The COUP-TFII (chicken ovalbumin upstream promoter-transcription factor II) nuclear receptor, which is composed of a DNA-binding domain and a ligand-binding domain, exerts pleiotropic effects on development and cell differentiation by regulating the transcription of its target genes, including Cyp7a1 (cytochrome P450, family 7, subfamily a, polypeptide 1), which plays important roles in catabolism of cholesterol in the liver. Although multiple variants of COUP-TFII exist, their roles in the regulation of Cyp7a1 expression have not been elucidated. In the present study, we investigated the roles of COUP-TFII-V2 (variant 2), which lacks a DNA-binding domain, in the regulation of the transcriptional control of the Cyp7a1 gene by COUP-TFII in hepatocellular carcinoma cells. We found that COUP-TFII-V2 was significantly expressed in Huh7 cells, in which Cyp7a1 was not expressed. Furthermore, knockdown of COUP-TFII-V2 enhanced endogenous Cyp7a1 expression in Huh7 cells. Although COUP-TFII activates the Cyp7a1 promoter through direct binding to DNA, this activation was affected by COUP-TFII-V2, which physically interacted with COUP-TFII and inhibited its DNA-binding ability. Chromatin immunoprecipitation assays showed that COUP-TFII-V2 inhibited the binding of endogenous COUP-TFII to the intact Cyp7a1 promoter. The results of the present study suggest that COUP-TFII-V2 negatively regulates the function of COUP-TFII by inhibiting its binding to DNA to decrease Cyp7a1 expression.
Cyp7a1 (cytochrome P450, family 7, subfamily a, polypeptide 1; also known as cholesterol 7α-hydroxylase) catalyses the first and rate-limiting step in the catabolism of cholesterol to bile acid in the liver . Previous reports have indicated that the expression of Cyp7a1 is tightly regulated in order to eliminate cholesterol from the body and maintain cholesterol homoeostasis. Cyp7a1 expression is regulated by several nuclear receptors, including LXRα (liver X receptor-α) [2,3], LRH-1 (liver receptor homologue 1) , HNF4α (hepatocyte nuclear factor 4α) , RXR (retinoid X receptor) and COUP-TFII (chicken ovalbumin upstream promoter-transcription factor II) .
COUP-TFII [also known as ARP-1 (apolipoprotein A-I regulatory protein 1) or NR2F2 (nuclear receptor subfamily 2, group F, member 2)], which belongs to the steroid/TR (thyroid hormone receptor) superfamily, is an orphan nuclear receptor [7,8]. COUP-TFII has been implicated in many biological processes, such as angiogenesis [9,10], lymphangiogenesis [11–13], organogenesis [14–16], cell fate determination  and metabolic homoeostasis .
In addition, COUP-TFII has been implicated in the induction of Cyp7a1 expression in liver through direct binding to the DR (direct repeat) sequences (AGGTCA) in the Cyp7a1 promoter [6,19]. Following DNA binding, COUP-TFII forms a homodimer  and/or heterodimers by interacting with other nuclear receptors such as RXR , HNF4α  and GRs (glucocorticoid receptors) .
COUP-TFII has the typical structure of a nuclear receptor (Figure 1A), and contains an N-terminal-ligand-independent activation domain [AF-1 (activation function 1); amino acids 1–79], a highly conserved DBD (DNA-binding domain; amino acids 79–144) and a LBD (ligand-binding domain; amino acids 195–414). The DBD and LBD are separated by a hinge region. The LBD contains a ligand-dependent AF-2 (amino acids 394–401) . The LBD of nuclear receptors is important for their functions, including receptor dimerization, ligand recognition and ligand-dependent activation. The conformational states of AF-2 determine the nuclear receptor activity.
COUP-TFII-V2 regulates Cyp7a1 expression in HepG2 and Huh7 hepatocarcinoma cell lines
A previous report revealed that the COUP-TFII gene produces at least five isoforms . Two of them (COUP-TFII-001 and COUP-TFII-201; Genbank® number NM_021005) encode full-length COUP-TFII (hereafter referred to as COUP-TFII or COUP-TFII-V1), whereas the others (COUP-TFII-203, COUP-TFII-202 and COUP-TFII-204; Genbank® numbers NM_001145155, NM_001145156 and NM_001145157 respectively) encode a shorter coding sequence (hereafter COUP-TFII-203, COUP-TFII-202 and COUP-TFII-204 are referred to as COUP-TFII-V2, COUP-TFII-V3 and COUP-TFII-V4 respectively), resulting in the loss of AF-1 and DBD. On the other hand, although the coding sequences of COUP-TFII-V3 and COUP-TFII-V4 are completely identical (Supplementary Figure 1A at http://www.biochemj.org/bj/452/bj4520345add.htm), the nucleotide sequences of their untranslated regions are different. Among these isoforms, only COUP-TFII-V2 has been reported in mice. Although COUP-TFII-V2 contains the LBD, which includes the AF-2 domain, it has a truncated hinge region and a COUP-TFII-V2-specific N-terminal sequence (Figure 1A). The transcriptional activities of COUP-TFII-V2 and its physiological roles in the regulation of Cyp7a1 expression have not yet been elucidated.
In the present study, we showed that COUP-TFII-V2 interacted with COUP-TFII-V1 and inhibited its binding to DNA, leading to the down-regulation of the COUP-TFII-V1-mediated transcriptional activation of Cyp7a1. Furthermore, COUP-TFII-V2 acts as an endogenous repressor of Cyp7a1 in hepatocellular carcinoma cells. The results of the present study suggest that the transcriptional activity of COUP-TFII is negatively regulated by its variant through physical interactions.
