Characterization of the Human OATP-C (SLC21A6) Gene Promoter and Regulation of Liver-specific OATP Genes by Hepatocyte Nuclear Factor 1 (cid:1) *

OATP-C ( SLC21A6 ) is the predominant Na (cid:2) -independ-ent uptake system for bile salts and bilirubin of human liver and is expressed exclusively at the basolateral (si-nusoidal) hepatocyte membrane. To investigate the ba-sis of liver-specific expression of OATP-C, we studied promoter function in the two hepatocyte-derived cell lines HepG2 and Huh7 and in nonhepatic HeLa cells. OATP-C promoter constructs containing from 66 to 950 nucleotides of 5 (cid:1) -regulatory sequence were active in HepG2 and Huh7 but not HeLa cells, indicating that determinants of hepatocyte-specific expression reside within the minimal promoter. Deoxyribonuclease I footprint analysis revealed a single region that was protected by HepG2 and Huh7 but not HeLa cell nuclear extracts. The liver-enriched transcription factor hepatocyte nuclear factor 1 (cid:1) (HNF1 (cid:1) ) was shown by mobility shift assays to bind within this footprint. Coexpression of HNF1 (cid:1) stimulated OATP-C promoter

Although no disease state that is associated with mutations in the OATP-C gene has been identified so far, OATP-C probably represents the chief Na ϩ -independent bile salt and bilirubin uptake system of human liver. Reduced expression of OATP-C has been observed in the chronic cholestatic liver disease primary sclerosing cholangitis (7). The rat orthologue of OATP-C, called Oatp4 (Slc21a6), exhibits reduced mRNA levels after bile duct ligation or cecum ligation and puncture, two experimental models of cholestasis (8). Cecum ligation and puncture leads to bacteremia and consequent endotoxinemia, which causes intrahepatic cholestasis through decreased expression levels of hepatocellular bile salt and organic anion transporters (9). In rats, endotoxinemia decreases nuclear activity of hepatocyte nuclear factor 1 (HNF1) 1 , a critical factor for basal expression of the Na ϩ -taurocholate cotransporting polypeptide (Ntcp 1 (Slc10a1)) (10). Whether reduced nuclear binding of HNF1 is also the cause of decreased hepatic Oatp4 expression in endotoxemic rats is unknown.
Using oligonucleotide microchip expression analysis, mRNA levels of Ntcp (Slc10a1), Oatp1 (Slc21a1), Oatp2 (Slc21a5), and Oatp4 (Slc21a6) were found to be decreased or absent in the livers of Tcf1 Ϫ/Ϫ (HNF1␣ Ϫ/Ϫ ) mice with a null mutation in the HNF1␣ gene (11). These data together with the known dependence of rat Ntcp (Slc10a1) gene expression upon HNF1␣ (12), suggested a general role for HNF1␣ as a transcriptional acti-vator of hepatic bile salt and bilirubin transporters. Although a direct role for HNF1␣ in regulating human liver OATP genes has not been shown, two lines of evidence suggested OATP-C as a likely candidate for regulation by HNF1 as follows. (i) OATP-C is expressed only in hepatocytes and binding sites for HNF1 have been shown in the promoters or enhancers of numerous genes expressed exclusively in the liver (13,14); (ii) the phenotype of increased serum bile acid concentrations and a 3-10-fold elevation of serum bilirubin in HNF1␣-deficient (Tcf1 Ϫ/Ϫ ) mice (11,15) is consistent with a reduced function of Oatp4 (Slc21a6), the liver-specific mouse orthologue of human OATP-C (SLC21A6). In this study, we report the isolation of the human OATP-C promoter and demonstrate the critical role of HNF1␣ for liver-specific OATP gene expression.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]Adenosine triphosphate (3,000 Ci/mmol) was purchased from Amersham Pharmacia Biotech. Restriction enzymes and proteinase K were from Roche Molecular Biochemicals, PfuTurbo DNA polymerase were from Life Technologies, Inc., and T4 polynucleotide kinase was from Stratagene (Amsterdam, Netherlands). Polyacrylamide was obtained from Bio-Rad. Unless stated otherwise, all chemicals were purchased from Sigma-Aldrich.
