The transcription factor hepatocyte nuclear factor (HNF)-4α, is a member of the steroid hormone receptor superfamily expressed in the liver, pancreatic islets, kidney and small intestine [1]. It binds to DNA as a homodimer and activates the transcription of target genes. HNF-4α consists of several functional domains: an N-terminal transactivation domain (AF-1), a DNA binding domain and a functionally complex C-terminal region that forms a ligand binding domain, a dimerization interface and a transactivation domain (AF-2) [2, 3]. HNF-4α plays an important role in the regulation of glucose and lipid metabolism. Late-onset Type 2 diabetes is considered to be a multifactorial disease with both genetic and environmental factors and has an uncertain mode of inheritance. In contrast, MODY is a monogenic form of Type 2 diabetes characterized by autosomal dominant inheritance, early-onset (usually before 25 years of age), and impaired insulin secretion. We have identified previously that mutations in the HNF-4α gene cause a form of MODY (MODY1) [4]. Of interest, the G319S variant of the HNF-1α/MODY3 gene was found to be a common cause of Type 2 diabetes in the Canadian Oji-Cree population [5]. This is an example of a different allele of the MODY gene also predisposing late-onset Type 2 diabetes. Several recent genetic studies, including a Japanese study, have obtained evidence for the linkage of an HNF-4α containing region on chromosome 20q12-q13 with Type 2 diabetes [6, 7, 8, 9, 10]. Thus, genetic variation in the HNF-4α gene might contribute to the susceptibility to Type 2 diabetes in such populations.

The T130I mutation is a missense mutation affecting a residue of the DNA binding region (A-box region). The A-box region is important for homodimerization and high-affinity DNA binding [2]. A threonine residue at position 130 is conserved in human, mouse, rat and Xenopus although the functional properties of T130I-HNF-4α have yet to be determined. The T130I mutation can be found in the general population (0–5%), so this variant alone does not cause MODY [4]. In one Danish study [11], the frequency of this allele was higher in subjects with Type 2 diabetes than in control subjects, but this finding was not confirmed in other studies of Caucasian subjects [8, 12]. In a Japanese study, the T130I mutation was only present in Type 2 diabetic patients (2/100) and was not found in 100 control subjects, although the difference was not statistically significant [13]. Our study was designed to further examine the diabetogenic effects of the T130I mutation by using genetic and functional analyses.

Materials and methods

Subjects

For the association study, 423 unrelated Japanese subjects with Type 2 diabetes (253 men and 170 women of 63.9±8.9 years of age, BMI 23.0±3.2 kg/m2, HbA1c 7.6±1.6%, means±SD) and 354 unrelated Japanese non-diabetic control subjects (158 men and 196 women of 63.8±18.9 years of age, BMI 22.4±2.8 kg/m2, HbA1c 4.9±0.4%) were enrolled. Type 2 diabetes was diagnosed in accordance with the World Health Organization criteria. All patients were diagnosed as having Type 2 diabetes after the age of 35 years. Patients with Type 1 diabetes and other types of diabetes (such as MODY) were excluded from this study. A written informed consent was obtained from each participant.

Screening for the T130I mutation in HNF-4α gene

Exon 4 and flanking intron were amplified using PCR with specific primers (5′-CCACCCCCTACTCCATCCCTGT-3′ and 5′-CCCTCCCGTCAGCTGCTCCA-3′) [4]. The T130I mutation generates a BsmI site and it was detected by PCR-restriction fragment length polymorphism (Thr: 271 bp, Ile: 81+190 bp). The mutation was tested in 423 diabetic and 354 non-diabetic subjects.

Plasmids

T130I mutant HNF-4α was generated from human HNF-4α2 cDNA using a Chameleon Double-Stranded Site-Directed Mutagenesis kit (Stratagene, La Jolla, Calif., USA) and cloned in pcDNA3.1 expression vector (Invitrogen, San Diego, Calif., USA). The construct was verified by DNA sequencing. The reporter construct with a heterologous TK promoter and HNF-4α binding sequences as an enhancer (pHNF4-tk-Luc), HNF-1α reporter and PKL reporter constructs have been described [14].

