Potential conflict of interest: Nothing to report.
Hepatocyte nuclear factor-4 alpha (HNF4A) is an essential transcriptional regulator for many genes that are expressed preferentially in the liver. Among the important functions of the liver is drug metabolism in response to xenobiotic exposure. Recent studies have suggested that HNF4A regulates the expression of cytochrome P450 (CYP), including CYP2D6 and CYP3A4, which show large individual variations in their activities. To understand the genetic factors that influence individual CYP activities, a genetic variant of HNF4A and the effects of genetic variants of HNF4A on CYP activity were investigated. Here, we report the identification of a novel coding variant of HNF4A that influences CYP2D6 activity in humans. After direct sequencing, a polymorphism search revealed the HNF4A G60D variant in Koreans. This variant was unable to bind to the recognition site in the CYP2D6 promoter and therefore lacked the regulatory function for this gene. Human liver specimens with the heterozygous HNF4A G60D genotype showed a tendency toward lower levels of CYP2D6 activity than the wild-type genotype in the same genetic background of CYP2D6. Furthermore, human subjects with the HNF4A G60D genotype tended to have lower CYP2D6 activity than those with the wild-type HNF4A. The HNF4A G60D variant was detected at low frequency in Asian populations, including Koreans, Chinese, and Vietnamese, and was not found in Africans or Caucasians. Conclusion: This is the first report to show that the genetic polymorphism of liver-enriched nuclear receptor HNF4A influences downstream CYP2D6 function in human subjects. (HEPATOLOGY 2008;48:635–645.)
The nuclear receptor hepatocyte nuclear factor 4 alpha (HNF4A; also called NR2A1) is a transcriptional regulator of genes in the transport and metabolism of glucose and lipids. Through a complex hierarchical transcriptional network, HNF4A plays an essential role in the normal physiologic functions of hepatocytes and pancreatic cells.1
In the pancreas, HNF4A regulates the expression of genes that are involved in glucose metabolism and insulin secretion.2 The important role of HNF4A in glucose homeostasis is exemplified by the identification of HNF4A mutants in certain types of diabetes. Maturity onset of diabetes in the young, which is a rare form of non–insulin-dependent diabetes mellitus, is a dominantly inherited disease that is characterized by defective glucose-dependent secretion of insulin by pancreatic β-cells. Nonsense, missense, and frameshift mutations in the human HNF4A gene have been identified in maturity onset of diabetes in the young patients. Those mutations include R154Stop, Q268Stop, F75delT, K99fsdelAA, G115S, R127W, V255M, E276Q, and V393I (reviewed by Ryffel3). Because HNF4A is also expressed in other tissues, such as the liver, small intestine, colon, and kidney, these mutant HNF4A forms may affect the normal physiology of these tissues.
In livers, HNF4A is known to be involved in the expression of many hepatic drug disposition-related proteins, such as cytochrome P450s (CYP), phase II enzymes, and transporters. HNF4A is required for the induction of CYP1A1, CYP1A2, CYP2C9, CYP2C8, CYP2A6, and CYP3A4, which have HNF4A binding sites in their promoters.4–7 For example, in hepatic cells, HNF4A determines the extent of CYP3A4 induction through the pregnane X receptor or constitutive androstane receptor–mediated pathways. Tirona et al.4 found no induction of CYP3A4 in HNF4A-deficient cells. A more dramatic demonstration of the important role of HNF4A in the regulation of CYP is the constitutive expression of CYP2D6 by this factor.8 It is well known that the expression or activity of CYP2D6 is not inducible by xenobiotics, including ligands for pregnane X receptor, CAR, and aromatic hydrocarbon receptor, but shows constitutive expression.9, 10 The CYP2D6 promoter incorporates a strong HNF4A binding site and requires functional HNF4A for its constitutive expression. In the absence of functional HNF4A, the transcriptional activity of the CYP2D6 promoter is negligible. Furthermore, Corchero el al.11 have found that a conditional knockout of HNF4A in mice decreases CYP2D6 activity by more than 50%, which suggests that HNF4A regulates CYP2D6 activity in vivo.
As previously mentioned, HNF4A is involved in the regulation of CYP expression. Although most of the CYPs under the regulation of HNF4A (for example, CYP1A1, CYP1A2, CYP3A4, CYP2A6, CYP2D6, CYP2C8, and CYP2C9) show some genetic polymorphisms (reviewed by Daly12), the reported genetic variants do not fully explain the interindividual differences seen for these enzymes. Given the important role of HNF4A, it is possible that genetic variants of HNF4A contribute to these differences. In the current study, we sequenced HNF4A genes in a Korean population, and found two novel nonsynonymous HNF4A variants. The functional consequences of these variants were tested both in vitro and in vivo.
aa, amino acid; cHRE, consensus sequence for HNF4A binding sites; CYP, cytochrome P450; EMSA, electrophoretic mobility shift assay; HNF4A, hepatocyte nuclear factor-4α; HWE, Hardy-Weinberg equilibrium; LA-PCR, long and accurate polymerase chain reaction; LD, linkage disequilibrium; MR, metabolic ratio; PCR, polymerase chain reaction; SNP, single nucleotide polymorphism; TBST, 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 0.1% (vol/vol) Tween-20; WT, wild-type.
Patients and Methods
Subjects and Human Livers.