Localization of the 5Ј-Region of the OATP-C Gene-Total RNA was isolated from human liver by the acid guanidinium/phenol/chloroform procedure (16). The 5Ј-end of the OATP-C mRNA was determined using rapid amplification of cDNA ends (5Ј-RACE 1 system, Roche Molecular Biochemicals). 1 g of total RNA was reverse-transcribed using the OATP-C-specific primer C-RT (see Table I). The 5Ј-end was subsequently amplified by PCR using primer C-PCRace and nested primer nC-PCRace as downstream primers (see Table I). The resulting 5Ј-RACE product was subcloned into pCRII-TOPO (Invitrogen BV, Groningen, Netherlands) and sequenced on an AlfExpress Sequencer (Amersham Pharmacia Biotech).
Plasmid Construction-Four fragments of the 5Ј-region of the OATP-C gene were PCR-amplified using human genomic DNA as a template, upstream primers p-950, p-424, p-128, or p-66, downstream primer pϩ21 (see Table I), and PfuTurbo DNA polymerase. Human OATP8 (SLC21A8) and mouse Oatp4 (Slc21a6) promoter fragments were amplified from genomic DNA using primers p8 -120/p8ϩ38 and p4 -137/p4ϩ15, respectively (see Table I). The upstream primers contained an internal SacI restriction site, and the downstream primers contained an internal BglII site. The resulting PCR products were digested with BglII and SacI and ligated into the luciferase reporter gene vector pGL3-Basic (Promega Catalys AG, Wallisellen, Switzerland) that had been predigested with BglII and SacI, yielding the following promoter constructs: C-950, C-424, C-128, and C-66 for OATP-C, 8-120 for OATP8, and m4-137 for mouse Oatp4. Sequence identity of all constructs with the OATP-C, OATP8, and mOatp4 genes was verified by sequence analysis. Plasmid DNA was prepared using the Qiagen system (Basel, Switzerland).
Site-directed Mutagenesis-An OATP-C-derived Ϫ128/ϩ21 construct containing staggered nicks was generated by PCR using two complementary oligonucleotides mutated in the HNF1 binding site (sequence mutHNF1 in Table I) and PfuTurbo DNA polymerase. The product was digested with DpnI to remove the parental DNA template and select for DNA containing the mutation. The mutated plasmid was termed mutC-128.
Cell Culture and Transfections-HepG2 1 , Huh7 1 and HeLa 1 cell lines were purchased from ATCC (Manassas, VA). Cells were maintained in RPMI 1640 (Sigma) supplemented with 10% fetal calf serum (Life Technologies, Inc.), 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.). Cells were seeded at 90 -95% density in 48well plates. To normalize for transfection efficiency, pRL-TK plasmid (Promega Catalys AG) coding for Renilla luciferase under the control of a thymidine kinase promoter was cotransfected. For transient transfections, a DNA/liposome mix containing 1 l of LipofectAMINE 2000 (Life Technologies, Inc.) and 0.5 g of plasmid DNA/well was used. The DNA mix contained 50 ng of HNF1␣ expression vector (17) where indicated or 50 ng of empty pBluescript vector (pB-SKII, CLONTECH, Basel, Switzerland) as carrier DNA. For transfection of competitor oligonucleotides, 1-10 pmol of the dimerized oligonucleotides wtHNF1 and mutHNF1 (see Table I) were added to the transfection mix. Cells were lysed with passive lysis buffer (PLB, Promega Catalys AG) 24 -36 h after transfection. Luciferase activities were assayed using the dual luciferase reporter system (Promega Catalys AG) and were quantified in a Lumat LB 9507-2 luminometer (Berthold, Bad Wildbad, Germany).
Preparation of Nuclear Extracts and Deoxyribonuclease Footprinting-Nuclear extracts were prepared as described (18). DNase I protection assays were performed with a newly developed non-radioactive approach. A single end-labeled PCR fragment was generated using 5Ј-Cy3-conjugated primer Cy3-pϩ30 and primer p-424 (see Table I) and was purified by agarose gel electrophoresis and QIAEX purification (Qiagen, Basel, Switzerland). 1 ng of labeled PCR fragment was incubated on ice for 10 min with 10 -20 g of nuclear extract and 1 g of poly(dI)poly(dC) (Amersham Pharmacia Biotech) in 50 l of buffer (0.5 M saccharose, 15 mM Tris/HCl, 60 mM KCl, 0.25 mM EDTA, 0.125 mM EGTA, 5 mM MgCl 2 , 2.5 mM CaCl 2 ). After the addition of 1 volume of digestion buffer (5 mM MgCl 2 , 2.5 mM CaCl 2 ), a 60-s digestion with 5 l of freshly diluted DNase I (Roche Molecular Biochemicals) was performed. Digestion was terminated by the addition of 1 volume of stop solution (50 mM EDTA, 0.1% SDS, 150 g/ml tRNA, 200 g/ml proteinase K). DNase I-digested samples and a control sample digested in the absence of nuclear proteins were extracted with phenol/chloroform and precipitated in ethanol. Resuspended DNA samples were loaded on a denaturating polyacrylamide gel alongside a sequencing reaction that had employed the Cy3-pϩ30 primer and the C-424 plasmid as template. Bands were detected and analyzed using the AlfExpress system (Amersham Pharmacia Biotech).