Huh7, HepG2 human hepatocarcinoma, HEK (human embryonic kidney)-293T and HaCaT human keratinocyte cells were obtained from A.T.C.C. (Manassas, VA, U.S.A). Huh7, HEK-293T and HaCaT cells were maintained in DMEM (Dulbecco's modified Eagle's medium) (Life Technologies) containing 10% FBS (fetal bovine serum), 50 units/ml streptomycin and 50 units/ml penicillin (Life Technologies). For HepG2 cells, non-essential amino acids (Life Technologies) and sodium pyruvate were also supplemented.
cDNA encoding COUP-TFII-V2, obtained by PCR cloning, was inserted into the pcDEF vector, and deletion mutants and point mutants of COUP-TFII were generated by PCR and subcloned into pcDEF vectors. All of the PCR products were sequenced before use. The sequences used for mutagenesis are shown in Supplementary Table S1 (at http://www.biochemj.org/bj/452/bj4520345add.htm).
Western blot and IP (immunoprecipitation) assays
Western blot analyses were performed as described previously  with antibodies against rabbit polyclonal COUP-TFII-V1 (ab64849; Abcam), mouse monoclonal COUP-TFII-V2 (generated at the Research Center for Advanced Science and Technology, The University of Tokyo, in accordance with the policies of the Animal Ethics Committee of the University of Tokyo), HA (haemagglutinin) epitope (3F10; Roche Applied Science), GFP (green fluorescent protein) (048-3; MBL International), HDAC (histone deacetylase) 1 (2E-10; Merck Millipore) or mouse monoclonal α-tubulin (Sigma–Aldrich). Nuclear and cytoplasm fractions of HaCaT cells were obtained using NE-PER Nuclear and Cytoplasmic Extraction Reagents according to the manufacturer's instructions (Thermo Scientific). For the co-IP assays, HEK-293T cells were co-transfected with FLAG-COUP-TFII-V2 and 6Myc-COUP-TFII-V1 expression plasmids using FuGENE® 6 reagent (Roche Applied Science). The cells were lysed 24 h later, and extracts were incubated with an anti-FLAG antibody (Sigma–Aldrich) for 1 h at 4°C. The resulting mixture was incubated with Protein G–Sepharose (GE Healthcare) for 30 min, washed three times with Nonidet P40 lysis buffer, and subjected to SDS/PAGE (12% gel).
Isolation of RNA and quantitative RT (reverse transcription)–PCR
Total RNA was extracted from cells using the RNeasy mini kit (Qiagen). First-strand cDNAs were synthesized by a PrimeScript RT–PCR kit (Takara Bio) using oligo(dT) primers according to the manufacturer's protocol. Quantitative RT–PCR was conducted as described previously . To quantify the relative levels of mRNA expression between COUP-TFII-V1 and COUP-TFII-V2, a titration curve was generated with the human COUP-TFII-V1 and COUP-TFII-V2 plasmids (gifts from Dr Toshiya Tanaka, The University of Tokyo, Tokyo, Japan). The diluted concentration curve was used to calculate the amount of COUP-TFII-V1 and COUP-TFII-V2 in Huh7 and HepG2 cell lines, as described previously . The primer sequences used are listed in Supplementary Table S1.
Immunostaining and PLA (proximity ligation assay)
Cells were fixed in 50% acetone/methanol for 10 min, washed three times with PBS and incubated with an anti-FLAG antibody diluted in Blocking One solution (Nacalai Tesque) overnight at 4°C. Subsequently, cells were incubated with Alexa Fluor® 488-conjugated secondary antibody (Life Technologies) for 1 h at room temperature (25°C) and stained with TOTO®-3 (Life Technologies) for 10 min at room temperature. The Duolink® in situ PLA kits were purchased from Olink Bioscience. Cells were fixed as described above and incubated with anti-rabbit polyclonal COUP-TFII-V1 (ab64849; Abcam) and anti-mouse monoclonal COUP-TFII-V2 primary antibodies as listed above, which were diluted in Blocking One solution overnight at 4°C. Subsequently, a pair of secondary antibodies that were conjugated to oligonucleotides (PLA probes) were used according to the manufacturer's instructions. For nuclear staining, cells were incubated with TOTO®-3. Stained specimens were analysed with an LSM 510 META confocal microscope (Carl Zeiss). The number of PLA signals from five images was quantified with the Duolink Image Tool (Olink Bioscience).
Generation and infection of lentivirus
HA-tagged mouse COUP-TFII-V2 was inserted into pCSII-EF-RfA using pENTR according to the manufacturer's instructions (Life Technologies). The generation of lentiviral vectors was performed as described previously . HepG2 cells were infected with the lentivirus twice. HepG2 cells expressing GFP were used as a control.
siV1 [siRNA (small interfering RNA) against human COUPTFII-V1], siV2 (siRNA against COUP-TFII-V2) and siV1V2 (siRNA against both COUP-TFII-V1 and COUP-TFII-V2) were synthesized by Life Technologies. The target sequences of siRNA duplexes were as follows: siV1, 5′-CAGUUCCGGUUGGCGCGGCACGUGU-3′; siV2, 5′-UGCAAGCGGUUUGGGACCUUGAACA-3′; and siV1V2, 5′-AACAAUUGCUCUAUGACUGAGGAGG-3′. siNTC (negative control siRNA) was obtained from Life Technologies. siRNAs were transfected with Lipofectamine™ RNAiMax reagent (Life Technologies) according to the manufacturer's instructions and the cells were harvested 48 h later.
