Cloning and Characterization of a Novel Human Hepatocyte Transcription Factor, hB1F, Which Binds and Activates Enhancer II of Hepatitis B Virus*

Enhancer II (ENII) of hepatitis B virus (HBV) is one of the essential cis-elements for the transcriptional regulation of HBV gene expression. Its function is highly liver-specific, suggesting that liver-enriched transcriptional factors play critical roles in regulating the activity of ENII. In this report, a novel hepatocyte transcription factor, which binds specifically to the B1 region (AACGACCGACCTTGAG) within the major functional unit (B unit) of ENII, has been cloned from a human liver cDNA library by yeast one-hybrid screening, and demonstrated to trans-activate ENII via the B1 region. We named this factor hB1F, for human B1-binding factor. Amino acid analysis revealed this factor structurally belongs to nuclear receptor superfamily. Based on the sequence similarities, hB1F is characterized to be a novel human homolog of the orphan receptor fushi tarazu factor I (FTZ-F1). Using reverse transcription-polymerase chain reaction, a splicing isoform of hB1F (hB1F-2) was identified, which has an extra 46 amino acid residues in the A/B region. Examination of the tissue distribution has revealed an abundant 5.2-kilobase transcript of hB1F is present specifically in human pancreas and liver. Interestingly, an additional transcript of 3.8 kilobases was found to be present in hepatoma cells HepG2. Fluorescent in situ hybridization has mapped the gene locus of hB1F to the region q31-32.1 of human chromosome 1. Altogether, this study provides the first report that a novel human homolog of FTZ-F1 binds and regulates ENII of HBV. The potential roles of this FTZ-F1 homolog in tissue-specific gene regulation, in embryonic development, as well as in liver carcinogenesis are discussed.

Enhancer II (ENII) of hepatitis B virus (HBV) is one of the essential cis-elements for the transcriptional regulation of HBV gene expression. Its function is highly liver-specific, suggesting that liver-enriched transcriptional factors play critical roles in regulating the activity of ENII. In this report, a novel hepatocyte transcription factor, which binds specifically to the B1 region (AACGACCGACCTTGAG) within the major functional unit (B unit) of ENII, has been cloned from a human liver cDNA library by yeast one-hybrid screening, and demonstrated to trans-activate ENII via the B1 region. We named this factor hB1F, for human B1-binding factor. Amino acid analysis revealed this factor structurally belongs to nuclear receptor superfamily. Based on the sequence similarities, hB1F is characterized to be a novel human homolog of the orphan receptor fushi tarazu factor I (FTZ-F1). Using reverse transcriptionpolymerase chain reaction, a splicing isoform of hB1F (hB1F-2) was identified, which has an extra 46 amino acid residues in the A/B region. Examination of the tissue distribution has revealed an abundant 5.2-kilobase transcript of hB1F is present specifically in human pancreas and liver. Interestingly, an additional transcript of 3.8 kilobases was found to be present in hepatoma cells HepG2. Fluorescent in situ hybridization has mapped the gene locus of hB1F to the region q31-32.1 of human chromosome 1. Altogether, this study provides the first report that a novel human homolog of FTZ-F1 binds and regulates ENII of HBV. The potential roles of this FTZ-F1 homolog in tissue-specific gene regulation, in embryonic development, as well as in liver carcinogenesis are discussed.
Hepatitis B virus (HBV) 1 is the major cause of acute and chronic hepatitis, also closely associated with the development of hepatocellular carcinoma (1). HBV predominantly infects hepatocytes, which is a prominent characteristic of hepadnaviruses. The genome of HBV is a small, circular, partially doublestranded DNA of about 3.2 kb, which contains four partially overlapping open reading frames (ORF) encoding the surface antigens (preS/S), the core antigen/e antigen (preC/C), the polymerase (P) and the X protein (X), respectively (2,3). These HBV genes express specifically in liver, controlled by the combinatorial action of the promoters and enhancers. To date, four promoters (Sp1, Sp2, Cp, and Xp) (4) have been identified to be responsible for the transcription of the viral mRNAs, and they are regulated by two HBV enhancers. Enhancer I (ENI) functions in a relatively tissue independent manner (5,6), while enhancer II (ENII) shows strong hepatocyte specificity.