In total, 562 unrelated Korean subjects were recruited for the HNF4A genotyping study. Genomic DNA samples from 139 Vietnamese, 94 Chinese, and 153 Caucasian subjects were obtained from the Biomedical Resource Bank of the Pharmacogenomics Research Center, Inje University.13, 14 All Koreans had ethnic background of Korean for at least three generations by self-identification. The racial backgrounds of the Vietnamese and Chinese subjects were Viet Kinh and Han, respectively. DNA samples from African Americans and Caucasians were originally provided by Dr David A. Flockhart at the Indiana University School of Medicine. All of the participants were judged to be healthy, based on their medical histories, physical examinations, and routine laboratory tests. All participants provided written informed consent for the current study, which was approved by the Institutional Review Board of Busan Paik Hospital (Busan, Korea). Forty-three human livers were obtained from the Biomedical Resource Bank of the Pharmacogenomics Research Center. The livers were surgically resected, and only nontumor parts of the tissues were collected for tissue banking by experienced pathologists at the Department of Anatomical Pathology, Busan Paik Hospital. All tissues in the Bank were confirmed to be histologically normal without evidence of liver cirrhosis, which was reviewed by the pathologists. The patients including 23 women (aged 40–72) and 20 men (aged 39–76) who were diagnosed for hepatic biliary stone (n = 18), liver metastasis (n = 12), cholangiocarcinoma (n = 8), hepatic cyst (n = 2), hemangioma (n = 1), benign tumor (n = 1), or gallbladder cancer (n = 1).
DNA Purification and Sequencing.
Of 562 Koreans, 50 randomly selected subjects were analyzed by direct sequencing for the identification of HNF4A gene polymorphisms. The other 512 subjects were genotyped for only two novel variants. Genomic DNA was isolated from whole blood cells using the Qiagen DNA Extraction Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. Specific primers were designed to amplify 10 exons, the 2-kb proximal promoter region, the exon-intron boundaries, and the 3′-UTR of the HNF4A gene. Polymerase chain reaction (PCR) was performed in a reaction volume of 20 μL that contained 100 ng of genomic DNA, 1 PCR buffer, 0.2 mM deoxyribonucleotide triphosphates, 0.2 μM of each primer, 1.5 mM MgCl2, and 1 U Taq polymerase (Takara, Shiga, Japan). PCR was performed in the GeneAmp PCR 9700 (Applied Biosystems, Foster City, CA), with an initial denaturation step of 95°C for 5 minutes, followed by 35 cycles of denaturation at 95°C for 30 seconds, annealing at 49.3°C to 62.7°C for 30 seconds, and extension at 72°C for 30 to 60 seconds. A final termination of elongation step was performed at 72°C for 7 minutes. The primer sequences and annealing temperatures used are listed in Supplementary Table 1. The amplified PCR products were sequenced using an automated sequencer and the BigDye Terminator Sequencing Kit (Applied Biosystems).
Linkage Disequilibrium Analysis and Haplotype Prediction.
Two population genetics analysis programs, the SNPAlyze 4.1 (Dynacom, Yokohama, Japan) and Haploview software,15 were used for pairwise linkage disequilibrium (LD) analysis, haplotype prediction, and Hardy-Weinberg equilibrium (HWE) testing. The LD for each of the pairs of segregating sites was quantified using the values of D′ and r square.
Genotyping of HNF4A 4676G>A (G36S) and 4749G>A (G60D) Variants by Duplex Pyrosequencing.
We found two novel HNF4A variants in our genetic screening of 50 Korean subjects. For these variants, further genotyping was performed by pyrosequencing the genomic DNA samples from 562 Korean (as well as 50 human livers of Korean origin), 139 Vietnamese, 94 Chinese, and 153 Caucasian subjects. To detect the HNF4A variants, partial fragments of the HNF4A gene were amplified using the primer pair, with one primer being biotinylated at the 5′-end. The sequences of the primers were: 5′-biotin-AAGGCTCCCTTAGATGCCT-3′ and 5′-TCAGGGAGAAGACAGACCTTG-3′. The biotinylated PCR products were immobilized on streptavidin-coated beads (Streptavidin Sepharose High Performance; Amersham Biosciences, Uppsala, Sweden) following strand separation for the PSQ 96 Sample Preparation kit (Pyrosequencing; Amersham Biosciences). Briefly, Sepharose bead slurry (3 μL) was mixed with binding buffer (37 μL), PCR product (15 μL), and distilled water (25 μL) and incubated at room temperature for 10 minutes on a shaker. The beads were transferred to a filter plate and the liquid was removed by vacuum filtration (Multiscreen Resist Vacuum Manifold; Millipore, Billerica, MS). The DNA strands were separated in denaturation solution (0.5 M NaOH) for 5 seconds. The immobilized template was washed with 10 mM Tris-acetate (pH 7.6) (washing buffer) and transferred to a PSQ 96 plate, resuspended in annealing buffer [20 mM Tris-acetate (pH 7.6)] that contained the sequencing primer mix. The sequencing primer mix included the primer 5′-CGTTGAGGTTGGTGC-3′ for 4676G>A variant and the primer 5′-GCACCGTAGTGTTT-3′ for the 4749G>A variant. The sequencing primer was annealed at 80°C to 90°C for 3 minutes. The sequence was analyzed using the PSQ 96 system (Biotage AB, Uppsala, Sweden) with the SNP Reagent Kit (Amersham Biosciences).
