Transcriptional Repression of Human Hepatitis B Virus Genes by a bZIP Family Member, E4BP4

Abstract

Box α is an essential element of both the upstream regulatory sequence of the core promoter and the second enhancer, which positively regulate the transcription of human hepatitis B virus (HBV) genes. In this paper, we describe the cloning and characterization of a box α binding protein, E4BP4. E4BP4 is a bZIP type of transcription factor. Overexpression of E4BP4 represses the stimulating activity of box α in the upstream regulatory sequence of the core promoter and the second enhancer in differentiated human hepatoma cell lines. E4BP4 can also suppress the transcription of HBV genes and the production of HBV virions in a transient-transfection system that mimics the viral infection in vivo. Expression of an E4BP4 antisense transcript can, instead, elevate the transcription of the core promoter. A low abundance of E4BP4 protein and mRNA in differentiated human hepatoma cell lines is detected, and E4BP4 is not a major component of box α binding proteins in untransfected differentiated human hepatoma cell lines. C/EBPα and C/EBPβ, in contrast, are major components of the box α binding activity present in nuclear extracts. E4BP4 has a stronger binding affinity towards box α than the endogenous box α binding activity present in nuclear extracts. Structure and function analysis of E4BP4 reveals that DNA binding activity is sufficient to confer the negative regulatory function of E4BP4. These results indicate that binding site occlusion is the mechanism whereby E4BP4 suppresses transcription in HBV.

Hepatitis B virus (HBV) is a small DNA virus with a partially double-stranded 3.2-kb genome. The genome organization of HBV is very compact, with four overlapping open reading frames coding for the surface, core, polymerase, and X proteins. The transcription of these open reading frames is under the control of four promoters: two for surface, one for core and polymerase, and one for X. Two enhancers, enhancer I and enhancer II, play important roles in the transcription regulation of these viral genes. The core promoter is composed of the basal core promoter and its upstream regulatory sequence (70). This promoter produces two 3.5-kb RNAs, i.e., the precore and pregenomic RNAs. Pregenomic RNA has dual functions: (i) it can be packaged into nucleocapsids (core particles) along with viral polymerase and serve as the template for reverse transcription, and (ii) it can serve as mRNA that encodes the core and polymerase proteins. Regulated expression of pregenomic RNA plays a pivotal role in the control of the viral replication cycle. A detailed understanding of the transcription control of viral genes may reveal new targets for therapeutic intervention.

The second enhancer of HBV has a unique bipartite structure in that the cooperation of two noncontiguous elements, box α and box β, is required for its enhancer function. It stimulates the transcriptional activity of the simian virus 40 (SV40) early promoter and the HBV surface and X promoters (67, 68). The second enhancer is colocalized with the core upstream regulatory sequence (CURS). Box α, for example, is not only an essential component of the second enhancer but also a potent stimulatory element of the CURS. In other words, box α can activate the basal core promoter from an upstream position in differentiated human hepatoma cell lines (HepG2 and HuH-7) (69, 70). A negative regulatory element, designated NRE, which represses the activities of enhancer II and the core promoter, was identified upstream of the CURS (44). The core promoter provides a valuable system to study the positive and negative transcription regulation of eukaryotic promoters.

Transcription regulation is governed by a constellation of trans-acting cellular factors that bind to specific cis-acting elements that act in either a positive or negative manner. Transcription initiation by RNA polymerase II involves a stepwise assembly of general transcription factors or a holoenzyme on a promoter template to form a preinitiation complex. Transcription activators may stimulate transcription by increasing the assembly of a preinitiation complex. Several distinct models have been proposed as the mechanism of transcription repression (20, 21, 24, 25, 29, 42, 52). In the competition model, repressors may bind directly at or near a transcription start site and compete with the formation of a preinitiation complex in the promoter. Alternatively, activators and repressors may compete for overlapping or closely linked binding sites. In the activator-sequestering model, repressors stoichiometrically bind to particular activators through protein-protein interactions, leading to the formation of complexes with reduced or no DNA binding activity. In the quenching model, repressors and activators may bind to adjacent, nonoverlapping DNA sequences, but the repressors neutralize the ability of the activators to transmit stimulatory signals to the basal transcription machinery. In the fourth model, direct repression, repressors may bind to any of the basal transcription factors, with RNA polymerase II itself, or with a corepressor that ultimately targets the basal machinery. Such interaction may interfere with the formation or the activity of the basal transcription preinitiation complex.

