HNF1α and SREBP2 are important regulators of NPC1L1 in human liver

Niemann-Pick C1-like 1 (NPC1L1), a key regulator of intestinal cholesterol absorption, is highly expressed in human liver. Here, we aimed to gain more insight into mechanisms participating in its hepatic regulation in humans. Correlation analysis in livers from Chinese patients with and without gallstone disease revealed strong positive correlations between NPC1L1 and sterol regulatory element binding protein 2 (SREBP2) (r = 0.74, P < 0.05) and between NPC1L1 and hepatic nuclear factor α (HNF4α) (r = 0.53, P < 0.05) mRNA expression. HNF4α is an upstream regulator of HNF1α; thus, we also tested whether HNF1α participates in the regulation of NPC1L1. We showed a dose-dependent regulation by SREBP2 on the NPC1L1 promoter activity and mRNA expression in HuH7 cells. Chromatin immunoprecipitation assay confirmed the binding of SREBP2 to the promoter in vivo. Surprisingly, HNF4α slightly decreased the NPC1L1 promoter activity but had no effect on its gene expression. By contrast, HNF1α increased the promoter activity and the gene expression, and an important HNF1 binding site was identified within the human NPC1L1 promoter. ChIP assays confirmed that HNF1α can bind to the NPC1L1 promoter in vivo.

Cell experiments, mutagenesis, and transfections. HuH7 and HEK293 cells were cultured as described ( 19 ). For all cell experiments, HuH7 or HEK293 cells were plated out on 6-well tissue culture plates so that they reached ‫ف‬ 70% confl uence after 24 h (i.e., ‫ف‬ 700,000 cells/well). The human NPC1L1 promoter ( Ϫ 1570 to +137) was screened using the transcription factor TESS database ( www.cbil.upenn.edu at University of Pennsylvania, Philadelphia, PA) to search for putative HNF1 binding sites, and specifi c point mutations were generated as described (primer sequences are available upon request) ( 19 ).
Transfections of HuH7 cells were performed using 2 µg promoter construct (or mutated constructs) and 2 µg pSV-␤galactosidase control vector (Promega, Madison, WI) with or without increasing concentrations of SREBP2, HNF1 ␣ , and HNF4 ␣ expression vectors or with 0.5 g of each expression vector using Lipofectin reagent (Invitrogen, Carlsbad, CA) at a ratio of 3:1 (Lipofectin:DNA). pGL3 empty vector (Promega) was used to adjust for differences in amount of DNA added to the cells. Transfections of HEK293 cells were performed like those for HuH7 cells, except that Lipofectamine 2000 reagent (Invitrogen) at a ratio of 0.25:1 (Lipofectamine:DNA) was used. The pSV-␤ -galactosidase control vector was used to correct for variation in transfection effi ciency. Forty-eight hours after transfection, cell lysates were prepared in reporter lysis buffer (Promega). ␤ -Galactosidase and luciferase activities were determined using ␤ -galactosidase or luciferase assay kits, respectively, according to the manufacturer's instructions (Promega). All transfection data are expressed as luciferase activity corrected by ␤ -galactosidase activity.
To study the effect of cholesterol on NPC1L1 gene expression, HuH7 cells were incubated for 12 h with 0, 0.1, 0.5, 1, and 2 mM LDL cholesterol or with 10% LPDS prior to RNA extraction.
All cell experiments were performed in quadruplicates, and data represent means ± SEM.
Electrophoretic mobility shift assay and supershift assay. Nuclear extracts were prepared from HuH7 cells as described by Azzout-Marniche et al. ( 20 ). Forward sequences for primers creas ( 11 ). Defi ciency of HNF1 ␣ in mice ( 12 ) results in defect bile acid transport, increased bile acid and liver cholesterol synthesis, and impaired HDL metabolism. HNF4 ␣ knockout mice die before birth ( 13 ), and conditional liver-specifi c disruption of HNF4 ␣ ( 14 ) results in hepatomegaly, lipid deposition in the liver, reduced serum cholesterol and triglyceride levels, and elevated serum bile acid concentrations. Thus, both HNF1 ␣ and HNF4 ␣ play important roles in lipid homeostasis.