The wild-type HNF4A-coding plasmid, pRC/CMV2-hHNF-4-WT, was provided by Dr. Terumasa Tsuchiya (Research Center for Molecular Medical Science, Tokyo, Japan). The HNF4A G36S and G60D variants were generated from pRC/CMV2-hHNF-4-WT using the Strategene QuickChange XL Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's protocol. The following pairs of primers (with the altered nucleotide) were used for the site-directed mutagenesis: for HNF4A G36S, 5′-CGTCCCCATCAGAAAGCACCAACCTCAACG-3′ and 5′-CGTTGAGGTTGGTGCTTTCTGATGGGGACG-3′; and for HNF4A G60D, 5′-CCGGGCCACGGACAAACACTACGGTGC-3′ and 5′-GCA- CCGTAGTGTTTGTCCGTGGCCCGG-3′. HNF4A expression plasmids were then constructed using the pcDNA3.1 vector, to ensure high expression of the recombinant HNF4A in transfected cells. The expression plasmids pcDNA3.1(+)-hHNF-4-WT, -G36S, and -G60D were generated by PCR using pRC/CMV2-hHNF-4-WT, -G36S, and -G60D, respectively, as template, together with the upstream oligonucleotide 5′-AGTAGGATCCATGGACATGGCCGACTACAG-3′ (contains a BamHI site) and downstream oligonucleotide 5′-GCATGAATTCCTAGA TAACTTCCTGCTTGG-3′ (contains an EcoRI site). The resulting expression fragment was subcloned into the BamHI and EcoRI sites of pcDNA3.1(+) (Invitrogen, Carlsbad, CA). All of the constructs were confirmed by DNA sequencing.
For the functional assay of HNF4A, a CYP2D6 promoter-driven luciferase reporter system was developed, because the CYP2D6 promoter includes a high-affinity binding site for HNF4A. A 1.8-kb fragment from the 5′-flanking region of the CYP2D6 gene was isolated by PCR using human genomic DNA as template and the following primers: 5′-GATCCCTCGAGCACTGGCTCCAAGCATGGCAG-3′ (contains an XhoI site) and 5′-ATTCAAGCTTACCTGCCTCACTACCAAATG-3′ (contains a HindIII site). The resulting promoter fragment (-1849 to +1 of the CYP2D6 gene, nucleotides numbered relative to the transcriptional start site) was inserted into the XhoI and the Hind III sites of pGL2 (Promega, Madison, WI).
Transient Transfection and Promoter Activity Analysis.
Previous studies have shown that human kidney cell line, 293, does not express endogenous HNF4A protein while it retains substantial function of all other coactivators required for transactivation by ectopic HNF4A.16 For these reasons, the effect of mutant HNF4A proteins on its target genes has been studied in this cell line without the cotransfection of coactivators or corepressors.17 Therefore, we selected this cell line to test the functionality of novel HNF4A variants. The 293F cells were cultured at 37°C in 5% CO2in Dulbecco's modified Eagle's medium that was supplemented with 10% heat-inactivated fetal bovine serum and 100 IU/mL penicillin-streptomycin (Gibco, Gaithersburg, MD).To measure the transactivation of the CYP2D6 promoter by the HNF4A-WT, -G36S, and -G60D proteins, the 293F cells were transfected with HNF4A together with the CYP2D6 promoter-reporter plasmids. In brief, the 293F cells were grown to a density of 5 × 105 cells/well in six-well plates and cotransfected with 3.6 μg CYP2D6 promoter construct, 0.3 μg β-galactosidase reporter plasmid, and 0.05 to 0.3 μg HNF4A wild-type (WT) and variant expression constructs using Lipofectamine 2000 (Invitrogen), according to the manufacturer's protocol. After incubation for 24 hours, the cells were harvested and analyzed for luciferase and β-galactosidase activities using the Luciferase Reporter Gene Assay (Roche, Basel, Switzerland) and CPRG (Roche), respectively. The relative luciferase values were normalized to the β-galactosidase activities. In all experiments, the values given represent the mean ± standard deviation of at least three experiments. A minimum of two plasmid preparations were used for each construct.
Immunoblot Analysis of HNF4A Expression.
To measure the expression of the WT and variant HNF4A proteins in transiently transfected 293F cells, an immunoblot assay was performed. The 293F cells were first washed with 1 mL phosphate-buffered saline, then scraped into 1 mL phosphate-buffered saline, and pelleted twice by centrifugation. The cell pellet was resuspended in cold radio immunoprecipitation assay buffer [150 mM NaCl, 1% (vol/vol) nonyl phenoxylpolyethoxylethanol, 0.5% (wt/vol) deoxycholic acid, 0.1% (wt/vol) sodium dodecyl sulfate, 50 mM Tris-HCl (pH7.5)], and the cells were allowed to swell on ice for 15 minutes. The homogenate was centrifuged, and the supernatant fractions were collected. Protein concentrations were determined by the Bradford assay (Bio-Rad Laboratories, Hercules, CA), according to the manufacturer's recommendations. Whole cell lysates were separated on a NuPAGE 4% to 12% Bis-Tris Gel (Invitrogen) and then transferred to a polyvinylidene fluoride membrane (Amersham Life Science, Arlington Heights, IL). The membrane was preincubated for 2 hours at room temperature in 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 0.1% (vol/vol) Tween-20 (TBST) in the presence of 5% (wt/vol) nonfat dried milk. The membrane was incubated overnight at 4°C in a 1:200 dilution of the specific anti-HNF4A antiserum (Santa Cruz Biotechnology, Santa Cruz, CA) in TBST that contained 1.3% nonfat dried milk. After thorough washing, the membrane was incubated with peroxidase-conjugated anti-goat antibody (diluted 1:2000 in TBST that contained 1.7% nonfat dried milk) for 1 hour at room temperature. The blot was washed again in TBST and peroxidase activity was revealed with the enhanced chemiluminescence system (Amersham Life Sciences).
