Thyroid Hormone Receptor Interacting Protein 3 (Trip3) Is a Novel Coactivator of Hepatocyte Nuclear Factor-4α

The following plasmids were used in the yeast two-hybrid screening assay: pBTM116-HNF-4α lacking the NH₂-terminal 139 amino acid residues fused to the DNA-binding domain of LexA as bait; pAS2–1-HNF-4α lacking the NH₂-terminal 139 amino acid residues fused to the DNA-binding domain of Gal4, also as bait; pACT2-HNF-4α lacking the NH₂-terminal 139 amino acid residues fused to the GAL4 activation domain (amino acids 768–881) as the positive control; and pACT2-no insert or pACT2-RAD51 as the negative control. The pCAGGS expression vector bearing a β-actin promoter and cytomegalovirus enhancer or the pcDNA3.1 expression vector (Invitrogen, Carlsbad, CA) were used. PCR was performed to introduce the nucleotide sequence encoding the HA epitope (YPYDVPDYA) in-frame at the NH₂ terminus. The HNF-4α reporter plasmid (pHNF4-tk-Luc) used has been described previously (10). HNF-4α cDNA was cloned in-frame to the BamHI site of pGEX-3X (Amersham Pharmacia Biotech, Tokyo, Japan).

The yeast strain L40 (MATa trp1 leu2 his3 LYS2::lexA-HIS3 URA3::lexA-lacZ) was used. L40 harboring pBTM116-HNF4α was transformed with 100 μg of an oligo(dT) and/or random-primed human adult liver cDNA fusion library in a Gal4-activating domain vector (pACT2; Clontech, Tokyo, Japan) using lithium acetate. Transformants were selected by culture for 3 days on SD/-Trp/-Leu/-His plates containing 10 mmol/l 3-aminotriazole, a competitive inhibitor of the HIS3 gene product (Sigma, Tokyo, Japan). Colonies with a diameter >2 mm were streaked out onto a fresh master plate. After 3 days, the β-galactosidase filter assay was performed as described previously (11). Quantative assays for determining β-galactosidase activity using chlorophenol red-β-d-galactopyranoside (CPRG) were performed as described in the manufacturer’s protocol (Matchmaker Library Protocol; Clontech, Palo Alto, CA).

The partial sequence of Trip3 cDNA (accession number L40410) has been reported, and our Trip3 clone originally isolated also lacked 5′ nucleotides. To determine the sequence of full-length Trip3 cDNA, 1 μl of human kidney cDNA library was amplified by PCR with a specific primer for Trip3 (5′-cgagtttcagggttgcactgttct-3′) and a forward primer for λgt10 (forward; 5′-agcaagttcagcctggttaag-3′). PCR products were separated by electrophoresis on 2% agarose gels, transferred to nylon membranes, and hybridized with a radiolabeled Trip3-specific internal oligonucleotide (5′-ttccggaagcagactaccgag-3′). After the identified band was excised from the gel, DNA was extracted and subcloned into the TA vector for DNA sequencing. Mouse dbEST (a database of expressed sequence tags) was screened with the human sequence corresponding to mouse Trip3 cDNA.