Enhancer II of HBV is located within the X ORF, about 600 bp downstream of ENI (7)(8)(9). In our previous study, ENII was mapped in a 148-bp region from nt 1627 to 1774 (HBV adr1 subtype), partially overlapped with the core promoter (Cp) (7). According to the functional analyses, it can be divided into two parts, A and B, with part B as the basal functional unit, which could be further subdivided into three regions B1 (nt 1687-1704), B2 (nt 1705-1735), and B3 (nt 1736 -1774) ( Fig. 1) (7,10). ENII plays critical roles in the regulation of the liverspecific viral gene transcription. It is a major cis element regulating the liver-specific transcription initiated by Cp, which is critical for viral replication and morphogenesis (11). ENII also participates in the regulation of other promoters of HBV genome, and is able to activate heterologous promoters in a highly liver-specific manner (7,12).
To elucidate the regulatory mechanism of the liver-specific activity of ENII, a considerable amount of work has been focused on the study of trans-acting factors that interact with ENII. As shown in Fig. 1, many transcription factors have been identified to interact with different regions of ENII. Among these factors, a set of liver-enriched transcription factors such as HNF1, C/EBP, HNF3, and HNF4 are believed to be involved in the determination of hepatocyte specificity of ENII. Recently, we found the B1 region (nt 1688-AACGACCGACCTT-GAG-nt 1703) was bound specifically by a novel hepatocyte nuclear factor (B1-binding factor, Fig. 1). 2 This binding activity is present in differentiated hepatoma cell line HepG2 but not in the nonhepatic cell line HeLa. Mutations in the B1 region not only aborted the specific binding by this factor, but also significantly reduced the activity of ENII in co-transfection analysis. In addition, the disruption of the specific binding was shown to cause a decrease in viral gene transcription initiated from the core promoter, resulting in a reduction of the 3.5-kb pregenomic RNA (13,14). Therefore, this binding site has an important function in regulating the ENII activity, and consequently, it affects the global HBV gene expression. Our previous efforts to identify the B1-binding factor have suggested it not to be any members of known liver-enriched factors such as HNF1, HNF3, HNF4, or C/EBP (13,15), but a novel hepatocyte transcription factor uncharacterized before; it was thus considered of great interest to clone and characterized this factor.
For this purpose, we employed a yeast one-hybrid screen (16 -18) to clone the cDNA of the B1-binding factor. By screening GAL4 activation domain (AD) fused-cDNA library of human adult normal liver, we isolated a cDNA named hB1F (human B1-binding factor) and investigated both its interaction with the B1 region in vitro and its trans-regulatory effects on ENII in vivo. hB1F was further characterized as a novel human homolog of FTZ-F1, an orphan nuclear receptor that was originally identified as a transcriptional factor involved in the regulation of fushi tarazu (ftz) expression in early Drosophila melanogaster embryos (19).

EXPERIMENTAL PROCEDURES
cDNA Cloning by Yeast One-hybrid Screen-The two oligonucleotides, 5Ј-gatcAACGACCGACCTTGAG-3Ј and 3Ј-TTGCTGGCTGGAAC-TCctag-5Ј, contained the B1 fragment in capital letters with extra nucleotides in lowercase letters added for cloning purpose, were chemically synthesized and annealed. Four tandem repeats of the B1 fragment were placed upstream of the E1b minimal promoter in HIS3 reporter plasmid (p6012), and the CYC1 minimal promoter in lacZ reporter plasmid (pCZII), respectively (CLONTECH). These two plasmids were linearized and sequentially integrated into Saccharomyces cerevisiae YM4271 (MATa,112, (CLONTECH) to obtain a reporter yeast strain YMB1-HB. The yeast was then transformed with human liver MATCHMAKER cDNA library (CLONTECH) and selected on histidinedeficient (His Ϫ ) plates containing 3-aminotriazole (45 mM was shown optimal for suppressing HIS3 background growth). Large colonies (His ϩ ) were transferred onto Hybond N filters (Amersham Pharmacia Biotech) and further screened for ␤-galactosidase activity (20). After being placed in liquid nitrogen for 30 s, the filters were incubated in a buffer containing 0.8 mM 6-bromo-4-chloro-3-indolyl-␤-D-galactosidase at 30°C. The positive interaction was determined by the appearance of blue colonies. The LacZ ϩ colonies were selected and plasmids were recovered (21). The candidate plasmids isolated from the positive clones were transformed into YMB1-HB to retest for His ϩ phenotype and ␤-galactosidase activity. Those that could reproduce the positive phenotypes were called true positive clones and their cDNA inserts were sequenced and further characterized.