Electrophoretic Mobility Shift Assays-Double-stranded oligonucleotide probes were obtained by hybridizing single-stranded complementary oligonucleotides (Microsynth, Balgach, Switzerland). Dimers with the sense sequences shown in Table I were labeled with [␥-32 P]ATP using T4 polynucleotide kinase (Stratagene, Amsterdam, Netherlands). Sequence wtHNF1 corresponded to the wild type HNF1␣ sequence found in the OATP-C gene promoter, perHNF1 corresponded to a perfect consensus sequence for HNF1, and mutHNF1 corresponded to the wild type sequence mutated within the HNF1 recognition site. For gel mobility shift assays, 5 g of nuclear extracts were incubated on ice for 20 min with 2-5 fmol of ␥-32 P-end-labeled dimerized oligonucleotide and 1 g of poly(dI)poly(dC) (Amersham Pharmacia Biotech) in 20 mM HEPES-KOH, pH 7.9, 20% glycerol, 100 mM KCl, 2 mM MgCl 2 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. For competition assays, a 10 -500-fold excess of unlabeled dimerized oligonucleotides was added. For supershift experiments, 1 l of antibody against HNF1␣ (17), 1 l of antibody against HNF1␤ (19), or 2 l of antibody against HNF1 (H-205: sc-8986, Santa Cruz Biotechnology Inc., Heidelberg, Germany) was added to the reaction mix. Reactions were analyzed by electrophoresis through 3.4% polyacrylamide gels in 0.25ϫ Trisborate EDTA buffer at 120V for 2 h.
Statistical Analysis-Reporter gene activities are expressed as the mean Ϯ 1 S.D. of at least three individual transfection experiments. All data were reproduced at least once using two different preparations of plasmid DNA.  (4). These additional nucleotides were also present in the genomic SLC21A6 sequence derived from BAC clone RP11-125O5 (AC022335) and were consequently designated as the transcription start site (nt ϩ1 in Fig. 1B). Based on the genomic SLC21A6 sequence, an additional exon with a length of 39 bp (exon Ϫ1 in Fig. 1A) is separated by a 10277-bp intronic sequence from the 66-bp untranslated region of exon 1 (AJ400749) (5). The transcription start site of the OATP-C cDNA is located 105 bp upstream of the ATG initiation codon.

Localization of the Transcription Initiation Site and Promoter Region of the OATP-C (SLC21A6) Gene-To
Analysis of the 5Ј-Flanking Region of the Human OATP-C (SLC21A6) Gene-The 5Ј-flanking region of the OATP-C gene was PCR-amplified from human genomic DNA and showed sequence identity with clone Homo sapiens 12p BAC RP11-125O5 (AC022335). Fig. 1B displays the sequence from nt Ϫ950 to ϩ40 relative to the transcription initiation site. Homologies to known gene regulatory elements were identified using the program Mat Inspector (Genomatix Software, Munich, Germany). Several potential transcription factor recognition sites were found including ubiquitously expressed factors such as the activator protein 1 (AP-1) at nt Ϫ58 to Ϫ68 and Ϫ557 to Ϫ567. A TATA motif was identified at nt Ϫ82 to Ϫ96 but not within the immediate 5Ј-flanking region of the transcriptional start site. Other more generally represented DNA elements included potential binding sites for the nuclear factor 1 (NF-1) at nt Ϫ15 to Ϫ32 and Ϫ111 to Ϫ125, a CAAT element at Ϫ199 to Ϫ206, and an octamer binding site (Oct-1) at Ϫ322 to Ϫ336.
