Journal of Diabetes & Metabolism

Type 2 Diabetes Mellitus (T2DM) is a substantial health issue worldwide with increasing prevalence at alarming rate [1]. India leads the world with largest number of diabetic patients (around 40.9 million in 2006 which is expected to rise to ~70 million by 2025) and is therefore termed as “Diabetes capital of the world” [2]. T2DM is characterized by pancreatic beta cell dysfunction and insulin resistance as a result of the interaction of genetic and environmental factors. Association of a number of genes with T2DM has been shown in the recent past [3]. Recent GWAS have identified several unsuspected genes (with previously unknown functions in pathology of T2DM) associated with T2DM [4-8]. However, the underlying molecular mechanisms in the development of diabetes remain poorly understood. In the present study, we have examined the association of the most significant genetic variant of loci CDKN2A/2B (rs10811661) and WFS1 (rs6446482) with diabetes in the population of Eastern Uttar Pradesh, India. These variants have been found to be convincingly associated with T2DM in Caucasian populations [6-14]. However, recently Nemr et al. [15] have found that CDKN2A/2B rs10811661 is not associated with T2DM in Lebanese.

CDKN2A and CDKN2B are adjacent Cyclin dependent kinase inhibitor genes on chromosome 9p. SNP rs10811661 located upstream of CDKN2A/2B showed genome-wide significant association to T2DM (OR 1.20, P=7.8×10^-15) in all data meta-analysis. SNP rs10811661 may have a long-range effect on one of the genes, or may influence a gene not yet annotated [6]. The region of association is limited to a 9 kb region flanked by strong recombination hot-spots, in which there are multiple conserved non-coding sequences but no known genes or microRNAs [9]. CDKN2A and CDKN2B encode p16^INK4a and p15^INK4b respectively. p16^INK4a inhibits CDK4, a powerful regulator of pancreatic beta cell replication. In mice, CDKN2A over-expression leads to islet hypoplasia and diabetes [8].

WFS1 encodes wolframin, a transmembrane glycoprotein that maintains calcium homeostasis of the endoplasmic reticulum. Mutations in this gene causes Wolfram syndrome (OMIM 222300), characterized by diabetes insipidus, juvenile-onset non-autoimmune diabetes mellitus, optic atrophy and deafness. Disruption of WFS1 gene in mice causes overt diabetes or impaired glucose tolerance depending on genetic background. Both humans and mice deficient in wolframin show pancreatic beta cell loss. Thus, WFS1 is critical for survival and function of insulin-producing pancreatic beta cells. Four of the SNPs rs10010131, rs6446482, rs752854 and rs734312 (H611R) in the WFS1 locus have been convincingly shown to be associated with T2DM in Caucasians, having odds ratios (ORs) 0.90~0.92 and P values 1.3×10^-4 ~ 1.4×10^-7 [10]. Replication of the association between variants in WFS1 and risk of type 2 diabetes was also shown by Franks et al. [16] in European population. Statistically significant associations between the major alleles of three variants of WFS1 rs10010131, rs1801213/ rs7672995 and rs734312 and prevalent T2D in the DESIR cohort at baseline was observed by Cheurfa et al. [17]. Also, the most frequent haplotype at the haplotype block containing the WFS1 gene modulated insulin secretion and was associated with an increased risk of type 2 diabetes.

Sample collection

Samples were collected from Eastern Uttar Pradesh in this casecontrol study. Blood from diabetic patients and normal healthy controls (>35 years) was collected after informed consent according to the approved protocol by the Institutional Ethical Committee of Banaras Hindu University from the patients attending out-patient departments of Institute of Medical Sciences, Banaras Hindu University, Heritage Hospital and Prakash Pathology, Varanasi.

Screening of the study subjects

We have genotyped two single nucleotide polymorphisms, CDKN2A rs10811661 and WFS1 rs6446482 in 517 unrelated individuals from Eastern Uttar Pradesh, India, including 271 type 2 diabetic patients and 246 ethnically matched control subjects. Subjects were diagnosed diabetic according to WHO criteria (1999). Subjects were included in the diabetes group if they had fasting glucose concentrations ≥ 126 mg/dl or 2-hour glucose concentrations ≥ 200 mg/dl after a 75 g Oral Glucose Tolerance Test (OGTT). Clinical history of diabetes and associated complications as well as the family history were recorded. Non-diabetic control subjects were chosen based on the absence of a history of diabetes in the subject and among first-degree relatives, as well as normal glucose tolerance, confirmed by a 75 g oral glucose tolerance test. After screening with standard OGTT, age, gender and Body Mass Index (BMI) matched 246 normal healthy controls were enrolled from the population undergoing routine health check-up.

