Biochemical Genetics

, Volume 50, Issue 3, pp 298–308

Association of Hepatocyte Nuclear Factor 4 Alpha Polymorphisms with Type 2 Diabetes With or Without Metabolic Syndrome in Malaysia

Authors

    • Department of Molecular Medicine, Faculty of MedicineUniversity of Malaya
    • Faculty of MedicineSana’a University
  • Roslan Harun
    • UKM Medical CenterUniversity Kebangsaan Malaysia
    • UKM Medical Molecular Biology Institute (UMBI)University Kebangsaan Malaysia
  • Nor Azmi Kamaruddin
    • UKM Medical CenterUniversity Kebangsaan Malaysia
  • Saad Al-Jassabi
    • Institute of Biological ScienceUniversity of Malaya
  • Wan Zurinah Wan Ngah
    • UKM Medical CenterUniversity Kebangsaan Malaysia
    • UKM Medical Molecular Biology Institute (UMBI)University Kebangsaan Malaysia
Article

DOI: 10.1007/s10528-011-9472-2

Cite this article as:
Saif-Ali, R., Harun, R., Kamaruddin, N.A. et al. Biochem Genet (2012) 50: 298. doi:10.1007/s10528-011-9472-2

Abstract

This study investigated the association of hepatocyte nuclear factor 4 (HNF4) alpha single nucleotide polymorphisms (SNPs) with type 2 diabetes with or without metabolic syndrome in Malaysia. Nine HNF4 alpha SNPs were genotyped in 390 type 2 diabetic subjects with metabolic syndrome, 135 type 2 diabetic subjects without metabolic syndrome, and 160 control subjects. The SNPs rs4810424, rs1884613, and rs2144908 were associated with protection against type 2 diabetes without metabolic syndrome (recessive P = 0.018, OR 0.32; P = 0.004, OR 0.25; P = 0.005, OR 0.24, respectively). The 6-SNP haplotype2 CCCGTC containing the risk genotype of these SNPs was associated with higher risk for type 2 diabetes with or without metabolic syndrome (P = 0.002, OR 2.2; P = 0.004, OR 3.1). These data suggest that HNF4 alpha SNPs and haplotypes contributed to increased type 2 diabetes risk in the Malaysian population.

Keywords

Hepatocyte nuclear factor 4 alphaType 2 diabetesMetabolic syndromeSingle nucleotide polymorphismsHaplotype

Introduction

Hepatocyte nuclear factor 4 alpha (HNF4 alpha) is a transcriptional factor expressed at high levels in the liver and to a lesser degree in several other tissues, including pancreatic beta cells (Jiang et al. 2003). Two promoters, P1 and P2 (located approximately 45 kb apart on chromosome 20), drive the transcription of HNF4 alpha in liver cells, whereas the expression of HNF4 alpha in pancreatic cells is principally controlled by P2 (Briancon et al. 2004). Fine mapping studies of chromosome 20q near the HNF4 alpha gene region suggested that type 2 diabetes is associated with single nucleotide polymorphisms (SNPs) in the P2 and P1 regions of HNF4 alpha (Love-Gregory et al. 2004; Silander et al. 2004). Subsequent studies (Bagwell et al. 2005; Barroso et al. 2008; Bento et al. 2008; Bonnycastle et al. 2006; Cauchi et al. 2008; Hansen et al. 2005; Hara et al. 2006; Johansson et al. 2007; Takeuchi et al. 2007; Tanahashi et al. 2006; Vaxillaire et al. 2005; Weedon et al. 2004; Winckler et al. 2005) showed inconsistent results, possibly because of population variations and possibly not accounting for the presence of metabolic syndrome in diabetes mellitus. The aim of this research is to study the associations of HNF4 alpha SNPs, haplotypes and diplotypes, with type 2 diabetes with or without metabolic syndrome in Malaysia.