In addition, several liver-enriched transcription factor binding sites were detected, including two sites for HNF1 at Ϫ39 to Ϫ51 and Ϫ165 to Ϫ177, two CCAAT-enhancer binding protein (CEBP-␤) binding sites at Ϫ188 to Ϫ202 and Ϫ473 to Ϫ486, and three HNF3␤ binding sites at Ϫ206 to Ϫ220, Ϫ639 to Ϫ653, and Ϫ829 to Ϫ842. Regarding the HNF1 binding sites, the element nearer the transcriptional start site (nt Ϫ39 to Ϫ51) showed a better match with the consensus sequence for HNF1 binding than the element located further upstream (nt Ϫ165 to Ϫ177). Analysis of Basal OATP-C (SLC21A6) Gene Promoter Activity in Cell Lines of Hepatocellular and Non-hepatocellular Origin-Human OATP-C gene expression is limited strictly to hepatocytes (2). To identify the cis-acting elements in the OATP-C promoter that control liver-specific transcriptional regulation, the hepatocyte-derived cell lines HepG2 and Huh7 and the nonhepatic HeLa cell line were transfected with plasmids C-950, C-424, C-128, and C-66 that contained deletion fragments of the OATP-C promoter sequence (see Fig. 1B). In HepG2 and Huh7 cells, all four constructs conferred significant reporter gene activity as compared with the promoterless pGL3-Basic plasmid (Fig. 2). In contrast to the hepatocytederived cell lines HepG2 and Huh7, no luciferase activity was induced by any of the promoter constructs in HeLa cells. These data supported the notion that a liver-enriched factor present only in HepG2 and Huh7 but not in HeLa cells is required for function of the OATP-C promoter. Whereas the fragment from Ϫ66 to ϩ21 was sufficient to confer residual promoter activity in HepG2 and Huh7 cells, maximum luciferase activity was induced by the fragment from Ϫ424 to ϩ21 (Fig. 2), which likely contains all necessary regulatory elements for full human OATP-C promoter activity in hepatocyte-derived cells.
DNase I Footprint Analysis of the OATP-C Promoter-To identify which factor is responsible for the hepatocyte-specific function of the OATP-C gene promoter, DNA-protein interaction in HepG2 and Huh7 cells was compared with that in HeLa cells by DNase I footprinting. A non-radioactive technique was developed using a 454-bp Cy3 single end-labeled PCR fragment spanning nt Ϫ424 to ϩ30 of the OATP-C gene promoter. This fragment was chosen since the C-424 construct conferred maximum promoter activity in HepG2 and Huh7 cells (Fig. 2). DNase I digestion of the Ϫ424/ϩ30 region, performed in the presence of nuclear extracts from HepG2, Huh7, or HeLa cells, revealed a single region from nt Ϫ37 to Ϫ59 that was protected by HepG2 and Huh7 but not HeLa cell nuclear proteins (Fig. 3). This region contained the consensus recognition site for binding of HNF1. Thus, the footprint pattern discriminated between hepatocyte-derived and non-liver HeLa cells only with respect to the lack of HNF1 site protection by HeLa cell nuclear extracts. No other HepG2 or Huh7 specific footprints were visible in the Ϫ424 to ϩ30 region of the OATP-C promoter.
A second footprint from nt Ϫ111 to Ϫ125 contained the consensus recognition site for binding of nuclear factor 1 (NF-1) (Fig. 3). This sequence was protected by nuclear extracts from all three cell lines, indicating that the transcription factor binding within this region was not hepatocyte-specific. A third footprint localized in the Ϫ206 to Ϫ220 region was also protected by nuclear extracts from all three cell lines and corresponded to the consensus binding site for HNF3␤ (data not shown).
HNF1 Binds to the OATP-C Promoter-The identity of the factor that was found to bind in the Ϫ37/Ϫ59 region by footprint analysis was investigated by electrophoretic mobility shift assays. A 32 P-labeled dimerized oligonucleotide corresponding to nt Ϫ32 to Ϫ58 (wtHNF1) was incubated with nuclear proteins from HepG2, Huh7, and HeLa cells. Binding of nuclear proteins was only found with nuclear extracts from HepG2 and Huh7 cells, resulting in a shifted band (Fig. 4). No binding occurred with HeLa cell nuclear extracts. To study whether the bound protein exhibited the same electrophoretic migration pattern as HNF1, a 32 P-labeled dimerized oligonucleotide with a perfect palindromic HNF1 binding site (perHNF1) was incubated with HepG2, Huh7, and HeLa cell nuclear extracts. The binding pattern of the perfect HNF1 consensus oligonucleotide was identical to that of the OATP-Cderived oligonucleotide, strongly suggesting that the protein binding to the OATP-C sequence was HNF1. The DNA-protein complex was not formed in the presence of an excess of unlabeled perHNF1 oligonucleotide but was unaffected by the presence of excess mutHNF1 oligonucleotide (Fig. 4) with a mutated HNF1 binding site (Table I). These competition experiments were repeated using increasing amounts of unlabeled wild type (wtHNF1) or mutated (mutHNF1) oligonucleotide. As shown in Fig. 5A, reduced binding of HepG2 nuclear extracts to labeled wtHNF1 oligonucleotide was already seen in the presence of a 25-fold excess of unlabeled wtHNF1 oligonucleotide and was maximal with a 200-fold excess, whereas a 100-fold or 200-fold excess of mutated mutHNF1 oligonucleotide had no effect on protein binding.