Anthropometric and biochemical evaluation

Anthropometric measurements including weight, height, and waist were obtained using standard protocol. The BMI was calculated as the weight in kilograms divided by the square of height in meters. Clinical and biochemical data (fasting plasma glucose (FPG) and 2 hour postprandial plasma glucose (PPPG)) were obtained as part of our study protocol.

DNA analysis and genotyping

Blood sample (4-5 ml) was taken in 0.5 M EDTA (Sigma, USA) vials. Genomic DNA was extracted from peripheral blood leucocytes using the standard salting-out method.

CDKN2A/2B: We genotyped 271 diabetic subjects and 246 healthy controls. Polymorphic region of CDKN2A/2B (rs10811661) was PCR amplified using a forward primer: 5’-ATAAGCGTTCTTGCCCTGTC- 3’ and reverse: 5’-GTCAAAAACCTTCCCCATCC-3’. The cycling conditions were 94°C for 5 minutes, followed by 32 cycles of 94°C for 30 seconds, 58°C for 30 seconds, 72°C for 30 seconds and final extension at 72°C for 7 minutes. The PCR products of 121 bp were digested with BspHI (Bangalore Genie, India) for the rs10811661 (C/T) polymorphism. The resulting products were electrophoresed on a 3 % agarose gel.

WFS1: We genotyped 234 diabetic subjects and 234 normal healthy controls (smaller sample size than CDKN2A/B is due to the genotyping failure in some of the samples). A 136 bp polymorphic region of rs6446482 was PCR-amplified using forward primer: 5’-TGGTCACCTCACAGCTCAGT-3’ and reverse primer: 5’-TGCAAGAGGAGAGGAGTCG-3’ followed by RsaI (Bangalore Genie, India) digestion for the rs6446482 (C/G) polymorphism. The cycling conditions were 94°C for 5 minutes, 35 cycles at 94°C for 30 seconds, 58°C for 30 seconds, 72°C for 30 seconds and final extension at 72°C for 7 minutes. The resulting products were electrophoresed on a 3% agarose gel.

Statistical analysis

Clinical information of the T2DM patients and healthy controls was recorded as shown in Table 1. Data on quantitative characteristics are expressed as mean ± SD. Allele and genotype frequencies were compared between T2DM patients and controls using 2×2 and 2×3 contingency tables respectively with χ² test. Odds ratio (OR) with 95% Confidence Interval (CIs) was determined to describe the strength of association. P values of less than 0.05 were considered statistically significant.

Table 1: Clinical characteristics of the study populations for CDKN2A/2B and WFS.

The study group cases had significantly higher values of BMI (P=0.024 for CDKN2A/2B; P=0.0027 for WFS1); waist circumference (P=0.0128 for CDKN2A/2B; P < 0.0001 for WFS1); FPG (P<0.0001 for both the SNPs) and 2h-PPPG (P<0.0001 for both the SNPs) compared to controls. The cases were also found to be of significantly higher age group than healthy controls (Table 1). CDKN2A/2B (rs10811661) and WFS1 (rs6446482) polymorphisms show Hardy- Weinberg distributions in our study group. The genotype and allele distributions of the SNPs, rs10811661 and rs6446482 are significantly different between the T2DM and control groups. The major allele T of rs10811661 of CDKN2A/2B (OR 1.50, 95% CI 1.109-2.032; P=0.009) shows modest effect size and the major allele C of rs6446482 of WFS1 also shows modest effect size (OR 1.43, 95% CI 1.074-1.896; P=0.014) (Tables 2 and 3) indicating that these variants are risk factors associated to type 2 diabetes in this population. Our data support the association of rs10811661 of CDKN2A/2B to T2DM under both dominant (P=0.0270; OR 1.50, 95% CI 1.05 - 2.14) and recessive (P=0.0308; OR 2.62, 95% CI 1.06-6.49) models, while rs6446482 of WFS1 shows association under only dominant model (P=0.0205; OR 1.54, 95% CI 1.07-2.22) (Table 4).