Materials and Methods

Subjects

Subjects aged 30 years and above were randomly recruited from patients previously diagnosed with type 2 diabetes, who received treatment at the Hospital University of Kebangsaan Malaysia (HUKM). After providing written informed consent, 525 patients agreed to participate, and 10 ml blood was collected. For the control group, brochures were distributed to offices around the Cheras area and to staff of our institution, the Hospital and Faculty of Medicine of University Kebangsaan Malaysia. Brochures were also given to participants for distribution to their relatives and friends. In 2 years, about 900 subjects responded, but after several tries, only 262 participated. After biochemical analysis of their blood, 160 of the 262 were found to be without diabetes and/or metabolic syndrome; they formed the control group. Ethics approval was obtained from the University Kebangsaan Malaysia Research and Ethics Committee.

Biochemical Analysis

Commercially available kits were employed to measure glucose, triglyceride, total cholesterol, and HDL cholesterol (reference nos. 10260, 10724, 10028, and 10018; Human GmbH, Wiesbaden, Germany). Elevated control sera (Humatrol P ref no. 13512) were used as quality controls.

Genotyping of SNPs

HNF4 alpha SNPs rs4810424, rs1884613, rs1884614, rs2144908, rs6031551, rs6031552, rs1885088, rs1028583, and rs3818247 were selected for genotypic analysis based on published results from the Finnish and Ashkenazi Jewish studies (Love-Gregory et al. 2004; Silander et al. 2004). All SNPs in this research were amplified by an Icycler thermocycler (Bio-Rad Laboratories, Richmond, Calif., USA) using a 96-microwell plate. Touchdown PCR was applied for all SNPs except rs1884613, which was amplified and identified by allele-specific PCR. Restriction enzymes (New England Biolabs, USA) were employed to identify the genotypes of rs4810424 (BstXI), rs1884614 (BsgI), rs6031551 (MseI), rs6031552 (CvikI-1), rs1885088 (Sau96), and rs1028583 (PstI).

A recommendation for running rs2144908 at 60°C and rs3818247 at 58°C was obtained when these PCR-amplified sequences were uploaded to the DHPLC melt program (http://insertion.stanford.edu/melt.html). The recommended temperature was confirmed experimentally by running 5 μl PCR products of these SNPs in DHPLC (Varian, Palo Alto, CA, USA) at 2°C below and above. The separated peaks of heteroduplex and homoduplex DNA from the heterozygous samples of rs2144908 and rs3818247 were detected at 59 and 57.5°C, respectively, using a DHPLC program gradient buffer and time. Known heterozygous, recessive homozygous, and dominant homozygous samples of rs2144908 and rs3818247 (these samples were sequenced) were included in the runs as quality controls, at 59 and 57.5°C, respectively. Some samples were sequenced by an automated 3130×l Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) using a Terminator Cycle Sequencing Kit version 3.1 to confirm the allele-specific PCR and DHPLC results.

Statistical Analysis

HelixTree 6.0.1 SNP and Variation Suite for Genetic Statistics was employed to study the deviation of SNPs from Hardy–Weinberg equilibrium, to study linkage disequilibrium (LD), and to construct haplotypes and diplotypes of related SNPs. The other statistical analyses were done using the SPSS 11.5 program. The associations of the HNF4 alpha SNPs, recessive, dominant, and additive models, as well as the association of haplotypes and diplotypes with type 2 diabetes with and without metabolic syndrome were evaluated by hierarchical logistic regression controlled for propensity score (PS) as a calculated probability prediction to control the confounding factors. Two propensity scores were calculated by logistic regression, for type 2 diabetes with and without metabolic syndrome. The independent variables of gender, race, and history of diabetes may affect the inherence of SNPs, haplotypes, and diplotypes (genetic exposure related variables), and so all these variables were included for calculating the propensity scores. The logistic regression showed that age, systolic and diastolic blood pressures, total cholesterol, and triacylglycerol (variables related to outcome) were significantly associated with type 2 diabetes without metabolic syndrome. Therefore, these variables were also included as independent variables to calculate the PS for this group.

Type 2 diabetes with metabolic syndrome comprises the four factors of abdominal obesity, high triglycerides, low levels of HDL cholesterol, and high blood pressure, in addition to high fasting blood glucose. These factors were outcome variables based on the objective of this research, and not considered related to type 2 diabetes with metabolic syndrome. Therefore, these factors were not included in the PS calculations for this model. Because the logistic regression showed that total cholesterol (a variable related to outcome) was associated with type 2 diabetes with metabolic syndrome, it was included in the PS for this group.