Total RNA was isolated from mouse tissues (kidney, intestine, brain, and muscle) and from MIN6 cells using TRIzol reagent (Life Technologies, Rockville, MD) according to the manufacturer’s protocol. Pancreatic islets were isolated from 10- to 20-week-old C57BL/6 mice by the collagenase digestion method (12), and 10 μg total RNA was isolated from the islets. For RT-PCR analysis, contaminating genomic DNA was removed by treatment with DNase I. cDNA synthesis was performed with 1 μg total RNA using Moloney murine leukemia virus reverse transcriptase (Life Technologies) with dNTPs and oligo(dT) primers. The following forward and reverse primers were used for the specific PCR amplification of mouse Trip3: 5′-ttgcccgacttgccgcgtgcccta-3′ and 5′-cctaaacagcagtctgcaaactcc-3′ (annealing temperature, 60°C; product size, 360 bp). One microliter of a human pancreatic islet cDNA library (1) was also amplified with specific primers for human Trip3 (P2F, 5′-agaacagtgcaaccctgaaactcg-3′; P3R, 5′-actgaccaagtttaactccat-3′; product size, 673 bp). Ten micrograms of RNA was separated on 1% agarose gel and blotted onto Hybond N⁺ membranes (Amersham). The membranes were prehybridized and then hybridized to ³²P-labeled probes in QuikHyb hybridization solution (Stratagene, La Jolla, CA). cDNA probe for mouse Trip3 was prepared by RT-PCR as described above. Mouse β-actin probe was prepared by RT-PCR with specific primers (13) and confirmed by sequencing. The Northern images were scanned and quantified using ScanningImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

Glutathione S-transferase (GST) fusion HNF-4α protein was expressed in Escherichia coli BL21 (DE3) and purified using glutathione-Sepharose 4B beads (Amersham). Trip3 protein was synthesized using a TNT T7 coupled reticulocyte lysate system (Promega, Tokyo) with [³⁵S]cysteine. ³⁵S-labeled proteins were incubated with ∼1 μg GST fusion proteins in NETN (100 mmol/l NaCl, 1 mmol/l EDTA, 20 mmol/l Tris-HCl, pH 8.0, and 0.5% NP40) for 1.5 h at 4°C. After washing, specifically bound proteins were subjected to 15% SDS-PAGE. Then, the radiolabeled proteins were detected using autoradiography and a BAS-2000 image analysis system (FujiFilm, Tokyo).

COS-7 and MIN6 cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% and 15% fetal bovine serum, respectively. When 80% confluence was reached, the cells were detached and plated in 6-well plates at a density of 6 × 10⁴ and 3 × 10⁵ cells per well, respectively. Transfection was performed for 3 h in serum-free OPTI-MEMI (Life Technologies) with the indicated amounts of the expression vectors and 0.5 μg of the reporter genes together with 10 ng pRL-SV40, the internal control plasmid (Promega), using Lipofectamine Plus Reagent (Life Technologies) according to the manufacturer’s protocol. After incubation for another 24 h (COS-7) or 48 h (MIN6), luciferase activity was measured with the Dual-Luciferase Reporter Assay System (Promega) (14).

We studied 41 unrelated Japanese subjects with early-onset type 2 diabetes (diagnosed at <30 years of age) with a family history of type 2 diabetes in second-degree relatives (15). Nine of the 41 patients were considered to have MODY, which was defined as type 2 diabetes occurring in at least three generations, consistent with autosomal dominant transmission and onset before 25 years of age in at least one affected subject from the family. Eleven subjects were diagnosed as having type 2 diabetes before 15 years of age. These 11 subjects might have also had MODY, since it is common in Japanese of this age group (16). BLAST homology searches of human genomic sequences were performed to determine the exon-intron boundaries of the Trip3 gene. Based on the database sequences, specific PCR primers were designed to amplify the coding regions and flanking introns, and then mutation screening was performed. The mutations/polymorphisms that were identified in the diabetic patients were also assessed in 50 control subjects by DNA sequencing. Written informed consent was obtained from all of the participants.

To isolate proteins interacting with HNF-4α, we used a LexA DNA-binding domain fused to the AF-2 region of human HNF-4α (amino acids 140–474) as bait in the yeast two-hybrid screening assay (Fig. 1).

