Diabetologia

, 53:299 | Cite as

Common variants at the GCK, GCKR, G6PC2–ABCB11 and MTNR1B loci are associated with fasting glucose in two Asian populations

  • F. Takeuchi
  • T. Katsuya
  • S. Chakrewarthy
  • K. Yamamoto
  • A. Fujioka
  • M. Serizawa
  • T. Fujisawa
  • E. Nakashima
  • K. Ohnaka
  • H. Ikegami
  • T. Sugiyama
  • T. Nabika
  • A. Kasturiratne
  • S. Yamaguchi
  • S. Kono
  • R. Takayanagi
  • Y. Yamori
  • S. Kobayashi
  • T. Ogihara
  • A. de Silva
  • R. Wickremasinghe
  • N. Kato
Article

Abstract

Aims/hypothesis

To test fasting glucose association at four loci recently identified or verified by genome-wide association (GWA) studies of European populations, we performed a replication study in two Asian populations.

Methods

We genotyped five common variants previously reported in Europeans: rs1799884 (GCK), rs780094 (GCKR), rs560887 (G6PC2–ABCB11) and both rs1387153 and rs10830963 (MTNR1B) in the general Japanese (n = 4,813) and Sri Lankan (n = 2,319) populations. To identify novel variants, we further examined genetic associations near each locus by using GWA scan data on 776 non-diabetic Japanese samples.

Results

Fasting glucose association was replicated for the five single nucleotide polymorphisms (SNPs) at p < 0.05 (one-tailed test) in South Asians (Sri Lankan) as well as in East Asians (Japanese). In fine-mapping by GWA scan data, we identified in the G6PC2–ABCB11 region a novel SNP, rs3755157, with significant association in Japanese (p = 2.6 × 10-8) and Sri Lankan (p = 0.001) populations. The strength of association was more prominent at rs3755157 than that of the original SNP rs560887, with allelic heterogeneity detected between the SNPs. On analysing the cumulative effect of associated SNPs, we found the per-allele gradients (β = 0.055 and 0.069 mmol/l in Japanese and Sri Lankans, respectively) to be almost equivalent to those reported in Europeans.

Conclusions/interpretation

Fasting glucose association at four tested loci was proven to be replicable across ethnic groups. Despite this overall consistency, ethnic diversity in the pattern and strength of linkage disequilibrium certainly exists and can help to appreciably reduce potential causal variants after GWA studies.

Keywords

Asians Association study Ethnicity Fasting plasma glucose Polymorphisms 

Abbreviations

CEU

Utah residents with northern and western European ancestry from the Centre d’Etude du Polymorphisme Humain collection

FPG

Fasting plasma glucose

GWA studies

Genome-wide association studies

JPT

Japanese in Tokyo

LD

Linkage disequilibrium

RAF

Risk allele frequency

SNP

Single nucleotide polymorphism

Introduction

Fasting plasma glucose (FPG) levels are associated with the future risk of type 2 diabetes and cardiovascular diseases [1, 2] and are tightly regulated despite considerable variation in food intake [3]. It has been reported that genetic effects explain 54.8% of the variance of glucose levels in a European population [4]. Recent progress in complex-trait genetics has allowed the identification of loci regulating FPG levels [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20].

Several loci influencing FPG levels have been identified or verified by genome-wide association (GWA) studies of Europeans; these include glucokinase (GCK) [5, 6, 7], glucokinase regulatory protein (GCKR) [8, 9, 10, 13], glucose-6-phosphatase catalytic subunit 2 (G6PC2), the ATP-binding cassette, subfamily B (MDR/TAP), member 11 (ABCB11) [14, 15, 16, 17], and melatonin receptor 1B (MTNR1B) [16, 17, 18].

All the associations were originally identified in populations of European ancestry. While some studies have shown reproducible associations [9, 11, 12, 19, 20], it remains to be further defined to what degree loci discovered in Europeans will show an association in populations of different ancestries. In addition, to localise the variant(s) responsible for an association signal, we need to generate a comprehensive list of potential causal variants in the regions of interest, i.e. to conduct fine-mapping after GWA studies. As discussed elsewhere [21], this fine-mapping will be challenging and genetic information from populations of different ancestries is expected to be useful [21, 22, 23, 24].

Apart from assessing the previously identified variants in two Asian populations, Japanese of East Asian ancestry and Sri Lankan of South Asian ancestry, we also explored index single nucleotide polymorphism (SNP) markers, which either tag the SNPs attaining a locus-wise significance level in the GWA scan of Japanese or were previously reported in Europeans. This was done to advance the fine-mapping of the associated loci [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20].

Methods

Study populations

A replication study of the previously identified variants was performed in the general Japanese and Sri Lankan populations (Electronic supplementary material [ESM] Table 1, ESM Study samples for continuous traits), using 5,456 Japanese samples (including 4,813 non-diabetic participants) consecutively enrolled in a population-based setting as described elsewhere [25] and 3,012 Sri Lankan samples (including 2,319 non-diabetic participants) who had participated in the baseline survey of the Ragama Health Study [26] in Sri Lanka. Complementary to this replication study, we organised genetic studies of FPG levels as part of an ongoing GWA scan for cardiometabolic disorders among the Japanese population (ESM Study samples for continuous traits). We used 776 population-based, non-diabetic Japanese samples for preliminary screening of association with FPG levels. Then, the association signals were examined in the general populations mentioned above. In addition to quantitative trait analysis, type 2 diabetes associations were tested for index SNPs at G6PC2–ABCB11 in a Japanese case–control study panel comprising 5,629 cases and 6,406 controls as previously reported [27], and in a Sri Lankan case–control study panel (ESM Study samples for type 2 diabetes case–control studies). All participants from these different studies provided written informed consent and the local Ethics Committees approved the protocols.

Type 2 diabetes was diagnosed according to the WHO criteria as described in ESM Study samples for type 2 diabetes case–control studies.

