Introduction

Top Summary Introduction Experimental Procedures Results Discussion References

In mammals, detoxification of the hydrophobic bilirubin occurs mainly in the liver via its conjugation with uridine diphosphate glucuronic acid. This reaction, which is catalyzed by microsomal UDP-glucuronosyltransferases (UGTs) (EC 2.4.1.17) produces hydrophilic bilirubin-mono- and diconjugates that are excreted at high rates in the bile (Hauser and Gollan, 1990; Jansen et al., 1992). UGTs are produced by a multigene superfamily with specificity for a variety of endogenous substrates and xenobiotics. Mammalian UGTs are subdivided into two groups, the UGT1 and the UGT2 families. UGT1 family members are encoded by a particularly complex gene of about 500 kb ("the UGT1 cluster," Mackenzie et al., 1997). It consists of multiple related homologous unique first exons (exon 1), which encode the isoform specific N terminus of the enzymes. The C-terminal part is encoded in a single set of four exons (exons 2-5), which is common to all UGT1 isoforms (Ritter et al., 1992; Emi et al., 1995; Mackenzie et al., 1997). A promoter region with its regulatory elements is thought to be located upstream of each unique first exon, but, for several of them, this region has not yet been characterized (Ritter et al., 1992).

The UGT1A1 isoform is likely to be the major isoform involved in bilirubin detoxification in human (Bosma et al., 1994). Inherited disorders that decrease or suppress UGT1A1 expression result in unconjugated bilirubin accumulation. Mild forms of unconjugated hyperbilirubinemia are known as Gilbert and Crigler-Najjar type II syndromes, whereas in more severe forms, the Crigler-Najjar type I disease can be fatal (Aono et al., 1995; Bosma et al., 1995). Some of the mutations causing these disorders are localized in the common exons 2 to 5, leading to an alteration of all isoforms produced by the UGT1 locus (reviewed in Mackenzie et al., 1997). Others are in exon 1A1 and thus alter only UGT1A1 activity. Thus far, the UGT1A1*28 polymorphism is the only genetic variation reported in a regulatory region of the gene. It is responsible for most Gilbert's disease cases (Bosma et al., 1995). This polymorphism concerns the six-thymidine/adenine (TA) track in the wild-type UGT1A1 promoter, which is replaced by five, seven, or eight TA repeats in the altered genes, whose activity is inversely correlated with the length of the track (Beutler et al., 1998). Expression of the human UGT1A1 mRNA is primarily in the liver but can also be detected in biliary and colon epithelia (Ritter et al., 1992; Strassburg et al., 1997, 1998). In rat, the tissue distribution is similar in liver and intestine. In addition, UGT1A1 is also expressed at low levels in kidney, which is not the case in human (Ritter et al., 1992; Emi et al., 1995).

Surprisingly, only the promoter of the human UGT1A1 gene has been sequenced thus far. However, with the exception of the TA track, no functional promoter characterization has been performed (Brierley et al., 1996). The broad use of mice and rats as laboratory animals for pharmacological studies has prompted us to investigate UGT1A1 transcriptional regulation in the mouse, not least because the gene knockout technology is now well mastered for this animal and might be applied later to the UGT1 locus once characterized. Moreover, studies using this technology provided evidence for a role of liver-enriched transcription factors in UGT1A1 expression. The inactivation of the HNF1 gene resulted in a mild hyperbilirubinemia, which, according to the above-mentioned observations, might reflect an altered UGT1A1 gene expression (Pontoglio et al., 1996). Similarly, knock-out of C/EBP in mouse liver affects UGT1A1 gene expression resulting in a severe toxic hyperbilirubinemia (Lee et al., 1997). These observations, together with UGT1A1 being the main isoform from the UGT1 locus expressed in rat liver (Emi et al., 1995), argue strongly for a crucial role of the UGT1A1 isoform in bilirubin detoxification in rodents. Recently, other studies have highlighted the important role of HNF1 and C/EBP in regulating the transcription of the rat UGT2B1 isoform (Hansen et al., 1997, 1998). In contrast, nothing is presently known of the mechanisms of transcriptional regulation of the UGT1A1 gene by these factors, although their involvement has been suggested, as mentioned above.

To better understand the physiological role of the UGT1A1 isoform in the mouse, we have isolated the UGT1A1 gene. Its promoter was then compared with that of the human gene, which revealed the lack of a TATA box in the mouse promoter. Furthermore, evidence is provided for a direct involvement of HNF1 in the regulation of the mouse and human UGT1A1 promoters.

