Diabetologia

, Volume 50, Issue 4, pp 747–751

A genetic variation of the transcription factor 7-like 2 gene is associated with risk of type 2 diabetes in the Japanese population

  • M. Horikoshi
  • K. Hara
  • C. Ito
  • R. Nagai
  • P. Froguel
  • T. Kadowaki
Short Communication

DOI: 10.1007/s00125-006-0588-6

Cite this article as:
Horikoshi, M., Hara, K., Ito, C. et al. Diabetologia (2007) 50: 747. doi:10.1007/s00125-006-0588-6

Abstract

Aims/hypothesis

It has been suggested that transcription factor 7-like 2 protein (TCF7L2) plays an important role in glucose metabolism by regulating the production level of glucagon-like peptide-1, a hormone which modifies glucose-dependent insulin secretion. Recently, variants of TCF7L2 gene were reported to confer an increased risk of type 2 diabetes in three different samples from European and European-origin populations. We studied whether the single nucleotide polymorphisms (SNPs) in TCF7L2 were associated with type 2 diabetes in samples from a Japanese population.

Methods

Five SNPs were genotyped in three different sample sets. Association with type 2 diabetes was investigated in each, as well as in combined sample sets.

Results

The SNP rs7903146 was nominally associated with type 2 diabetes in the initial (p = 0.08) and two replication sample sets (p = 0.05 and 0.06). For the combined sample set, in which we successfully genotyped 1,174 type 2 diabetes patients and 823 control subjects, rs7903146 showed a significant association with type 2 diabetes (odds ratio = 1.69 [95% CI 1.21–2.36], p = 0.002) with the same direction as the previous reports in samples from European and European-origin populations. SNPs rs7903146 and rs7901695 were in complete linkage disequilibrium. The rest of the five SNPs (rs7895340, rs11196205 and rs12255372) did not show any significant associations with type 2 diabetes.

Conclusions/interpretation

The consistent association between rs7903146 in TCF7L2 and type 2 diabetes in different ethnic groups, including the Japanese population, suggests that TCF7L2 is a common susceptibility gene for type 2 diabetes.

Keywords

AssociationSusceptibility geneType 2 diabetes

Abbreviations

GLP-1

glucagon-like peptide-1

HOMA

homeostasis model assessment

LD

linkage disequilibrium

OR

odds ratio

PAR

population attributable risk

SNP

single nucleotide polymorphism

TCF7L2

transcription factor 7-like 2

Introduction

Transcription factor 7-like 2 protein (TCF7L2) regulates the production level of proglucagon, which is the precursor of the insulinotropic hormone glucagon-like peptide 1 (GLP-1) in enteroendocrine cells [1]. GLP-1 exerts critical effects on blood glucose homeostasis by increasing insulin secretion. TCF7L2 also has an essential role in the developmental and growth regulatory mechanisms of intestinal epithelial cells which secrete GLP-1, because TCF7L2-deficient mice lack an intestinal epithelial stem cell compartment [2]. TCF7L2 could influence susceptibility to type 2 diabetes by altering levels of GLP-1 and/or other hormones. Moreover, TCF7L2 is located in a chromosomal region that has been reported to be linked to type 2 diabetes in the Icelandic population [3]. Therefore, TCF7L2 is a plausible candidate gene for type 2 diabetes. Recently, the Icelandic group reported that genetic variations located in introns of TCF7L2 were significantly associated with type 2 diabetes in samples from Icelandic, Danish and US populations [4].

In this study, we investigated whether the previously demonstrated genetic variations shown to be strongly associated with type 2 diabetes in the samples from European and European-origin populations are also associated with type 2 diabetes in samples from a Japanese population.

Subjects and methods

Subjects

We performed association studies in three different sample sets (Table 1). In the initial and replication sample sets, diabetic patients were randomly recruited from among those attending the outpatient clinic of the Department of Metabolic Diseases, Graduate School of Medicine, University of Tokyo (Tokyo, Japan), and the non-diabetic subjects from among those undergoing annual health check-ups at the Hiroshima Atomic Bomb Casualty Council Health Management Center (Hiroshima, Japan). Another unrelated 356 diabetic patients and 192 control subjects were recruited from the same region and same facility in Hiroshima to avoid bias due to population stratification. The inclusion criteria for non-diabetic subjects were as follows: (1) >60 years of age; (2) HbA1c values <5.8%; and (3) no family history of type 2 diabetes in first- and second-degree relatives. Diabetes was diagnosed in accordance with WHO criteria [5]. All participants gave informed consent, and the Ethics Committee of the University of Tokyo approved this study.
Table 1

Clinical characteristics (means ± SD) of the subjects

 