SNP genotyping and quality control

In the replication study, samples were genotyped using the TaqMan assay (Life Technologies Japan, Tokyo, Japan) for five SNPs from four gene loci previously identified in European-descent populations [5, 6, 7, 8, 9, 10, 13, 14, 15, 16, 17, 18]. These included GCK (rs1799884), GCKR (rs780094), G6PC2–ABCB11 (rs560887) and MTNR1B (rs1387153 and rs10830963).

In the GWA scans, genotyping was performed with a bead array (Infinium HumanHap550; Illumina, San Diego, CA, USA) as described elsewhere [27] (ESM Fig. 1, ESM SNP genotyping, ESM Quality control of the GWA scan data). After the GWA scan, three additional SNPs in the G6PC2–ABCB11 region, rs483234, rs3755157 and rs853778, were genotyped with the TaqMan assay for follow-up.

Statistical analysis

SNP association analysis

SNPs were tested for association with FPG levels by using linear regression analysis in the additive genotype model (ESM SNP-based association analysis). A p value of <0.05 was considered statistically significant. For an association to be considered significant, it had to involve the same risk allele as that reported in Europeans and was accordingly assessed with a one-tailed test. To assess the proportion of variance for FPG that could be explained by a SNP, we calculated the coefficient of determination R2. The per-allele gradients, which correspond to the increase in FPG levels by additional ‘high FPG’ alleles of associated SNPs, were calculated in the linear regression model (including age, sex and BMI as covariates) as previously reported [14, 18] (ESM Evaluation of cumulative effect of multiple loci on FPG). We used PLINK (http://pngu.mgh.harvard.edu/∼purcell/plink/), the R software (version 2.8.1; www.r-project.org) and the rmeta package (http://cran.r-project.org) for association test and meta-analysis (websites accessed 15 October 2009).

Haplotype analysis

In the G6PC2–ABCB11 region, we selected SNPs attaining a locus-wise significance level (p < 0.002 by Bonferroni’s correction for 23 SNPs genotyped in the relevant region) or reported in European studies, inferring the haplotypes using PLINK [28] and PHASE [29] software (http://depts.washington.edu/ventures/UW_Technology/Express_Licenses/PHASEv2.php). We then tested which haplotypes were strongly associated with the trait. In parallel, haplotypes were inferred from the genotype data of the SNPs in HapMap (www.hapmap.org) Utah residents with northern and western European ancestry from the Centre d’Etude du Polymorphisme Humain collection (CEU) and Japanese in Tokyo (JPT) categories using HaploView software (www.broad.mit.edu/mpg/haploview/) [30] and in South Asians from the Human Genome Diversity Panel (http://hagsc.org/hgdp/files.html) [31]. Haplotype-tagging SNPs were selected and characterised in the large study panels.

Stepwise regression analysis for testing of independent associations

To test the most likely explanation for the signal of association among the index SNPs and their genotyped correlates, we performed stepwise linear regression analysis for FPG levels by forward selection (ESM Index SNPs showing an independent association). If two SNPs simultaneously included in the model each attained significance (p < 0.05), they could have independent associations. Further, when two haplotype classes that are distant in the phylogeny have an opposite effect and are tagged by two SNPs showing independent associations, the haplotype classes are presumably linked to different causative variants, thus implying allelic heterogeneity (ESM Haplotype explaining index association).

Cross-population filtering of causal variants

To appreciably narrow the location of potential causal variants, we closely inspected subsets of SNPs and haplotypes shared by multiple ethnicities. We partitioned all the HapMap SNPs located in the G6PC2ABCB11 region (Fig. 1) so that SNPs in the strongest linkage disequilibrium (LD) with one index SNP (e.g. rs3755157) rather than with other index SNPs (e.g. rs560887, rs483234 and rs853778) were grouped into a bin of rs3755157 correlates. We then narrowed target intervals by investigating a subset (or subsets) of variants that could show a consistent pattern of trait association across different ethnic groups (ESM Narrowing target intervals in fine-mapping).
Fig. 1

Plots of FPG association, and LD and SNP partitioning for the G6PC2–ABCB11 region in Japanese (a, d, g, j), Sri Lankan (b, e, h, k) and Europeans (c, f, i, l). Association results for Europeans are drawn from the published studies [14, 17]. a–c Bar graphs of all genotyped SNPs that passed the quality control (ESM Table 6) in the Japanese GWA scan (a) and those in the published European GWA scan [14] (c) with −log10 (p values) for FPG plotted against chromosome position in Mb. Red diamonds, p values for the genotypes of general populations (n = 4,813 in Japanese, n = 2,319 in Sri Lankan) and meta-analysis data (>12,000 in the ENGAGE consortium [17]). d–f Genomic location of G6PC2 and ABCB11 genes with intron and exon structure (NCBI Build 36) in the relevant populations. g–i WGAViewer (http://people.genome.duke.edu/∼dg48/WGAViewer/whatis.php) [36] plot of LD (r2) for all SNPs across the regions for the HapMap populations JPT (g) and CEU (i), and for the Human Genome Diversity Panel, ethnic groups belonging to South Asia (h). j–l SNPs in the LD block were partitioned into four subsets using the extent of LD with lead SNPs and/or haplotype-tagging SNPs, i.e. rs560887, rs483234, rs3755157 and rs853778 (see Methods). Red circles, SNP with 0.8 ≤ r2 ≤ 1.0 to the index SNP; orange circles, SNP with 0.6 ≤ r2 < 0.8 to the index SNP; green bars, intervals where causative variants are most likely to be located

Results

Association with FPG and metabolic traits at four loci

Significant (p < 0.05 by one-tailed test) association was replicated for all five SNPs from four tested loci in the Sri Lankan and Japanese populations (Table 1). Together with the previous reports in Europeans and Chinese [8, 11, 12, 17, 18, 20], we performed meta-analysis of FPG associations to compare the effect sizes among different ethnic groups (ESM Fig. 2). Among the four loci, significant cross-population heterogeneity was detected for rs1387153 (MTNR1B; p = 0.03). The variance for FPG that was explained by the associated SNPs totalled 2% in both the Japanese and Sri Lankan populations (ESM Table 2). Per-allele gradients in the two Asian populations (β = 0.055 and 0.069 mmol/l in Japanese and Sri Lankans, respectively) were almost equivalent to those reported in populations of European descent (β = 0.07 mmol/l) [18] (ESM Fig. 3, ESM Evaluation of cumulative effect of multiple loci on FPG).
Table 1