	Experimental Procedures

Top Summary Introduction Experimental Procedures Results Discussion References

Materials. The expression plasmids pRSV-HNF1 and pRSV-HNF1 were a kind gift from Dr. Moshe Yaniv (Unité des Virus Oncogènes, Institut Pasteur, Paris, France). The -HCt-284 polyclonal antiserum directed against HNF1 (Chouard et al., 1997) was a kind gift from Dr. Marco Pontoglio (Unité des Virus Oncogènes, Institut Pasteur, Paris, France) and Dr. Moshe Yaniv. All UGT1A1 nucleotide sequences used were numbered relative to the major transcription start site determined in this study.

Isolation and Characterization of Genomic and cDNA Clones. A portion (1 × 10⁶ plaque-forming units) of a mouse L129/SVJ genomic DNA library in FIXII (Clontech Laboratories, Palo Alto, CA) were screened with two oligonucleotides complementary to the mouse UGT1A1 cDNA (Chu et al., 1997) as probes. Two clones (mPGT1A1-1, mPGT1A1-2) were isolated, plaque-purified, and characterized by polymerase chain reaction (PCR) and restriction enzyme analysis. From the mPGT1A1-1 clone, a 2.4-kb XbaI restriction fragment containing the UGT1A1 first exon was subcloned in the pBluescript KSII vector, yielding the pKS-1A1 genomic subclone. DNA was sequenced on both strands by the dideoxy termination method using T7 DNA polymerase (T7 sequencing kit; Pharmacia Biotech Europe, Dübendorf, Switzerland). The cDNA clones were obtained by screening 1 × 10⁶ plaque-forming units of an adult mouse liver 5'-extended cDNA library in gt10 with the same oligonucleotides as indicated above, yielding 10 independent clones. Similarly, a 15-day-old embryonic mouse 5'-extended cDNA library in gt10 was screened following the same procedure, yielding one clone. After PCR amplification using phage-specific oligonucleotides, the 11 independent clones were subcloned into the pBluescript vector for subsequent sequencing. The human UGT1A1 promoter was cloned by PCR on human genomic DNA and subcloned into the pBLCAT3 vector yielding the 617/+15 hUGT1A1-chloramphenicol acetyltransferase (CAT) construct. The construct was sequenced on both strands. Its sequence is identical with that published previously (Brierley et al., 1996) except for the length of the TA repeat in the polymorphic TATA box, which is (TA)₅TAA instead of (TA)₆TAA (Beutler et al., 1998).

RNA Isolation and Primer Extension Analysis. Total RNA was isolated from mouse liver by the phenol and guanidine isothiocyanate method following the supplier's instructions (Trizol reagent; Life Technologies, Basel, Switzerland). An end-labeled antisense oligonucleotide primer complementary to nucleotides +95/+120 of the UGT1A1 gene was hybridized to 50 µg of total RNA for 1 h at 42°C in 50 mM Tris-HCl (pH 8.3), 8 mM MgCl₂, 30 mM KCl, 10 mM DTT in a final volume of 50 µl. The primer-annealed RNA was then subjected to extension by adding 2 mM each dNTP and 25 units of avian myeloblastosis virus-reverse transcriptase (Pharmacia Biotech Europe, Dübendorf, Switzerland) for 1 h at 42, 44, or 46°C. The reaction was stopped by addition of 50 mM EDTA and 15 µg of Rnase Avian myeloblastosis virus (Boehringer Manheim, Rotkreuz, Switzerland) and incubated for 1 h at 37°C. The reaction products were resolved on a 7 M urea, 8% polyacrylamide-1× Tris-borate-EDTA sequencing gel.

RNase Protection. A DNA fragment corresponding to nucleotide 76/+336 of the UGT1A1 gene was generated by PCR (Expand High Fidelity PCR system; Boehringer Mannheim, Rotkreuz, Switzerland) using the pKS-1A1 genomic subclone as a template and cloned into the pBluescript vector to obtain the pKS-RPA-1A1 construct. A 293-bp radiolabeled antisense riboprobe was generated from this plasmid. The riboprobe (0.5 × 10⁵ cpm) was mixed with 20 µg of total mouse liver RNA in 30 µl containing 40 mM PIPES (pH 6.4), 1 mM EDTA, 400 mM NaCl, and 80% formamide. After overnight hybridization at 42°C, the samples were digested with 300 µl of RNase mix containing 300 mM NaCl, 10 mM Tris-HCl (pH 7.4), 5 mM EDTA, 2 µg/ml RNase T1 (Life Technologies, Inc., Basel, Switzerland), and 40 µg/ml RNase A (Boehringer Mannheim) for 30 min at 30°C. The reaction was stopped by addition of 100 µg of proteinase K (Boehringer Mannheim). The samples were resolved on a 7 M urea, 8% polyacrylamide-1× Tris-borate-EDTA sequencing gel.