The rat Cyp7a1 (−416/+32) luciferase reporter plasmid was a gift from Dr John Y.L. Chiang (Northeast Ohio Medical University, OH, U.S.A.). HepG2 cells were grown in duplicate in 24-well plates to 50% confluence and transfected with FuGENE® HD (Roche Applied Science). After 48 h, luciferase activities were measured by a Dual-Luciferase Assay kit (Promega). Luciferase activities were normalized to the Renilla luciferase activity of the co-transfected phRL-CMV plasmid. The lysates were subjected to Western blotting.
EMSA (electrophoretic mobility-shift assay)
EMSAs were performed as described previously . Recombinant COUP-TFII-V1 and COUP-TFII-V2 were generated using the TNT®-coupled reticulocyte lysate systems (Promega) with pBS-SK-COUP-TFII-V1 and pBS-SK-COUP-TFII-V2 according to the manufacturer's instructions. Nuclear extracts from HepG2 cells were prepared as described previously  and dialysed using a Slide-A-Lyzer G2 Dialysis Cassette (Thermo Scientific). To reduce non-specific binding, 1 μg of poly(dI/dC) (Life Technologies) was added to the reaction mixture. Double-stranded oligonucleotides were labelled with [α-32P]dCTP and Klenow fragments and purified by a spin column (GE Healthcare). For the supershift assay, recombinant proteins were preincubated with an anti-COUP-TFII-V1 antibody (Perseus Proteomics), the anti-COUP-TFII-V2 antibody listed above or anti-(mouse IgG) antibody (Santa Cruz Biotechnology) for 30 min at room temperature. In competition experiments, recombinant proteins were preincubated with 100-fold molar excess of unlabelled wild-type or mutant oligonucleotides for 20 min at room temperature and then added to the reaction mixture. The oligonucleotide sequences that were used as probes and competitors are shown in Supplementary Table S1.
ChIP (chromatin IP) assay
HepG2 cells were seeded into 15-cm-diameter dishes and transfected with FuGENE® HD according to the manufacturer's instructions. At 48 h later, cells were fixed in 1% formaldehyde for 10 min, and 2.5 M glycine was added to result in a final concentration of 0.125 M. Then, cells were washed twice with ice-cold PBS and collected. The cell pellets were suspended in SDS lysis buffer containing P8340 protease inhibitor (Sigma–Aldrich). The cross-linked chromatin was subjected to sonication using BioRuptor (Cosmo Bio) to obtain DNA fragments. The output was high, and the cycle of 30 s on and 30 s off was repeated three times. The chromatin complexes were collected by centrifugation at 19630 g at 8°C and diluted 10-fold in ChIP dilution buffer. The samples were immunoprecipitated with 10 μg of anti-COUP-TFII-V1 antibody (Perseus Proteomics), anti-HNF4α antibody (C-19, Santa Cruz Biotechnology) or anti-(mouse IgG) antibody and 150 μl of Dynabeads M-280 sheep anti-(mouse IgG) antibody (Life Technologies). Precipitated chromatin complexes were washed and eluted with a PCR purification kit (Qiagen). The primer sequences are listed in Supplementary Table S1.
Results were statistically examined using two-sided Student's t tests. Differences were considered significant at P<0.05.
Expression of a COUP-TFII variant lacking the DBD is negatively correlated with that of Cyp7a1 in hepatocellular carcinoma cell lines
COUP-TFII-V1 (full-length COUP-TFII) contains the highly conserved N-terminal DBD, a hinge region and a C-terminal LBD, whereas Rosa and Brivanlou  reported that COUP-TFII-V2 lacks the DBD (Figure 1A). As COUP-TFII has been implicated in the transcriptional control of Cyp7a1 expression in the liver through its direct interaction with the Cyp7a1 promoter [6,31], we first examined whether COUP-TFII-V2 transcripts are endogenously expressed in two hepatocellular carcinoma cell lines, Huh7 and HepG2. We compared the expression levels of COUP-TFII-V1, COUP-TFII-V2 and Cyp7a1 in HepG2 and Huh7 cells. COUP-TFII-V1 was expressed both in HepG2 and Huh7 cells (Figure 1B). In contrast, COUP-TFII-V2 was highly expressed in Huh7 cells, but not in HepG2 cells (Figure 1C). To our interest, endogenous Cyp7a1 expression was much lower in Huh7 cells than in HepG2 cells (Figure 1D), suggesting that COUP-TFII-V2 expression is inversely correlated with Cyp7a1 expression in these cell lines. In order to examine whether the expression levels of COUP-TFII-V2 are comparable with that of COUP-TFII-V1, we generated titration curves using the plasmids encoding COUP-TFII-V1 and COUP-TFII-V2 and performed quantitative RT–PCR analyses using specific primers for COUP-TFII-V1 and COUP-TFII-V2. Although the expression levels of COUP-TFII-V2 were significantly lower than those of COUP-TFII-V1 in HepG2 cells, COUP-TFII-V1 and COUP-TFII-V2 were equivalently expressed in Huh7 cells (Figure 1E).