Association of SNPs with fasting plasma glucose level

SNP

Neighbouring gene(s)

Alleles

Japanese panel (n = 4,813)

Sri Lankan panel (n = 2,319)

Europeans

FPG+a

FPG−a

Allele frequencyb

Per-allele effectc

p value

Allele frequencyb

Per-allele effectc

p value

Allele frequency d

Per-allele effectc,e

Reported in European studies

rs780094

GCKR

G

A

0.44

0.032 (0.012, 0.052)

0.002

0.80

0.074 (0.035, 0.113)

2.1 × 10−4

0.62

0.067 (0.045, 0.090)

rs560887

G6PC2–ABCB11

G

A

0.97

0.103 (0.044, 0.162)

6.4 × 10−4

0.91

0.050 (−0.005, 0.105)

0.075

0.67

0.064 (0.056, 0.072)

rs1799884

GCK

A

G

0.18

0.075 (0.049, 0.101)

1.1 × 10−8

0.12

0.076 (0.028, 0.123)

0.002

0.20

0.062 (0.048, 0.076)

rs1387153

MTNR1B

T

C

0.41

0.058 (0.038, 0.078)

9.7 × 10−9

0.38

0.036 (0.005, 0.068)

0.024

0.28

0.07 (0.05, 0.08)

rs10830963

MTNR1B

G

C

0.42

0.056 (0.036, 0.075)

2.9 × 10−8

0.45

0.064 (0.033, 0.094)

3.6 × 10−5

0.30

0.072 (0.062, 0.082)

Tested in additionf

rs483234

G6PC2–ABCB11

A

G

0.51

0.046 (0.026, 0.065)

3.9 × 10−6

0.42

0.043 (0.012, 0.074)

0.007

0.70

rs3755157

G6PC2–ABCB11

T

C

0.38

0.057 (0.037, 0.078)

2.6 × 10−8

0.16

0.069 (0.027, 0.111)

0.001

0.07

rs853778

G6PC2–ABCB11

A

G

0.40

0.044 (0.024, 0.064)

1.5 × 10−5

0.35

0.026 (−0.006, 0.058)

0.117

0.46

Type 2 diabetes participants were excluded from the analysis; FPG association of each SNP was tested using linear regression models with adjustment for BMI, age and sex

aFPG-increasing (+) and decreasing (−)

bOf FPG-increasing allele

cEffect (95% CI) (mmol/l)

dOf FPG-increasing allele in HapMap CEU panel

eAssociation results were drawn from the previous studies: rs560887, rs4607517 (in substitution for rs1799884, r2 = 1 in CEU) and rs10830963 [17]; rs1260326 (in substitution for rs780094, r2 = 0.93 in CEU) [8]; rs1387153 [18]

fIn the G6PC2–ABCB11 region

Besides FPG levels, we analysed the relationship of SNPs with lipid traits (ESM Tables 3 and 4). Notably, rs780094 (GCKR) significantly and consistently modulated triacylglycerol levels in both ethnic groups (p = 2.2 × 10-10 in Japanese, p = 1.4 × 10-4 in Sri Lankan populations), where glucose-increasing alleles were associated with lower triacylglycerol levels as previously reported [8, 9, 11, 13]. Furthermore, glucose-increasing alleles at rs1799884 (GCK) and rs10830963 (MTNR1B) were significantly associated with reduced beta cell function (HOMA-B; p = 0.037 for rs1799884, p = 2.6 × 10-4 for rs10830963 in the Sri Lankan population), with no appreciable effect on fasting insulin or insulin sensitivity (ESM Table 5).

Refinement of genetic association in the G6PC2–ABCB11 region

In fine-mapping with Japanese GWA scan data, we identified in the G6PC2–ABCB11 region a novel associated SNP rs3755157, which was proven to be independent of the SNPs previously reported by GWA studies in Europeans [14, 15, 16, 17].

In our GWA scan, multiple and significant SNPs were found in the G6PC2–ABCB11 region (p = 0.0004 to 0.002; Fig. 1, ESM Table 6) but not in the other candidate regions (ESM Tables 79). We therefore performed a detailed investigation of the G6PC2–ABCB11 region. With reference to the LD and haplotype data (ESM Tables 1012), we chose four haplotype-tagging SNPs (rs3755157, rs483234, rs853778 and rs560887) for genotyping the general Japanese population (n = 4,813), which resulted in concordant evidence of associations between the tests of individual SNPs and those of haplotypes (Tables 1 and 2, ESM Haplotype explaining index association). The most significantly associated haplotype, class 5 (frequency = 0.35, p = 2.8 × 10-7) (Table 2), was almost unequivocally tagged by rs3755157, which showed the strongest association by SNP-based test in the general Japanese population (p = 2.6 × 10-8; Table 1).
Table 2

Fasting glucose association according to haplotypes in the G6PC2–ABCB11 region

Haplotype class

Tested SNPsa

Japanese panelb

Sri Lankan panelc

Europeans

rs560887

rs483234

rs3755157

rs853778

Frequency

Effect (mmol/l)

p value

Frequency

Effect (mmol/l)