In Vitro Translation-EcoRI-digested 2.5-kb cDNA of the isolated positive clone (named pGAD-16) containing the complete hB1F coding sequence was inserted downstream of the T7 promoter of the pBS(ϩ) (Stratagene), and used for in vitro translation in a typical 50-l reaction volume of TNT coupled reticulocyte lysate systems (Promega), containing 25 l of rabbit reticulocyte lysate, 1 l of 1 mM amino acids, and 1 g of DNA template.
Expression and Purification of Glutathione S-Transferase Fusion Protein-The BglII-NdeI fragment of pGAD-16 (nt 158 -696) (see Fig.  3B) containing the suspected DNA-binding domain of hB1F was made blunt-ended and in-frame fused with glutathione S-transferase by inserting into the SmaI digested pGEX-3X vector (Amersham Pharmacia Biotech). The glutathione S-transferase fusion protein was expressed in Eschericha coli. BL21 essentially as described by Smith and Johnson (22), then was purified with glutathione-Sepharose 4B (Amersham Pharmacia Biotech).
Preparation of Nuclear Extract-The nuclear extract of HepG2 cells and HeLa cells was prepared according to the method of Andreas and Faller (23). The detailed procedure has been described previously (24). Briefly, cells were first suspended in 400 l of buffer A (10 mM HEPES-KOH pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride), kept on ice for 15 min, and then homogenized in a Dounce homogenizer with a B pestle about 20 -25 strokes. After centrifugation, the nuclear pellet was resuspended in 200 l of buffer B (20 mM HEPES-KOH, pH 7.9, 420 mM NaCl, 1.5 mM MgCl 2, 0.2 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl, fluoride, 25% glycerol), and kept on ice for 20 min. The supernatant was collected after centrifugation, immediately frozen in small aliquots in liquid nitrogen, and stored at Ϫ70°C.
DNA Transfection and CAT Assays-To construct the CAT reporter plasmid pENII(B)/CpCAT, the ENII(B)/Cp fragment (nt 1686 -1878) containing the basic functional unit of ENII (part B) and the Cp (10), was amplified by PCR and inserted into the XbaI and HindIII sites of  (7,10). The underlined sequences represent the Sp1 recognition sequences (60), the C/EBP recognition sequences (46), the HNF4 recognition sequences (48,49), the HNF3 recognition sequences (24,47), the HNF1 recognition sequences (15), and the recognition sequences of members of nuclear receptors (49). The double underlined sequence represents the specific binding site of a novel hepatocyte nuclear factor (B1-binding factor). pBS(ϩ) (Stratagene). The CAT reporter gene was then cloned downstream of pENII(B)/Cp. Four point mutations of the B1 region (sequence was listed in an EMSA procedure) were introduced by PCR to construct mutant reporter plasmid pENII(B1m)/CpCAT. To construct the eukaryotic expression plasmid of hB1F, pCMV-hB1F, 2.5-kb hB1F cDNA from pGAD-16 released by EcoRI digestion was inserted downstream of CMV promoter in pCMV-poly vector (a gift from E. Lai). The antisense expression plasmid pAnti-hB1F, which transcribes hB1F in the antisense orientation, was constructed by inserting the hB1F cDNA in the reverse orientation downstream of CMV promoter in the same vector.