To confirm the specificity of HNF1 binding, supershift analysis was performed. The addition of antibody specifically recognizing HNF1␣ (17) produced a single supershifted complex (upper arrow in Fig. 5B). For control purposes, a commercial competitive antibody recognizing HNF1 was also employed and completely inhibited DNA-protein complex formation (HNF1 in Fig. 5B). A residual lower band was not shifted by the HNF1␣ antibody and remained visible in the presence of the competitive HNF1 antibody (lower arrow in Fig. 5B). To exclude the possibility that this residual band represented binding of HNF1␤, an antibody targeted against HNF1␤ (19) was employed. The HNF1␤ antibody also failed to shift the lower band, which thus corresponded to an unknown factor. The slight shift of the HNF1␣-DNA complex induced by HNF1␤ antibody was attributable to the known cross-reactivity of the HNF1␤ antibody with HNF1␣. Taken together, the data indicate that the protein complex binding to the OATP-C wtHNF1 sequence is mostly HNF1␣.
In addition to supershift analysis, specificity of HNF1 binding was studied in a transfection assay. HepG2 cells were transfected with the C-128 construct and the HNF1␣ expression plasmid together with 1 or 10 pmol of either the wtHNF1 (Ϫ32/Ϫ58) or the mutHNF1 dimerized oligonucleotides used previously in mobility shift assays. In the presence of 10 pmol of wtHNF1 oligonucleotide, the activity of the C-128 construct was reduced to 32% of the value measured in the presence of 1 pmol wtHNF1 (data not shown). In contrast, cotransfection of the mutHNF1 oligonucleotide had no effect on promoter activity, which was in agreement with the absence of DNA-protein complex formation by mutHNF1 (Figs. 4 and 5A). These data indicate that the Ϫ32/Ϫ58 sequence of the OATP-C promoter binds HNF1␣ not only in mobility shift assays but also in the intact environment of living HepG2 cells.
HNF1␣ Is a Potent Transactivator of the OATP-C Promoter-To investigate whether exogenously expressed HNF1␣ affects OATP-C promoter function in hepatocyte-derived and nonhepatic cells, an expression plasmid coding for HNF1␣ was introduced into HepG2 and HeLa cells together with several OATP-C promoter constructs. As shown in Fig. 6A, cotransfection of the HNF1␣ plasmid enhanced OATP-C promoter-driven luciferase activity 2.7-fold (C-66 construct) to 30-fold (C-424 construct) in HepG2 cells. A similar effect was observed in Huh7 cells (data not shown). Interestingly, cotransfection of HeLa cells with HNF1␣ plasmid produced significant activity of the OATP-C promoter (Fig. 6B). The factor of induction over luciferase activity measured in the absence of HNF1␣ ranged from 9.7-fold (C-66 construct) to 49-fold (C-128 construct). No additional enhancement compared with the C-128 construct was seen with the C-424 construct in HeLa cells. These data indicate that expression of HNF1␣ is sufficient to confer basal promoter activity of the liver-specific OATP-C gene even in the nonhepatic HeLa cell line.
The fact that the C-128 construct was induced more strongly than the C-66 construct may have been attributable to the requirement of an additional essential factor binding to the region between nt Ϫ66 and Ϫ128. This could represent nuclear factor 1, shown by footprint analysis to bind to the Ϫ111 to Ϫ125 region and to be present in nuclear extracts from both HepG2 and HeLa cells (Fig. 3).