p value

Frequencyd

1

Ga

G

C

G

0.48

−0.047

2.3 × 10−6

0.56

−0.041

9.6 × 10−3

0.26

2

Ga

G

C

Aa

0

0.02

−0.029

0.65

0

3

Ga

Aa

C

G

0.09

0.004

0.81

0.08

0.035

0.23

0.28

4

Ga

Aa

C

Aa

0.01

0.048

0.31

0.09

0.030

0.28

0.07

5

Ga

Aa

Ta

Aa

0.35

0.054

2.8 × 10−7

0.16

0.072

9.8 × 10−4

0.08

6

Ga

Aa

Ta

G

0.02

0.053

0.12

0

0

7

A

Aa

C

Aa

0.02

−0.098

2.5 × 10−3

0.08

−0.078

8.1 × 10−3

0.28

FPG association was tested with adjustment for BMI, age and sex

aAllele increasing fasting glucose

b4,792 complete observations

c2,306 complete observations

dHaplotype frequency estimated in the HapMap CEU panel

We then performed a stepwise linear regression (for FPG levels) to test whether one of the four haplotype-tagging SNPs was necessary and sufficient to explain the association signal (ESM Tables 1317, ESM Fig. 4, ESM Index SNP showing an independent association). The FPG association remained significant (p < 0.05) when two haplotype-tagging SNPs, rs3755157 and rs560887, were included in the regression model (ESM Table 13). This independent association had gone unnoticed among more significant associations of SNPs that were in strong LD with a leading SNP, rs560887, among Europeans [14, 15]. Thus the presence of a novel SNP, rs3755157, and of allelic heterogeneity (ESM Fig. 5) has become evident in the G6PC2–ABCB11 region for the first time, as a result of comparing the GWA scan data between European and Japanese populations.

In the G6PC2–ABCB11 region, G6PC2 and ABCB11 are both biologically plausible candidate genes [15, 32, 33]. During fine-mapping, we attempted to partition the LD block into intervals, each containing SNPs strongly correlated with an index SNP, in the hope that correlation coefficients r2 would reflect phylogenic closeness once the index SNPs were selected from a reasonably dense set of SNP markers. This partitioning approach helped to prioritise the target interval for fine-mapping, thereby reducing the potential candidate variants to manageable proportions. For the G6PC2–ABCB11 region, the target intervals were estimated to be 14 kb (in the ABCB11 gene) for rs3755157 and 14 kb (in the G6PC2 gene and between the genes) for rs560887 when the LD threshold was set at r2 ≥ 0.6 in the HapMap data (Fig. 1, ESM Table 18, ESM Narrowing target intervals in fine-mapping).

Concordance of association for FPG levels and type 2 diabetes

In addition to the quantitative trait analysis of FPG, we performed case–control analysis of type 2 diabetes for three of four haplotype-tagging SNPs, rs3755157, rs483234 and rs560887, in the G6PC2–ABCB11 region and confirmed a significant (p < 0.05) association in a relatively large study panel comprising 5,629 cases and 6,406 controls. The strongest association was found for rs3755157 (OR 1.09, 95% CI 1.03–1.15, p = 1.7 × 10-3; Table 3), which was in good agreement with the FPG association mentioned above. To confirm the consistency of associations with increases in FPG levels and the risk of type 2 diabetes, we examined the changes in risk allele frequency (RAF) between the diabetes subgroup and non-diabetic participants in quartiles of FPG levels stratified in the general populations for unbiased estimates (ESM Fig. 6, ESM Table 19). In the Japanese and Sri Lankan populations, the RAF in the diabetes subgroup reached the second highest quartile, supporting the concordant association for FPG levels and the risk of type 2 diabetes in the G6PC2–ABCB11 region.
Table 3

Association of FPG-altering SNPs with type 2 diabetes in the G6PC2–ABCB11 region

SNP

rs560887 (FPG-increasing allele: G)

rs483234 (FPG-increasing allele: A)

rs3755157 (FPG-increasing allele: T)

Japanese (JPN)a

   

 Frequency, cases (n = 5,629)

0.974

0.508

0.380

 Frequency, controls (n = 6,406)

0.969

0.490

0.360

 OR (95% CI)

1.20 (1.03–1.41)

1.07 (1.02–1.13)

1.09 (1.03–1.15)

 p value for trend

0.019

0.0056

0.0017

Sri Lankan (SL)b

   

 Frequency, cases (n = 599)

0.917

0.422

0.169

 Frequency, controls (n = 515)

0.896

0.408

0.159

 OR (95% CI)

1.28 (0.95–1.73)

1.06 (0.89–1.26)

1.08 (0.86–1.36)

 p value for trend

0.08

0.51

0.52

JPN and SL combined

   

 OR (95% CI)

1.22 (1.06–1.40)

1.07 (1.02–1.13)

1.09 (1.03–1.15)

European-descent

   

 OR (95% CI)

0.93 (0.89–0.97)c

1.05 (0.99–1.13)d

1.00 (0.91–1.09)d

 p value

0.0017

0.12

0.93

Type 2 diabetes association was tested with the Cochran–Armitage trend test in the case–control analysis

aSNPs were genotyped in a Japanese case–control study panel independently of the general Japanese population [27]

bIn the Sri Lankan population, 515 controls are part of those used for FPG association analysis, whereas 599 cases were independent participants

cResults for 18,236 cases and 64,453 controls from a previous study [17]

dResults for 4,549 cases and 5,579 controls from the DIAGRAM consortium [37]

Discussion

The present study has proven that common variant loci influencing FPG levels are reproducible in two populations of Asian descent, Japanese (East Asians) and Sri Lankan (South Asians). To our knowledge, this is the first study investigating the genetic associations with FPG and related metabolic traits at four candidate loci, GCK, GCKR, G6PC2–ABCB11 and MTNR1B, in South Asians, who are known to have high prevalence of type 2 diabetes [34]. The combined impact of associated SNPs is almost equivalent across the ethnic groups despite some cross-population diversity in the effect size of individual loci (ESM Figs 2 and 3, ESM Table 2). Other novel aspects of the present study include a fine-mapping approach using Japanese GWA scan data and consistent associations of FPG and type 2 diabetes in the G6PC2–ABCB11 region.

According to genome-wide patterns of SNPs examined in the Human Genomic Diversity Panel [31], much of sub-Saharan Africa, Europe, South and Central Asia (including Sri Lanka), and East Asia appear to be homogeneous and individuals from these populations can be distinguished from each other. Although limited in the number of examples, our study has provided evidence supporting the importance of human genetic diversity in complex disease studies. For instance, beside replicating FPG association at four candidate loci in two Asian populations, our data also clarified the genetic architecture of the G6PC2–ABCB11 region with regard to ethnic diversity. Using the GWA scan data, we found a novel SNP, rs3755157, to be a leading SNP among Japanese and independent of a leading SNP, rs560887, in Europeans (ESM Table 1315).