Since the physiological signifi cance in human liver remains to be clarifi ed, the aim of this study was to gain more insight into the mechanisms that participate in the transcriptional regulation of hepatic NPC1L1.

Materials
2×SYBRGreen Mastermix was purchased from MedProbe (Oslo, Norway). HuH7 and HEK293 cells were purchased from American Type Culture Collection (Manassas, VA). SREBP2, HNF1 ␣ , and IgG antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The human NPC1L1 promoter (an ‫ف‬ 1,700 bp fragment, ranging from Ϫ 1570 to +137 bp, cloned into pGL3 promoter vector) was a generous gift from Dr. Charlotte Murphy and Dr. Mats Gåfvels (Karolinska Institutet, Sweden). The SREBP2 expression vector, which is not the full length but encodes the transcriptionally active form of the protein, was from American Type Culture Collection (pCMV-SREBP2-468 No.63452). The HNF4 ␣ expression vector ( 15 ) was a generous gift from Dr. Theodore C. Simon (Washington University School of Medicine, St. Louis, MO). The HNF1 ␣ expression vector ( 16 ) was a generous gift from Professor Pal R. Njølstad and Dr. Lise Bjørkhaug Gundersen (Haukeland University Hospital, Norway). Lipoprotein defi cient serum (LPDS) and LDL were isolated using FBS or plasma from a healthy blood donor, respectively, by density gradient ultracentrifugation ( 17 ).

Methods
Subjects. Liver biopsies were from subjects who have previously been investigated in a study by Jiang et al. ( 18 ). In brief, 22 normolipidemic and nonobese Chinese patients (11 females and 11 males) with cholesterol gallstone disease (GS) and 12 Chinese gallstone-free patients (GSF; nine females and three males) were included. None of the patients were subjected to any lipid-lowering treatment. Informed consent to participate in the study and to collection of a liver biopsy was obtained. Patients were fasted overnight before a biopsy of ‫ف‬ 0.5-1.0 g was taken from the right lobe of the liver, snap-frozen in liquid nitrogen, and stored at -70°C. used to generate double-stranded probes were as follows: The underlined bases in the forward sequences were deleted for the mutated HNF1 binding sites. Unlabeled double-stranded probes were generated by mixing 1 g forward and 1 g reverse oligonucleotide, 5 l 1 M NaCl, and ddH 2 O up to 50 l, and annealed at 95°C for 10 min. Labeled probes were generated by mixing 5 l unlabeled probe, 2 l 10× PNK buffer (Promega), 1 l 0.1 M DTT, 1.5 l T4 polynucleotide kinase (Promega), 2 l [ ␥ -32 P ATP], and ddH 2 O up to 50 l and incubated at 37°C for 1 h, before passing through a Sephadex G-50 column (Amersham, Uppsala, Sweden). The labeled probes (40,000 cpm) were combined with 10 g nuclear extract along with 2 g poly (dI-dC) (Amersham, Piscataway, NJ), 7.5 l DNA binding buffer (80 mM HEP-ES-NaOH, pH 7.6, 0.2 M NaCl, 40 mM DTT, 20% v/v glycerol, 2 mM EDTA, and 1.2 mg/ml BSA), 1 l 0.1 M DTT, 4 l Buffer C (10 mM HEPES-KOH, pH 7.9, 0.4 M NaCl, 0.1 mM EDTA, 5% v/v glycerol, 1 mM DTT, and protease inhibitors), and ddH 2 O up to 30 l. Binding reaction mixtures were incubated for 20 min at room temperature and resolved on a nondenatured (4% w/v) acrylamide gel in 1× TBE at 4°C for 3 h. For supershift assays, 2 g HNF1 ␣ (sc-6547×; Santa Cruz Biotechnology) or 2 g SREBP2 (sc-8151×; Santa Cruz Biotechnology) antibodies were added to the binding reaction mixtures. After electrophoresis, gels were dried and exposed over night to X-ray fi lm at Ϫ 80°C.