Mutagenesis of the HNF1 Binding Site Results in a Complete
Loss of Activation-To assess the importance of HNF1 binding for basal promoter function, mutations were introduced into the HNF1 binding site of the Ϫ128/ϩ21 construct by sitedirected mutagenesis. The resulting reporter gene plasmid mutC-128 contained the same mutations in the Ϫ40/Ϫ49 region as the mutHNF1 oligonucleotide (Table I). Introduction of these mutations had been shown to abolish HNF1 binding in mobility shift assays (Figs. 4 and 5A).
As illustrated in Fig. 7, mutation of the HNF1 binding site resulted in a complete loss of basal promoter activity of the Ϫ128/ϩ21 construct in HepG2 cells. Furthermore, OATP-C promoter activity could no longer be stimulated by cotransfection of HNF1␣ expression plasmid. In HeLa cells, activation of basal activity of the Ϫ128/ϩ21 construct by coexpressed HNF1␣ was similarly absent when the mutC-128 plasmid was used (data not shown). These data provide definite evidence for the critical role of HNF1␣ binding to the OATP-C gene for basal promoter function. single end-labeled 454-bp fragment spanning nt Ϫ424 to ϩ30 relative to the transcription start site was incubated in the presence (HepG2, Huh7, HeLa) or absence (no extract) of nuclear extracts from the indicated cell lines and digested with DNase I. Samples were loaded alongside a sequencing reaction primed by the Cy3-pϩ30 oligonucleotide (A, C, G, T). Compared with the pattern obtained with nonhepatic HeLa cell nuclear extracts, a single protected region spanning nt Ϫ37 to Ϫ59 was found using nuclear extracts from HepG2 or Huh7 cells, the sequence of which is shown (HNF1). A second region spanning nt Ϫ111 to Ϫ125 was protected by nuclear extracts from all three cell lines and corresponded to the consensus recognition site of nuclear factor 1 (NF-1).

FIG. 4.
Electrophoretic mobility shift assay using oligonucleotides of the ؊32/؊58 region and nuclear extracts from HepG2, Huh7, and HeLa cells. The wtHNF1 oligonucleotide (Table I) corresponded to the wild type OATP-C-derived sequence. A DNA-protein complex that resulted in a band shift (arrow) was formed in the presence of HepG2 or Huh7 but not HeLa cell nuclear proteins. The same complex was found using a modified oligonucleotide that contained the perfect HNF1 consensus binding site (perHNF1). Competition experiments using a 500-fold excess of unlabeled perHNF1 oligonucleotide showed complete inhibition of wtHNF1 binding to nuclear proteins, whereas a 500-fold excess of an oligonucleotide that contained a mutated HNF1 recognition site (mutHNF1, see Table I) had no effect.

HNF1␣ Transactivates the Human OATP8 (SLC21A8) and
Mouse Oatp4 (Slc21a6) Promoters-A homologue of OATP-C that exhibits 80% identity at the amino acid level and is also expressed exclusively in hepatocytes is human OATP8 (SLC21A8). In rodents, the only known orthologue of these two closely related human OATPs is Oatp4 (Slc21a6). Both human OATP8 and mouse Oatp4 possess a consensus recognition site for HNF1 in their promoter regions (Table II). To study whether HNF1␣ also affects promoter function of hOATP8 and mOatp4, HepG2 cells were transfected with promoter constructs together with the HNF1␣ expression plasmid. As shown in Fig. 8, coexpression of HNF1␣ increased both hOATP8 and mOatp4 promoter activity 3.5-fold. These data indicate positive regulation not only of OATP-C but also of other liver-specific OATP genes by HNF1␣. DISCUSSION The present study reports the characterization of the promoter region of the human OATP-C (SLC21A6) gene and the identification of a binding site for the liver-enriched transcription factor HNF1␣. Initially the transcription initiation site was localized by 5Ј-RACE and was shown to reside eight nu-  A, competition of HepG2 nuclear protein binding to the 32 P-labeled wtHNF1 oligonucleotide was already evident in the presence of a 25-fold excess of unlabeled wtHNF1 oligonucleotide and was maximal using a 200-fold excess (arrow). In contrast, a 100-or 200-fold excess of mutated mutHNF1 oligonucleotide had no effect. B, supershift analysis using a specific antibody against HNF1␣ (HNF1␣), a commercial competitive antibody against HNF1 (HNF1), or an antibody targeted against HNF1␤ with known cross-reactivity against HNF1␣ (HNF1␤-cr.