As a fine-mapping approach, we listed HapMap SNPs having the strongest r2 (in the range of r2 ≥ 0.6) with each of the index SNPs in the G6PC2–ABCB11 region (Fig. 1, ESM Table 18). We performed cross-population filtering, which appreciably decreased the number of potential causal variants from 79 to 8 in the G6PC2–ABCB11 region (ESM Narrowing target intervals in fine-mapping). The novel SNP rs3755157 and its correlated SNPs are located in the 3′-side (introns) of the ABCB11 gene. While four different mRNAs, two alternatively spliced variants and two unspliced forms, are known to be transcribed from the ABCB11, it is possible that the potential causal variant(s) will influence the selective production of any of the 3′-side mRNA variants or the alteration of mRNA expression. Thus, closer inspection of subsets of SNP haplotypes shared by multiple ethnicities may allow us to appreciably narrow the field of potential causal variants before starting in-depth resequencing and functional follow-up studies, as demonstrated for the G6PC2–ABCB11 region. During the preparation of our manuscript, replication of the G6PC2 association was also reported in a Chinese population [19], where four SNPs were selected from the HapMap database so as to tag common variations near the G6PC2 gene. Although the index SNP (rs560887) originally detected in Europeans was not tested, three (of four) SNPs appeared to show significant association in the Chinese population, in agreement with our findings in Japanese.

Our data also verified concordance of association for FPG levels and type 2 diabetes risk by using a systematic study design; i.e. unbiased estimates with stratification of general populations plus large-scale case–control studies involving 12,035 Japanese and 1,114 Sri Lankan samples (Table 3, ESM Fig. 6). It has been debated whether the genetic determinants regulating FPG levels in physiological states differ from those increasing type 2 diabetes risk. Some studies report that carriers of glucose-increasing alleles at three loci (MTNR1B, GCK and GCKR) show a higher risk of type 2 diabetes [8, 17, 18], although there is no significant association between G6PC2–ABCB11 variants and type 2 diabetes in populations of European descent [14, 17, 35]. In this context, our data not only supported the concordant association of G6PC2–ABCB11 variants for FPG and type 2 diabetes in two Asian populations, but also indicated that genetic determinants regulating FPG levels could, at least in part, differ from those increasing type 2 diabetes risk (ESM Fig. 6, ESM Consistent association of fasting glucose and type 2 diabetes in the G6PC2–ABCB11 region). It is likely that genetic susceptibility for FPG levels increases type 2 diabetes risk in the population at large, but that some diabetic patients will develop the disease independently of a predisposition to elevated FPG levels.

In summary, despite the overall reproducibility of FPG association across the populations, ethnic diversity in allele frequencies led to the discovery of allelic heterogeneity in the G6PC2–ABCB11 region. The diversity in the LD pattern also helped to reduce the probable causative variants in the corresponding region. The prevalence of the phenomena described here in human complex trait genetics is another research area warranting investigation. For applicable cases, the use of ethnic diversity in genetic studies can constitute an efficient approach subsequent to GWA scan.

Notes

Acknowledgements

This work was supported by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation Organization (NIBIO) and the Manpei Suzuki Diabetes Foundation. Support also came from the Ministry of Education, Culture, Sports, Science and Technology of Japan. We acknowledge the outstanding contributions of the International Medical Center of Japan (IMCJ) employees, who provided technical and infrastructural support for this work. Above all, we thank the patients and study participants who made this work possible and who give it value. We also thank all the people who continuously support the Hospital-based Cohort Study at IMCJ, the Amagasaki Study and the Kyushu University Fukuoka Cohort Study in Japan, and the Ragama Health Study in Sri Lanka. We also thank C. Makibayashi and the many physicians of the Amagasaki Medical Association, as well as M. Makaya, T. Mizoue, H. Janaka de Silva, U. Ranawaka and other staff at the University of Kelaniya for their help with collection of DNA samples and accompanying clinical information. The DNA samples of type 2 diabetes cases used for this research were partly provided by the Leading Project for Personalized Medicine in the Ministry of Education, Culture, Sports, Science and Technology, Japan. The GWA study conducted by NIBIO GWA Study Group was organised to clarify the pathogenesis of diabetes and associated metabolic disorders as well as cardiovascular complications. The collaborating institutions that constitute the NIBIO GWA Study Group are: International Medical Center of Japan; Kyushu University; Osaka University; Nagoya University; Kinki University; Shimane University; Tohoku University; the Institute for Adult Diseases, Asahi Life Foundation; Chubu Rosai Hospital; Amagasaki Health Medical Foundation; collaborating groups in the Amagasaki Medical Association; and collaborating groups in the Kyushu region.

Duality of interest

The authors declare that there is no duality of interest associated with this manuscript.