In Vitro Translation of Wild-Type and Variant HNF4A Proteins.
The wild-type and variant HNF4A proteins were synthesized in vitro using the TNT Quick Coupled Transcription/Translation System (Promega) according to the manufacturer's instructions. For these systems, 1 μg circular plasmid DNA of pcDNA3.1(+)-hHNF-4-WT, -G36S or -G60D was added to an aliquot of the TNT Quick Master Mix and incubated in a 50-μL reaction volume for 90 minutes at 30°C. The expression of HNF4A proteins was confirmed by immunoblotting.
Electrophoretic Mobility Shift Assay.
To test the DNA-binding ability of the wild-type and variant HNF4A proteins, electrophoretic mobility shift assay (EMSA) was performed using double-stranded oligonucleotides for the CYP2D6 promoter region and a commercial consensus-binding motif for the HNF4A protein. The oligonucleotide sequence of the CYP2D6 promoter is 5′-AGCAGAGGGCAAAGGCCATCATCAG-3′, which corresponds to the sequence between nucleotides -56 and -36 of the human CYP2D6 promoter. The sequence specificity of HNF4A binding was confirmed using a commercial oligonucleotide for the consensus HNF4A binding site (Santa Cruz Biotechnology). The probes were end-labeled with T4-polynucleotide kinase and [−P]adenosine triphosphate. The DNA binding reaction was performed at room temperature for 30 minutes with 10 μg nuclear extracts in 20 μL of a mixture that contained 5% glycerol (vol/vol), 20 mM Tris-HCl (pH 8.0), 60 mM KCl, 25 mM MgCl2, and [32P]-labeled oligonucleotide probes. DNA–protein complexes were separated electrophoretically from unbound DNA probe on nondenaturing 4% polyacrylamide gels. For competition experiments, unless otherwise indicated, 1 to 10 pmol of unlabeled oligonucleotide was mixed with 0.1 pmol of the labeled probe before addition to the nuclear extracts.
Blood samples for genotyping were collected from 130 volunteers. Genomic DNA was isolated from the blood samples using the QIAamp Blood Kit (Qiagen) and stored at −20°C until use. Of the 130 genomic DNAs, 51 randomly selected subjects were used for complete DNA sequencing analysis. For direct DNA sequencing analysis, the CYP2D6 gene was amplified by Long and Accurate PCR (LA-PCR, TaKaRa LA-PCR kit; TaKaRa) using the primers 5′-CACTGGCTCCAAGCATGGCAG-3′ and 5′- ACTGAGCCCTGGGAGGTAGGTA-3′. The length of the LA-PCR product was 6.7 kb, and it contained all nine exons, the intron/exon boundaries, and the 5′ untranslated region (UTR) of the CYP2D6 gene. The genomic DNA (200 ng) was dissolved in LA-PCR buffer that contained 3.0 mM MgCl2, 0.2 μM primers, 400 mM deoxyribonucleotide triphosphates, and 2.5 U LA-Taq polymerase. After initial denaturation at 94°C for 1 minute, the samples were subjected to 30 cycles of 10 seconds at 98°C, 30 seconds at 64°C, and 7 minutes at 72°C, with a final elongation step at 72°C for 10 minutes. The PCR products were directly sequenced using the Applied Biosystems Model 377A DNA Sequencer. For the genotyping of the remaining subjects, the SNaPshot method was performed to discriminate the CYP2D6 *1, *2, *2L, *3, *4, *5, *10B, *14, *18, *21, *41, *49, *52, and *60 alleles (http://www.cypalleles.ki.se/cyp2d6.htm). The details of the SNaPshot method are contained in a pending patent (PCT, KR2007/003102) (unpublished data).
In Vitro Enzymatic Activity Assay.