Statistical analysis. Data

NPC1L1 correlates with SREBP2 and HNF4 ␣ in human livers
Gene expression analyses were performed in liver samples from 22 Chinese patients with cholesterol GS and 12 Chinese GSF patients. As we recently reported ( 18 ), no signifi cant differences were observed in NPC1L1, SREBP2, and HNF1 ␣ mRNA expression between GS and GSF patients, though GS patients had 43% ( P < 0.05) higher HNF4 ␣ mRNA expression compared with GSF patients.
A strong positive correlation between NPC1L1 and SREBP2 mRNA expression was present. As shown in Fig.  1A , when all patients were considered the regression coef- fi cient was 0.74 ( P < 0.05). When the patients were divided by the presence of gallstone disease, the correlation was still present ( r = 0.71 in GS, and r = 0.89 in GSF, P < 0.05).
Also, a positive correlation between NPC1L1 and HNF4 ␣ mRNA expression was observed. When all the patients were considered, the regression coeffi cient was 0.53 ( P < 0.05) ( Fig. 1B ), and when the patients were divided by the presence of gallstone disease, the correlation was still present ( r = 0.44 in GS, and r = 0.79 in GSF, P < 0.05) also for this transcription factor.

SREBP2 interacts with the human NPC1L1 promoter in vitro and in vivo
The above observations prompted us to investigate whether SREBP2, HNF4 ␣ , and HNF1 ␣ may participate in the hepatic regulation of NPC1L1 in humans. Cotransfection experiments in human hepatoma cells (HuH7) using the human NPC1L1 promoter ( Ϫ 1,570 to +137 bp) and the SREBP2 expression vector were performed. A strong dose-dependent regulation by SREBP2 on the human NPC1L1 promoter activity ( Fig. 2A ) was observed. Also, the NPC1L1 mRNA expression increased ‫ف‬ 60% ( P < 0.05) using 0.5 g SREBP2 expression vector ( Fig. 2B ). Activation of SREBP2 is dependent on the free cholesterol levels inside the cell. To study the effect on the endogenous NPC1L1 gene expression under more physiological conditions, HuH7 cells were depleted or loaded with cholesterol. Loading of the cells with LDL cholesterol decreased both the NPC1L1 and the SREBP2 mRNA expression ( Fig.  2C ). A strong positive correlation ( r = 0.94, P < 0.05) between NPC1L1 and SREBP2 mRNA expression was also observed (data not shown). Cholesterol depletion, on the other hand, resulted in an insignifi cant trend toward increased NPC1L1 mRNA levels (data not shown). Recently, Alrefai et al. ( 23 ) identifi ed two SREs, SRE1 and SRE2, in Fig. 2. SREBP2 binds to and regulates the human NPC1L1 promoter. A: The human NPC1L1 promoter ( Ϫ 1,570 to +137) was used for cotransfection analysis along with 0.1 or 0.5 g SREBP2 expression vector in HuH7 cells. HuH7 cells were transfected with increasing amounts of SREBP2 expression vector (B) or treated with increasing concentrations of LDL cholesterol (C). Gene expressions were normalized to GAPDH mRNA. Experiments in A-C were performed in quadruplicates, and data are expressed as means ± SEM. Differences between groups were tested by one-way ANOVA followed by posthoc comparisons according to the Dunnett