HepG2 and HeLa cells were propagated in Dulbecco's modified Eagle's medium with 10% fetal bovine serum at 5% CO 2 and 37°C. DNAs were transfected into HepG2 or HeLa cells by the calcium phosphate precipitation method (26). Cells were harvested 48 h after transfection and the protein extracts were prepared. CAT assay was performed according to the method of Gorman et al. (27). In the co-transfection assay, different amounts of hB1F expression plasmid (pCMV-hB1F) and 1 g of the reporter plasmid pENII(B)/CpCAT or pENII(B1m)/ CpCAT were used; 1 g of plasmid pCMV/SEAP (28), which expresses secreted placental alkaline phosphatase (SEAP), was included as internal control. The total amount of DNA plasmids used for each transfection experiment was 15 g, and the plasmid pBS(ϩ) (Stratagene) was used to make up the difference. Each transfection experiment was performed at least three times.
To examine the distribution of hB1F isoforms with difference within A/B regions, 1 g of total RNA isolated separately from fetal liver, adult liver, HepG2 cells, and HeLa cells by TRIzol reagent (Life Technologies, Inc.) was reverse transcribed into cDNA with primer VI (5Ј-CTTGGA-TCACCTGAGACATGGCTTCTAGC-3Ј) (nt 521-493). Subsequently, first round PCR was performed using primer VI and primer IV, and second round using primer VII (5Ј-GTCTTTAAAGCACGGACTTACAC-CTATTG-3Ј) (nt 91-119) and primer VIII (5Ј-CTTCTAGCTTAAGTCC-ATTGGCTCGGATG-3Ј) (nt 500 -472). The resulting PCR fragments were expected to include part of the A/B region and the C region, and were analyzed by electrophoresis on a 5% nondenaturing polyacrylamide gel.
Northern Blot Analysis-Multiple Tissue Northern blots containing 2 g of poly(A) ϩ RNA per lane from multiple human adult tissues were purchased from CLONTECH. Blots were hybridized with random radiolabeled 2.5-kb hB1F cDNA probe (Prime-a-Gene labeling system, Promega) in ExpressHyb hybridization solution (CLONTECH) at 68°C for 1 h. The filters were washed twice with wash solution I (0.3 M NaCl, 0.03 M sodium citrate (pH 7.0), 0.05% SDS) for 20 min at room temperature, and twice with wash solution II (15 mM NaCl, 1.5 mM sodium citrate (pH 7.0), 0.1% SDS) for 30 min at 50°C. The same filters were hybridized separately with a ␤-actin cDNA control probe. Twenty micrograms of total RNA isolated from HepG2 and HeLa cells by TRIzol reagent (Life Technologies, Inc.) were electrophoresed in 1.0% agarose-formaldehyde gels, blotted onto Hybond N ϩ nylon membranes (Amersham Pharmacia Biotech), and subjected to Northern blot hybridization as described above.
Human Chromosomal Localization by Fluorescent in Situ Hybridization (FISH) Mapping-A 2.5-kb hB1F cDNA probe was biotinylated with dATP using the Life Technologies, Inc. BioNick labeling kit. The FISH detection was performed in See DNA Biotech. Inc. (Canada), according to Heng and Tsui (29) and Heng et al. (30).
Nucleotide Sequence Accession Number-The sequence of the hB1F cDNA determined in this study has been deposited in the GenBank TM data base under accession no. HSU80251.

Cloning of cDNA Encoding B1-binding Factor hB1F by Yeast
One-hybrid Screen-Our previous study has revealed that B1 region of HBV ENII is bound specifically by a novel liverenriched nuclear factor (13)(14)(15). As shown in Fig. 2, using the 16-bp B1 fragment (AACGACCGACCTTGAG) as probe in EMSA, we detected a specific shift (band s) with the nuclear extract of the hepatoma HepG2 cell, but not with that of nonhepatic HeLa cell. When four mutations (AACtACaGAtCTc-GAG) were introduced into the B1 fragment, the specific binding was no longer detected. Although HNF4 DNA binding consensus oligonucleotide in large excess amount seemed to be able to compete the B1 specific binding (Fig. 2), the possibility of HNF4 binding was ruled out since HNF4-specific antiserum failed to affect this specific band in a supershift assay (13). As the interaction of the B1-binding factor with B1 fragment was of functional significance (13,14), it was of great importance to identify this unknown factor.