Results

International Diabetes Federation criteria were used to categorize the type 2 diabetic subjects with and without metabolic syndrome (Table 1). The nine SNPs included in this study did not deviate from Hardy–Weinberg equilibrium in normal subjects; thus, the associations of the nine SNPs with type 2 diabetes with or without metabolic syndrome were investigated.
Table 1

Demography and biochemical parameters of type 2 diabetic and control subjects

Parameter

Control group (N = 160)

Type 2 diabetes patient group

Without metabolic syndrome (N = 135)

With metabolic syndrome (N = 390)

Gender (%)

   

 Male

41.3

38.5

35.6

 Female

58.8

61.5

64.4

Race (%)

 Malay

63.8

52.6

53.3

 Chinese

26.9

35.6

26.2

 Indian

9.4

11.9

20.5

Age (year)

46.8 ± 9.2

44.2 at diagnosis

45.2 ± 9.33 at diagnosis

 

53.4 at sample collection

54.1 ± 9.19 at sample collection

Weight (kg)

63.7 ± 10.3

61.1 ± 9.79

72.2 ± 13.3

Height (m)

1.60 ± 0.08

1.6 ± 0.87

1.59 ± 0.081

Body mass index (kg/m2)

24.4 ± 2.90

23.9 ± 3.27

28.5 ± 4.38

Waist (cm)

82.1 ± 9.2

83.6 ± 9.2

96.6 ± 9.4

Blood pressure (mmHg)

 

 Systolic

124 ± 14.6

128 ± 18.3

142 ± 18.9

 Diastolic

78 ± 7.6

72 ± 10.0

78.8 ± 10.1

Fasting C-peptide (pmol/l)

807 ± 445

600 ± 350

875 ± 505

Fasting glucose (mmo/l)

4.9 ± 0.49

8.4 ± 3.1

8.8 ± 3.29

Triglyceride (mmol/l)

1.05 ± 0.45

1.13 ± 0.41

1.79 ± 0.89

Cholesterol (mmol/l)

 

 Total

4.7 ± 0.74

4.3 ± 0.87

4.6 ± 0.97

 HDL

1.29 ± 0.30

1.28 ± 0.32

1.14 ± 0.30

Note: Values are mean ± standard deviation

The association analysis under dominant and additive models for intronic 3 SNP rs1885088 showed a strong association with type 2 diabetes with metabolic syndrome (OR = 8.0, 7.24; P = 0.005, 0.008, respectively). In the recessive model for P2 promoter SNP, rs4810424, rs1884613, and intronic 1D SNP rs2144908 showed protection for type 2 diabetes without metabolic syndrome (OR = 0.32, 0.25, 0.24; P = 0.018, 0.004, 0.005, respectively) while not significantly protecting for type 2 diabetes with metabolic syndrome (Table 2). The other SNPs (rs1884614, rs6031551, rs6031552, rs1028583, and rs3818247) were not associated with type 2 diabetes.
Table 2

Association of HNF4 SNP recessive, dominant, and additive genetic models with type 2 diabetes without metabolic syndrome

SNP

Genetic modela

Control group (N = 160) n (Frequency)

Type 2 diabetes patient group

Without metabolic syndromeb (N = 135)

With metabolic syndromec (N = 390)

n (Frequency)

Odds ratio

P

n (Frequency)

Odds ratio

P

rs4810424

Recessive

36 (0.23)

20 (0.15)

0.32

0.018

68 (0.18)

0.70

0.23

Dominant

41 (0.26)

36 (0.30)

0.67

0.32

107 (0.28)

0.79

0.34

Additive

  

0.59

0.049

 

0.76

0.12

rs1884613

Recessive

35 (0.23)

19 (0.14)

0.25

0.004

64 (0.17)

0.63

0.13

Dominant

52 (0.33)

35 (0.27)

1.61

0.22

130 (0.34)

0.97

0.89

Additive

  

0.78

0.33

 

0.82

0.25

rs1884614

Recessive

33 (0.22)

19 (0.15)

0.50

0.13

65 (0.17)

0.64

0.12

Dominant

47 (0.31)

36 (0.28)

1.34

0.45

117 (0.31)

0.66

0.13

Additive

  

0.91

0.72

 