test. * P < 0.05, ** P < 0.01, *** P < 0.001, and **** P < 0.0001. D: Soluble chromatins ( ‫ف‬ 60 µg DNA, ‫ف‬ 500 bp in length) were prepared from human liver and immunoprecipitated with 4 µg antibodies against SREBP2 or IgG and amplifi ed by real-time RT-PCR using primers for detection of SRE1 ( Ϫ 91/ Ϫ 81 bp) and SRE2 ( Ϫ 748/735 bp) in the NPC1L1 promoter. The IgG antibody was used as a baseline control and used to compare the relative fold enrichment of the NPC1L1 promoter by the specifi c DNA fragments. Before immunoprecipitation, a small aliquot of chromatin was saved and used as an input control. Experiments were performed in triplicate, and data are expressed as means ± SEM. the human NPC1L1 promoter. To determine whether SREBP2 can bind to these SREs in the NPC1L1 promoter in vivo, we performed ChIP assay using human liver. Immunoprecipitation with a specifi c antibody against SREBP2 and primers designed to target the two previously identifi ed SREs were used in the PCR analysis. This led to >36fold enrichment of SRE1 ( Ϫ 91/ Ϫ 81 bp) and ‫ف‬ 17-fold enrichment of SRE2 ( Ϫ 748/ Ϫ 738 bp) sequences present in the human NPC1L1 promoter ( Fig. 2D ). Collectively, these results show that SREBP2 can regulate and bind to the NPC1L1 promoter in human liver.

Effects of HNF4 ␣ overexpression on NPC1L1 promoter activity and gene expression
Due to the positive correlation between NPC1L1 and HNF4 ␣ in the patients, we also performed cotransfections in HuH7 cells using the human NPC1L1 promoter along with the HNF4 ␣ expression vector. Surprisingly, HNF4 ␣ decreased the human NPC1L1 promoter activity in a dosedependent fashion ( Fig. 3A ), whereas HNF4 ␣ overexpression had no effect on its mRNA expression ( Fig. 3B ). Iwayanagi, Takada, and Suzuki ( 24 ) showed that the transcription of NPC1L1 was stimulated by HNF4 ␣ together with SREBP2, but not by HNF4 ␣ alone. Cotransfection of HuH7 cells with both SREBP2 and HNF4 ␣ expression vectors was performed to test a possible synergism in the activation of the NPC1L1 promoter. However, no further activation of the promoter activity occurred ( Fig. 3C ).

HNF1 ␣ interacts with the human NPC1L1 promoter in vitro and in vivo
HNF4 ␣ is an essential positive regulator of HNF1 ␣ ( 25 ). HNF4 ␣ and HNF1 ␣ can bind directly to one another, and both contain a binding site in their promoter region for the other ( 26 ), suggesting a reciprocal regulation. Thus, we wanted to test whether HNF1 ␣ might participate in the regulation of NPC1L1. We performed cotransfections in HuH7 cells using the human NPC1L1 promoter along with the HNF1 ␣ expression vector. These experiments revealed a dose-dependent regulation by HNF1 ␣ on the NPC1L1 promoter activity ( Fig. 4A ) and increased mRNA expression ( Fig. 4B ). Next, we screened the sequence using TESS to search for putative HNF1 binding sites in the human NPC1L1 promoter. Six HNF1 cis -elements were found located Ϫ 119/ Ϫ 114, Ϫ 158/ Ϫ 144, Ϫ 360/ Ϫ 354, Ϫ 598/ Ϫ 593, Ϫ 665/ Ϫ 660, and Ϫ 769/ Ϫ 763 bp upstream of the ATG start codon ( Fig. 5 ). We performed mutagenesis