Generation of UGT1A1 Promoter Deletion Constructs and Site-Directed Mutagenesis. Four reporter genes encompassing different lengths of the UGT1A1 promoter were generated by PCR using the pKS-1A1 genomic subclone as a template. The 5'-sense oligonucleotide primers used were complementary to the nucleotides 455 to 436, 93 to 74, and 44 to 25 of the cloned genomic sequence. The longest construct (from 1,074) was obtained using a vector-specific oligonucleotide primer. The common 3'-antisense oligonucleotide used was complementary to nucleotides 2 to +19 of the cloned genomic sequence. The PCR products were subcloned into the promoterless vector pBLCAT3, yielding the following plasmids: 1074/+19-mUGT1A1-CAT; 455/+19-mUGT1A1-CAT; 93/+19-mUGT1A1-CAT; and 44/+19-mUGT1A1-CAT. Point mutations were introduced into the 455/+19 mUGT1A1-CAT plasmid (designated WT-mUGT1A1-CAT in Fig. 8), at specific regulatory elements by PCR using the QuikChange site-directed mutagenesis kit (Stratagene, Basel, Switzerland) following the manufacturer's instructions. The three mutant plasmids were designated HNF1-mUGT1A1-CAT, Sp1-mUGT1A1-CAT, and Sp1/HNF1-mUGT1A1-CAT; the integrity of each was validated by sequencing. The same procedure was used to mutate the HNF1 site of the 617/+15-hUGT1A1-CAT plasmid (designated WT-hUGT1A1-CAT in Fig. 8), yielding HNF1-hUGT1A1-CAT plasmid.

Preparation of Nuclear and Whole-Cell Extracts (WCEs). Nuclear extracts from mouse liver were prepared as described (Dignam et al., 1983). Briefly, the final suspension buffer containing 20 mM HEPES (pH 7.9), 20 mM KCl, 2 mM MgCl₂, 0.2 mM EDTA, 0.5 mM DTT, 20% glycerol, and protease inhibitors (Complete; Boehringer Mannheim) was aliquoted, frozen in liquid nitrogen, and stored at 70°C. WCE was prepared from transfected and nontransfected human embryonic kidney (HEK) 293 cells by disrupting the cell membrane with three steps of freezing/thawing. The residual debris were pelleted by centrifugation for 10 min at 12,000g at 4°C. The extracts were aliquoted and stored at 70°C.

DNase I Footprint Analysis. The DNA fragment from 253 to +81 of the proximal UGT1A1 promoter sequence was subcloned into the pBLCAT3 vector. This plasmid possesses restriction sites to generate probes, with either the coding or noncoding strands labeled. End-labeled DNA fragments (3 × 10⁴ cpm) were incubated with 10 to 50 µg of mouse liver nuclear extracts (MLNEs) or with 10 µg of whole HEK293 cell extracts, and when indicated, a 100-fold molar excess of oligonucleotide competitor was added. After 30 min of incubation on ice, the mixtures were digested with 1.5 U of RQ1DNase I (Promega, Madison, WI) for 1 min at room temperature. The reaction was stopped by addition of 200 µl of 50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 2% SDS, 0.4 mg/ml proteinase K, and 100 µg/ml glycogen. Samples were separated on a 7 M urea, 8% polyacrylamide-1× Tris-borate-EDTA sequencing gel. The position of specific DNase I-protected fragments was determined by chemical sequencing of the probe (Maxam and Gilbert, 1977).

Electrophoretic Mobility-Shift Assay (EMSA). EMSAs were performed as described (Ijpenberg et al., 1997) with either 10 µg of MLNEs or 2.5 µg of whole HEK293 cell extracts. For supershift experiments, 1 µl of HNF1-specific antibody was used. Competition assays were performed by adding a 100-fold excess of unlabeled double-stranded oligonucleotide to the reaction 10 min before the addition of labeled probe.