0.73

0.064

rs2144908

Recessive

35 (0.22)

18 (0.13)

0.24

0.005

70 (0.18)

0.92

0.77

Dominant

54 (0.34)

43 (0.32)

1.34

0.43

111 (0.28)

1.48

0.13

Additive

  

0.81

0.39

 

1.17

0.37

rs6031551

Recessive

0 (0.0)

1 (0.01)

5 (0.01)

Dominant

127 (0.80)

104 (0.78)

1.94

0.14

317 (0.84)

0.93

0.81

Additive

  

1.94

0.13

 

1.02

0.95

rs6031552

Recessive

1 (0.01)

2 (0.02)

4 (0.01)

Dominant

131 (0.32)

102 (0.77)

2.03

0.12

351 (0.91)

1.10

0.76

Additive

  

1.85

0.14

 

1.09

0.77

rs1885088

Recessive

0 (0.0)

0(0.0)

49 (0.13)

Dominant

152 (0.97)

132 (0.98)

2.98

0.30

166 (0.44)

8.03

0.005

Additive

  

2.98

0.30

 

7.24

0.008

rs1028583

Recessive

22 (0.14)

25 (0.19)

0.67

0.42

49 (0.13)

0.66

0.18

Dominant

70 (0.44)

54 (0.40)

0.94

0.86

166 (0.44)

0.93

0.76

Additive

  

0.91

0.73

 

0.79

0.18

rs3818247

Recessive

23 (0.14)

25 (0.19)

1.44

0.44

85 (0.22)

1.32

0.38

Dominant

54 (0.34)

53 (0.40)

0.82

0.60

124 (0.33)

0.90

0.70

Additive

  

1.00

0.99

 

1.07

0.70

aRecessive model: Odds ratio [homozygous recessive vs. (heterozygous + homozygous dominant)]. Dominant model: Odds ratio [(homozygous recessive + heterozygous) vs. homozygous dominant]. Additive model: Common odds ratio (homozygous recessive vs. heterozygous vs. homozygous dominant); additive model was recoded for the count of the minor allele d, which is 0 within genotype homozygous dominant, 1 within genotype heterozygous, and 2 within genotype homozygous recessive

bControlled for age, gender, race, history of diabetes, systolic and diastolic blood pressures, cholesterol, and triacylglycerol

cControlled for age, gender, race, history of diabetes, and cholesterol

Bold & Italics P-value indicates significant (< 0.05)

A block of six SNP haplotypes and diplotypes was identified with significant LD. This block was constructed from rs4810424, rs1884613, rs1884614, rs2144908, rs6031551, and rs6031552 (Fig. 1). The possible haplotypes for each individual were adjusted to more than 0.5, resulting in 26 haplotypes1, 15 haplotypes2, and 75 diplotypes. Furthermore, the common haplotypes1, haplotypes2, and diplotypes (frequencies in the total sample greater than 0.02) were further analyzed for their associations with type 2 diabetes with and without metabolic syndrome.
https://static-content.springer.com/image/art%3A10.1007%2Fs10528-011-9472-2/MediaObjects/10528_2011_9472_Fig1_HTML.gif
Fig. 1

Linkage disequilibrium correlation among SNPs rs24, rs4810424; rs13, rs1884613; rs14, rs1884614; rs08, rs2144908; rs51, rs6031551; and rs52, rs6031552

Haplotypes1 were not associated with type 2 diabetes, either with or without metabolic syndrome. Haplotype2 (GGTATC) contained the protected genotypes of P2 promoter SNPs; rs4810424, rs1884613, and intronic 1D SNP rs2144908 were associated with protection for type 2 diabetes without metabolic syndrome (OR = 0.37, P = 0.048; Table 3). A similar result was seen with the diplotype that was constructed from this protective haplotype2 (OR = 0.28, P = 0.034). The most common haplotype2 (CCCGTC) contained the risk genotypes of these SNPs and was associated with increased risk of type 2 diabetes without metabolic syndrome (OR = 3.1, P = 0.004) and type 2 diabetes with metabolic syndrome (OR = 2.25, P = 0.002). Furthermore, the diplotype that was constructed from the protective and risk haplotypes2 (GGTATC-CCCGTC) showed association with type 2 diabetes with and without metabolic syndrome (OR = 3.48, P = 0.003; OR = 1.95, P = 0.022).
Table 3