In accordance with previous reports, when we knocked down the expression of COUP-TFII-V1 in HepG2 cells, the expression of Cyp7a1 was decreased (Supplementary Figure S2 at http://www.biochemj.org/bj/452/bj4520345add.htm), suggesting that endogenous Cyp7a1 expression is maintained by COUP-TFII-V1. Interestingly, when COUP-TFII-V1 expression was silenced, COUP-TFII-V2 expression was elevated (Supplementary Figure S2). Therefore we next examined whether endogenous Cyp7a1 expression is regulated by COUP-TFII-V2 in multiple types of hapatocellular carcinoma cells. As shown in Figures 1(B)–1(E), the expression of Cyp7a1 in HepG2 cells was higher than that in Huh7 cells whereas the expression of COUP-TFII-V2 in HepG2 cells was lower than that in Huh7 cells. In order to examine the causal relationship between the expression of Cyp7a1 and COUP-TFII-V2 in HepG2 cells, we established HepG2 cells that stably express control GFP or COUP-TFII-V2 proteins by lentiviral infection (Figure 1F). As shown in Figure 1G, the endogenous Cyp7a1 expression level in COUP-TFII-V2-expressing cells was lower than that of control cells. To further examine the physiological relevance of endogenous COUP-TFII-V2 in the regulation of Cyp7a1 expression, we used siV2, which decreased the COUP-TFII-V2 expression by approximately 90% without significant alteration of COUP-TFII-V1 expression in Huh7 cells (Figures 1H and 1I). The expression of Cyp7a1 was significantly elevated in Huh7 cells when the expression of COUP-TFII-V2 was decreased (Figure 1J). The results suggest that COUP-TFII-V2 suppresses the endogenous Cyp7a1 expression in hepatocellular carcinoma cells.
COUP-TFII-V2 inhibits the Cyp7a1 promoter activity induced by COUP-TFII-V1
The results of the present study showing that the expression levels of COUP-TFII-V2 and Cyp7a1 are inversely correlated prompted us to investigate the roles of COUP-TFII-V2 in the regulation of the Cyp7a1 promoter activity. We conducted luciferase assays using HepG2 cells because they do not express a high level of COUP-TFII-V2. As shown in Figure 2, COUP-TFII-V1 activated the rat Cyp7a1 promoter in HepG2 cells. In contrast, COUP-TFII-V2 alone did not affect the Cyp7a1 promoter activity. However, when COUP-TFII-V2 was co-expressed with COUP-TFII-V1, it repressed the Cyp7a1 promoter activity that was stimulated by COUP-TFII-V1 in a dose-dependent manner (Figure 2). The results suggest that COUP-TFII-V2 inhibits the Cyp7a1 promoter activity that is induced by COUP-TFII-V1.
COUP-TFII-V2 inhibits Cyp7a1 promoter activity induced by COUP-TFII-V1
COUP-TFII-V2 interferes with the DNA binding of COUP-TFII-V1
In order to study the molecular mechanisms underlying the inhibition of COUP-TFII-induced Cyp7a1 promoter activity by COUP-TFII-V2, we performed EMSA using DR1 (DR sequences separated by one nucleotide)  as a probe. Incubation of the 32P-labelled DR1 probe with recombinant COUP-TFII-V1 proteins that were generated by the reticulocyte lysate systems resulted in the appearance of specific DNA–protein complexes (Figure 3A, lane 3). The specificity of these DNA–protein complexes was confirmed by competition assays (Figure 3A, lanes 4 and 5) and supershift assays (Figure 3A, lanes 6 and 7), which showed that COUP-TFII-V1 bound to the DR1 sequence. In contrast, incubation of the 32P-labelled DR1 probe with comparable levels of recombinant COUP-TFII-V2 proteins did not result in the appearance of DNA–protein complexes (Figure 3A, lane 10), suggesting that COUP-TFII-V2 does not bind to the DNA sequences that are bound by COUP-TFII-V1.
COUP-TFII-V2 does not have DNA binding ability and interferes with the DNA binding of COUP-TFII-V1
We then examined whether COUP-TFII-V2 may affect the interactions of COUP-TFII-V1 with its target sequences. When 32P-labelled DR1 probe was incubated with COUP-TFII-V1 proteins in combination with increasing amounts of COUP-TFII-V2 proteins, the amount of DNA–protein complexes consisting of COUP-TFII-V1 was decreased by the addition of COUP-TFII-V2 in a dose-dependent manner (Figure 3B). In order to confirm the results of EMSAs using in vitro translated recombinant proteins, we further performed EMSAs using nuclear extracts from HepG2 cells. As reported previously, we confirmed that endogenous COUP-TFII-V1 interacts with the DR1 probe in HepG2 cells, and the specificity of the complexes was checked by competition assays (Figure 3C, lanes 3 and 4) and supershift assays (Figure 3C, lanes 5, 6 and 7). The nuclear extracts from HepG2 cells expressing exogenous COUP-TFII-V2 yielded weaker binding of COUP-TFII-V1 to the DR1 probe (Figure 3C, lane 8, and Figure 3D), which is consistent with the results of EMSAs using in vitro translated recombinant COUP-TFII-V1 and COUP-TFII-V2 proteins. Together, with the results of the luciferase assay, these results suggest that COUP-TFII-V2 inhibits COUP-TFII-V1-induced activation of the Cyp7a1 promoter by inhibiting the binding of COUP-TFII-V1 to DR sequences in the Cyp7a1 promoter.
Endogenous COUP-TFII-V2 physically interacts with COUP-TFII-V1
Nuclear receptors, including COUP-TFII, dimerize to bind DNA . In order to elucidate further the molecular mechanisms of how COUP-TFII-V2 inhibits the COUP-TFII-V1-mediated activation of the Cyp7a1 promoter, we tested the hypothesis that these two variants form heterodimers. When we conducted IP assays using the lysates prepared from HEK-293T cells that were transfected with COUP-TFII-V1 and COUP-TFII-V2, we found that COUP-TFII-V2 interacted with COUP-TFII-V1 to form a heterodimer as well as with COUP-TFII-V2 to form a homodimer (Figure 4A).