The enzymatic activity of CYP2D6 in human livers was determined using CYP2D6-catalyzed dextromethorphan O-demethylation as a probe for the metabolic pathway. It is well established that dextromethorphan is demethylated to dextrorphan exclusively by CYP2D6.18 The incubation mixtures (0.25 mL) were composed of 12.5 μg microsomal protein with 1 μg cytochrome b5 in 50 mM potassium phosphate buffer (pH 7.4) that contained probe substrates in the range of 2 to 100 μM. After a 5-minute preincubation at 37°C, the reactions were initiated by the addition of the nicotinamide adenine dinucleotide phosphate, reduced form–generating system (3.3 mM glucose-6-phosphate, 1.3 mM nicotinamide adenine dinucleotide phosphate, 3.3 mM MgCl2, 1.0 U/mL glucose-6-phosphate dehydrogenase) and incubated at 37°C for 30 minutes. The reactions were stopped by placing the incubation tubes on ice. The incubation mixtures were then centrifuged at 20,000g for 10 minutes at 4°C. Aliquots (2 mL) of the supernatants were injected into the liquid chromatography with tandem mass spectrometry detection system, which comprised the Agilent 1100 series high=performance liquid chromatography (Agilent, Wilmington, DE) and the API 3000 tandem mass spectrometer (Applied Biosystems). The liquid chromatography with tandem mass spectrometry detection methods for the quantification of dextrorphan were based on the method of Eichhold et al.,19 with slight modifications. The liquid chromatography chromatograms were obtained by separation on a Luna C18 column (250 mm, 3 mm; Phenomenex, Torrance, CA) in an isocratic mobile phase that consisted of acetonitrile and water (3:7, vol/vol and 4:6, vol/vol, respectively) at a flow rate of 0.2 mL/minute. The turbo-ion spray interface was operated in the positive mode at 5000 V and 400°C. The optimum collision energy for the ionization of dextrorphan was 35 eV. The multiple reaction monitoring mode using specific precursor/product ion transition was employed for the quantification. The detection of dextrorphan was performed by monitoring the transitions of m/z, 258/157.
Phenotyping was performed on 130 CYP2D6-genotyped subjects, who gave written informed consent to take part. The subjects were admitted to overnight housing in the Clinical Trial Center at Pusan Paik Hospital and were prohibited from taking alcohol or medicines for 1 week before and during the trial period. After emptying the bladder, the subjects took a single oral dose of dextromethorphan 30 mg (Romilar; Roche Korea, Ansung, Korea). Food was not allowed for 4 hours before and 4 hours after dosing. All urine was collected for up to 8 hours after the administration of dextromethorphan. The urine volume was measured and aliquots were stored at −20°C until assayed. The dextromethorphan and dextrorphan levels in the urine samples were determined by the reverse-phase high-performance liquid chromatography method previously reported by Bartoletti et al.20 The mean values of urinary metabolic ratio (MR) were used as an index of CYP2D6 activity.
For the CYP2D6 activity and MR values in the group of the same CYP2D6 genotypes in the phenotype study, normality tests were performed to check the normality of the data points. In the groups that did not show the normality, the CYP2D6 activity or MR values were compared using the Mann-Whitney U test for pairwise comparison between different HNF4A genotypes. Extent of correlation between CYP2D6 expression and its activity in human livers was evaluated by Pearson's correlation test. All statistical analysis was performed using SAS (SAS version 9.1.3 software; SAS Institute, Cary, NC). A P value of less than 0.05 was considered statistically significant.
Identification of the HNF4A G60D SNP in a Korean Population.
The HNF4A gene was amplified from 50 Korean subjects using a set of specific primers that span a DNA region that encodes amino acids, exon-intron boundaries, and a 2-kb 5′-flanking region. Bulk sequencing of each PCR product identified 20 single-nucleotide polymorphisms (SNPs) and one deletion (del) mutation in the HNF4A gene (Table 1). Among the SNPs, the 4676G>A and 4749G>A variations caused amino acid (aa) conversions from glycine at aa 36 to serine and from glycine at aa 60 to aspartate, respectively. The other two exonic SNPs, 4767G>C and 28152G>T, were synonymous variations without aa changes. In the promoter region, seven SNPs and one del mutation were found with frequencies that ranged from 1% to 56%. Four promoter SNPs appeared at frequencies greater than 10%. Two nonsynonymous SNPs were confirmed by repeated PCR amplification, followed by direct sequencing of multiple clones. All of the HNF4A SNPs exhibited HWE at P > 0.05. After the initial sequencing of DNA from 50 subjects, the frequency of HNF4A G36S was found to be 7%, although this frequency decreased to 3.8% after the screening of up to 612 subjects. The frequency of HNF4A G60D was 1.3% in 612 subjects (Table 2). To compare the frequencies of these HNF4A variants in diverse ethnic groups, the allelic frequencies of the HNF4A G36S and G60D SNPs were determined in large populations of subjects from various ethnic backgrounds, such as Korean, Chinese, Vietnamese, Caucasian, and African American. The HNF4A G36S SNP was found in all ethnic groups, except African Americans, at frequencies of 1.0% to 3.8%. The HNF4A G60D SNP was found in Korean (1.3%), Chinese (0.5%), and Vietnamese (0.7%) populations, but it was not observed in Caucasians or African Americans (Table 2).
Table 1. Genetic Variations of the HNF4A Gene Identified in Korean Subjects
The LD was analyzed using the D′ values across the HNF4A SNPs that were found with allelic frequencies of greater than 5% in the initial screening. Among the 15 possible pairs of variants, only two pairs (−2003G>A and −1048delGGG; −1650A>G and −1048delGGG) showed significantly strong LD (ID′I = 1; red boxes in Fig. 1). Moderate LD was found between −1048delGGG and −755A>C. The other 12 pairs clearly showed no or unknown LD (Fig. 1; white boxes and blue boxes, respectively). Because of the rarity of LD among the HNF4A SNPs, haplotype analysis with only six common variants predicted more than 25 haplotypes with frequency greater than 1% (data not shown).
HNF4A G60D, But Not G36S, Shows Loss of Transcriptional Activity.