Cell culture and luciferase assay

Mouse primary hepatocytes were prepared using collagenase perfusion method [15] and plated in six-well tissue culture plates at a density of 3×105 cells per well. HeLa, MIN6, HepG2 cells and primary hepatocytes were transfected with indicated amounts of expression and reporter vectors together with 10 ng of pRL-TK (Promega, Madison, Wis., USA) as an internal control using Lipofectamin Plus reagent. Transactivation activities were measured after 48 h using the Dual Luciferase Reporter Assay system (Promega) and Lumat LB9501 Chemiluminescence (Berthold Japan, Osaka, Japan) [16].

Western blot analysis and electrophoretic mobility shift assay (EMSA)

Western blot analysis was done using anti-HNF-4α antibody (Santa Cruz Biotechnology, Santa Cruz, Calif., USA) as described previously [16]. Wild-type and T130I-HNF-4α proteins were synthesized from one µg of pcDNA3.1 expression vector using TNT T7 Quick Coupled transcriptionranslation system (Promega). In vitro translated proteins were incubated with 32P-labelled oligonucleotides containing HNF-4α binding site of HNF-1α gene (5′-GGCTGAAGTCCAAAGTTCAGTCCCTTCGC-3′) in a 20 µl reaction mixture [17]. In the competition binding assay, 50-fold of unlabelled probes were used as the competitor. DNA-protein complexes were analysed on 5% polyacrylamide gels using 0.5× TBE buffer. The polyclonal anti-HNF-4α antiserum (α445) was used for supershift analysis [1]. To determine the relative binding affinity of T130I-HNF-4α, increasing amounts of unlabelled competitor (10-, 25- and 50-fold) was added at non-equilibrium conditions (5-min incubation) [17].

Statistical analysis

Differences in allele frequencies were tested for statistical significance by the chi-square test and Fisher's exact test when appropriate. All clinical data are expressed as means ± SD. Comparison of variables between groups of genotypes was done by using the two-tailed Student's t test and/or the Mann-Whitney non-parametric test. Statistical analysis was done by using StatView 5.0 software (SAS institute, Cary, N.C., USA). A p value less than 0.05 was considered significant.

Results

Screening for the T130I mutation in Japanese subjects

We screened 423 unrelated Japanese subjects, who had been diagnosed as having Type 2 diabetes after the age of 35 years, and 354 unrelated Japanese non-diabetic control subjects for the T130I mutation in the HNF-4α gene. T130I mutation was found in 15 out of 423 diabetic patients (3.5%) and in 3 out of 354 non-diabetic subjects (0.8%), indicating that the frequency of the mutation was higher (p=0.015, odds ratio 4.3, 95%CI 1.24–14.98) in the Type 2 diabetic group compared with the non-diabetic Japanese subjects. A previous study has showed that the frequencies of the same mutation were 2.0% (2/100) and 0% (0/100) in Japanese Type 2 diabetic and control subjects, respectively [13]. Combining the results of these two Japanese studies indicates that the T130I mutation is more strongly associated with Type 2 diabetes [diabetes; 17/523 (3.3%), control; 3/454 (0.7%), p=0.0053, odds ratio 5.05, 95%CI 1.47–17.35]. The clinical features of the diabetic patients with the Ile codon at 130 are shown (Table 1). The average age at diagnosis of diabetes with MODY1 was 28.2±15.2 years [18, 19, 20, 21, 22, 23, 24, 25, 26]. The mean age at diagnosis was higher for patients with the T130I mutation (47.1±8.8 years, p=1.0×10−6) than for those with other HNF-4α mutations. There were no significant differences of the age at the onset of diabetes, BMI, maximum BMI, fasting plasma glucose concentrations, HbA1c, homeostasis model assessment of insulin resistance (HOMA-IR) [27], total cholesterol, and triglyceride concentrations between two diabetic groups with and without the T130I mutation. However, the serum HDL-cholesterol concentration was lower in the group with the T130I mutation (p=0.006). The glucagon-stimulated C-peptide response was variable (0.30, 0.43, 1.32 and 1.56 nmol/l, normal >0.66 nmol/l) in four diabetic patients with the T130I genotype from whom data were available. The urinary C-peptide (uCPR) concentrations in six patients tested were 4.87, 15.8, 22.2, 31.0, 33.4, 49.3 nmol/24 h (normal: 16.6 to 39.7 nmol/24 h). These findings suggest that there was variable insulin secretion among these subjects with the T130I-HNF-4α mutation. This is in sharp contrast with previous reports that MODY1 mutations lead to a characteristic defect of insulin secretion [4, 19, 28].