The DNA-binding domain was eliminated to avoid nonspecific binding with DNA. More than 10⁶ primary transformants from an adult human liver cDNA library were screened, yielding seven potentially positive clones judged by β-galactosidase filter assay. Sequence analysis of these seven clones revealed that two independent clones encoded Trip3. Analysis of the other five clones is still in progress. Trip3 has previously been reported to interact with the thyroid hormone receptor and also with the retinoid X receptor in a ligand-dependent manner, but its function has not been clarified (9). To confirm that the Trip3 clone is a true positive, we retransformed the clone with the bait or control plasmids and compared the resultant β-galactosidase activity. Because HNF-4α generates a homodimer, it was used as a positive control. Trip3 transformants cotransformed with HNF-4α showed about half the β-galactosidase activity of the positive control, whereas no β-galactosidase activity was seen in those with no insert or Rad51 negative control vectors (Fig. 2). A different bait (pAS2–1, with the Gal4 DNA-binding domain fused to HNF-4α) and a different yeast strain (CG1945) were also used to test the interaction, and similar results were obtained (data not shown).

The clones of Trip3 isolated in the two-hybrid screening assay contained a partial fragment of the COOH-terminal 119–amino acid residues. Because the previously published sequence of Trip3 also lacked the NH₂-terminal residues, we cloned the entire coding sequence of Trip3 by PCR-based cloning from a human kidney cDNA library. The cDNA contained an open reading frame for a protein of 155 amino acids (accession number AF400652). This human sequence was used to screen mouse dbEST, and three full-length mouse clones (AI194289, AK014350, and AK004259) were found. Mouse Trip3 encoded a protein consisting of 151 amino acids that showed 79.4% identity with human Trip3 protein at the amino acid level. Trip3 has no similarity to known proteins in mammalians. However, it has a cysteine-rich region at the NH₂ terminus, and this region was recognized as a potential zinc finger, metal-binding, and DNA-binding domain, according to a motif scan in PROSITE of the ExPASy Molecular Biology Server (on-line at http://www.expasy.ch). Besides, human and mouse Trip3 both contained a short conserved LXXLL binding motif with HNF-4α (17,18). A BLAST search also revealed that Trip3 showed weak homology (38%) with the Drosophila melanogaster gene product CG8204 (19).

Human Trip3 was ubiquitously expressed in all tissues examined, including muscle, kidney, and pancreas (9). Trip3 gene is expressed at higher levels in muscle and kidney than in pancreas. Trip3 expression in pancreatic islets (β-cells) has not been investigated previously. RT-PCR and Northern blot analysis demonstrated Trip3 expression in human and mouse pancreatic islets as well as in a mouse β-cell line (MIN6 cells) (Fig. 3). The expression level of mouse Trip3 in pancreatic islets was about 64 and 47% of that in kidney and MIN6 cells, respectively, adjusted by the internal control, β-actin. These data suggested that the Trip3 gene is expressed in pancreatic islets (β-cells), a site where HNF-4α is also expressed.

Direct interaction of Trip3 with HNF-4α was examined using a GST pull-down assay (Fig. 4). Trip3 did not bind to GST protein, but it did bind to GST-HNF-4α in the absence of its ligand. This result confirmed that Trip3 can interact with HNF-4α.

To examine whether Trip3 is capable of modifying the transcriptional activity of HNF-4α, cotransfection experiments were performed. Expression of Trip3 protein was confirmed by Western blot analysis using anti-HA monoclonal antibody (data not shown). In COS-7 cells, Trip3 enhanced the transcription activity of HNF-4α in a concentration-dependent manner by two- to threefold compared with HNF-4α alone (Fig. 5A). Similar results were also obtained using MIN6 cells (Fig. 5B). These findings suggest that Trip3 is a coactivator of HNF-4α.

A homology search through human genomic sequences indicated that human Trip3 cDNA matched the human sequence segment (AC023133) localized to chromosome 17. The Trip3 gene consisted of five exons spanning ∼10 kb. The coding region and flanking introns were screened for mutations in 41 unrelated Japanese subjects with early-onset type 2 diabetes. Sequences of the primers used for amplification and sequencing are shown in Table 1.