Initial

Replication

Hiroshima

All subjects

Diabetes

Control

Diabetes

Control

Diabetes

Control

Diabetes

Control

n

192

272

657

360

356

192

1,205

824

Male

123

129

421

141

217

93

761

363

Female

69

143

236

219

139

99

444

461

Age at onset of diabetes (years)

48.4 ± 11.2

NA

49.8 ± 10.9

NA

57.5 ± 11.0

NA

51.9 ± 11.6

NA

Age at examination (years)

62.8 ± 9.6

68.6 ± 6.7

63.2 ± 9.4

70.5 ± 6.8

70.6 ± 7.1

68.6 ± 6.7

65.3 ± 9.4

69.4 ± 6.8

BMI (kg/m2)

23.9 ± 3.9

24.0 ± 3.8

24.4 ± 3.9

23.6 ± 3.5

24.0 ± 3.3

23.8 ± 4.0

24.2 ± 3.7

23.8 ± 3.7

Fasting glucose (mmol/l)

8.9 ± 3.0

5.2 ± 0.5

8.9 ± 3.0

5.0 ± 0.8

8.0 ± 3.2

5.2 ± 0.9

8.6 ± 3.1

5.1 ± 0.7

HbA1c (%)

7.4 ± 1.4

5.2 ± 0.3

7.6 ± 1.7

5.2 ± 0.2

6.9 ± 1.2

5.2 ± 0.2

7.4 ± 1.6

5.2 ± 0.2

NA Not applicable

Single nucleotide polymorphism genotyping

We genotyped all five TCF7L2 single nucleotide polymorphisms (SNPs) previously described by Grant et al. [4]. rs7901695 and rs7895340 were genotyped by direct sequencing, performed with a BigDye terminator (Applied Biosystems, Foster City, CA, USA) and resolved using an ABI 3700 automated DNA sequencer (Applied Biosystems). rs7903146, rs11196205 and rs12255372 were genotyped using Taqman SNP Genotyping Assays by means of an ABI 7900HT (Applied Biosystems) according to the manufacturer’s protocol. Genotyping success rates were 99% (rs7901695), 98% (rs7903146), 94% (rs7895340), 99% (rs11196205) and 99% (rs12255372). Concordance rate, based on duplicate comparisons in 192 control subjects and 192 type 2 diabetes, was 100%. All the SNPs were in accordance with Hardy–Weinberg equilibrium in type 2 diabetes subjects (p > 0.17), control subjects (p > 0.5) and the whole subject group (p > 0.13).

Biological measurements

Insulin resistance and beta cell function were quantified using homeostasis model assessment (HOMA-IR and HOMA-beta, respectively); HOMA-IR = (fasting insulin [pmol/l]) × glucose [mmol/l])/22.5 × 6 and HOMA-beta = (fasting insulin [pmol/l]) × 20)/(glucose [mmol/l]−3.5) × 6 as described elsewhere [6] (original equation modified to incorporate SI units, with mU/l converted to pmol/l). Data were expressed as means ± SD.

Statistical analysis

The proportions of genotypes or alleles were compared between type 2 diabetic and non-diabetic subjects using a multiple logistic regression analysis adjusted for age, sex and BMI. Differences in HOMA according to genotypes were determined by analysis of covariates in non-diabetic subjects, after adjustment for age, sex and BMI. The statistical analyses were performed using JMP for Windows version 4.00 software (SAS Institute, Cary, NC, USA). To examine the pairwise linkage disequilibrium (LD) structure, r2 between the SNPs were estimated via the method of maximum likelihood from two-locus genotype data using the expectation–maximisation algorithm under the assumption of Hardy–Weinberg equilibrium. The calculations were performed with SNPAlyze v3.2 Pro software (Dynacom, Yokohama, Japan). We considered p < 0.05 to be significant. The odds ratio (OR) was assessed by counting the number of risk alleles for each individual, and we used the number of risk alleles to predict case/control status using logistic regression. Population attributable risk (PAR) was calculated as PAR = (p[OR − 1])/(1 + p[OR − 1]), where p is the prevalence of subjects with the risk allele.