Cell Culture and Transfection. HEK293 cells were seeded at 3 × 10⁵ cells/well in six-well plates, cultured overnight, and transfected with plasmid DNAs by the calcium phosphate-DNA coprecipitation method (Jordan et al., 1996). The reporter deletion constructs were at 250 ng/well, and expression plasmids for HNF1 (pRSV-HNF1) and HNF1 (pRSV-HNF1) were each at 500 ng/well. Cytomegalovirus--galactosidase expression vector (50 ng/well; Stratagene, Basel, Switzerland) was included as an internal standard for transfection efficiency. After 12 h, the DNA precipitates were removed, and the cells were further incubated in fresh culture medium for 24 h. The CAT activity was assayed using ¹⁴C-labeled chloramphenicol (Amersham Pharmacia Biotech, Dübendorf, Switzerland) and n-butyryl coenzyme A substrates. The -galactosidase activity was determined as described (Sambrook et al., 1989). All transfections were performed in duplicate. Activities reported are an average ± S.D. of three independent experiments.

	Results

Top Summary Introduction Experimental Procedures Results Discussion References

Cloning of the Mouse UGT1A1 Gene. A mouse genomic DNA library was screened using two oligonucleotides derived from the previously reported mouse UGT1A1 cDNA (Chu et al., 1997). Two clones (mPGT1A1-1, mPGT1A1-2) were isolated and characterized (Fig. 1). Together, they form a contig of about 25 kb, but only clone mPGT1A1-1 (19-kb insert) contains the entire gene, including several kb of 5'- and 3'-flanking sequences. The 5'-end of clone mPGT1A1-2 (20-kb insert) lies within the isoform specific exon 1A1. This cloned genomic DNA fragment also contains all common exons and extends far into the 3'-flanking region of the gene. The common exons 2, 3, 4, and 5 of the UGT1 locus are localized about 2 kb downstream of exon 1A1, and their position relative to each other as determined by PCR analysis (Fig. 1) is similar to that observed in rat and human (Ritter et al., 1992; Emi et al., 1995).

View larger version (9K): [in this window] [in a new window]   Fig. 1.   Structure of the mouse UGT1A1 locus. Schematic representation of the mouse UGT1A1 gene deduced from the characterization of the two genomic DNA clones mPGT1A1-1 and mPGT1A1-2 that are shown. Restriction enzyme cleavage sites were mapped only in the region given in bold. The position of the exons was determined by restriction enzyme mapping and PCR analysis of both clones. Only the mPGT1A1-1 clone contains the entire gene with the complete isoform-specific exon 1A1 and promoter sequences. X, XbaI; S, StuI; E, EcoRI; B, BamHI.

View larger version (45K): [in this window] [in a new window]   Fig. 2.   Determination of the transcription start sites of the mouse liver UGT1A1 gene. A, sequence of the 5'-UTR of the UGT1A1 gene with the transcription initiation sites (T1, T2, *) and the position of the 5'-end of 11 cDNA clones (vertical arrows). B, primer extension analysis was carried out at three different temperatures using a 32P-end-labeled primer complementary to nucleotides +95/+120 of the UGT1A1 exon annealed to either 50 µg of total mouse liver RNA (lanes 2-4) or yeast tRNA as a control (lane 1). The primer-extended products were analyzed on an 8% denaturing polyacrylamide gel in parallel with sequencing reactions (lanes 5-8) carried out on the genomic subclone using the same primer as above. Arrows indicate the extended products. T1 and T2 indicate the major and minor transcription initiation site, respectively; Short-stop, prematurely ended extension products (see Results). C, RNase protection assay using an antisense riboprobe corresponding to nucleotides 76/+336 of the UGT1A1 gene annealed to 20 µg of total mouse liver RNA (lane 5) or yeast tRNA (lane 6) as control. After RNase digestion, protected fragments were resolved on an 8% polyacrylamide gel together with sequencing reactions (lanes 1-4). T1, T2, and * indicate the major, minor, and weak transcription initiation site, respectively. In positioning these initiation sites on the sequencing ladder, a correction of 1 base has been made because of the lower mobility of RNA than DNA of the same length (Sambrook et al., 1989). The open arrow indicates residual nondigested riboprobe.

