Lack of association between genetic polymorphisms within DUSP12 - ATF6 locus and glucose metabolism related traits in a Chinese population

Background Genome-wide linkage studies in multiple ethnic populations found chromosome 1q21-q25 was the strongest and most replicable linkage signal in the human chromosome. Studies in Pima Indian, Caucasians and African Americans identified several SNPs in DUSP12 and ATF6, located in chromosome 1q21-q23, were associated with type 2 diabetes. Methods We selected 19 single nucleotide polymorphisms (SNPs) that could tag 98% of the SNPs with minor allele frequencies over 0.1 within DUSP12-ATF6 region. These SNPs were genotyped in a total of 3,700 Chinese Han subjects comprising 1,892 type 2 diabetes patients and 1,808 controls with normal glucose regulation. Results None of the SNPs and haplotypes showed significant association to type 2 diabetes in our samples. No association between the SNPs and quantitative traits was observed either. Conclusions Our data suggests common SNPs within DUSP12-ATF6 locus may not play a major role in glucose metabolism in the Chinese.


Background
Type 2 diabetes is a complex disease caused by both genetic and environmental factors. Although recent genome-wide association studies have identified several novel, possibly causative genes, the contribution of them to disease risk is still very limited [1]. Thus the genetic architecture of type 2 diabetes remained largely unknown. Previous genome-wide linkage studies in multiple ethnic populations, including Caucasians, Chinese and Pima Indian, showed that chromosome 1q21-q25 was the strongest and most replicable linkage signal in the human chromosome [2][3][4][5][6][7][8][9]. Although genome-wide association studies show no strong association signal in this region, whether variants harbored in this region conferred modest effect to the disease are worthy to be analyzed.
Dual specificity phosphatase 12 (DUSP12) and activating transcription factor 6 (ATF6) were two neighbored genes locating on the chromosome 1q21-q23. DUSP12 is a glucokinase -associated protein identified from rat hepatic cDNA library through yeast two-hybrid, using glucokinase as bait. It may participate in glycolysis in the liver and pancreatic beta-cell through dephosphorylation of glucokinase in the cytoplasm [10]. ATF6 is a key sensor of endoplasmic reticulum stress. It activates unfolded protein response through regulating a group of genes encoding molecular chaperones and folding enzymes [11]. Previous studies identified several single nucleotide polymorphisms (SNPs) in this region associated with type 2 diabetes in different populations. Among them, rs2070150 (P145A) was firstly identified to be associated with type 2 diabetes in Pima Indian, while rs4579731, rs3820449 and rs10918215 were reported later in studies focusing on Caucasians and African Americans [12][13][14][15]. However, International Type 2 Diabetes 1q Consortium failed to detect any association signal on DUSP12 and ATF6 in a fine mapping study in multiethnic samples [16]. Even though, only 285 East Asian origin samples were included in the previous studies and no one analyzed the association between SNPs from this region and type 2 diabetes in large Asian samples. Therefore, we performed the present study, aiming to test if variants from DUSP12 and ATF6 played a role in the genetic susceptibility of type 2 diabetes in the Chinese.

Participants
In this case-control study, we recruited 3,700 unrelated Chinese Han subjects, including 1,892 case and 1,808 controls. All the subjects were eastern Chinese Han ancestry, residing in Shanghai and nearby region. In the present study, all case subjects were type 2 diabetes patients selected from Shanghai Diabetes Institute inpatient database. Control subjects were community-based populations enrolled from the Shanghai Diabetes Studies [17]. The inclusion and exclusion criteria for the cases and controls were described previously [17]. Briefly, all cases were type 2 diabetes patients defined according to 1999 WHO criteria (fasting plasma glucose ≥7.0 mmol/l and/or 2-h plasma glucose ≥11.1 mmol/l) and were treated with oral hypoglycemic agents and/or insulin. The control subjects were normal glucose tolerance defined based on fasting plasma glucose <6.1 mmol/l and 2-h plasma glucose <7.8 mmol/l. This study was approved by the institutional review board of Shanghai Jiao Tong University Affiliated Sixth People's Hospital. Written informed consent was obtained from each participant.

Clinical measurement
All subjects underwent detailed clinical investigations, as described previously [17]. Briefly, anthropometric parameters such as height, weight, waist and hip circumference (for the control subjects only) were measured. For the control subjects, blood samples were obtained at 0 and 120 min during the oral glucose tolerance tests (OGTTs) to measure plasma glucose and serum insulin levels. Lipid profiles such as total cholesterol and triglyceride were also obtained. Insulin resistance and pancreatic β-cell function were assessed by homeostasis model assessment (HOMA) [18]. HOMA-IR = fasting insulin × fasting plasma glucose ÷ 22.5, HOMA-B = 20 × fasting insulin ÷ (fasting plasma glucose ÷ 3.5).

SNPs selection, genotyping and quality control
We selected 19 SNPs that spanning 197 kb of DUPS12 and ATF6 region, from 10 kb 5' upstream the DUSP12 to 2 kb 3' downstream the ATF6. These SNPs could tag 98% of the SNPs with MAF over 0.1 derived from Hap-Map Phase III Chinese Han database under the threshold of r 2 ≥0.7. Among them, 7 SNPs located in the coding region. The SNPs previously reported were either directly genotyped or in linkage disequilibrium (LD) with genotyped SNPs. All the SNPs were genotyped using Sequenom's MassARRAY iPLEX system (MassARRAY Compact Analyzer, Sequenom, San Diego, CA, USA).
The key quality control requirements were: 1) sample call rates ≥75%; 2) SNP call rate ≥85%; 3) less than two discrepant genotypes of 100 duplicate samples; and 4) Hardy-Weinberg equilibrium test ≥0.05 in controls and cases respectively. After the quality control procedures of the genotypes, 71 individuals were excluded. And one SNP (rs3767635) failed Hardy-Weinberg equilibrium test. The average call rate for the remaining 18 SNPs was 97.5%, and the average concordance rate based on 100 duplicate comparisons for each SNP was 99.4%. Detailed information of the call rates and concordance rates for the SNPs was shown in the Additional file 1 Table S1.