Endogenous COUP-TFII-V2 interacts with COUP-TFII-V1 proteins
To further confirm the results of the co-IP assays, we examined the physical interactions of endogenous COUP-TFII-V1 and COUP-TFII-V2 proteins using in situ PLA in HaCaT cells, since endogenous COUP-TFII-V1 and COUP-TFII-V2 are highly expressed in HaCaT cells at mRNA and protein levels (Supplementary Figures S1 and S3 respectively at http://www.biochemj.org/bj/452/bj4520345add.htm). We first examined whether the antibodies against COUP-TFII-V1 and COUP-TFII-V2, which were used for PLA, can specifically recognize them using the lysates prepared from HaCaT cells that were transfected with siV1, siV2 and siV1V2. The specificities of the antibodies to COUP-TFII-V1 and COUP-TFII-V2 were confirmed by Western blot analysis showing that the signals detected by the antibodies were specifically decreased by the corresponding siRNAs (Figures 4B and 4C).
COUP-TFII-V3 and COUP-TFII-V4 are also abundantly expressed in HaCaT cells (Supplementary Figure S1). In order to assess whether the anti-COUP-TFII-V2 antibody does not react with COUP-TFII-V3 and COUP-TFII-V4 proteins, whose sizes are close to that of COUP-TFII-V2, we studied the expression of COUP-TFII-V3 and COUP-TFII-V4 in HaCaT cells transfected with various siRNAs (Supplementary Figure S4 at http://www.biochemj.org/bj/452/bj4520345add.htm). Since siV1V2 is designed for the common sequence for all four variants, the expression of COUP-TFII-V3 and COUP-TFII-V4 was decreased only by siV1V2, whereas siV2 did not alter their expression (Supplementary Figure S4). However, as shown in Figure 4C, the band detected by the anti-COUP-TFII-V2 antibody disappeared completely by silencing only COUP-TFII-V2 expression alone, suggesting that the antibody against COUP-TFII-V2 specifically detects COUP-TFII-V2.
Using the antibodies examined above, we carried out in situ PLAs to detect physical interaction between endogenous COUP-TFII-V1 and COUP-TFII-V2 proteins in HaCaT cells. In HaCaT cells transfected with control siRNA, we detected a number of definite fluorescent signals that were restricted to the nuclei, not the cytoplasm, which indicates the endogenous interactions between COUP-TFII-V1 and COUP-TFII-V2 proteins were mainly in the nuclei (Figure 4D). When we decreased the COUP-TFII-V1 expression by specific siRNA, the number of signals that were detected in the nuclei was significantly decreased (Figure 4D). These results suggest that endogenous COUP-TFII-V1 and COUP-TFII-V2 physically interact with each other.
Hinge and LBD regions within the COUP-TFII-V1 is required for interaction with COUP-TFII-V2
In order to determine which regions of COUP-TFII-V1 play important roles in its interaction with COUP-TFII-V2, we generated a series of deletion mutants of COUP-TFII-V1 (Figure 5A) and performed IP analyses using HEK-293T cells (Figure 5B). Although a deletion mutant lacking the C-terminal AF-2 domain (COUP-TFII-V1; amino acids 1–393) interacted with COUP-TFII-V2, deletion of the C-terminal regions containing the LBD (COUP-TFII-V1; amino acids 1–195) and the hinge (COUP-TFII-V1; amino acids 1–144) completely abolished their interaction.
COUP-TFII-V1 interacts with COUP-TFII-V2 through its hinge and LBD
In order to examine whether the COUP-TFII-V1 mutants display differential intracellular localization, we performed immunostaining of HaCaT cells transfected with the expression plasmids encoding FLAG-tagged COUP-TFII variants. We found that COUP-TFII-V1 was localized to the nuclei, whereas COUP-TFII-V2 was present in both the nuclei and cytoplasm of HaCaT cells (Figure 5C and 5F). We confirmed that endogenous COUP-TFII-V2 was also localized in the nuclei and cytoplasm by performing nuclear and cytoplasm fractionation (Figure 5G). We also found that the deletion mutants (COUP-TFII-V1; amino acids 1–144 and COUP-TFII-V1; amino acids 144–414) were localized to the nuclei and cytoplasm (Figures 5D and 5E), suggesting that loss of interaction between COUP-TFII-V1 (amino acids 1–144) and COUP-TFII-V2 is not caused by the alteration of intracellular localization. A deletion mutant lacking an N-terminal region containing AF-1 and DBD (COUP-TFII-V1; amino acids 144–414) retained a weak binding ability with COUP-TFII-V2, whereas deletion of the hinge region (COUP-TFII-V1; amino acids 193–414) abolished their interaction. As the AF-2 domain plays important roles in transactivation and repression , we next examined whether the AF-2 domain was required for the interaction of the COUP-TFII-V1 (amino acids 144–414) mutant with COUP-TFII-V2. Deletion of the AF-2 domain (amino acids 144–393) failed to abolish its interaction with COUP-TFII-V2. These results suggest that the fragment consisting of the hinge region and the LBD (amino acids 144–393) is essential and sufficient for the interaction of COUP-TFII-V1 with COUPTFII-V2.