Determination of the Mouse UGT1A1 Transcription Start Site. The transcription start site of the mouse UGT1A1 gene was identified by primer extension and RNase protection assays. Primer extension using mouse liver RNA as template was performed at 42, 44, and 46°C (Fig. 2B). Three extension products were obtained in different relative amounts, depending on the reaction temperature. The upstream end of the short-stop product is in the coding sequence in a region with potential secondary structure, which explains a decrease in the amount of this product at higher temperature in favor of the longer extension products. The upstream end of these latter products is 34 and 38 bp upstream of the ATG translation start codon at sites we named T1 and T2, respectively. T1 is considered the major transcription start site (+1) because it appears stronger than T2 at all three temperatures used in the extension assay (see Fig. 3A). The 5'-end of the longest cDNA clone we have obtained is three nucleotides upstream of T2 and most likely corresponds to a minor transcription start site. To confirm the position of transcription initiation, we carried out RNase protection analysis with mouse liver RNA and a radiolabeled antisense riboprobe, which corresponds to the region between nucleotides 76 and +336 with respect to the major transcription start site T1. The protected fragments also indicate transcriptional start sites at positions T1 and T2 (Fig. 2C). The bands between these positions in Fig. 2C most likely reflect an incomplete digestion of the riboprobe-RNA duplexes (as indicated by the presence of residual nondigested riboprobe; open arrow). A minor fragment indicating the same start site as that of the longer cDNA clone could be seen after longer exposure (* in Fig. 2, A and C). All three initiation start sites (*, T1, T2) are shown with respect to the 5'-ends of the 11 cDNA clones mentioned above (Fig. 2A) and in the context of the promoter sequence in Fig. 3A.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Characteristics of the mouse and human UGT1A1 promoter. A, nucleotide sequence of the mouse UGT1A1 promoter. Nucleotides are numbered with respect to the major transcription start site as determined herein (T1). The minor and a very weak transcription start sites are indicated with a small arrow (T2) and an asterisk, respectively. The translation-initiation codon ATG is boxed, and the first eight amino acids are given. The nucleotide sequence corresponding to the L1 retrotransposon element is indicated in italics. Footprinted regions A, B, C, and D obtained in Fig. 4 are indicated in bold letters. B, comparison of human (Brierley et al., 1996) and mouse UGT1A1 promoter sequences. The human UGT1A1 nucleotide sequence (top) was aligned with its mouse counterpart (bottom) using the Gap program from the GCG package (Wisconsin Package Version 9.1; Genetics Computer Group, Madison, WI). The transcription initiation site (T1) of the mouse gene and the conserved HNF1 and Sp1 sites are indicated on both mouse and human sequence. The putative human TATA box is indicated. The transcription-initiation site of the human promoter (open arrow) has been determined previously (Ritter et al., 1992).

Structural Features of the Mouse UGT1A1 TATA-Less Promoter. Analysis of the promoter region of the mouse UGT1A1 gene revealed several structural features of interest, some of them highlighted for comparison with the human sequence (Fig. 3, A and B). The mouse promoter has neither a TATA box (Matsui et al., 1980) nor an identifiable initiator-type sequence, often found in eukaryotic TATA-less promoters at the transcription start site (Emami et al., 1998). In contrast, the human promoter has a TATA box-like element associated with the polymorphic TA repeat responsible for some of the cases of Gilbert's disease (Clarke et al., 1997). Otherwise, this TATA box has not yet been characterized functionally.

Liver Nuclear Proteins Bind to the Mouse UGT1A1 Promoter. UGT1A1 has a major detoxification function in the liver; therefore, it was of interest to test whether a factor present in MLNEs is able to bind to the putative HNF1 and Sp1 sites described above. For this purpose, we performed a DNase I footprinting analysis. As shown in Fig. 4, a total of four independent protected areas, designated A to D, were identified in the proximal promoter region (253/+81; noncoding strand). Footprints A (+32/+41) and B (20/+24) are close to and comprise the transcription initiation site, respectively; they are separated by DNase I-hypersensitive sites (*). It is reasonable to speculate that these footprints result from the binding of factors belonging to the machinery responsible for transcription initiation, and therefore they were not further characterized herein. Footprint C (68/40) corresponds to the putative HNF1 recognition site and footprinted region D between nucleotides 145 to 122, comprising the putative Sp1 site to which purified Sp1 can indeed bind in gel-retardation assays, as will be shown later. The same results were obtained with the coding strand (data not shown). These data, together with the observation that these binding sites are well conserved in the mouse and human promoters, suggested that these factors may play a role in UGT1A1 regulation.