This analysis revealed two polymorphisms in intron 1: IVS1+148–49insCGGGAGGG (insertion of CGGGAGGG between nucleotides 48 and 49 of intron 1; the G of the acceptor site invariant GT was denoted as nt +1) and IVS1 + 56G→C (G to C substitution at nucleotide 56 of intron 1). The two polymorphisms were always found simultaneously, suggesting linkage disequilibrium. The frequencies of these polymorphisms were similar between the diabetic group (6 of 82, 7.3%) and the control group (3 of 100, 3.0%).

Mutations of the HNF-4α gene are a cause of MODY1 (2). HNF-4α is a transcription factor belonging to the steroid receptor superfamily, and it plays an important role in regulating the expression of genes by pancreatic β-cells. It has been reported that HNF-4α regulates the genes for GLUT2, l-pyruvate kinase, aldolase B, and 2-oxoglutarate dehydrogenase E1 (20). Cofactors that interact with HNF-4α and modify its effect on transcriptional activation could also be involved in the regulation of such genes. Previous studies have identified steroid receptor coactivator (SRC)-1, SRC-2/glucocorticoid receptor interacting protein-1 (GRIP-1), p300, and CREB binding protein as proteins that interact with HNF-4α (21,22). In the present study, using a yeast two-hybrid system, we isolated Trip3 as another protein that interacted with HNF-4α, and direct interaction between HNF-4α and Trip3 was also confirmed using the GST pull-down assay. Earlier studies have indicated that Trip3 interacts with the thyroid hormone receptor and also with the retinoid X receptor, but not with the glucocorticoid receptor (9). However, the function of Trip3 has been unclear. Our transfection data strongly suggest that Trip3 is a novel coactivator of HNF-4α. Many cofactors bind to a common region (the COOH-terminal helix 12 of the AF-2 region) of nuclear receptors to cause transcriptional activation (23,24). These cofactors contain a short conserved LXXLL motif (L denotes leucine and X denotes any amino acid), which appears to be both necessary and sufficient for such interactions (17,18). Interestingly, both human and mouse Trip3 contained this motif, suggesting that Trip3 may bind to HNF-4α via this motif. A protein database search revealed that Trip3 has a zinc finger, DNA-binding domain–like sequence at the NH₂ terminus, suggesting that Trip3 might bind to DNA directly. Further studies are necessary to clarify the molecular mechanism underlying the activation of HNF-4α by Trip3.

In this study, we also screened for Trip3 gene mutations in Japanese subjects with MODY/early-onset type 2 diabetes. Although there was no evidence that Trip3 gene mutations are a common cause of MODY/early-onset type 2 diabetes in the Japanese population, Trip3 gene mutations may be contributory in other populations. Otherwise, since Trip3 interacts with the thyroid hormone receptor, its mutations might be involved in thyroid disease. Furthermore, the mechanism of transcriptional regulation of the Trip3 gene is unknown. For example, the expression of PGC-1 (peroxisome proliferator–activated receptor-γ [PPAR-γ] coactivator-1), a coactivator of PPAR-γ, is altered by β₃-adrenergic agonists (25) and insulin (26). Trip3 expression and the regulation of glucose metabolism in pancreatic β-cells by HNF-4α might also be influenced by various stimuli.

In summary, we cloned a novel coactivator of HNF-4α. Interaction between HNF-4α and Trip3 might play an important role in the regulation of glucose homeostasis. The information presented here should facilitate studies regarding the roles of Trip3 and HNF-4α in normal β-cell function and in the pathophysiology of MODY/type 2 diabetes.

This work was supported by grants from the Japanese Ministry of Science, Education and Culture (13204054 and 13671190); the Japan Diabetes Foundation; and the Research for the Future Program of the Japan Society for the Promotion of Science (97L00801).

We thank J. Miyazaki and A. Shinohara for their kind gifts of pCAGGS and pBTM116/Rad51, respectively; Graeme Bell (University of Chicago) for providing the human pancreatic islet cDNA library; and Y. Tsujimoto, Y. Eguchi, and T. Kodama for their helpful comments.