From the SNP screening, the HNF4A G36S and G60D variants were identified initially. To assess the functional consequence of these novel variants of HNF4A, the mutations were introduced into a full-length HNF4A expression construct by site-directed mutagenesis. The transcriptional activities of the HNF4A G36S and G60D SNPs were then compared with that of wild-type HNF4A. Various doses of the HNF4A vectors were transfected into 293F cells together with a firefly luciferase reporter gene regulated by a CYP2D6 promoter that included a strong HNF4A binding site, and with a control cytomegalovirus-regulated β-galactosidase vector. As anticipated, the wild-type HNF4A consistently showed high transactivational activity for the CYP2D6 promoter in a gene-dosage–dependent manner (Fig. 2). Similar transactivation activity was observed for the HNF4A G36S variant. In contrast, the HNF4A G60D variant did not activate transcription from the CYP2D6 promoter. To rule out the possibility that the HNF4A G60D construct failed to produce a protein, immunoblotting was carried out to measure the wild-type and variant HNF4A protein levels in HNF4A-transfected cells. The results showed normal expression of HNF4A proteins in G60S-transfected cells, as well as in the wild-type HNF4A-transfected and G36S-transfected cells. The protein levels correlated well with the dosages of transfected DNA. Given that the protein level of G60D was comparable to that of the wild-type HNF4A, it appeared that the G60D variant had reasonably high protein stability, similar to that of the WT protein, in the transfected cells. Therefore, our results show that G60D HNF4A is transcriptionally inactive, and that the loss of transcriptional activity of this mutant is not attributable to altered protein stability.
HNF4A G60D Shows Loss of DNA-Binding Activity.
HNF4A forms a homodimer that binds to its recognition sites. To examine the DNA-binding properties of HNF4A variants, EMSA was performed using in vitro synthesized HNF4A, the [32P]-labeled HNF4A binding site located in the CYP2D6 promoter (HRE-CYP2D6), and the commercially available consensus sequence for HNF4A binding sites (cHRE). Similar amounts of each HNF4A protein were obtained after in vitro synthesis, as confirmed by immunoblotting (data not shown). The wild-type HNF4A showed shifted bands after incubation with HRE-CYP2D6 or cHRE (Fig. 3). Competition experiments with fivefold and 50-fold excesses of unlabeled HRE-CYP2D6 or cHRE confirmed that the shifted bands contained HNF4A–HRE complexes. Similar amounts of HNF4A–HRE complexes were detected after incubation with HNF4A G36S and [32P]-labeled HREs. In contrast, the HNF4A G60D variant did not bind to either HRE-CYP2D6 or cHRE.
CYP2D6 Activities in Human Livers with the HNF4A G60D Genotype.
In the in vitro experiments, the HNF4A G60D variant showed loss of transcriptional activity for the CYP2D6 promoter. We investigated further the effects of this variant on CYP2D6 expression and activity in human livers. Because it is well established that the CYP2D6 gene shows strong genetic polymorphism and exhibits ethnic diversity, we genotyped CYP2D6 using SNaPshot analysis of the genomic DNAs from 50 human livers. The CYP2D6 alleles found in the livers included CYP2D6*2, *2N, *5, *10B, *14, *21, and *41 (see http://www.cypalleles.ki.se/for allele nomenclature). Based on the known enzymatic activity of each allelic variant (see http://www.cypalleles.ki.se/cyp2d6.htm), CYP2D6*1 and *2 were grouped as ”normal” alleles, and CYP2D6*5, *14, and *21 were grouped as ”null” alleles (Fig. 4). We then measured the CYP2D6 protein levels by immunoblotting and the levels of CYP2D6 activity by measuring dextromethorphan O-desmethylation activity. The results showed that CYP2D6 activity varied with CYP2D6 genotype, although the range of activity within the same genotype groups was less wide than that between the different genotype groups. We also determined the presence of the HNF4A G36S and G60D variations in the same set of livers using duplex pyrosequencing. The pyrosequencing-based HNF4A genotyping method was validated with 50 DNA samples, which included at least three subjects who were heterozygous for each variant, by blind genotype calling by the third person. From the HNF4A genotyping, seven livers were found to be heterozygous for G36S and three livers were heterozygous for G60D. To assess the effects of HNF4A variants on CYP2D6 activity in the livers of the same CYP2D6 genotypes, the CYP2D6 activities of seven HNF4A G36S livers were compared with those of wild-type HNF4A livers in the same background of the CYP2D6 genotype. One liver with HNF4A G36S and the CYP2D6*1/*1 genotype showed the highest CYP2D6 activity among the CYP2D6 normal/normal livers, and two livers with HNF4A G36S and the CYP2D6 normal/null genotype showed relatively low CYP2D6 activities (Fig. 4A). Four HNF4A G36S livers with the CYP2D6 genotype of *10B/normal or *10B/*10B showed no evidence of altered CYP2D6 activity compared with other livers with the same CYP2D6 genotypes. Two HNF4A G60D livers showed the lowest CYP2D6 activity within the CYP2D6 normal/*10B livers. CYP2D6 activities of HNF4A G60D livers (12.5 and 60.5 nmol/minute/mg) were below the lower limit of the 99% confidence interval for all data of this group (136.9–263.3 nmol/minute/mg). The data within the CYP2D6 normal/*10B livers did not show normal distribution. When we performed the Mann-Whitney U test, the HNF4A G60D livers and the others showed a significant difference between the two groups (P = 0.0338). Because HNF4A G60D showed loss of transcriptional activity, the effect of this variant on CYP2D6 expression was also examined in the human livers. The CYP2D6 protein levels were determined by immunoblotting. Pearson's correlations test showed a good correlation between the amounts of CYP2D6 protein and the CYP2D6 activities (r = 0.806, P < 0.0001, n = 43; data not shown). However, the expression pattern was more heterogeneous in relation to the CYP2D6 genotype than in relation to the CYP2D6 activity profile. The CYP2D6 expression levels of the HNF4A variant genotypes were compared with that of the HNF4A WT. Similar to the results obtained for CYP2D6 activity, the HNF4A G36S livers showed no consistent tendency for altered CYP2D6 protein levels compared with the WT HNF4A livers (Fig. 4B). The HNF4A G60D livers ranked first and fourth in terms of the lowest amounts of CYP2D6 protein in the CYP2D6 normal/*10B livers, and ranked in the middle in terms of the lowest amounts of CYP2D6 in the CYP2D6 normal/null livers.