A yeast one-hybrid screen was employed to clone the B1binding factor from human liver MATCHMAKER cDNA library (CLONTECH). In a total of approximately 4 ϫ 10 6 transformants, 19 His ϩ colonies were selected, of which one colony showed strong ␤-galactosidase activity (LacZ ϩ ). The plasmid of this dual positive colony was recovered from the yeast, and named pGAD-16. When retransformed into the reporter yeast strain, pGAD-16 restored His and LacZ activity, indicating it was a true positive clone.
Subsequent sequencing of the 2.5-kb cDNA insert of pGAD-16 revealed a 1.5-kb ORF. Surprisingly, the ORF is in an orientation reverse to the GAL4 AD (Fig. 3A). This cDNA was probably transcribed by a cryptic promoter located around the ADH1 terminator region of the vector, 3  tions had been reported previously (31). The complete sequence of the cDNA insert in pGAD-16 is shown in Fig. 3B. The presence of an in-frame stop codon TAA upstream of the predicted initiation codon indicated that the cDNA insert contained the complete ORF encoding a protein of 495 amino acids. We named it hB1F for human B1-binding factor. The predicted molecular mass of 54 kDa for hB1F was verified by in vitro translation (data not shown).
The amino acid sequence analysis showed that hB1F shared a common modular structure with the nuclear receptors (32). It consists of subdomains A to E or F (Fig. 3C), among which the C region (DNA-binding domain) and the E/F region (ligandbinding domain) are the two conserved domains of the nuclear receptor superfamily (33,34). The homology comparisons suggested hB1F to be a human homolog of orphan nuclear receptor FTZ-F1 (see below). Since the whole sequences including the DNA-binding domain and the activation domain have been encompassed in this factor, it is then not surprising that hB1F could activate the B1-driven HIS3 and lacZ reporter genes independent of GAL4 AD in the yeast screening.
hB1F Binds Specifically to B1 Sequence-To confirm the binding specificity of the cloned hB1F for the B1 target sequence, in vitro translated hB1F was incubated with labeled B1 probe and analyzed by EMSA. As shown in Fig. 4A, the formation of shifted complex (band s) could be efficiently competed by unlabeled B1 oligonucleotides, but not by B1m, or other nonspecific oligonucletides. Quite notably, the shifted complex (band s) migrated to a position similar to that of the complex found with the HepG2 nuclear extract (Fig. 2), suggesting that the binding fashion of the cloned hB1F to the B1 sequence was consistent with that of the B1-binding factor observed in the HepG2 nuclear extract.
Since hB1F shares the common structure with nuclear receptors (Fig. 3, B and C), a partial hB1F cDNA (nt 158 -696, encoding amino acids 35-215) containing the putative DNAbinding domain was fused with glutathione S-transferase and expressed in E. coli. The purified protein glutathione S-transferase-hB1F(DBD) was able to bind to the B1 target region specifically (data not shown). To localize further the hB1F specific binding site(s) in HBV ENII and core promoter region, we performed a DNase I footprinting analysis, using the labeled fragment of HBV DNA (nt 1634 -1816) as target. After incubation with the purified glutathione S-transferase-hB1F(DBD), one region of ENII (nt 1689-ACGACCGACCTT-GAGGCA-nt 1706) was shown to be completely protected from DNase I digestion (Fig. 4B). It agrees well with the B1 region of ENII (nt 1688-AACGACCGACCTTGAG-nt 1703). No other binding site of hB1F was found in ENII and Cp region.
The above results provided strong evidence of the specific binding of hB1F to the B1 sequence. Together with the results obtained from a yeast one-hybrid screen, we have demonstrated that hB1F is the human cellular factor binding specifically to the B1 region of enhancer II of HBV.