Table 1. Clinical characteristics of the Type 2 diabetic subjects with and without T130I-HNF-4α mutation
Full size table

Functional properties of T130I-HNF-4α mutant

We investigated the function of T130I-HNF-4α. For this purpose, HeLa cells were transfected with wild-type (WT) HNF-4α or T130I-HNF-4α expression vectors. The levels of WT and T130I-HNF-4α expression were similar (Fig. 1A). The transactivation ability of WT and T130I-HNF-4α was assessed using the luciferase reporter gene assay. There were no significant differences of the transactivation activity between WT-HNF-4α and T130I-HNF-4α in HeLa cells as well as MIN6 cells, a mouse insulinoma cell line (Fig. 1B, C). However, when the same amount of expression vector was transfected into HepG2 cells, a human hepatoma cell line, the transcriptional activity of T130I-HNF-4α was decreased by 46.2% (p<0.001) compared with that of WT-HNF-4α (Fig. 1D). Figure 1E shows that impaired transactivation of T130I-HNF-4α in HepG2 cells was found at all doses tested, whereas T130I-HNF-4α achieved similar transactivation compared with WT-HNF-4α in MIN6 cells at these doses (data not shown). HepG2 cells have heterogenous D69A mutation in the HNF-4α gene [29]. To determine whether D69A mutation caused the reduced transactivation of T130I-HNF-4α in HepG2 cells, we carried out a reporter gene assay using primary cultured mouse hepatocytes. Reduced transactivation of T130I-HNF-4α (27.9% of WT-HNF-4α, p=9.7×10−5) was also found in primary hepatocytes (Fig. 1F). HNF-1α and L-type pyruvate kinase (PKL) are target genes for HNF-4α [30]. Transcriptional activation of the HNF-1α gene (78.2%, p=0.024) and the PKL gene (77.1%, p=0.002) by T130I-HNF-4α was impaired (Fig. 1G). These data strongly suggest that T130I-HNF-4α acts as a loss-of-function mutation in the hepatic cell environment. Since the A-box region is considered to be important for DNA binding [2], we also tested the DNA binding ability of T130I-HNF-4α. When WT-HNF-4α and T130I-HNF-4α proteins were translated in vitro and used for EMSA, WT-HNF-4α and T130I-HNF-4α specifically bound to the oligonucleotide (Fig. 2A). The binding of T130I-HNF-4α relative to WT-HNF-4α was similar under both equilibrium (Fig. 2A) and non-equilibrium conditions (Fig. 2B), suggesting that this amino acid change does not alter DNA binding in vitro. However, further studies will be necessary to examine the DNA binding ability of T130I-HNF-4α in vivo (e.g., in hepatocytes).