Supplementary material

125_2009_1595_MOESM1_ESM.pdf (181 kb)
ESM 1(PDF 184 kb)
125_2009_1595_MOESM2_ESM.xls (22 kb)
ESM Table 1(XLS 21.5 kb)
125_2009_1595_MOESM3_ESM.xls (26 kb)
ESM Table 2(XLS 25.5 kb)
125_2009_1595_MOESM4_ESM.xls (20 kb)
ESM Table 3(XLS 19.5 kb)
125_2009_1595_MOESM5_ESM.xls (20 kb)
ESM Table 4(XLS 20 kb)
125_2009_1595_MOESM6_ESM.xls (19 kb)
ESM Table 5(XLS 19 kb)
125_2009_1595_MOESM7_ESM.xls (24 kb)
ESM Table 6(XLS 23.5 kb)
125_2009_1595_MOESM8_ESM.xls (32 kb)
ESM Table 7(XLS 32.5 kb)
125_2009_1595_MOESM9_ESM.xls (24 kb)
ESM Table 8(XLS 24 kb)
125_2009_1595_MOESM10_ESM.xls (21 kb)
ESM Table 9(XLS 21 kb)
125_2009_1595_MOESM11_ESM.xls (28 kb)
ESM Table 10(XLS 28.5 kb)
125_2009_1595_MOESM12_ESM.xls (28 kb)
ESM Table 11(XLS 27.5 kb)
125_2009_1595_MOESM13_ESM.xls (23 kb)
ESM Table 12(XLS 23 kb)
125_2009_1595_MOESM14_ESM.xls (20 kb)
ESM Table 13(XLS 19.5 kb)
125_2009_1595_MOESM15_ESM.xls (19 kb)
ESM Table 14(XLS 19 kb)
125_2009_1595_MOESM16_ESM.xls (19 kb)
ESM Table 15(XLS 19 kb)
125_2009_1595_MOESM17_ESM.xls (18 kb)
ESM Table 16(XLS 18 kb)
125_2009_1595_MOESM18_ESM.xls (18 kb)
ESM Table 17(XLS 18 kb)
125_2009_1595_MOESM19_ESM.xls (37 kb)
ESM Table 18(XLS 37 kb)
125_2009_1595_MOESM20_ESM.xls (20 kb)
ESM Table 19(XLS 19.5 kb)
125_2009_1595_MOESM21_ESM.xls (22 kb)
ESM Table 20(XLS 21.5 kb)
125_2009_1595_MOESM22_ESM.pdf (231 kb)
ESM Fig. 1a, b Multi-dimensional scaling plots representing similarity among samples, with plot (b) zooming in around points for the GWA study samples (see ESM Quality control of the GWA scan data). HapMap YRI, Yoruba (PDF 236 kb).
125_2009_1595_MOESM23_ESM.pdf (183 kb)
ESM Fig. 2Meta-analysis of FPG association for five SNPs rs780094 (GCKR) (a), rs560887 (G6PC2) (b), rs1799884 (GCK) (c), rs1387153 (MTNR1B) (d) and rs10830963 (MTNR1B) (e). The forest plots are coloured separately for the East Asians (orange), South Asians (Sri Lankans, green) and Europeans (blue). When the per-allele effect sizes of each SNP were compared between the Japanese (present study) and the Chinese [17, 18, 19], no significant heterogeneity was found between the two populations. When the per-allele effect sizes of each SNP were compared among the three ethnic groups, heterogeneity was detected for rs1387153 in MTNR1B (p = 0.03). See ESM Text for references (PDF 187 kb).
125_2009_1595_MOESM24_ESM.pdf (221 kb)
ESM Fig. 3Cumulative effects of GCK rs1799884, GCKR rs780094, G6PC2–ABCB11 rs560887 and rs3755157, and MTNR1B rs10830963 variants on FPG levels in the general populations of Japanese (a) and Sri Lankans (b). The size of solid squares is proportional to the number of individuals in each category. To adjust for the inter-population difference in SE, z score of FPG (%) is plotted on the y-axis, as well as FPG in millimoles per litre. Data are presented as mean (95% CI); the β coefficient in the linear regression model (including age, sex and BMI as covariates) corresponds to the increase of FPG levels by additional FPG-increasing alleles (PDF 226 kb).
125_2009_1595_MOESM25_ESM.pdf (192 kb)
ESM Fig. 4Fasting plasma glucose level according to combined genotypes of rs560887 and rs3755157 in the Japanese (a) (n = 4,813) and Sri Lankan (b) (n = 2,319) populations. For each subgroup stratified by the combined genotypes, the estimate of mean FPG is depicted with a square and a vertical line (95% CI). The size of square is proportional to the frequency of each subgroup in the population. Pairwise comparison indicates the independence of FPG association signals at the two SNP loci in Japanese. For example, T allele of rs3755157 is significantly associated with increased FPG level when the genotype of rs560887 is fixed as either GG or AG. While a similar trend is observed, the corresponding effect does not attain statistical significance in the Sri Lankan population, probably due to the modest sample size (PDF 196 kb)
125_2009_1595_MOESM26_ESM.pdf (504 kb)
ESM Fig. 5The phylogeny of haplotypes in the G6PC2–ABCB11 region, defined by a selected set of SNPs (ESM Table 11). A haplotype class is depicted as a black circle, which is labelled with a unique number (as in ESM Table 11) and the frequency in each population. Alleles of a SNP that distinguish two haplotype classes are labelled on the line connecting the adjacent circles (i.e. haplotypes). The four underlined SNPs were characterised in the Japanese and Sri Lankan populations of the present study. The phylogeny of haplotypes/SNPs is almost consistent across three populations: HapMap JPT, Human Genome Diversity Panel (HGDP) South Asians and HapMap CEU. As for the combination of rs853774 and rs483234, four gametes appeared to exist presumably due to recombination (represented by dashed lines in the figure); as a consequence, this part of the phylogeny formed a 'rectangular' network in the HGDP South Asians and HapMap CEU. Two SNPs, rs853787 and rs853789, were not assayed in the HGDP South Asians and not listed in the phylogeny (PDF 516 kb).
125_2009_1595_MOESM27_ESM.pdf (449 kb)
ESM Fig. 6Comparison of RAF between the type 2 diabetes subgroup (T2D) and non-diabetic persons in quartiles of fasting glucose level as labelled, adjusted for BMI, age and sex. Results are shown for: rs560887 in (a) Japanese and (b) Sri Lankan; rs483234 in (c) Japanese and (d) Sri Lankan; and rs3755157 in (e) Japanese and Sri Lankan (f) populations, separately. A risk allele is defined as one correlated with increase in FPG. Sample size in each class from left to right was 1,200, 1,200, 1,200, 1,201, 919 in Japanese and 600, 600, 600, 600, 599 in Sri Lankans. Estimated mean RAF is depicted with a square and a vertical line (95% CI) (PDF 459 kb).