Table 2: Genotype /Allele frequency distribution of CDKN2A/2B rs10811661(C/T) variant among control subjects and type 2 diabetes patients and their Odds Ratio (OR).

Table 3: Genotype /Allele frequency distribution of WFS1 rs6446482 (C/G) variant among control subjects and type 2 diabetes patients and their Odds Ratio (OR).

Table 4: Comparison of dominant and recessive models for CDKN2A/2B and WFS1.

Further, in Normal Glucose-Tolerant (NGT) subjects, FPG and 2h- PPPG levels were found to be significantly higher in carriers of at risk allele (genotypes TT+CT) of rs10811661 of CDKN2A/2B (C/T) polymorphism (P=0.0031 and P=0.049 for FPG and 2h-PPPG, respectively) compared to non-carriers (genotype CC) (Table 5). Similarly, carriers of at risk allele of rs6446482 of WFS1(C/G) polymorphism (genotypes CC+CG) also showed significantly higher FPG and 2h-PPPG levels (P=0.007 and P=0.029 for FPG and 2h-PPPG, respectively) compared to non-carriers (genotype GG) (Table 6). Compared to the carriers of the risk genotypes of both rs1081166 of CDKN2A/2B and rs6446482 of WFS1 together in T2DM group, the FPG and 2h-PPG levels were not significantly different from the carriers of risk genotypes of either CDKN2A/2B (rs1081166) or WFS1 (rs6446482) separately (Table 7).

Table 5: Correlation of the CDKN2A/2B rs10811661(C/T) genotype with FPG and 2h- PPPG.

Table 6: Correlation of the WFS1 rs6446482 (C/G) genotype with FPG and 2h-PPPG.

Table 7: Combined effect of risk genotypes of CDKN2A/2B and WFS1 in T2DM group on FPG and 2h PPG compared with the risk genotypes of CDKN2A/2B and WFS1 separately (Mean ± S.D.).

Asian Indians have an increased risk of developing T2DM. Characteristic “Asian Indian Phenotype” (higher body fat percent but a lower lean mass for a given BMI, central obesity leading to high insulin resistance) make them more susceptible to the disease [2]. Most of the studies related to search for T2DM susceptibility loci have been conducted in populations of European decent. Since different gene-environment interactions operate in different populations to increase risk of developing diabetes, the association studies of western populations are required to be replicated in different Asian Indian populations, especially populations with a high prevalence of diabetes such as India. Moreover, Indian populations are not genetically homogeneous and are unique in their genetic origin and the mutations underlying susceptibility to complex disease [18]. The result shows the evidence for modest effect of CDKN2A/2B (rs10811661) and WFS1 (rs6446482) on susceptibility to T2DM in the population from Eastern Uttar Pradesh. The CDKN2A/2B (OR 1.5) shows modest effect size similar to other reports in some populations [6-9,11]. In our study WFS1 (OR 1.43) shows larger effect size compared to European populations [10,16,17] and the susceptible variant attain statistical significance in the present study.

Recent GWAS and meta-analysis provide convincing evidence for CDKN2A/2B gene region to be involved in T2DM [6-9,11,13,14]. Metaanalysis of genotype data from GWAS in northern Europeans have confirmed that SNPs rs10811661 and rs564398 in the CDKN2A/2B region are T2DM susceptibility variants, although the combined evidence for rs10811661 is far stronger than that for rs564398 [6,8,9]. GWAS in French-Canadian obtained nominal association signals for proxies (r² ≥ 0.9) of rs10811661 (rs2383208, P=2×10^-4) and rs564398 (rs1063192, P=2×10^-2) [7]. A strong association between the major allele of rs10811661 and T2DM was reported in French Europids [19], Chinese Hans population [20], and Korean population [21]. In a Danish population, variants of CDKN2A/2B was found to be highly associated with T2DM with an OR of 1.30 per risk allele and the SNPs within CDKN2A/2B loci impaired glucose induced insulin release in healthy Danes [22]. Association of variant of CDKN2A/2B was modestly replicated in Asians but not replicated in African Americans and Pima Indians [23]. Recent GWAS in Diabetes Prevention Program (DPP) have shown CDKN2A/2B (rs10811661) as a potential intervention-interaction site showing response to treatment with Troglitazone by improving insulin sensitivity [24]. A strong association between CDKN2A/2B (rs10811661) and T2DM has been reported in Chinese population with an Odds ratio 1.406 [25]. Similar Odds ratio has been reported for CDKN2A/2B (rs10811661) OR 1.3 by Chauhan et al. [26] in the populations of a region of North-Western India. We report a higher OR 1.5 for the same SNP rs10811661 of CDKN2A/2B in the Eastern region of India. India represents one of the largest human diversity, consisting of 4635 culturally and anthropologically welldefined populations with little or no gene flow between them [18]. The deviation in our results from those of Chauhan et al. [26] may probably be due to this genetic heterogeneity. Bao et al. [13] have also reported OR 1.28 for CDKN2A/2B rs10811661 in a recent meta-analysis.