CYP2D6 Activity in Human Subjects Heterozygous for HNF4A G60D.
The results of the assays of liver CYP2D6 activities in relation to HNF4A genotype suggested the possibility of in vivo functionality of this variant in human subjects. To assess the functional consequences of HNF4A variants in human subjects, a clinical trial was carried out in the subjects who were genotyped for both the CYP2D6 and HNF4A genes. Healthy volunteers (n = 130) were recruited for the study. All of these subjects were genotyped either by direct sequencing of full-length CYP2D6 genes or by the multiplex SNaPshot method. The CYP2D6 alleles found in the 130 Korean subjects included CYP2D6*2, *2N, *5, *10B, *14, *18, *21, *41, *49, and *52. CYP2D6*1 and *2 were grouped as ”normal” alleles and CYP2D6*5, *14, *18, and *21 were grouped as ”null” alleles. All of the volunteers were also genotyped for the HNF4A G36S and G60D variants. Among the 130 subjects, six heterozygous HNF4A G36S, one homozygous G36S, and three G60D subjects were identified. The HNF4A G36S subjects had the CYP2D6 genotype of *1/*2N, *10B/*10B, *41/*41, *14/*41, *10B/*21, *5/*41, or *2/*2, and the three HNF4A G60D subjects had the CYP2D6 genotype of *1/*1, *10B/*21, or *10B/*14. We also measured in vivo CYP2D6 activity in all subjects with HNF4A variant genotypes, as well as subjects with WT HNF4A genotypes with the same genetic background of CYP2D6 as the HNF4A variant genotypes, to assess the effects of HNF4A G36S and G60D using dextromethorphan as a probe drug. The HNF4A G36S subjects showed no trend of differential CYP2D6 activity among those with the same CYP2D6 genotype (data not shown). In contrast, all three HNF4A G60D subjects showed relatively low-level CYP2D6 activity within the same CYP2D6 genotype group (Fig. 5). The lowest CYP2D6 activity was detected in a HNF4A G60D subject in the CYP2D6 normal/normal genotype group. Similarly, relatively low-level CYP2D6 activity was found in two of the HNF4A G60D subjects in the CYP2D6 *10B/null genotype group. Because of the limited number of subjects with the HNF4A G60D variant, we were unable to test the statistical significance of these outcomes.
It is well known that HNF4A plays an essential role in the regulation of the hepatic expression of drug-metabolizing enzymes and drug transporters.4–7 Kamiyama et al.21 have shown that down-regulation of HNF4A expression in human hepatocytes decreases the messenger RNA levels of CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP3A4, UGT1A1, UGT1A9, SULT2A1, ABCB1, ABCB11, ABCC2, OATP1B1, and OCT1.21 In addition, conditional knockout of HNF4A in mice significantly decreases CYP2D6 activity.11 These results raise the possibility that functional genetic variants of HNF4A affect the fate of drugs in humans by regulating the levels of drug disposition–related proteins. However, almost all of the genetic studies conducted on HNF4A polymorphisms have focused on the correlation between genetic variations and diabetes. From the findings that the functional HNF4A variants are found in human subjects and that the loss of HNF4A gene in mice decreases in vivo CYP2D6 activity, we hypothesized that the functional genetic variants of HNF4A in human may affect in vivo CYP2D6 activity. To test this hypothesis, we identified the loss of function mutation in human HNF4A gene and tested its contribution to human CYP2D6 activity.
In the current study, we sequenced all the exons and the promoter region of the HNF4A gene in healthy Korean subjects, thereby identifying 21 genetic variations, including the nonsynonymous G36S and G60D polymorphisms. In the promoter region, we found seven SNPs and one del mutation at frequencies between 1% and 56%. Among these SNPs, four promoter variations were commonly observed at frequency greater than 19%. LD analysis of the common promoter variations revealed that −1048delGGG was in strong LD with −2003G>A and −1650A>G, and in moderate LD with −755A>C. Other than this, there is no apparent LD between promoter SNPs and coding region SNPs. The absence of gene-wide LD block (or haplotype block) suggests that the effect of genetic variants on HNF4A function or expression should be tested separately for each variant, but not for the haplotypes. The regulatory function of the promoter SNPs in the HNF4A expression remains to be studied.