Fig. 1A–H.
figure 1

Transactivation activities of wild-type (WT) and T130I-HNF-4α. A Expression of WT and T130I-HNF-4α in HeLa cells. Eight µg of expression vectors were transfected and Western blot was done after 48 h. Lane 1: WT-HNF-4α; Lane 2: T130I-HNF-4α; Lane 3: empty vector. B–D Transactivation activities of WT and T130I-HNF-4α in HeLa (B), MIN6 (C) and HepG2 (D) cells. Cells were transfected with 500 ng of expression vectors together with 100 ng of pHNF4-tk-Luc. E Transactivation activities of T130I-HNF-4α in HepG2 cells. Increasing amounts of expression vectors (50–200 ng) were transfected with pHNF-4-tk-Luc. Data are means ± SD values of three independent experiments. F–H Mouse primary hepatocytes were transfected with 0.5 µg of expression vectors together with 100 ng of reporter genes (F: pHNF-4-tk-Luc; G: HNF-1α promoter; H PKL promoter). Data are means ± SD values of six independent experiments

Full size image
Fig. 2A, B.
figure 2

DNA binding ability of WT and T130I-HNF-4α. A HNF-4α proteins were in vitro translated and used for EMSA. Equal expression levels of in vitro translated proteins were confirmed by Western blot analysis. An excess (50-fold) of unlabelled oligonucleotide was used as a competitor. Lane 1: WT-HNF-4α; Lane 2: WT-HNF-4α and competitor; Lane 3: T130I-HNF-4α; Lane 4: T130I-HNF-4α and competitor; Lane 5: empty vector; Lane 6: empty vector and competitor; Lane 7: WT-HNF-4α and ant-HNF-4α antibody; Lane 8: T130I-HNF-4α and anti-HNF-4α antibody. Arrow shows the position of HNF-4α/DNA complex. Lanes 7 and 8 show supershift of bands. B DNA binding ability of T130I-HNF-4α in non-equilibrium conditions. WT-HNF-4α and T130I-HNF-4α were bound to the labelled oligonucleotide in the presence of increasing amounts (10, 25 and 50 molar excess) of unlabelled competitor in non-equilibrium conditions

Full size image

Discussion

In this study, we showed that the T130I mutation in HNF-4α gene was associated with late-onset Type 2 diabetes in Japanese subjects and found that this is a loss-of-function mutation, at least in hepatocytes. It is not clear why the T130I mutation only affects transactivation activity in hepatic cells. One possible explanation is a post-translational modification that only occurs in those cells. For example, protein kinase B is a downstream molecule of the insulin signalling pathway in hepatocytes. Threonine at codon 130 might be phosphorylated by protein kinase B in hepatocytes (amino acids: 125-RDRIST-130; the PKB recognition motif is RXRYZS/T [31]) and phosphorylation might affect the DNA binding of T130-HNF-4α. Alternatively, hepatocytes could contain proteins that inhibit binding of the T130I mutant. Further studies will be necessary to clarify the mechanism involved. A recent genome-wide linkage study of Japanese Type 2 diabetes identified a susceptible locus around the HNF-4α gene on chromosome 20q (D20S119 and D20S855) [10]. The T130I mutation of the HNF-4α gene might be involved in this linkage. The molecular mechanism by which low HNF-4α activity in hepatocytes, but not in pancreatic beta cells, that leads to late-onset diabetes is not known. Liver specific HNF-4α knockout mice show severe fatty change of the liver [32], but no information is available on whether these mice develop Type 2 diabetes with aging. Liver is an important insulin target organ and accumulation of fat metabolites could cause insulin resistance [33]. We also found that the serum HDL-cholesterol concentration was lower in the group with the T130I mutation. Reduced triglyceride concentrations were observed in some MODY1 patients [34, 35]. The serum concentration of HDL-cholesterol was dramatically decreased in the knockout mice [32]. Hepatic HNF-4α function could be important for lipid metabolism. Clinical data on patients with the T130I genotype, as well as examination of glucose and lipid metabolism in HNF-4α liver-specific knockout mice, are needed to obtain a better understanding of the role of the T130I mutation in the development of Type 2 diabetes. In addition, several rare but not MODY-causing variants have also been described in the HNF-1α/MODY3 gene [36, 37, 38]. Some have also been found more often in Type 2 diabetic patients than controls subjects although the difference was not statistically significant for the low frequencies of the variants. Some mutations in the HNF-4α and HNF-1α genes could predispose to Type 2 diabetes.