of NR2C2 (nuclear receptor subfamily 2, group C, ember 2). Nr2c2−/− knock out in mice retards growth and displays phenotypes of low serum concentrations of IGF1, hypoglycemia and reduced gluconeogenesis via PEPCK inactivation.7, 35 Zeggini et al.7 have previously reported a strong association between rs864745 (in intron 1 of the JAZF1 gene) and type 2 diabetes risk in Europeans (OR=1.10, P=5.0 × 10−14). In the study by Grarup et al.,9 it has been estimated that carriers of the T allele of JAZF1 rs864745 had a 2.6% (0.9–4.3%; P=0.003) decreased insulin release, as assessed by the BIGTT acute insulin response index. In this article, the association studies and meta-analysis further confirmed the association between JAZF1 rs864745 and type 2 diabetes in Chinese population. In addition, compared with the Europeans, the frequency of this risk variant in Chinese appeared to be much higher (T allele, 0.77 vs 0.50).

We also observed significant association between TSPAN8/LGR rs7961581 and type 2 diabetes with a modest OR (OR=1.11, 95% CI 1.05–1.17) in East Asian populations, similar to that in European descents. In the study by Grarup et al.,9 it was also reported that rs7961581 in TSPAN8 was associated with various OGTT-based surrogate measures of insulin release, suggesting a role of TSPAN8 in pancreatic β-cell function. In our meta-analysis, this variant also showed significant association with type 2 diabetes in Japanese population (OR=1.11, 95% CI 1.03–1.19, P=5.54 × 10−3). Therefore, rs7961581, which resides 110 kb upstream of TSPAN8 (tetraspanin 8), is associated with type 2 diabetes across the boundary of race.

In the current study, polymorphisms (rs12779790, rs4607103 and rs9472138) showed any significant relations neither with type 2 diabetes risk nor with quantitative metabolic traits. Aside from the fact that genetic background may differ between Chinese and Europeans, differences in statistical power across studies may also explain such inconsistent observations, especially when the SNP-type 2 diabetes relation was modest. For example, we had a power of 99% to detect an OR of 1.15 for rs11187007 (given minor allele frequency=0.30 and 4000 samples), but power declined to 79% for detecting an OR of 1.10.

In summary, in this large case–control study of type 2 diabetes among Han Chinese, the IDE rs11187007 and HHEX rs1111875 variants are significantly associated with increased type 2 diabetes risk. Our meta-analysis further confirmed the modest effects of the JAZF1, TSPAN8 and HHEX-IDE loci on type 2 diabetes risk in Chinese and other East Asians.