Association of haplotypes1, haplotypes2, and diplotypes with type 2 diabetes with and without metabolic syndrome

Haplotype

Control group (N = 160) n (Frequency)

Type 2 diabetes patient group

Without metabolic syndromea (N = 135)

With metabolic syndromeb (N = 390)

n (Frequency)

Odds ratio

P

95% CI

n (Frequency)

Odds ratio

P

95% CI

Haplotypes1

 GGTATC

68 (0.42)

61 (0.45)

1.54

0.24

0.75–3.19

196 (0.50)

1.24

0.32

0.81–1.90

 CCCGTC

39 (0.24)

25 (0.18)

0.59

0.15

0.26–1.22

76 (0.19)

0.74

0.24

0.44–1.22

 CCCGCA

26 (0.16)

25 (0.18)

1.79

0.23

0.70–4.56

46 (0.12)

0.54

0.07

0.28–1.04

 CCCATC

8 (0.05)

3 (0.02)

0.27

0.24

0.03–2.44

22 (0.06)

1.36

0.56

0.49–3.75

 GCTATC

1 (0.01)

2 (0.02)

0.61

0.79

0.02–24.24

14 (0.06)

2.96

0.30

0.34–23.25

 GGCATC

6 (0.04)

5 (0.04)

0.70

0.64

0.15–3.19

7 (0.02)

0.27

0.06

0.07–1.05

Haplotypes2

 CCCGTC

67 (0.46)

69 (0.56)

3.1

0.004

1.43–6.7

219 (0.60)

2.25

0.002

1.36–3.73

 GGTATC

46 (0.30)

31 (0.24)

0.37

0.048

0.14–0.99

78 (0.21)

0.62

0.09

0.36–1.07

 GCCGTC

13 (0.09)

5 (0.04)

0.17

0.018

0.04–0.70

20 (0.05)

0.61

0.28

0.24–1.51

 GGTGTC

10 (0.07)

6 (0.05)

0.62

0.52

0.14–2.72

16 (0.04)

0.41

0.07

0.15–1.08

 CCTGTC

9 (0.06)

7 (0.06)

1.85

0.36

0.50–6.92

13 (0.03)

0.32

0.024

0.12–0.86

 CGCGTC

1 (0.01)

6 (0.05)

2.50

0.45

0.24–26.15

16 (0.04)

6.09

0.13

0.59–62.49

Diplotypes

 GGTATC-CCCGTC

34 (0.22)

41 (0.32)

3.48

0.003

1.52–7.99

125 (0.32)

1.95

0.022

1.1–3.47

 CCCGTC-CCCGTC

21 (0.13)

16 (0.12)

0.92

0.87

0.34–2.48

55 (0.14)

1.12

0.77

0.56–2.20

 GGTATC-GGTATC

24 (0.16)

9 (0.07)

0.28

0.034

0.09–0.91

43 (0.11)

0.81

0.55

0.41–1.60

 CCCGCA-GGTATC

15 (0.09)

13 (0.1)

1.75

0.34

0.55–5.58

19 (0.05)

0.39

0.03

0.17–0.92

 CCCGCA-CCCGTC

6 (0.04)

9 (0.07)

2.43

0.32

0.42–14.16

12 (0.03)

1.45

0.64

0.32–6.63

 CCCGTC-GCCGTC

9 (0.06)

1 (0.01)

0.04

0.008

0.004–0.43

9 (0.02)

0.28

0.03

0.09–0.88

aControlled for age, gender, race, history of diabetes, systolic and diastolic blood pressure, cholesterol, and triacylglycerol

bControlled for age, gender, race, history of diabetes, and cholesterol

Bold & Italics P-value indicates significant (< 0.05)