The functionalities of the two nonsynonymous HNF4A variants were tested by assaying transactivation activity and DNA-binding properties. Because the CYP2D6 promoter is strongly activated by functional HNF4A in cultured cells,8 the functionalities of HNF4A variants were tested in CYP2D6 promoter-transfected cells. The G36S SNP increased CYP2D6 promoter activity to an extent that was similar to that of the wild-type HNF4A. This SNP bound to the HNF4A binding site in the CYP2D6 promoter as strongly as wild-type HNF4A, as revealed by EMSA. Because G36S also bound to consensus HNF4A binding sites, this variant is expected to have no effect on the expression of other HNF4A-regulated genes. In contrast, the HNF4A G60D SNP did not increase CYP2D6 promoter activity in the transfected cells. Furthermore, the loss of transactivation activity of this SNP in the transfected cells was not attributable to the loss of HNF4A expression. Instead, the G60D SNP did not bind to the HNF4A binding site in the CYP2D6 promoter, as revealed by EMSA. The HNF4A G60D SNP is located within the DNA binding domain, which is crucial for the transcriptional activity of HNF4A. Interestingly, the aa sequences surrounding the G60D SNP are highly conserved across various species, such as human, rat, mouse, frog, and nematode. This conservation of the G60 residue implies a crucial role for this residue in the function of HNF4A. In contrast to the G60D SNP, the G36S SNP is not conserved in other species. In the frog, this residue is valine, and the residue next to G36 is altered from threonine in the human to alanine in the rat and mouse. Therefore, the G60D amino acid conversion may cause structural changes that prevent the HNF4A variant protein from binding to its recognition sites for transcriptional activation.
The effect of the HNF4A G60D variant on hepatic CYP2D6 expression and activity was tested in human liver microsomes. Although genetic polymorphism of the CYP2D6 gene is common, the frequency of HNF4A G60D is low. Thus, it was not possible to obtain a sufficient number of samples for a statistical analysis of the differences in CYP2D6 function between HNF4A G60D livers and wild-type livers. Nevertheless, livers of the HNF4A G60D genotype showed a tendency for reduced CYP2D6 function, regardless of the CYP2D6 genotypes of the livers. To investigate further whether the low CYP2D6 activity is associated with the G60D genotype, we recruited volunteers for an in vivo phenotyping study. Similar to the results obtained for the human livers, subjects with the HNF4A G60D SNP showed a tendency for lower CYP2D6 metabolic activity, as measured by dextromethorphan metabolism. In contrast to the HNF4A G60D SNP, the G36S variant showed no consistent tendency for altered CYP2D6 function, either in human livers or in human subjects. These results were to some extent expected from the results of in vitro studies performed on the variants. Further clinical studies with a larger population of subjects are needed to test whether there is a significant association between G60D SNP and the CYP2D6 activity.
Standard genetic knockout of HNF4A results in embryonic lethality in experimental mice.22 Studies with conditional knockout mice have revealed that HNF4A plays a crucial role in various developmental processes, such as liver morphogenesis, as well as in lipid homeostasis.23, 24 These findings imply that the homozygous genotype for null alleles of HNF4A should not be observed in humans. In our in vitro experiments, HNF4A G60D was characterized as a null allelic variant with complete loss of its regulatory function. Based on the HWE, 1 in 6000 Korean subjects is expected to have the homozygous G60D genotype. We also evaluated ethnic variations in the frequencies of the two HNF4A variants. HNF4A G36S, which is the most common HNF4A coding variant, with a frequency of 3.8% in Koreans, was also found in other ethnic groups, such as Chinese, Vietnamese, and Caucasians (frequencies of 1% to 3.6%), but not in African Americans. The functional G60D variant was found only in Asians, such as Koreans, Chinese, and Vietnamese, at frequencies of 0.5% to 1.3%. Despite the limited size of each study population, our results show that the G60D SNP is a rare functional variant, at least in several Asian populations. Previous studies of Caucasian populations have identified several nonsynonymous HNF4A SNPs that are different from the SNPs described here. For example, the T130I variant has been identified in 1 of 100 healthy Caucasian subjects,25 and at a frequency of 3.4% in 4726 Danish subjects.26 The V255M variant has been found in 10% of the same Danish population. These variants decrease, more or less, the transcriptional activity of HNF4A in vitro. In our initial screening of HNF4A variants in 50 Korean subjects, the V255M variant was not detected. Together with our data, these results demonstrate the ethnic variability of common HNF4A genetic polymorphisms. Considering that the in vitro functional G60D variant of HNF4A may influence CYP2D6 function in human livers or human subjects, the functional consequences in vivo of V255M or T130I in Caucasian populations needs to be tested. Furthermore, because HNF4A is involved in the regulation of many drug disposition–related proteins, the effects of nonsynonymous variants or functional promoter variants on protein function may help to explain the interindividual variability of drug disposition. Our results suggest that HNF4A G60D SNP genotyping may provide additional information for CYP2D6 pharmacogenetics and for prediction of the CYP2D6 activity, at the level of the individual, at least in Asian populations. Further studies are required to reveal the clinical relevance of the pharmacogenetic aspects of HNF4A variants.