␣). The HNF1␣ antibody shifted most of the wtHNF1 bound protein complex (top arrow), indicating that HNF1␣ is the major protein binding to this region. cleotides upstream of the published start of the OATP-C cDNA sequence (Fig. 1B) (4). The 5Ј-untranslated region extends over two exons that are separated by a 10,277-bp intron (Fig. 1A). The promoter region contains several consensus recognition sites for both ubiquitously expressed and liver-enriched transcription factors including HNF1, HNF3, CCAAT-enhancer binding protein, and activator protein 1. Basal promoter activity was observed in two hepatocyte-derived cell lines, HepG2 and Huh7, whereas the nonhepatic HeLa cell line did not express reporter gene activity from an OATP-C promoter construct (Fig. 2). To identify which nuclear factor could be responsible for liver-restricted expression of OATP-C, DNase I footprinting was performed. A clear discrimination between HeLa cells and the hepatocyte-derived cell lines was only possible with respect to a single region from nt Ϫ37 to Ϫ59 that was not protected by HeLa cell nuclear extracts (Fig. 3). This region contained the consensus sequence for HNF1. Binding of HNF1 to this region of the OATP-C promoter was confirmed in mobility shift assays (Figs. 4 and 5A) and by supershift analysis using an HNF1␣ antibody (Fig. 5B). No nuclear protein binding was observed using HeLa cell nuclear extracts (Fig. 4). Coexpression of exogenous HNF1␣ stimulated OATP-C promoter activity up to 30-fold in HepG2 cells and even produced basal promoter activity in HeLa cells (Fig. 6). Targeted mutation of the HNF1 binding site abolished not only inducibility of the OATP-C promoter by HNF1␣ but even basal promoter function in HepG2 cells (Fig. 7). These data indicate that HNF1␣ is critical for OATP-C gene expression in human liver.
Human OATP-C is the second basolateral transport system next to rat Ntcp (12) for which direct binding of HNF1␣ to the promoter region has been demonstrated. Both genes appear to be critically dependent upon HNF1␣ for basal promoter func-

FIG. 6. Effect of exogenous HNF1␣ overexpression on OATP-C promoter function in HepG2 and HeLa cells.
Cells were transfected with the indicated OATP-C promoter constructs together with either an HNF1␣ expression plasmid (HNF1␣, black bars) or with the empty pBluescript vector as a control (pB-SK, gray bars). A, coexpression of HNF1␣ led to a 2.7-30-fold increase in OATP-C promoter activity in HepG2 cells. B, HeLa cells showing no basal activity of the OATP-C promoter constructs (Fig. 2) produced significant OATP-C promoter function upon coexpression of HNF1␣. The factor of induction compared with cells not expressing HNF1␣ ranged from 9.7fold (C-66 construct) to 49-fold (C-128 construct). Cotransfection of the pBluescript vector had no effect on reporter activity. Luciferase activity is shown as the factor of induction over background measured in cells transfected with pGL3-Basic alone (Basic). Results are given as the mean Ϯ 1 S.D. of three transfections.  Ϫ70 TAATAAGATGGTTAATCATCAtTGGACT--CAtAAAaaCAA Ϫ76 TgAagagataGTTAATCATCACTGaACT-G-aCAAAAAGCagA Regulation of Liver-specific OATPs by HNF1␣ tion, since (i) mutation of the HNF1 binding site abolishes OATP-C promoter activity in HepG2 cells (Fig. 7) and (ii) the reduction of nuclear activity of HNF1␣ secondary to endotoxinemia in rats is associated with an 86% decrease in steadystate Ntcp mRNA levels (10). Thus, HNF1␣ appears to be essential for bile salt transport across the basolateral hepatocyte membrane. The onset of unconjugated hyperbilirubinemia that accompanies septicemia in man may in part be attributable to an endotoxin-mediated reduction in nuclear activity of HNF1␣ and, consequently, OATP-C gene expression. Several lines of evidence point toward a role for HNF1␣ as a "master regulator" of hepatic organic anion transport proteins. First, HNF1␣ null mice (Tcf1 Ϫ/Ϫ ) exhibit a 55.6-fold reduction of Oatp4 (Slc21a6) expression (11). Second, the promoters of the mouse Oatp4 (Slc21a6) and the human OATP8 (SLC21A8) genes, both of which belong to the "liver-specific" OATP-C (SLC21A6) subfamily, possess consensus binding sites for HNF1 in the Ϫ53/Ϫ66 and the Ϫ47/Ϫ60 region, respectively (Table II). Third, coexpression of HNF1␣ with promoter constructs of the mouse Oatp4 (Slc21a6) or human OATP8 (SLC21A8) genes in HepG2 cells produces significant transactivation (Fig. 8), further supporting the role of HNF1␣ in activating expression of liver-specific basolateral transporters. A list of bile salt and organic anion transporter genes that are regulated by HNF1␣ is shown in Table III.