References

  1. 1.
    Tirosh A, Shai I, Tekes-Manova D et al (2005) Normal fasting plasma glucose levels and type 2 diabetes in young men. N Engl J Med 353:1454–1462CrossRefPubMedGoogle Scholar
  2. 2.
    Coutinho M, Gerstein HC, Wang Y, Yusuf S (1999) The relationship between glucose and incident cardiovascular events. A metaregression analysis of published data from 20 studies of 95,783 individuals followed for 12.4 years. Diabetes Care 22:233–240CrossRefPubMedGoogle Scholar
  3. 3.
    Xiang AH, Wang C, Peters RK, Trigo E, Kjos SL, Buchanan TA (2006) Coordinate changes in plasma glucose and pancreatic beta-cell function in Latino women at high risk for type 2 diabetes. Diabetes 55:1074–1079CrossRefPubMedGoogle Scholar
  4. 4.
    Pilia G, Chen WM, Scuteri A et al (2006) Heritability of cardiovascular and personality traits in 6, 148 Sardinians. PLoS Genet 2:e132CrossRefPubMedGoogle Scholar
  5. 5.
    Weedon MN, Frayling TM, Shields B et al (2005) Genetic regulation of birth weight and fasting glucose by a common polymorphism in the islet cell promoter of the glucokinase gene. Diabetes 54:576–581CrossRefPubMedGoogle Scholar
  6. 6.
    Rose CS, Ek J, Urhammer SA et al (2005) A -30G>A polymorphism of the beta-cell-specific glucokinase promoter associates with hyperglycemia in the general population of whites. Diabetes 54:3026–3031CrossRefPubMedGoogle Scholar
  7. 7.
    Weedon MN, Clark VJ, Qian Y et al (2006) A common haplotype of the glucokinase gene alters fasting glucose and birth weight: association in six studies and population-genetics analyses. Am J Hum Genet 79:991–1001CrossRefPubMedGoogle Scholar
  8. 8.
    Vaxillaire M, Cavalcanti-Proenca C, Dechaume A et al (2008) The common P446L polymorphism in GCKR inversely modulates fasting glucose and triglyceride levels and reduces type 2 diabetes risk in the DESIR prospective general French population. Diabetes 57:2253–2257CrossRefGoogle Scholar
  9. 9.
    Orho-Melander M, Melander O, Guiducci C et al (2008) Common missense variant in the glucokinase regulatory protein gene is associated with increased plasma triglyceride and C-reactive protein but lower fasting glucose concentrations. Diabetes 57:3112–3121CrossRefPubMedGoogle Scholar
  10. 10.
    Sparso T, Andersen G, Nielsen T et al (2008) The GCKR rs780094 polymorphism is associated with elevated fasting serum triacylglycerol, reduced fasting and OGTT-related insulinaemia, and reduced risk of type 2 diabetes. Diabetologia 51:70–75CrossRefPubMedGoogle Scholar
  11. 11.
    Tam CH, Ma RC, So WY et al (2009) Interaction effect of genetic polymorphisms in glucokinase (GCK) and glucokinase regulatory protein (GCKR) on metabolic traits in healthy Chinese adults and adolescents. Diabetes 58:765–769CrossRefPubMedGoogle Scholar
  12. 12.
    Qi Q, Wu Y, Li H et al (2009) Association of GCKR rs780094, alone or in combination with GCK rs1799884, with type 2 diabetes and related traits in a Han Chinese population. Diabetologia 52:834–843CrossRefPubMedGoogle Scholar
  13. 13.
    Saxena R, Voight BF, Lyssenko V et al (2007) Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science 316:1331–1336CrossRefPubMedGoogle Scholar
  14. 14.
    Bouatia-Naji N, Rocheleau G, van Lommel L et al (2008) A polymorphism within the G6PC2 gene is associated with fasting plasma glucose levels. Science 320:1085–1088CrossRefPubMedGoogle Scholar
  15. 15.
    Chen WM, Erdos MR, Jackson AU et al (2008) Variations in the G6PC2/ABCB11 genomic region are associated with fasting glucose levels. J Clin Invest 118:2620–2628PubMedGoogle Scholar
  16. 16.
    Sabatti C, Service SK, Hartikainen AL et al (2009) Genome-wide association analysis of metabolic traits in a birth cohort from a founder population. Nat Genet 41:35–46CrossRefPubMedGoogle Scholar
  17. 17.
    Prokopenko I, Langenberg C, Florez JC et al (2009) Variants in MTNR1B influence fasting glucose levels. Nat Genet 41:77–81CrossRefPubMedGoogle Scholar
  18. 18.
    Bouatia-Naji N, Bonnefond A, Cavalcanti-Proenca C et al (2009) A variant near MTNR1B is associated with increased fasting plasma glucose levels and type 2 diabetes risk. Nat Genet 41:89–94CrossRefPubMedGoogle Scholar
  19. 19.
    Hu C, Zhang R, Wang C et al (2009) A genetic variant of G6PC2 is associated with type 2 diabetes and fasting plasma glucose level in the Chinese population. Diabetologia 52:451–456CrossRefPubMedGoogle Scholar
  20. 20.
    Rönn T, Wen J, Yang Z et al (2009) A common variant in MTNR1B, encoding melatonin receptor 1B, is associated with type 2 diabetes and fasting plasma glucose in Han Chinese individuals. Diabetologia 52:830–833CrossRefPubMedGoogle Scholar
  21. 21.
    McCarthy MI, Hirschhorn JN (2008) Genome-wide association studies: potential next steps on a genetic journey. Hum Mol Genet 17:R156–R165CrossRefPubMedGoogle Scholar
  22. 22.
    Saccone NL, Saccone SF, Goate AM et al (2008) In search of causal variants: refining disease association signals using cross-population contrasts. BMC Genet 9:58CrossRefPubMedGoogle Scholar
  23. 23.
    Waters KM, Le Marchand L, Kolonel LN et al (2009) Generalizability of associations from prostate cancer genome-wide association studies in multiple populations. Cancer Epidemiol Biomarkers Prev 18:1285–1289CrossRefPubMedGoogle Scholar
  24. 24.
    Helgason A, Palsson S, Thorleifsson G et al (2007) Refining the impact of TCF7L2 gene variants on type 2 diabetes and adaptive evolution. Nat Genet 39:218–225CrossRefPubMedGoogle Scholar
  25. 25.
    Tsuchihashi-Makaya M, Serizawa M, Yanai K et al (2009) Gene-environmental interaction regarding alcohol-metabolizing enzymes in the Japanese general population. Hypertens Res 32:207–213CrossRefPubMedGoogle Scholar
  26. 26.
    Dassanayake A, Kasturiratne A, Rajindrajith S et al (2009) Prevalence and risk factors for non-alcoholic fatty liver disease (NAFLD) among adults in an urban Sri Lankan population. J Gastroenterol Hepatol 24:1284–1288CrossRefPubMedGoogle Scholar
  27. 27.
    Takeuchi F, Serizawa M, Yamamoto K et al (2009) Confirmation of multiple risk loci and genetic impacts by a genome-wide association study of type 2 diabetes in the Japanese population. Diabetes 58:1690–1699CrossRefPubMedGoogle Scholar
  28. 28.
    Purcell S, Neale B, Todd-Brown K et al (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81:559–575CrossRefPubMedGoogle Scholar
  29. 29.
    Stephens M, Smith NJ, Donnelly P (2001) A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 68:978–989CrossRefPubMedGoogle Scholar
  30. 30.
    Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21:263–265CrossRefPubMedGoogle Scholar
  31. 31.
    Li JZ, Absher DM, Tang H et al (2008) Worldwide human relationships inferred from genome-wide patterns of variation. Science 319:1100–1104CrossRefPubMedGoogle Scholar
  32. 32.
    Petrolonis AJ, Yang Q, Tummino PJ et al (2004) Enzymatic characterization of the pancreatic islet-specific glucose-6-phosphatase-related protein (IGRP). J Biol Chem 279:13976–13983CrossRefPubMedGoogle Scholar
  33. 33.
    Lang T, Haberl M, Jung D et al (2006) Genetic variability, haplotype structures, and ethnic diversity of hepatic transporters MDR3 (ABCB4) and bile salt export pump (ABCB11). Drug Metab Dispos 34:1582–1599CrossRefPubMedGoogle Scholar
  34. 34.
    Ramaraj R, Chellappa P (2008) Cardiovascular risk in South Asians. Postgrad Med J 84:518–523CrossRefPubMedGoogle Scholar
  35. 35.
    Reiling E, van’t Riet E, Groenewoud MJ et al (2009) Combined effects of single-nucleotide polymorphisms in GCK, GCKR, G6PC2 and MTNR1B on fasting plasma glucose and type 2 diabetes risk. Diabetologia 52:1866–1870CrossRefPubMedGoogle Scholar
  36. 36.
    Ge D, Zhang K, Need AC et al (2008) WGAViewer: software for genomic annotation of whole genome association studies. Genome Res 18:640–643CrossRefPubMedGoogle Scholar
  37. 37.
    Zeggini E, Scott LJ, Saxena R et al (2008) Meta-analysis of genome-wide association data and large-scale replication identifies additional susceptibility loci for type 2 diabetes. Nat Genet 40:638–645CrossRefPubMedGoogle Scholar