on these HNF1 cis -elements and used the mutant constructs in transfection experiments with or without the HNF1 ␣ expression vector. These experiments showed that mutation of one HNF1 binding site ( Ϫ 158/ Ϫ 144) almost completely abolished the regulatory effect of HNF1 ␣ on the NPC1L1 promoter activity ( Fig. 4C ). However, muta- Fig. 3. Effects of HNF4 ␣ on the human NPC1L1 promoter. A: The human NPC1L1 promoter ( Ϫ 1,570 to +137) was used for cotransfection analysis along with 0.5, 1, or 2 g HNF4 ␣ expression vector in HuH7 cells. B: HuH7 cells were transfected with increasing amount of HNF4 ␣ expression vector. The NPC1L1 gene expression was normalized to GAPDH mRNA. HuH7 (C) or HEK293 (D) cells were cotransfected with NPC1L1 promoter along with 0.5 g expression vectors. Experiments in A-D were performed in quadruplicates, and data are expressed as means ± SEM. Differences between groups were tested by one-way ANOVA followed by posthoc comparisons according to the Dunnett test. * P < 0.05, ** P < 0.01, and **** P < 0.0001. tion of two of the other HNF1 binding sites ( Ϫ 119/ Ϫ 114 and Ϫ 769/ Ϫ 763) also decreased the induction by HNF1 ␣ on the NPC1L1 promoter activity, although to a lesser extent. Furthermore, mutation of the Ϫ 665/ Ϫ 660 bp HNF1 site increased the induction by HNF1 ␣ , suggesting that this site may work as a repressor element. Also, to determine whether HNF1 ␣ can bind to the human NPC1L1 promoter in vivo, we performed ChIP analysis using human liver ( Fig. 4D ). Immunoprecipitation with a specifi c antibody against HNF1 ␣ and primers designed to target the region spanning over the six HNF1 sites ( Ϫ 769 to Ϫ 119 bp) were used in the PCR analysis. This led to >12fold enrichment of the human NPC1L1 promoter. Finally, to be able to distinguish which of the six putative HNF1 binding sites is responsible for the transactivation by HNF1 ␣ , we isolated nuclear extracts from HuH7 cells and performed electrophoretic mobility shift assay and supershift assays for each site. These experiments revealed a direct binding of HNF1 ␣ to the Ϫ 158/ Ϫ 144 bp HNF1 binding site in the NPC1L1 promoter ( Fig. 6 ) but no binding of HNF1 ␣ to the other fi ve HNF1 sites (data not shown). Specifi c bands, labeled with the HNF1 probe, were detectable (lanes 1 and 5). Supershift assay, with the labeled HNF1 probe and HNF1 ␣ antibody, shifted the specifi c bands even further (lanes 2 and 6). Furthermore, preincubation of 10 g nuclear extract with labeled mutated HNF1 probe did not result in appearance of any specifi c band (lanes 3), and no supershift occurred with either the labeled mutated HNF1 probe and HNF1 ␣ antibody (lane 4) or with the labeled HNF1 probe and SREBP2 antibody (lane 7). Collectively, these results show that HNF1 ␣ can regulate and bind to the NPC1L1 promoter in human liver and that these effects are mediated by the Ϫ 158/ Ϫ 144 bp HNF1 binding site.

DISCUSSION
In this study, we investigated whether SREBP2, HNF4 ␣ , and HNF1 ␣ might be important regulators of the NPC1L1 promoter in human liver in vitro and in vivo .