Statistical analyses
Observed genotypes were tested for fit to the expectation of Hardy-Weinberg equilibrium using χ 2 test. Pairwise LD was estimated from the combined data of cases and controls calulating |D'| and r 2 using Haploview (version 4.1) http://www.broadinstitute.org/haploview/haploview [19]. Haplotype block structure was determined using confidence interval algorithm [20] and haplotype frequencies were estimated by Expectation-Maximization algorithm [21] using Haploview (v 4.1). Allele, genotype and haplotype frequencies for cases and controls were compared using χ 2 test or Fisher's exact test. Odds ratios (ORs) with 95% confidence intervals (CIs) were presented. The genotype -disease association analyses were performed under the additive model adjusting age, gender and BMI as confounding factors by logistic regression. Quantitative traits with skewed distribution were natural logarithmically transformed to approximate univariate normality. Quantitative traits were analyzed under an additive genetic model by linear regression adjusted for age, sex, and BMI. All statistical analyses were performed by SAS (version 8.0; SAS Institute Inc., Cary, NC, USA) unless specified otherwise. A two-tailed P value <0.05 was considered significant. The allele frequencies in HapMap populations and statistic power of the SNPs were shown in the Additional file 2 Table S2.

Results
A total of 18 SNPs were successfully genotyped in 3,629 individuals in the present study. The LD pattern of these SNPs was shown in Figure 1. Three haplotype blocks were constructed in this region.
The single SNP association analysis showed that no SNP was significantly associated with type 2 diabetes in our samples. The minimum P value was 0.0954 for rs10799941. Logistic regression analysis adjusting age, gender and BMI as confounding factors also suggested no association between SNPs and type 2 diabetes (Table 1). For the haplotype analysis, we compared the haplotype distributions between cases and controls and observed that no haplotype was nominally associated with type 2 diabetes (Table 2).
We then analyzed the association between SNPs and quantitative traits related to glucose metabolism in the individuals with normal glucose regulation. No SNP was associated with plasma glucose and serum insulin levels at fasting status as well as 2-h after glucose stimulation. No significant association was detected between insulin sensitivity and beta cell function either (Table 3).

Discussion
Genome-wide linkage studies in various populations suggested the existence of multiple susceptibility gene(s) for type 2 diabetes at chromosome 1q21-q24 [2][3][4][5][6][7][8][9]. Several specific genes in this region, such as LMNA, NOS1AP and ATF6, were identified that they might confer risk for diabetes in some populations [12][13][14][15][16][22][23][24].  Among these genes, ATF6 is a strong candidate by its biological function in endoplasmic reticulum stress and unfolded protein response, which linked insulin demand with beta cell failure and diabetes. ATF6 is also the binding target of WFS1, a known type 2 diabetes susceptible gene, and mediates its effect on endoplasmic reticulum stress [25]. However, although we performed the association study by analyzing 18 SNPs in 3700 Chinese Han, we failed to find any evidence of association between SNPs from this locus and traits related to glucose metabolism in our samples. One possible explanation might be that the statistical power of our samples was not enough to detect the effects of this locus in the Chinese population. Although we had over 80% power to detect the association at the 0.05 level based on the previously reported ORs in non-Asian populations (1.2 1.3) and allele frequencies of reported SNPs in our Chinese samples, we could not exclude the possibility the reported effect size was overestimated due to the "winner's curse" effect or novel associated SNPs with lower minor allele frequencies in the Chinese existed, in this case our samples may not have sufficient power. Secondly, the relatively loose criteria for tagging SNP selection, which is the limitation of the current study, missed information for a group of SNPs in this region. As we used r 2 over 0.7 and minor allele frequency over 0.1 as SNPs selection criterion, we failed to capture 15 (6.8%) SNPs if the stringent criterion r 2 over 0.8 and minor allele frequency over 0.05 was adopted. Thirdly, the LD pattern and allele frequencies differed between Chinese Hans and previously studied populations, which suggested population differences in the genetic architecture between Chinese and other ethnic populations, may also partly explain the lack of association between this locus and previously reported phenotypes. Finally we cannot exclude the possibility that rare variants within   this region participated in the pathogenesis of diabetes as we only focused on the common ones.

Conclusion
Our data suggests common variants within DUSP12 and ATF6 genes may not play a major role in glucose metabolism in the Chinese. However, due to the limitation of the current study, the effects of SNPs from this locus on type 2 diabetes need to be tested in further studies with larger East Asian origin samples and higher marker density.

Additional material
Additional file 1: Call rates and concordance rates of SNPs genotyped. This file contains detailed information of quality control analysis of the SNPs, including call rates and concordance rates.
Additional file 2: Allele frequencies and statistic power of the SNPs. This file contains the allele frequencies of all SNPs in the HapMap populations and our samples. The statistic power of the SNPs in our samples was also shown in this file.