Discussion

This study showed that the HNF4 alpha SNPs rs4810424, rs1884613, and rs2144908 were associated with type 2 diabetes without metabolic syndrome in the Malaysian population. Associations of these SNPs with type 2 diabetes have been reported in Finnish (Silander et al. 2004), British (Weedon et al. 2004) Scandinavian (Johansson et al. 2007), and Ashkenazi Jewish (Barroso et al. 2008; Love-Gregory et al. 2004) populations. HNF4 alpha SNP rs1885088 was found to be associated with type 2 diabetes with metabolic syndrome in Malaysians. This association has also been reported in Finnish (Silander et al. 2004; Bonnycastle et al. 2006), Danish (Hansen et al. 2005), and American Caucasian (Bento et al. 2008) populations. Associations of HNF4 alpha SNPs could not be replicated in the French (Bagwell et al. 2005; Cauchi et al. 2008; Vaxillaire et al. 2005) and Americans (Bagwell et al. 2005), and a Japanese population yielded conflicting results (Takeuchi et al. 2007; Hara et al. 2006; Tanahashi et al. 2006). Most of the several interacting genetic determinants have small effects that contribute to complex diseases such as type 2 diabetes; it is difficult to identify the susceptibility locus (Wang et al. 2005). The discrepancy in the results of association of HNF4 alpha with T2DM might be due to the small effect that HNF4 alpha gene polymorphisms have on T2DM in some populations (Bonnycastle et al. 2006).

The International Haplotype Map (HapMap) project data showed differences in SNPs and allele frequencies across populations that affect the LD patterns of the human genome. Studies of many additional populations demonstrated that LD patterns could be highly variable among populations both across and within geographic regions (Gu et al. 2007). The factors that contribute to disequilibrium, for instance, gene conversion, allelic drift, and population history and structure (Ardlie et al. 2002; Reich et al. 2002), may result in discrepancies in the association results of HNF4 alpha SNPs with type 2 diabetes in different populations. Variant patterns of LD in populations could mirror the apparently conflicting results (Johansson et al. 2007). The haplotype variation in populations may indicate different SNPs that may be associated with type 2 diabetes, which might be located at a distance downstream or upstream of the P2 promoter of HNF4 alpha, and these regions need a comprehensive genotype study in homogeneous populations (Bonnycastle et al. 2006). In the recent French whole-genome association study, a weak association of rs2425637 (located 39 kb downstream of the P2 promoter) with type 2 diabetes was detected (Sladek et al. 2007). Another SNP, rs48123831, located 34 kb downstream from the P2 promoter, was associated with type 2 diabetes in Scandinavians (Johansson et al. 2007). Rs6103716 and rs6031558 (15 kb downstream from P2 promoter) were reported to be associated with type 2 diabetes in Finns (Bonnycastle et al. 2006). Rs2273618, rs6073435, and rs6031601 (at intron 7, 8, and 9, respectively) were associated with type 2 diabetes in a Japanese population (Tanahashi et al. 2006).

The association of variants in the HNF4 alpha P2 promoter region with type 2 diabetes was supported by meta-analysis involving 11,571 (Johansson et al. 2007) and 49,577 subjects (Sookoian et al. 2010). In a recent large cohort study consisting of 17,831 individuals from Sweden and Finland, it was reported that P2 promoter HNF4 alpha SNPs predicted further type 2 diabetes (Holmkvist et al. 2008). HNF4 alpha variants and haplotypes were found to be associated with elevated serum lipid levels and metabolic syndrome, as well as with elevated glucose parameters (Weissglas-Volkov et al. 2006). In conclusion, the HNF4 alpha P2 promoter variants in the Malaysian population were associated with protection for type 2 diabetes without metabolic syndrome, whereas the intronic 1D SNP rs1885088 was associated with increased risk for type 2 diabetes with metabolic syndrome. Haplotype2 containing the risk genotypes of the P2 promoter SNPs was associated with a higher risk for type 2 diabetes with or without metabolic syndrome. Furthermore, the diplotype constructed from protective and risk haplotypes was associated with a higher risk for type 2 diabetes.

Acknowledgments

This research was supported by a research grant from UKM Medical Molecular Biology Institute (UMBI). The authors thank Prof. Dr. Maznah Ismail and her laboratory staff, Laboratory of Molecular Biomedicine, Institute of Bioscience and University Putra Malaysia for allowing the use of their DHPLC. The authors are also grateful to HelixTree Team for their help and guidance in genetic analysis using the HelixTree program.

Copyright information

© Springer Science+Business Media, LLC 2011