Functional dimerization mutants of COUP-TFII-V2 are not capable of inhibiting the transcriptional activity of COUP-TFII-V1
The LBD of COUP-TFII, which forms a symmetric dimer, is required to bind to transcriptional cofactors to regulate the transactivation of target genes. A detailed deletion analysis and observation of the crystal structure of the COUP-TFII protein revealed that two leucine residues (Leu364 and Leu365) in the helix α10 within the LBD were important interface residues that form critical hydrophobic interactions and are required to form a functional DNA-binding dimer . A COUP-TFII mutant in which these two residues were replaced with alanine (functional mutant, COUP-TFII-V1-mut) failed to activate the NGFI-A [nerve growth factor inducible gene A; also known as EGR1 (early growth response 1)] promoter .
In order to examine whether these leucine residues were important for the COUP-TFII-V2-mediated inhibition of the Cyp7a1 promoter activation by COUP-TFII-V1, we constructed a functional mutant of COUP-TFII-V2. Because Leu364 and Leu365 in COUP-TFII-V1 correspond to Leu231 and Leu232 in COUP-TFII-V2, we replaced them with alanine in order to make a functional COUP-TFII-V2 mutant (V2-mut) (Figure 6A) and performed a Cyp7a1 luciferase assay. We found that the functional V1 mutant activated the Cyp7a1 promoter at a lower level (Figure 6B, lane 3). Of note, the COUP-TFII-V2 mutant completely lost its inhibitory effects on the transactivation of COUP-TFII-V1 (Figure 6B, lane 5), suggesting that the leucine residues that are involved in the formation of a functional dimer also play important roles in the inhibition of COUP-TFII-mediated transcription by COUP-TFII-V2.
Functional dimerization mutants of COUP-TFII-V2 have no ability to suppress the transcriptional activity of COUP-TFII-V1
The AF-2 domain is also required for the transcriptional activity of COUP-TFII . In order to test its roles in COUP-TFII-V2-mediated transcriptional repression, we generated deletion mutants of COUP-TFII-V2 that lack nine C-terminal residues (COUP-TFII-V2; amino acids 1–273) or the AF-2 domain (COUP-TFII-V2; amino acids 1–260) (Figure 6A). Whereas COUP-TFII-V2 (amino acids 1–273), lacking the nine C-terminal residues, still interfered with the transcriptional ability of COUP-TFII-V1 (Figure 6C, lane 4), the deletion of the AF-2 domain abolished its inhibitory effects (Figure 6C, lane 5). The results suggest that the two leucine residues and the AF-2 domain play critical roles in the inhibitory effects of COUP-TFII-V2 on the transcriptional activity of COUP-TFII-V1.
Interaction of endogenous COUP-TFII-V1 with the intact Cyp7a1 promoter is inhibited by COUP-TFII-V2
Previous studies have shown that COUP-TFII-V1 enhances Cyp7a1 transcriptional activity through the DR-binding site and that COUP-TFII-V1 can directly bind to the Cyp7a1 promoter in rat hepatoma cells . In order to elucidate whether COUP-TFII-V2 represses COUP-TFII-V1-induced Cyp7a1 transcriptional activity by preventing COUP-TFII-V1 from interacting with the Cyp7a1 promoter in chromatin, cross-linked chromatin samples prepared from HepG2 cells were subjected to ChIP assays. As shown in Figures 7(A) and 7(B), the Cyp7a1 promoter region containing two DR4 sequences (DR sequences that were separated by four nucleotides), but not the control HBB (haemoglobin β) promoter, was pulled down with an anti-COUP-TFII-V1 antibody, but not with control IgG. When the COUP-TFII-V2 expression was increased by the transient transfection of FLAG-tagged COUP-TFII-V2, the interaction of COUP-TFII-V1 with the intact Cyp7a1 promoter was significantly attenuated. We further found that COUP-TFII-V2 did not bind to the intact Cyp7a1 promoter using COUP-TFII-V2 and anti-FLAG antibody for IP (Figure 7A).
COUP-TFII-V2 inhibits the interaction of endogenous COUP-TFII-V1 and HNF4α with the intact Cyp7a1 promoter
Previous reports showed that HNF4α directly interacts with the Cyp7a1 promoter in juxtaposition with COUP-TFII-V1 and synergistically activates the Cyp7a1 promoter with COUP-TFII-V1 , suggesting that HNF4α serves as a co-activator of COUP-TFII-V1. These findings prompted us to examine whether COUP-TFII-V2 represses COUP-TFII-V1-induced activation of the Cyp7a1 promoter by preventing HNF4α from interacting with the Cyp7a1 promoter. ChIP assays revealed that the binding of HNF4α to the Cyp7a1 promoter in HepG2 cells was inhibited when COUP-TFII-V2 was expressed (Figure 7C). The results of the present study suggest that COUP-TFII-V2 inhibits the transcriptional activation of the Cyp7a1 promoter by COUP-TFII-V1 through the sequestration of COUP-TFII-V1 and HNF4α proteins from the intact Cyp7a1 promoter.
In the present study, we showed that an isoform of COUP-TFII (COUP-TFII-V2), which lacks DNA-binding ability, exerts inhibitory effects on the COUP-TFII-induced transcription of the Cyp7a1 gene by blocking its DNA binding through physical interactions. We further revealed that endogenous COUP-TFII-V2 functions as a physiological repressor of Cyp7a1 in hepatocellular carcinoma cells.