Association of the WFS1 with T2DM has been reported in several UK studies and one of the Ashkenazi Jewish populations where 1,536 SNPs in 84 candidate genes (including WFS1) that regulate pancreatic beta cell development, growth, function and survival, were genotyped in the association study of 9,533 cases and 11,389 controls. A haplotype with rs1801208 (R456H) and rs734312 (H611R) in WFS1 showed a nominal association with T2DM in Japanese subjects [27]. A metaanalysis in Caucasians for the WFS1 gene showed strong evidence of statistical association [16]. In an accompanying report from DPP showed that the carriers of the protective variants in WFS1 exhibited a trend towards increased insulin secretion [28]. Recent study in Korean population does not show association between WFS1 rs734312 and T2DM [21]. Another SNP for WFS1 (rs10010131) did not show significant association to T2DM in Chinese populations [25]. A casecontrol study in Japanese population has shown a marginal association between WFS1 rs12511742 and T2DM. However, association between rs6446482 and T2DM was not apparent in the same case-control study, although the OR was found to be in the same direction as that in Caucasians, probably due to lower minor allele frequency of the SNP in Japanese (0.02 in Japanese and 0.33~0.42 in Caucasians) [29]. We show association with modest effect size (OR 1.43) for WFS1 rs6446482 in the present study group.

We show modest effect size similar to other reports (loc. cit.) for CDKN2A/2B rs10811661 (OR 1.50, 95% CI 1.109-2.032; P=0.009) indicating that it can be a susceptible variant predisposing to T2DM in the population of Eastern Uttar Pradesh, India. For WFS1 rs6446482 (OR 1.43, 95% CI 1.074-1.896; P=0.014) the effect size is larger compared to other studies, implicating it to be a predisposing risk factor for T2DM in our population and the gene variant showed statistical significant association. The significantly higher values for fasting plasma glucose and 2-hour plasma glucose levels in the TT+CT genotype group of rs10811661 of CDKN2A/2B and CC+CG genotype group of rs6446482 of WFS1 polymorphisms among NGT subjects further support our finding that the ‘T’ allele of CDKN2A/2B and ‘C’ allele of WFS1 are risk alleles for type 2 diabetes mellitus. However, our data does not show any significant difference when the FPG and 2 h PPG values of the risk genotypes of both the variants combined together were compared with the risk genotypes of both CDKN2A/2B and WFS1 separately. Recently, Hribal et al. [30] also showed that TT genotype of rs10811661 polymorphism is associated with impaired insulin release and IGT suggesting that this variant may contribute to T2D by affecting beta cell function. Study at a larger scale with functional characterization is warranted to reliably confirm the association of the above variants and define their role(s) in future preventive measures for T2DM.

In conclusion, we replicate the association of well established common variants of CDKN2A/2B rs10811661(C/T) and WFS1 rs6446482 (C/G) with type 2 diabetes in the population of Eastern Uttar Pradesh, India. Interestingly, our data for both of the SNPs show larger effect size than those reported in European populations. To our knowledge this is the first report in this population and provides valuable information for comparison with other ethnic groups as well as in determining disease susceptibility in this population. However, in view of the genetic diversity of Indians, both these results need to be replicated in other ethnic groups.

This work was supported by Banaras Hindu University from 10^th plan and 11^th plan to GN and SS respectively. The support of Heritage Hospital and Prakash Pathology, Varanasi is acknowledged for diabetic and control sample collection. We also acknowledge KK Singh, Neha Singh (Department of Statistics, BHU) and Satya Piplani (Statistician, IMS, BHU) for statistical analysis.