View larger version (47K): [in this window] [in a new window]   Fig. 4.   Liver nuclear protein-binding sites within the mouse UGT1A1 promoter. DNase I footprinting experiment carried out with a 253/+81-proximal UGT1A1 promoter fragment (noncoding strand) and increasing amounts of MLNEs. , DNase I digestion of the probe without added protein; +, DNase I digestion of the probe with 50 µg of purified bovine serum-albumin. Amounts of nuclear extracts used are 10 µg (lane 5), 20 µg (lane 6), 30 µg (lane 7), and 40 µg (lane 8). Protected areas are indicated on the right of the autoradiogram. DNase I-hypersensitive sites present between the A and B regions are indicated by *. Also shown are chemical sequencing ladders G and G+A (lanes 1 and 2).

The Mouse UGT1A1 Promoter Is Transactivated by HNF1. The functional role of HNF1 was assayed by cotransfection assays performed with HNF1 and HNF1 expression vectors and a CAT reporter gene driven by the mouse UGT1A1 promoter comprising decreasing lengths of 5'-upstream sequences (1074/+19, 93/+19, 44/+19). The empty expression vector (Rous sarcoma virus) and promoterless reporter CAT vector (pBLCAT3) were used as controls. The recipient cells were the HEK293 cells in which HNF1 cannot be detected. The results show that both HNF1 and HNF1 are able to transactivate both the 1074/+19 mUGT1A1-CAT reporter gene and the 93/+19 mUGT1A1-CAT construct, both of which contain the HNF1 site (Fig. 5). The minimal promoter 44/+19 still has a relatively high basal promoter activity, a noteworthy feature for a TATA-less promoter. Some residual HNF1-dependent stimulatory activity was seen with this minimal promoter that is thought to be due to vector sequences brought close to the promoter as a result of upstream promoter deletion (see results with the promoterless CAT-vector). Instead of analyzing this effect further, we performed site-directed mutagenesis of the HNF1 site in the context of a relatively large promoter region (455/+19). A complete loss of HNF1-dependent stimulation of the mutated promoter confirmed that the HNF1 element is responsible for mediating induction by the liver-enriched transcription factor (Fig. 5, HNF1-mUGT1A1-CAT). We conclude that HNF1 is a transactivator of the UGT1A1 gene and that the stimulatory effect is mediated by the HNF1 binding site located at 68/40. It is noteworthy that removal of L1 element sequences did not influence HNF1 responsiveness.

View larger version (12K): [in this window] [in a new window]   Fig. 5.   HNF1 and HNF1 transactivate the mouse UGT1A1 promoter. HEK293 cells were cotransfected in six-well plates with 0.25 µg per well of the different deletion and mutant reporter constructs and 0.5 µg of expression vector for HNF1 (RSV-HNF1), HNF1 (RSV-HNF1), or empty vector (RSV) as a control. A pCMV-galactosidase (50 ng per well) expression vector was included as an internal control of transfection efficiency. Results are the mean of at least three independent experiments ± S.D. Deletion constructs are depicted, with the L1 element indicated as a black bar, and the position of the Sp1 and HNF1 sites are given according to the results of Fig. 4. In HNF1-mUGT1A1-CAT, the mutated HNF1 site is represented by a cross; this result is taken from Fig. 8. pBLCAT3 was used as a promoterless control reporter gene.

HNF1 Binds to the UGT1A1 Promoter. To assess whether the HNF1-mediated transactivation is due to a direct binding of the factor to the HNF1-binding site in the promoter, as suggested by the results presented above, we carried out EMSAs using the UGT1A1-HNF1 site as a probe. We first confirmed that a factor in the MLNE binds to this site, as shown previously by the DNase I footprinting experiment. A major retarded complex (C1) was detected in the presence of the nuclear extract (Fig. 6A, lane 2). This complex was competed away by a molar excess of a HNF1-specific oligonucleotide added to the binding reaction (lane 5), whereas binding sites for other transcription factors were unable to displace the complex (lanes 3, 4, 6, and 7). We also observed a faster-migrating minor complex (C2). This C2 complex most likely contains HNF3, because it was specifically displaced by a molar excess of HNF3 sites, which is in accordance with previously reported results indicating that HNF3 may bind to HNF1 sites (Gregori et al., 1993). Figure 6B shows that the C1 complex was disrupted by a polyclonal antibody directed against HNF1 (lane 3), allowing a reinforcement of C2 complex formation. The nonspecific band observed with the antibody alone (lane 4) has been previously reported (Chouard et al., 1997). Together, these results provide strong evidence that C1 is due to the occupancy of the UGT1A1-HNF1 site by HNF1, which is abundant in liver (Rey-Campos et al., 1991).