Results

Genotype and allele frequencies of the five SNPs in TCF7L2 are shown in Table 2. Genotype frequency of rs7903146 was nominally associated with type 2 diabetes in each sample set (p = 0.08, 0.05, 0.06; initial, replication and Hiroshima sample sets, respectively). However, when all the samples were combined, both genotype and minor allele frequencies were significantly associated with type 2 diabetes (OR 1.69 [95% CI 1.21–2.36], p = 0.0075 and p = 0.002; genotype and minor allele frequencies, respectively). This significant association did not change when adjustment for BMI was omitted from the multiple logistic regression analysis (data not shown). The minor allele frequency of rs7903146 in the samples from a Japanese population (0.03–0.05) was substantially smaller than that of the previously reported three samples from European and European-origin populations, in which minor allele frequencies ranged from 0.27 to 0.39. Association between rs7903146 and type 2 diabetes in a dominant model was also examined, as its minor allele frequency was very low, but did not reach significance in two of our three sample sets. However, when all the samples were combined, we could confirm the significantly increased risk of type 2 diabetes in subjects with CT or TT (10.5 vs 6.4%, type 2 diabetes vs controls, respectively; OR 1.75 [95% CI 1.23–2.48], p = 0.0018).
Table 2

Genotype and allele frequencies (n [%]) of TCF7L2 SNPs

 

Initial

Replication

Hiroshima

All subjects

Minor allele frequency

OR (95% CI)c

Diabetes

Control

p valuea

Diabetes

Control

p valuea

Diabetes

Control

p valuea

Diabetes

Control

p valuea

Diabetes/ control

p valueb

rs7901695d

               

 CC

165 (87)

251 (92.3)

             

 CT

22 (12)

21 (7.7)

             

 TT

2 (1)

0 (0)

0.08

            

rs7903146

               

 CC

165 (87)

251 (92.3)

 

584 (89.8)

338 (94)

 

302 (90)

181 (95)

 

1,051 (89.5)

770 (93.6)

    

 CT

22 (12)

21 (7.7)

 

64 (9.8)

20 (5.5)

 

33 (10)

10 (5)

 

119 (10.2)

51 (6.2)

 

0.05/0.03

0.002

 

 TT

2 (1)

0 (0)

0.08

2 (0.4)

2 (0.5)

0.05

0 (0)

0 (0)

0.06

4 (0.3)

2 (0.2)

0.0075

  

1.69 (1.21–2.36)

rs7895340

               

 GG

148 (85)

226 (90.4)

 

559 (87.6)

306 (90)

 

308 (91.7)

162 (90.5)

 

1,015 (88.4)

694 (90.2)

    

 GA

23 (13)

23 (9.2)

 

75 (11.8)

32 (9.4)

 

27 (8)

17 (9.5)

 

125 (10.9)

72 (9.4)

 

0.06/0.05

0.16

 

 AA

3 (2)

1 (0.4)

0.15

4 (0.6)

2 (0.6)

0.53

1 (0.3)

0 (0)

0.65

8(0.7)

3 (0.4)

0.37

  

1.18 (0.88–1.58)

rs11196205

               

 GG

161 (83.9)

244 (90)

 

578 (88)

319 (88.6)

 

310 (89)

172 (90.7)

 

1,049 (87.6)

739 (89.6)

    

 GC

28 (14.6)

27 (10)

 

76 (11.6)

38 (10.6)

 

35 (10)

18 (9.3)

 

139 (11.6)

83 (10)

 

0.06/0.05

0.13

 

 CC

3 (1.5)

0 (0)

0.03

3 (0.4)

3 (0.8)

0.67

3 (1)

0 (0)

0.40

9 (0.8)

3 (0.4)

0.29

  

1.21 (0.92–1.56)

rs12255372

               

 GG

175 (93)

254 (94)

 

615 (93.6)

343 (95.3)

 

330 (95.1)

185 (96.4)

 

1,120 (93.7)

782 (95.0)

    

 GT

16 (7)

17 (6)

 

41 (6.2)

16 (4.4)

 

16 (4.6)

7 (3.6)

 

73 (6.1)

40 (4.9)

 

0.03/0.02

0.21

 

 TT

0 (0)

0 (0)

0.46

1 (0.2)

1 (0.3)

0.43

1 (0.3)

0 (0)

0.64

2 (0.2)

1 (0.1)

0.44

  

1.21 (0.82–1.81)

ap values are based on genotype frequencies.

bp values are based on allele frequencies.

cORs were calculated using an additive genetic model that in logistic regression is multiplicative on the OR scale. OR for each SNP was adjusted simultaneously for age, sex and BMI.

dReplication and Hiroshima sample sets were not genotyped owing to the complete LD between rs7901695 and rs7903146.

We found a significant interaction between SNP rs7903146 and BMI (p = 0.031) in the logistic regression analysis, suggesting that the effect of SNP rs7903146 on the risk of type 2 diabetes was different according to BMI. When we restricted the subjects to those with BMI lower than the median (BMI < 23.8 and <23.5 kg/m2, for type 2 diabetes patients and control subjects, respectively), rs7903146 showed a higher OR (2.02 [95% CI 1.28–3.21], p = 0.0027; Electronic supplementary Table 1). The association was negative in subjects with BMI higher than the median (OR 1.32 [95% CI 0.81–2.17], p = 0.27).

rs7901695 showed a significant association with type 2 diabetes in the initial sample set (minor allele frequency: diabetes/control; 0.07/0.04, p = 0.04), and because it was in complete LD (r2 = 1.0; Electronic supplementary Fig. 1) with rs7903146, we did not conduct further genotyping in the replication and Hiroshima sample sets. There were no significant differences in genotype or allele frequencies between type 2 diabetes patients and control subjects regarding rs7895340, rs11196205 and rs12255372 (Table 2).