hB1F Activates the Enhancer II of HBV-To investigate the functional importance of hB1F to the activity of enhancer II, we performed co-transfection analysis in nonhepatic cells (HeLa) which do not express hB1F protein (see the results of Northern blot analyses, Fig. 8B). The CAT activity of the reporter plasmid pENII(B)/CpCAT is very low in HeLa cells due to the liver specificity of ENII (Fig. 5A, lane 1) (24). When eukaryotic expression plasmid pCMV-hB1F, which contains hB1F cDNA under the control of CMV promoter was co-transfected, the CAT level of pENII(B)/CpCAT could be stimulated with the increasing amount of pCMV-hB1F (Fig. 5A). However, if mutations were introduced into the B1 region, the activity of ENII could no longer be stimulated by hB1F (Fig. 5B). These results indicated that hB1F activates ENII via the B1 site. The transactivating effect of hB1F to ENII was also confirmed in the experiment of co-transfecting pENII(B)/CpCAT with the antisense expression plasmid pAnti-hB1F into HepG2 cells which express endogenous hB1F. In that experiment, the antisense hB1F mRNA was shown to be able to suppress the activity of ENII (data not shown).
hB1F Is a Human Homolog of FTZ-F1-Homology comparison revealed that hB1F was most closely related to members of the FTZ-F1 subfamily of orphan nuclear receptors. Since the identification of FTZ-F1 in D. melanogaster (19), its homologs or close relatives have been isolated in insects (35) and in vertebrate species (36 -43), and these factors constitute a distinct FTZ-F1 subfamily of nuclear receptors. In Fig. 6, the results of homology comparisons of hB1F with some representative vertebrate FTZ-F1 factors and Drosophila DmFTZ-F1 are summarized.
From the above sequence analysis, we categorized hB1F as a member of FTZ-F1 subfamily of nuclear receptors. It is a novel human homolog of FTZ-F1, distinct from the factor hSF-1 which was identified as a human FTZ-F1 homolog before (43). Based on the sequence similarities among the FTZ-F1 subfamily, we suggested that hB1F, together with its close relatives such as mLRH-1, rFTF, xFF1rA/xFF1rB, and zFF1A/zFF1B constituted one subgroup of vertebrate FTZ-F1; while hSF-1, bAD4BP and mSF-1/mELP, which have been also reported to share high homology with each other (45), composed the second subgroup.
An hB1F Isoform hB1F-2 Is Identified-Since the specific B1 binding activity was first identified in HepG2 cells and was found absent in HeLa cells, we used RT-PCR to examine the existence of this factor in these two cell lines. As expected, hB1F could be amplified from HepG2 cells but not from HeLa cells (data not shown). The amplified fragments from HepG2 cells were cloned into pcDNA3 vector and sequenced. We found two forms of amplified sequences, one of them matched with the cloned hB1F, while the second form had extra 138-bp nucleotides encoding 46 amino acid residues in the A/B region (Fig. 7A) with the rest of the coding sequence identical to hB1F. This factor was designated as hB1F-2. The 5Ј and 3Ј ends of the 138 bp nucleotides follows the "GT-AG" exon-intron junction rule (Fig. 7A), suggesting that hB1F-2 and hB1F might be two isoforms resulting from different splicing.
A panel of different pairs of primers was designed to conduct RT-PCR to investigate whether the isoforms with different A/B regions are also present in normal liver tissues. As indicated in Fig. 7B, these two forms of A/B region (with or without the 138 bp) were present not only in HepG2 cells, but also in human adult liver and fetal liver.
hB1F mRNA Is Restricted Mainly to Pancreas and Liver-The tissue distribution of hB1F gene expression was examined by Northern blot analysis of poly(A) ϩ RNA isolated from a variety of human adult tissues (multiple tissue Northern blot I and II, CLONTECH) with a labeled 2.5-kb hB1F cDNA probe. Among 16 tissues examined, a transcript of 5.2 kb was detected in an abundant amount in pancreas, less in liver, and very little in lung (Fig. 8A). No signal was detected in any other tissues tested, including heart, brain, placenta, skeletal muscle, kidney, spleen, thymus, prostate, testis, ovary, small intestine, colon, and peripheral blood leukocytes (Fig. 8A and data not  shown).