View larger version (41K): [in this window] [in a new window]   Fig. 6.   HNF1 proteins bind to the C site of the mouse UGT1A1 promoter. A, EMSA experiments using a MLNE were performed with a radiolabeled oligonucleotide corresponding to site C (HNF1-binding site) of the UGT1A1 promoter in the presence or absence of a 100-fold excess of different competitor oligonucleotides as indicated at the top of the autoradiogram. For the identification of C1 and C2 complexes, see text. * is an unspecific band. B, supershift experiments in which EMSA with the radiolabeled site C oligonucleotide was carried out with MLNE in the presence or absence of a specific antibody directed against HNF1 (Chouard et al., 1997). * is an unspecific band, as in A. ** is an unspecific band generated by the antibody, as described (Chouard et al., 1997). C, sequence of the different wild-type and mutated HNF1-binding sites comprised in the oligonucleotides used in EMSA, as compared with the consensus HNF1-binding site. Nucleotides corresponding to the consensus binding site are underlined in ALB-HNF1, UGT1A1-HNF1, and mut UGT1A1-HNF1. The matching index is given in brackets. D, EMSA using overexpressed HNF1 or HNF1 in HEK293 cells was performed with radiolabeled site C as probe (see above) in the presence of 100- or 500-fold excess of mutant site C (mutUGT1A1) or wild-type site C (UGT1A1) competitor oligonucleotides.

HNF1 and Sp1 Do Not Cooperate for the Control of the UGT1A1 Promoter. The footprinting experiment using MLNEs revealed a binding site for Sp1 in the mouse UGT1A1 promoter, which is conserved in the human promoter, 54 bp upstream of the HNF1 site. This observation was confirmed in EMSA experiments with whole HEK293 cell extracts as well as with purified Sp1 for both the mouse and human Sp1 sites (data not shown).

View larger version (98K): [in this window] [in a new window]   Fig. 7.   Noncooperative binding between Sp1 and HNF1 within the proximal mouse UGT1A1 promoter. DNase I footprinting was carried out using a 253/+81 UGT1A1 promoter fragment (noncoding strand) and protein extracts from mouse liver and HEK293 cells as indicated at the top of the gel in the presence of 100-fold excess of different competitor oligonucleotides. Mock, WCE from mock-transfected HEK293 cells; HNF1, WCE from HNF1-transfected HEK293 cells; HNF1, WCE from HNF1-transfected HEK293 cells. , DNase I digestion of the probe without added protein; +, DNase I digestion of the probe with 50 µg of purified BSA; H, HNF1 consensus oligonucleotide; S, Sp1 consensus; C, C/EBP consensus oligonucleotide. Protected areas corresponding to the Sp1- and HNF1-binding sites are indicated on the right of the gel. The sequencing ladder tracks G and G+A are indicated.

View larger version (14K): [in this window] [in a new window]   Fig. 8.   Absence of functional interaction between HNF1 and Sp1 in transactivation of the mouse UGT1A1 promoter. HEK293 cells were cotransfected in six-well plates with 0.25 µg per well of different mutant reporter constructs and 0.5 µg of expression vector for HNF1 (RSV-HNF1), HNF1 (RSV-HNF1), or empty vector as control (RSV). A pCMV-galactosidase (50 ng per well) expression vector was included as an internal control for transfection efficiency. Results are the mean of at least three independent experiments ± S.D. The reporter constructs used are depicted; mutated Sp1, and HNF1 sites are indicated by a cross. pBLCAT3 was used as promoterless control reporter.

	Discussion

Top Summary Introduction Experimental Procedures Results Discussion References

Glucuronidation of bilirubin is essential for its excretion from the liver into the bile. In human, UGT1A1 is the major UGT isoform involved in this process (Bosma et al., 1994). Consequently, alteration in UGT1A1 gene expression or mutations in the coding region of this gene can lead to a complete or partial absence of bilirubin glucuronidation activity, which results in toxic accumulation of unconjugated bilirubin (Hauser and Gollan, 1990). One of the ways to study and better understand bilirubin metabolism in human is to compare it to other mammalian species, such as the increasingly used mouse model. For this purpose, we have cloned and functionally characterized the mouse UGT1A1 promoter. Important structural differences with the human promoter were found. Surprisingly, the mouse UGT1A1 promoter is TATA-less and consequently does not contain any of the TA repeats found in human. In contrast, both human and mouse promoter activities are stimulated by HNF1 and HNF1 via a functional HNF1 site particularly well conserved between the two promoters.