We showed that Cyp7a1, which plays important roles in cholesterol catabolism, is abundantly expressed in HepG2, but not in Huh7 cells, and inversely correlated with the expression of COUP-TFII-V2. The findings of the present study raise the question of whether the COUP-TFII-V2 is actually involved in the regulation of cholesterol catabolism. ACAT2 (acyl-CoA:cholesterol acyltransferase 2) catalyses the synthesis of cholesteryl esters from free cholesterol and long-chain fatty acids. In addition to Cyp7a1, it has been reported that ACAT2 is expressed at higher levels in HepG2 than in Huh7 cells, and HepG2 cells contain more cholesterol than Huh7 . Furthermore, COUP-TFII acts not only as a transcription activator of Cyp7a1 expression in L35 rat hepatoma cells, but also as a transcriptional repressor of MTP (microsomal triglyceride transfer protein), which regulates lipoprotein secretion in FAO rat hepatoma cells . In that case, the dual role of COUP-TFII was mediated by RXRα. COUP-TFII is bound to the DR1 site in L35 cells, but not in FAO cells, whereas RXRα is bound to DR1 in FAO, but not in L35 cells, and COUP-TFII and a complex containing RXRα compete with each other for binding to the DR1 site. It is of interest to study whether COUP-TFII-V2 is involved in the transcriptional regulation of other players of cholesterol catabolism in the future.
Although COUP-TFII-V2 lacks a DBD, it is capable of binding to COUP-TFII-V1 and interferes with its transcriptional activity by sequestrating COUP-TFII-V1 proteins from promoters of target genes, suggesting that COUP-TFII-V2 functions as a dominant-negative mutant. A similar type of inhibition is conducted by Id HLH (helix–loop–helix) proteins [37–39]. They heterodimerize with members of the basic HLH family transcription factors. Whereas basic HLH transcription factors, such as E proteins, form homodimers and bind specific DNA sequences (E-box), heterodimers containing Id proteins are not capable of binding to DNA. Although Id and E proteins are encoded by distinct genes, it is noteworthy that COUP-TFII-V1 is inhibited by its variant that is encoded by the same gene. Furthermore, it remains to be studied why a COUP-TFII-V1–COUP-TFII-V2 heterodimer lacking a half-set of DBD is not able to bind to DNA. Chen and Young  reported the crystal structure of a homodimer TR DBD in complex with an inverted repeat class of TRE (thyroid response element). Since deletion of a part of DBD abolishes the formation of TR dimers on TRE and reduces transcriptional activation, the crystal structure of a TR dimer consisting of a TR mutant may address the mechanism of how an absence of the half-set of DBD of the dimers in the present study abolished its DNA-binding activity.
In addition to the mechanism where COUP-TFII-V2 sequestrates COUP-TFII-V1 from the Cyp7a1 promoter to suppress Cyp7a1 expression (Figures 7A and 7B), we showed that COUP-TFII-V2 also inhibited HNF4α binding to the Cyp7a1 promoter (Figure 7C). HNF4α has been shown to interact with COUP-TFII and synergistically activates transcription of the Cyp7a1 promoter by directly interacting with DR1 in the Cyp7a1 promoter , suggesting that it serves as a transcription co-activator of COUP-TFII-V1 in the activation of the Cyp7a1 promoter. Transcriptional regulation is orchestrated by co-activators and co-repressors, which recruit HATs (histone acetyl transferases) and HDACs respectively. In order to examine whether COUP-TFII-V2 plays any roles in the recruitment of co-repressors to the transcriptional complexes on the Cyp7a1 promoter, we treated HepG2 cells overexpressing COUP-TFII-V1 and various concentrations of COUP-TFII-V2 with TSA (trichostatin A), an inhibitor of HDACs, and performed the Cyp7a1 promoter assay. As shown in Supplementary Figure S5 (at http://www.biochemj.org/bj/452/bj4520345add.htm), treatment of TSA did not rescue the COUP-TFII-V2-induced inhibition of the Cyp7a1 promoter activity, suggesting that COUP-TFII-V2 does not affect HDAC activity during the transcriptional activation of Cyp7a1.
A functional dimerization mutant of COUP-TFII fails to activate the transcription of the NGFI-A promoter . We also showed that the functional dimerization mutant of COUP-TFII-V2 is not capable of inhibiting the COUP-TFII-V1-mediated activation of the Cyp7a1 promoter. The results of the present study suggest that the ability of COUP-TFII proteins to form functional dimers is required not only for activation of promoters, but also for the inhibition of transcriptional activities.
Rosa and Brivanlou  reported that the alternative splicing variants of COUP-TFII, including COUP-TFII-V2, are expressed during the differentiation of human embryonic stem cells, and interestingly, whereas COUP-TFII-V2 itself does not activate the NGFI-A promoter, it enhances the COUP-TFII-V1-mediated activation of the NGFI-A promoter. These differential effects of COUP-TFII-V2 on the COUP-TFII-V1-induced activation of Cyp7a1 and NGFI-A promoters can be explained by different mechanisms of the COUP-TFII-V1-mediated activation of their transcription. Although COUP-TFII-V1 directly binds to the Cyp7a1 promoter, COUP-TFII-V1 binds to the transcriptional complex containing the Sp1 (specificity protein 1) transcription factor that directly binds to the NGFI-A promoter [41,42]. In order to examine whether binding of COUP-TFII-V2 to COUP-TFII-V1 enhances the interaction between Sp1 and the NGFI-A promoter to increase the NGFI-A promoter activity, we performed EMSAs using nuclear protein extracts of HepG2 cells with a 32P-labelled oligonucleotide containing the Sp1-binding site in the NGFI-A promoter. Overexpression of COUP-TFII-V2 in HepG2 cells did not increase the binding of Sp1 to the NGFI-A promoter (results not shown), suggesting that the enhancement of the NGFI-A promoter activity by COUP-TFII-V2 is not mediated by the binding affinity of Sp1 to the NGFI-A promoter. Rosa and Brivanlou  proposed two hypotheses to explain this effect of COUP-TFII-V2. COUP-TFII-V2 may eliminate the transcriptional repressors that are involved in the inhibitory intermolecular interactions of COUP-TFII-V1. Alternatively, COUP-TFII-V2 may enhance the COUP-TFII-mediated transcriptional activation of the NGFI-A promoter by the formation of heterodimers.