To examine the distribution of hB1F in hepatoma cells and nonhepatic cells, total RNAs isolated from HepG2 and HeLa cells were hybridized with the cDNA probe of hB1F. As expected, no signal was detected in HeLa cells. Interestingly, however, there are two different sizes of hB1F transcripts in HepG2 cells. Besides the 5.2-kb transcript consistent with that seen in liver and pancreas, there was an additional transcript of 3.8 kb (Fig. 8B). hB1F Gene Is Located at Human Chromosome 1 q31-32.1-The human chromosomal localization of the hB1F gene was mapped by FISH, using the biotinylated 2.5-kb hB1F cDNA as a probe. Among 100 checked mitotic cell spreads, 81 showed signals on one part of the chromosomes (Fig. 9A), suggesting the hybridization efficiency was Ͼ80%. Using 4,6-diamidino-2phenylindole banding to identify the specific chromosome, we were able to assign the signals to the long arm of chromosome 1 (Fig. 9B). The detailed position was further determined based on the summary from 10 best photographs (Fig. 9C). No other locus was picked by FISH detection under the condition used. Therefore, the hB1F gene is located at human chromosome 1, region q31-32.1.

DISCUSSION
Enhancer II is a critical cis-element for the liver-specific transcriptional regulation of HBV gene expression. In order to elucidate the regulatory mechanism of liver-specific activity of ENII, it is of great importance to identify cellular factors interacting with this cis-element, especially the liver-specific or liver-enriched transcription factors. In this report, we described the cloning and identification of a novel hepatocyte nuclear factor that interacts with the ENII B1 region (nt 1688-AACGACCGACCTTGAG-nt 1703). Using the B1 sequence as a bait in yeast one-hybrid screening, the cloned factor named hB1F from human liver MATCHMAKER cDNA library has been proved by EMSA and DNase I footprinting to be specifically bound to the B1 sequence within ENII. The co-transfection assay demonstrated that hB1F trans-activated ENII via its direct interaction with the B1 site. The results with a survey of tissue specificity of its expression showed that it was restricted to liver and pancreas for normal tissues, present only in differentiated hepatoma cell line HepG2, but not in the nonhepatic HeLa cells. It is thus clear that we have successfully cloned the hepatocyte transcription factor which binds specifically to the B1 region and activates the enhancer II of hepatitis B virus.
It has been known that the strong hepatocyte specificity of ENII is due to the requirement of hepatocyte transcription factors. So far, several liver-enrich factors have been found to interact with ENII. These include HNF1 (15), C/EBP (46), HNF3 (24,47), and HNF4 (48,49). Now we can add hB1F to the list of liver-enriched transcription factors interacting with ENII. It should be noted that none of these liver-enriched factors is exclusively expressed in the liver, but the liver is the only tissue that expresses all of these factors. Therefore, it is probably the combination of these different liver-enriched factors that determines the hepatocyte specificity of ENII, and consequently direct the restricted hepatocyte expression of HBV genes.
hB1F is a novel nuclear receptor. Structurally, it belongs to the FTZ-F1 subfamily that binds target DNA as monomers. The binding sequence of hB1F within the B1 region (nt 1688-AACGACCGACCTTGAG-nt 1703) agrees well with the consensus FTZ-F1 response element (5Ј-PyCAAGGPyCPu-3Ј) derived from the DNA-binding sequence of known FTZ-F1 factors (19, 37, 40, 42, 50 -52). This binding site contains a single hexad half-site (5Ј-AGGPyCPu-3Ј) (53) and three 5Ј-flanking bases (5Ј-PyCA-3Ј) critical for the specific recognition of FTZ-F1 factors. In addition to the B1 region, two other binding sites of nuclear receptors within the ENII/Cp region have been previously described. One is located in the A unit of ENII, which binds HNF4 (48,49), while the other is located in the B3 region, which was recently found to bind HNF4, RXR, PPAR, COUPTF1, or ARP1 (49,54). Unlike the binding site of hB1F, both of these two sites are homologous to the nuclear receptor DR-1 element containing direct repeat of the hexad half-site (RGG/TTCA) (49), and the nuclear receptors interact with the target sites as homo-(i.e. HNF4) or hetero-dimers (i.e. RXR/ PPAR). As shown in DNase I footprinting assay (Fig. 4B), hB1F does not bind to those DR1 elements in A and B3 regions of ENII with high affinity. However, considering its binding sequence covers a half-site of the DR element, it is not surprising that the HNF4 binding consensus oligonucleotide in an excess amount was able to compete with the binding of hB1F in EMSA (Fig. 2).