SREBP2 regulates genes involved in cholesterol biosynthesis (e.g., HMG-CoA reductase) and uptake (e.g., LDLreceptor), and it is activated upon low cellular cholesterol levels ( 9 ). Sané et al. ( 4 ) reported that SREBP2 mRNA expression was increased in Caco-2 cells deprived of NPC1L1. Alrefai et al. ( 23 ) reported that NPC1L1 mRNA expres- with or without HNF1 ␣ expression vector. Experiments in A-C were performed in quadruplicates, and data are expressed as means ± SEM. Differences between groups were tested by one-way ANOVA followed by posthoc comparisons according to the Dun-nett test. * P < 0.05, ** P < 0.01, *** P < 0.001, and **** P < 0.0001. D: Soluble chromatins ( ‫ف‬ 60 µg DNA, ‫ف‬ 500 bp in length) were prepared from human liver and immunoprecipitated with 4 µg antibodies against HNF1 ␣ or IgG and amplifi ed by real-time RT-PCR using primers to target the region spanning over the six HNF1 binding sites ( Ϫ 769 to Ϫ 119 bp) in the NPC1L1 promoter. The IgG antibody was used as a baseline control and used to compare the relative fold enrichment of the NPC1L1 promoter by the specifi c DNA fragments. Before immunoprecipitations, a small aliquot of chromatin was saved and used as an input control. Experiments were performed in triplicate, and data are expressed as means ± SEM. which had no effect on the promoter activity in our experiments. In the cotransfection experiments, we used 0.5 g of each expression vector, whereas Iwayanagi, Takada, and Suzuki used 0.1 g SREBP2 and 0.5 g HNF4 ␣ expression sion decreased by cholesterol loading and increased in response to cholesterol depletion in Caco-2 cells, and two SREs in the NPC1L1 promoter region were identifi ed. These studies suggest that low levels of cholesterol activate the transcription of SREBP2, leading to an upregulation of the LDL-receptor and HMG-CoA reductase, which in turn increases NPC1L1 expression with a subsequent increase in cholesterol absorption. Furthermore, in miniature pigs treated with a combination of ezetimibe plus simvastatin, Telford et al. ( 27 ) reported increased NPC1L1 expression in liver and intestine, which was positively correlated with the SREBP2, HMG-CoA reductase, and LDL receptor expressions. In line with previous reports, we showed a ) strong positive correlation between NPC1L1 and SREBP2 mRNA expression in liver from Chinese patients with and without gallstone, as well as divided by the presence of gallstone disease; b ) a strong dose-dependent regulation of SREBP2 on the NPC1L1 promoter activity in HuH7 cells and that this fi nding is liver specifi c (>13-fold in HuH7 vs. <30% in HEK293 cells using 0.5 g SREBP2 expression vector; Fig. 3D ); c ) increased NPC1L1 mRNA expression following transfection with SREBP2 expression vector; d ) decreased NPC1L1 and SREBP2 mRNA expression following cholesterol loading of the cells with LDL; and, fi nally, e ) binding of SREBP2 to the human NPC1L1 promoter in human liver. The in vivo binding of SREBP2 to NPC1L1 may thus participate in a coordinated regulation wherein hepatocytes may acquire more cholesterol. Thus, SREBP2 seems to be an important regulator of both the intestinal and the hepatic NPC1L1 expression in humans.
Misawa et al. ( 28 ) reported a direct interaction between SREBP2 and HNF4 ␣ to enhance the sterol isomerase gene expression in Caco-2 and HepG2 cells. Iwayanagi, Takada, and Suzuki ( 24 ) showed that HNF4 ␣ small interfering RNA reduced the mRNA expression and abolished the cholesterol-dependent regulation of NPCL1 in HepG2 cells; also, the transcription of NPC1L1 was stimulated by HNF4 ␣ together with SREBP2, but not by HNF4 ␣ alone. In our study, cotransfection with the human NPC1L1 promoter along with HNF4 ␣ expression vector slightly decreased the promoter activity but had no effect on the NPC1L1 gene expression. Also, no synergistic activation of NPC1L1 promoter activity was seen in HuH7 cells after cotransfection with SREBP2 and HNF4 ␣ . The discrepancies might be due to different amounts of expression vector; we used up to 2 g, whereas Iwayanagi, Takada, and Suzuki ( 24 ) used 0.5 g of HNF4 ␣ expression vector,  Fig. 6. Functionality of the HNF1 binding sites in the human NPC1L1 gene. Nuclear extracts, prepared from HuH7 cells, were incubated with 32 P-labeled double-stranded probes, with or without mutation of the HNF1 binding site and resolved on nondenatured (4% w/v) acrylamide gel. Lanes 1 and 5, binding reactions between 10 or 20 g nuclear extract and 32 P-labeled HNF1 probe, respectively. Lanes 2 and 6, supershift reactions between labeled HNF1 probe and 2 g HNF1 ␣ antibody. Lane 3, binding reaction between 10 g nuclear extract and labeled mutated HNF1 probe. Lane 4, supershift reaction between labeled mutated HNF1 probe and 2 g HNF1 ␣ antibody. Lane 7, supershift reaction between labeled HNF1 probe and 2 g SREBP2 antibody. Lane 8, 10 g BSA and labeled HNF1 probe, serving as a negative control.