The first exons of COUP-TFII-V1 and COUP-TFII-V2 are different, suggesting that the transcriptional initiation of these variants is regulated by distinct mechanisms. We examined the histone modifications of COUP-TFII using the public database ENCODE (Encyclopedia of DNA Elements) (Supplementary Figure S6 at http://www.biochemj.org/bj/452/bj4520345add.htm). Covalent modification of a specific residue in N-terminal tails of histones alters chromatin structure and function. Among histone modifications, H3K4me3 (trimethylation of histone 3 Lys4) is a well-known epigenetic mark, which localizes at the TSS (transcription start site) of active genes . According to H3K4me3 modification enrichment and function, we can predict where RNA polymerase II begins transcription of an individual gene. We analysed the H3K4me3 ChIP-seq (ChIP-sequence) data in NHEK (normal human epidermal keratinocyte) cells instead of HaCaT cells, as H3K4me3 ChIP-seq data of HaCaT cells were not available. As shown in Supplementary Figure S6, at the COUP-TFII locus in primary keratinocyte NHEK cells, major signals of H3K4me3 were detected around the 5′ flanking regions of different isoforms which contain TATA or TATA-like sequences. The results suggest that COUP-TFII transcript variants might be regulated by different promoters, but not by alternative splicing.
Furthermore, the findings of the present study showing that differentially expressed COUP-TFII-V2 in HepG2 and Huh7 cells is involved in the regulation of endogenous expression of Cyp7a1 (Figure 1) suggest that the differential expression levels of COUP-TFII-V1 and COUP-TFII-V2 may play important roles in the regulation of the endogenous expression of their target genes. Although COUP-TFII-V1 and COUP-TFII-V2 were expressed in a similar manner in multiple tissues of mice, relative expression levels between COUP-TFII-V1 and COUP-TFII-V2 were not uniform among tissues (Supplementary Figure S7 at http://www.biochemj.org/bj/452/bj4520345add.htm). Of note, knockdown of COUP-TFII-V1 in HepG2 cells increased the expression of COUP-TFII-V2, suggesting that endogenous COUP-TFII-V1 may inhibit the expression of COUP-TFII-V2 in HepG2 cells (Supplementary Figure S2). Those results, together with the present finding that COUP-TFII regulates the endogenous Cyp7a1 expression in hepatocellular carcinoma cell lines, suggest that transcriptional activities of COUP-TFII-V1 are differentially regulated in various types of tissues depending on the levels of COUP-TFII-V2 expression.
In conclusion, we presented a unique function of COUP-TFII-V2 in the transcriptional regulation of the Cyp7a1 gene. As COUP-TFII has been implicated in the formation and maintenance of various organs, further analyses of the roles of COUP-TFII-V2 in the regulation of its target genes may lead to a better understanding of various physiological and pathological events in which COUP-TFII is involved.
chicken ovalbumin upstream promoter-transcription factor II
cytochrome P450, family 7, subfamily a, polypeptide 1
DR sequences separated by one nucleotide
electrophoretic mobility-shift assay
green fluorescent protein
trimethylation of histone 3 Lys4
human embryonic kidney
hepatocyte nuclear factor 4α
nerve growth factor inducible gene A
normal human epidermal keratinocyte
proximity ligation assay
retinoid X receptor
negative control siRNA
small interfering RNA
siRNA against COUP-TFII-V1
siRNA against both COUP-TFII-V1 and COUP-TFII-V2
siRNA against COUP-TFII-V2
specificity protein 1
thyroid hormone receptor
thyroid response element
Tomoko Yamazaki designed and performed the experiments, prepared the Figures, and wrote the paper. Jun-ichi Suehiro and Takashi Minami supervised the EMSA. Hideki Miyazaki prepared the deletion mutants of COUP-TFII. Tatsuhiko Kodama provided key materials. Takashi Minami, Kohei Miyazono and Tetsuro Watabe conceptualized the project, designed the experiment and revised the paper before submission. All authors contributed to and have approved the final paper.
We thank Dr John Y.L. Chiang for providing us with the rat Cyp7a1 promoter. We are grateful to Dr Takao Hamakubo, Dr Hiroko Iwanari, Dr Miyako Fujii and Ms Mai Miura (The University of Tokyo, Tokyo, Japan) for providing us with the anti-COUP-TFII-V2 antibody and Dr Toshiya Tanaka for the human COUP-TFII-V1 and COUP-TFII-V2 plasmids. We also thank Dr Hiroshi I. Suzuki, Dr Daizo Koimuma and the members of the Department of Molecular Pathology, The University of Tokyo, for their discussions.
This research was supported by Kakenhi [grants-in-aid for scientific research in innovative area (Integrative Research on Cancer Microenvironment Network)] [grant number 22112002] and the Global Center of Excellence Program (Integrative Life Science Based on the Study of Biosignaling Mechanisms) from the JSPS (Japan Society for the Promotion of Science). T.Y. was supported by a Research Fellowship of the JSPS for Young Scientists.