Since a number of trans-acting factors ( Fig. 1) have been identified to be required for the optimized function of ENII, an important question arises as to how these factors modulate the activity of ENII. First, the optimum activity of ENII probably requires cooperative interactions among these transcription factors. In the preliminary study, we have observed a synergism between hB1F and HNF1 in activating ENII (data not shown). Potential interaction of these two factors and with other factors is now under investigation. Second, it is conceivable that these factors may respond to different microenvironmental cues either within the cell or from extracellular stimuli, or may function at different stages of HBV transcription. Obviously, a clearer picture of the complex network of interaction and cross-talk of these factors will be needed to understand better the mechanism of transcriptional regulation of HBV ENII.
Based on the sequence similarity, hB1F is categorized as a novel human homolog of FTZ-F1. Another human FTZ-F1 homolog, hSF-1, has been identified recently (43). Although they share close sequence homology, hB1F and hSF-1 represent two subtypes of human FTZ-F1, which might evolve from one ancestor FTZ-F1 (55). They are encoded by two distinct genes, show different tissue distributions, and serve different roles. hSF-1, encoded by a gene at human chromosome 9 q33 (56), is expressed in all primary steroidogenic tissues and acts as a crucial transcription factor of enzyme involved in steroid production. It has been also delineated to be essential in the regulation of hypothalamic-pituitary-steroidogenic organ axis (reviewed in Parker and Schimmer (45)). In contrast, the gene of hB1F is located at human chromosome 1 q31-32.1. It is expressed specifically in pancreas and liver that do not express hSF-1. The function analysis has suggested hB1F is an important trans-activator accounting for liver-specific activity of enhancer II of HBV. As a hepatocyte transcription factor, hB1F is likely to be involved in regulating other liver-specific genes. This is supported by the report that the rat FTZ-F1 homolog rFTF, belonging to the same subgroup with hB1F, was able to activate the ␣-fetoprotein gene (41). On the other hand, since hB1F is most abundantly expressed in the pancreas, it might also contribute to the regulation of pancreas-specific genes. In addition, with respect to members of FTZ-F1 subfamily are highly conserved and some of them have been revealed to be crucial in embryonic development (57), it could be predicted that hB1F plays important roles in human embryonic development. As hB1F is expressed in liver and pancreas, two coderivatives of the gut endoderm (58), it is reasonable to propose this factor be involved in the early development of embryonic gut endoderm and tissue differentiation, which needs to be addressed by in vivo functional analyses.
We identified two isoforms, hB1F and hB1F-2, from HepG2 cells. They differ only in the A/B region. The A/B region of hB1F-2 is 46 amino acids longer than that of hB1F. These two isoforms might be derived from differential splicing, a common feature found in the nuclear receptor superfamily. It has been well known that an activation domain, activation function-1 is located in the highly varied A/B region of nuclear receptor, which usually synergizes with activation function-2 domain of nuclear receptors, and functions in a cell type-and promoterspecific manner. Isoforms of many nuclear receptors differing in their A/B region have different function properties (59), suggesting a similar possibility in the case of hB1F and hB1F-2. In our RT-PCR experiments, both isoforms seem to be present in adult liver, fetal liver, and HepG2 cells. But whether they are expressed at different levels or they have different functions needs to be further investigated.
Finally, it is noteworthy that two different sizes of hB1F transcripts were revealed by Northern blot in HepG2 hepatoma cells (Fig. 8B). The transcript of 5.2 kb is consistent with that detected in human adult normal liver and pancreas, whereas the other short transcript of 3.8 kb is specific for HepG2 cells. Compared with normal liver, there might be a different regulation of hB1F expression in HepG2 cells at the transcription level or post-transcription process, such as differential splicing of mRNA, and/or the usage of different promoters and/or different poly(A) ϩ sites. At the present time, the difference of these two hB1F transcripts in HepG2 cells and their respective function and biological significant are still unknown. However, the detection of an hB1F 3.8-kb transcript in HepG2 cells may allow us to study the variation of hB1F expression in light of liver carcinogenesis.