The importance of HNF1␣ in directing liver-specific gene expression has become evident from the observation that numerous liver-specific genes possess HNF1 binding sites in their promoter or enhancer regions (14,20). These include albumin, ␣ 1 -antitrypsin (14), cytochromes P450 2E1 (21), and 7A1 (22) and the basolateral bile salt transporter Ntcp (12). HNF1␣ is a homeodomain-containing transcription factor that is enriched in liver but is also expressed in kidney, intestine, stomach, and pancreas (15,17,24). The tissue distribution of HNF1␣ indicates that liver specificity of gene expression cannot be accomplished by HNF1␣ alone but rather by the concerted action of a selected group of liver-enriched proteins including the HNF1, HNF3, HNF4, and CCAAT-enhancer binding protein families of transcription factors (25,26). The OATP-C promoter was found to contain consensus recognition sites for several of these factors (Fig. 1B), which in combination are the likely effectors of liver-specific OATP-C expression.
A variant form of HNF1 termed HNF1␤ or vHNF1 has the same DNA binding specificity as HNF1␣. Whereas HNF1␣ functions mainly in terminally differentiated cells and has been implicated in the maintenance of a correct differentiation state (15,17), HNF1␤ is essential much earlier during development and is a key factor in visceral endoderm differentiation (14,19). The OATP-C-derived HNF1 sequence that binds HNF1␣ was shown not to bind HNF1␤ (Fig. 5B). HNF1␤ is expressed at only low levels in the liver and cannot fulfil the function of HNF1␣ in vivo, as suggested (i) by the severe phenotype of HNF1␣ null mice that exhibit marked liver enlargement, elevated plasma transaminases, hypercholesterolemia, and hyperbilirubinemia (15,24,27) and (ii) by the observation that Oatp4 mRNA levels are strongly decreased in mice with a disrupted HNF1␣ gene (11), but are normal in mice with a conditional liver-specific HNF1␤ inactivation. 3 The genomic organization of the OATP-C (SLC21A6) gene is similar to that of human OATP-A (SLC21A3) and OATP8 (SLC21A8) (5) and of mouse Oatp2 (Slc21a5) and Oatp4 (Slc21a6) (28,29). The human OATP-A, OATP-C, and OATP8 genes are all localized on chromosome 12p12. The only other mammalian OATP promoter that has been characterized to date is the human OATP-A (SLC21A3) promoter (30). Unlike OATP-C, the OATP-A gene does not contain an intronic sequence within its 5Ј-untranslated region, and the promoter is functional in both hepatocyte-derived and nonhepatic cell lines such as CHO-K1 1 and MDCK 1 . Based on sequence analysis and in accordance with its low expression level in the liver (23), no binding site for HNF1 is contained in the immediate 5Ј-flank-3 L. Gresh and M. Yaniv, unpublished observations. FIG. 8. Transactivation of the human OATP8 (SLC21A8) and mouse Oatp4 (Slc21a6) promoters by HNF1␣. HepG2 cells were transfected with the 8-120 (hOATP8) or m4-137 (mOatp4) promoter constructs or with the promoterless pGL3-Basic vector (Basic) together with either an HNF1␣ expression plasmid (HNF1␣, black bars) or with the empty pBluescript vector as a control (pB-SK, gray bars). Coexpression of HNF1␣ led to a 3.5-fold increase in both human OATP8 and mouse Oatp4 promoter activity. Cotransfection of the pBluescript vector had no effect on reporter activity. Luciferase activity is shown as the ratio firefly/Renilla luciferase (see "Experimental Procedures"), and data represent mean of Ϯ 1 S.D. of three transfection experiments. ing region of the OATP-A gene (30).
In conclusion, this study provides evidence for the critical role of HNF1␣ in directing basal expression of the human OATP-C (SLC21A6) gene promoter. The finding that HNF1␣ binds to the OATP-C promoter and transactivates other liverspecific OATP promoters supports the notion that HNF1␣ is an essential factor in mediating hepatocyte-specific gene expression and in maintaining the differentiated hepatocellular phenotype. Perhaps most importantly, it lends support to the hypothesis that HNF1␣ is a global regulator of hepatic uptake of bile salts and bilirubin from systemic blood.