The Mouse UGT1A1 Promoter Lacks a TATA Box. Two major new features occur only in the mouse UGT1A1 promoter in comparison to the human promoter. First, the mouse UGT1A1 promoter has no apparent TATA box, as observed previously for the rat UGT1A7 promoter (Metz and Ritter, 1998). In contrast, other UGTs belonging to family 1 (Emi et al., 1996) as well as to family 2 (Mackenzie and Rodbourn, 1990) possess a TATA-like element, indicating that there is no common promoter structure for all UGT isoforms. It is noteworthy that the rat UGT1A7 promoter has multiple transcription start sites clustered within 45 bp, whereas the mouse UGT1A1 has only two independent well detectable transcription start sites clustered in less than 10 bp, which can be detected by analysis of liver transcripts. Despite the lack of a TATA-binding protein binding site, it is clear that the mouse UGT1A1 transcriptional start site is well defined, suggesting that binding of transcription factors to this promoter directly interacts with the basal transcription machinery to help in positioning RNA polymerase II. HNF1 indeed is one of the transcription factors that have been shown to interact directly with components of the general transcription apparatus, such as transcription factor IID or transcription factor IIB (Ktistaki and Talianidis, 1997). The second characteristic feature of the mouse promoter is the presence of a retrotransposon-like sequence element, which belongs to the L1 family, at 480 bp upstream the transcription start site. The L1 element is found at more than 100,000 copies in mouse and represents about 10% of the genome (Smit et al., 1995). However, the majority of the L1 elements are truncated and not active in transposition (Kolosha and Martin, 1995). This is the case for the element identified in the mouse UGT1A1 promoter because its lacks the 5'-long terminal repeat. The recent characterization of the apolipoprotein(a) enhancer in mouse, which resides within a L1 element, illustrates the possibility of a regulatory action of these elements on neighboring promoters (Yang et al., 1998). However, further studies are needed, especially in hepatocytes rather than in HEK293 cells, to evaluate the relevance, if any, of this L1 element insertion for tissue-specific activity of the mouse UGT1A1 gene.

HNF1 Binds to and Activates the Mouse UGT1A1 Promoter. We have demonstrated that both HNF1 and HNF1 bind to the mouse UGT1A1 promoter and are able to transactivate it. However, because HNF1 is less expressed in adult mouse liver than HNF1, it is very likely that it is the latter factor that plays a regulatory role in this organ. Our results suggest that HNF3 also binds to the HNF1 site, although the signal of HNF1 is much stronger than that of HNF3 in liver nuclear extracts. It will be interesting to investigate whether there is a competition between these two factors relevant for expression of this gene in liver. The HNF1-binding site of the mouse UGT1A1 promoter is very similar to that of the rat albumin promoter, which displays a very well conserved half-site and a more divergent one with respect to the consensus sequence (Cereghini, 1996). Consistent with previous studies on the albumin-HNF1 site, HNF1 proteins failed to bind and transactivate the mouse UGT1A1 promoter when the conserved half-site is mutated. Our study provides strong evidence for HNF1 and HNF1 being involved in the expression of this member of the UGT superfamily in rodents. In adult mouse, HNF1 and HNF1 are expressed in the same tissues (kidney, liver, intestine, and pancreas; Rey-Campos et al., 1991) but at different relative levels. In liver, HNF1 is less expressed than HNF1, whereas both are present at similar levels in kidney. This pattern of expression correlates with that of the rat UGT1A1, which is predominantly found in liver, and less in kidney (Emi et al., 1995). Furthermore, HNF1 and HNF1 play an important role during development. It has been shown that HNF1, which systematically precedes HNF1 expression, is involved in early events of liver and kidney morphogenesis, whereas HNF1 is required for maintenance of the differentiated state (Cereghini, 1996). Further studies are required to elucidate to what extent these two factors contribute to the developmental expression of the UGT1A1 gene in mouse.

Is HNF1 Crucial for the UGT1A1 Gene Expression in Human and Mouse? The structural differences between human and mouse UGT1A1 promoters raised the question as to whether they are subjected to the same regulatory process. Comparison of the two promoters revealed conserved HNF1- and Sp1-binding sites, and our study also indicates that HNF1 and HNF1 transactivate both the human and mouse UGT1A1 promoter in the same manner. However, an extrapolation from cell culture experiments to the in vivo situation in human is not possible without further consideration. For instance, neither bilirubin glucuronidation nor UGT1A1 mRNA can be detected in human kidney (McGurk et al., 1998) in which HNF1 proteins are expressed and active. This indicates that HNF1 is not sufficient for the human UGT1A1 expression in this organ.