vector. Also, different plasmids were used to create the HNF4 ␣ expression vectors. Furthermore, we used the human hepatoma cell line HuH7, whereas Iwayanagi, Takada, and Suzuki ( 24 ) used HepG2 cells.
The strong correlation between NPC1L1 and HNF4 ␣ mRNA expression in our Chinese patients suggests that HNF4 ␣ may function by trans -activating NPC1L1 via binding to other transcription factors, including SREBP2. HNF4 ␣ is an upstream regulator of HNF1 ␣ , and both contain binding sites for each other in their promoter region ( 26,29 ), suggesting a reciprocal regulation. HNF1 ␣ was recently shown to be an important regulator of ACAT2 ( 19 ), which is another important gene involved in cholesterol homeostasis. By coimmunoprecipitation of nuclear extracts from human liver with an antibody raised against HNF1 ␣ and detection by Western blot with an HNF4 ␣ antibody, we were able to show that a protein-protein interaction between HNF4 ␣ and HNF1 ␣ occurs in human liver ( 30 ). Thus, it is possible that HNF4 ␣ also exerts its action on NPC1L1 by trans -activating NPC1L1 via binding to HNF1 ␣ or by increasing its expression. Interestingly, Odom et al. ( 29 ) reported that HNF4 ␣ can bind to the NPC1 gene but not to NPC1L1 in primary human hepatocytes. The NPC1L1 protein shares 42% identity and 51% similarity with NPC1 ( 31 ); despite this, mice defi cient in NPC1 did not show altered functional activity of NPC1L1 implying their independent roles and regulations ( 32 ).
Odom et al. ( 29 ) also reported that HNF1 ␣ can bind to the NPC1L1 gene in primary human hepatocytes. This is in line with our results in HuH7 cells, showing a dose-dependent regulation of HNF1 ␣ on the NPC1L1 promoter activity and increased mRNA expression following HNF1 ␣ overexpression. The liver specifi city of this fi nding was tested by cotransfection of HEK293 cells with the human NPC1L1 promoter and 0.5 g HNF1 ␣ expression vector ( Fig. 3D ); no induction by HNF1 ␣ on the promoter activity was identifi ed in these cells. Also, we identifi ed an HNF1 binding site located Ϫ 158/ Ϫ 144 bp upstream of transcription start site and showed that mutation of this site almost completely abolished the induction by HNF1 ␣ . Electrophoretic mobility shift assay and supershift assays showed a direct binding of HNF1 ␣ to this site but not to the other fi ve cis -elements. To show the in vivo binding of HNF1 ␣ to the NPC1L1 promoter in human liver, we performed ChIP analysis. The six putative HNF1 binding sites (located at Ϫ 119, Ϫ 158, Ϫ 360, Ϫ 598, Ϫ 665, and Ϫ 769 bp upstream of the ATG start codon) are closely located in the NPC1L1 promoter region, and the sonication procedure in the ChIP analysis generally generates DNA fragments of ‫ف‬ 500-800 bp. Thus, primers designed to span over the six HNF1 binding sites were used in the PCR analysis and revealed that HNF1 ␣ can bind to the human NPC1L1 promoter in vivo. Collectively, these data suggest that HNF1 ␣ may be an important regulator of the hepatic expression of NPC1L1 in humans.
In summary, in this study, we showed that SREBP2 and HNF1 ␣ are important transcription factors for the hepatic NPC1L1 promoter activity that can bind to and regulate its expression in humans.