Copyright information

© Springer-Verlag 2009

Authors and Affiliations

  • F. Takeuchi
    • 1
  • T. Katsuya
    • 2
  • S. Chakrewarthy
    • 3
  • K. Yamamoto
    • 4
  • A. Fujioka
    • 5
  • M. Serizawa
    • 1
  • T. Fujisawa
    • 6
  • E. Nakashima
    • 7
    • 8
  • K. Ohnaka
    • 9
  • H. Ikegami
    • 10
  • T. Sugiyama
    • 11
  • T. Nabika
    • 12
  • A. Kasturiratne
    • 3
  • S. Yamaguchi
    • 13
  • S. Kono
    • 14
  • R. Takayanagi
    • 15
  • Y. Yamori
    • 16
  • S. Kobayashi
    • 17
  • T. Ogihara
    • 18
  • A. de Silva
    • 3
  • R. Wickremasinghe
    • 3
  • N. Kato
    • 1
  1. 1.Department of Gene Diagnostics and Therapeutics, Research InstituteInternational Medical Center of JapanTokyoJapan
  2. 2.Department of Clinical Gene TherapyOsaka University Graduate School of MedicineSuitaJapan
  3. 3.Faculty of MedicineUniversity of KelaniyaKelaniyaSri Lanka
  4. 4.Department of Molecular Genetics, Medical Institute of BioregulationKyushu UniversityFukuokaJapan
  5. 5.Amagasaki Health Medical FoundationAmagasakiJapan
  6. 6.Department of Geriatric MedicineOsaka University Graduate School of MedicineSuitaJapan
  7. 7.Division of Endocrinology and Diabetes, Department of Internal MedicineNagoya University Graduate School of MedicineNagoyaJapan
  8. 8.Department of Metabolism and Endocrine Internal MedicineChubu Rosai HospitalNagoyaJapan
  9. 9.Department of Geriatric Medicine, Graduate School of Medical SciencesKyushu UniversityFukuokaJapan
  10. 10.Department of Endocrinology, Metabolism and DiabetesKinki University School of MedicineOsaka-SayamaJapan
  11. 11.Institute for Adult DiseasesAsahi Life FoundationTokyoJapan
  12. 12.Department of Functional PathologyShimane University School of MedicineIzumoJapan
  13. 13.Internal Medicine IIIShimane University School of MedicineIzumoJapan
  14. 14.Department of Preventive Medicine, Graduate School of Medical SciencesKyushu UniversityFukuokaJapan
  15. 15.Department of Medicine and Bioregulatory Science, Graduate School of Medical SciencesKyushu UniversityFukuokaJapan
  16. 16.Mukogawa Women’s UniversityInstitute for World Health DevelopmentNishinomiyaJapan
  17. 17.Shimane University HospitalIzumoJapan
  18. 18.Osaka General Medical CenterOsakaJapan

Personalised recommendations