We tested the SNP rs7903146 for quantitative trait association in non-diabetic subjects. rs7903146 did not show any association with age, sex, BMI and other clinical parameters related to type 2 diabetes such as HbA1c, fasting glucose, fasting insulin, HOMA-IR (CC vs CT/TT; 1.83 ± 1.2 vs 1.65 ± 0.8, p = 0.25), and HOMA-beta (CC vs CT/TT; 100.0 ± 65.7 vs 97.6 ± 49.8, p = 0.79) in the non-diabetic subjects.

Discussion

In this study, we found a significant association in samples from a Japanese population between the variation in TCF7L2 and type 2 diabetes, an association similar to that previously reported in samples from European and European-origin populations [4, 713]. It is noteworthy that the association between the SNP in TCF7L2 and type 2 diabetes has consistently been observed in different ethnic groups [14, 15], which supports the reliability of both previous studies as well as our present study. The mechanism of action of TCF7L2 in glucose metabolism and the pathogenesis of type 2 diabetes has yet to be elucidated, but it is possible that TCF7L2 has a role in regulating glucose-sensitive insulin secretion from beta cells. The prevalence of type 2 diabetes in the Japanese population is as high as in the USA [16], although the prevalence of obesity is much lower than that seen in Western countries [17]. One of the possible explanations is that fewer Japanese than European subjects are able to secrete enough insulin to compensate adequately for insulin resistance due to obesity [18]. Therefore, it is important to clarify the genetic components of susceptibility to insulin deficiency. Based on the present results, subjects having an at-risk allele account for 4% of the population, and the corresponding PAR is 3%, a value much lower than that in samples from European and European-origin populations (21%) [4]. It is possible that TCF7L2 plays a substantial role in genetic susceptibility to insulin deficiency in the Japanese population. Florez et al. [19] reported that TCF7L2 polymorphisms were associated with increased risk of developing type 2 diabetes in samples from a population of European origins. In that study, associations between TCF7L2 polymorphisms and type 2 diabetes in other ethnic groups including Asians were also investigated. However, no significant associations were identified, possibly due to the small sample size. The present study provides important information suggesting TCF7L2 is a type 2 diabetes susceptibility gene common to various ethnic groups including Japanese.

Acknowledgement

This work was supported by a Grant-in-aid (to T. Kadowaki) from the Organization for Pharmaceutical Safety and Research and by a Grant-in-aid for the 21st Century COE Program (to R. Nagai). We thank Y. Miyama for her technical assistance.

Duality of interest

There is no duality of interest.

Supplementary material

125_2006_588_MOESM1_ESM.doc (46 kb)
Table 1Genotype and allele frequencies (n [%]) of rs7903146 (DOC 47 kb)
125_2006_588_MOESM2_ESM.ppt (380 kb)
Figure 1Gene structure of TCF7L2 and pairwise LD among SNPs in Japanese subjects. The upper portion shows the gene structure of TCF7L2. The horizontal line represents the introns and the vertical lines represent the exons. The five SNPs genotyped in this study were located in intron 3 or intron 4. The lower portion shows the schematics of the pairwise LD (r2) in the Japanese type 2 diabetes and controls separately. The dotted lines connect each SNP name with the corresponding cell in the LD schematic. r2 values are shown in each cell and the level of LD is shown by colour(PPT 388 kb)

Copyright information

© Springer-Verlag 2007

Authors and Affiliations

  • M. Horikoshi
    • 1
    • 2
  • K. Hara
    • 1
    • 2
  • C. Ito
    • 3
  • R. Nagai
    • 4
  • P. Froguel
    • 5
  • T. Kadowaki
    • 1
    • 2
  1. 1.Department of Metabolic Diseases, Graduate School of MedicineUniversity of TokyoTokyoJapan
  2. 2.CRESTJapan Science and Technology Corporation (JST)TokyoJapan
  3. 3.Medical Court Life Care ClinicHiroshimaJapan
  4. 4.Department of Cardiovascular Medicine, Graduate School of MedicineUniversity of TokyoTokyoJapan
  5. 5.Genomics and Molecular Physiology of Metabolic DiseasesCNRS UMRLille CedexFrance