We have previously shown that C/EBP-like proteins can bind to box α. C/EBPs (CCAAT/enhancer binding proteins) are a family of highly conserved, leucine zipper-type (bZIP) DNA binding proteins. Members identified so far are C/EBPα, C/EBPβ (also known as NF-IL6, CRP2, LAP, and AGP/EBP), C/EBPδ (also known as NF-IL6β and CRP3), C/EBPγ, CRP1, Ig-C/EBP, and GADD153 (also known as CHOP) (1, 5, 6, 15, 31, 32, 36, 53, 54, 65). Different C/EBP family members are characterized by a high degree of sequence homology in the leucine zipper and basic regions. They have, however, much less conserved N-terminal regulatory and transactivation domains (5, 35). C/EBPs have the potential to form homo- and heterodimers with C/EBP family members or bZIP proteins or to interact with proteins that do not contain leucine zippers. Dimerization of C/EBPs is generally required for their DNA binding and transcription activation function (3, 10, 16, 18, 26, 33, 34, 37–41, 45, 46, 48, 58–65).

In this paper, we describe the cloning and characterization of a box α binding protein, E4BP4. Overexpression of E4BP4 represses the stimulating activity of box α in the CURS and the second enhancer. E4BP4 can also repress the transcription of HBV genes and the production of HBV virions in a transient-transfection system. Overexpression of an E4BP4 antisense transcript, on the other hand, can elevate the transcription of the core promoter. Though present in low abundance, E4BP4 can bind to the box α sequence with higher affinity than the box α binding activity present in nuclear extracts. Evidence that binding site occlusion is most likely the mechanism whereby E4BP4 suppresses transcription in HBV is presented.

MATERIALS AND METHODS

Isolation of cDNA clones.

A λZAPII cDNA library (prepared from Stratagene’s ZAP cDNA synthesis kit) of human hepatoma HepG2 cells was screened with concatemerized double-stranded synthetic oligonucleotides of box α by the method of Singh (57). The oligonucleotides contained the box α sequence gatCCAAGGTCTTACATAAGAGGACTCTT and its complement, which corresponded to the box α sequence extending from nucleotide (nt) 1644 to 1669 of HBV plus an MboI 5′ overhang. The oligonucleotides were concatemerized to ∼500 bp in size with T4 polynucleotide kinase and T4 DNA ligase and labeled with [α-³²P]dCTP by random prime labeling. All positive plaques were picked, replated, and clonally purified through secondary and tertiary screenings. cDNA inserts from positive clones were excised in the form of pBluescript plasmid (pSKP4) by coinfection with helper phage. DNA sequences were determined by dideoxy chain termination methods.

Plasmids.

The HBV sequence used in the study is of the adw subtype. Numbering of the HBV sequence begins at the unique EcoRI site, which is nt 1. All reporter plasmids used in transfection experiments contain a head-to-tail trimeric tandem repeat, referred to as A3, of a 237-bp BclI-BamHI fragment from the SV40 polyadenylation signal. A3 is placed 5′ of promoter sequences of interest and has been shown to stop transcription readthrough from spurious upstream initiation.

Plasmids pSV2CAT, pα/BCP-CAT, pCURS/BCP-CAT, p(1613-1851)CAT, p(1687-1851)CAT, pSVpCAT/ENII, pSVpCAT/αβ, and pHBV3.6 were described previously (44, 67, 68, 70).

The recombinant E4BP4 expression plasmid pXa-2-P4 was generated by cloning of the BamHI-KpnI fragment containing the E4BP4 open reading frame into the BamHI and KpnI sites of the PinPoint Xa-2 vector (Promega).

The plasmid pCMVP4 was generated by moving the cDNA inserts of pSKP4 into the BamHI and KpnI sites downstream of the cytomegalovirus (CMV) immediate-early promoter (CMVIE) in pCMVIE. To generate the FLAG-tagged E4BP4 construct, the BamHI-KpnI fragment containing the cDNA insert of pCMVP4 was cloned into the BglII and KpnI sites downstream of the CMVIE in the pFLAG-CMV2 expression vector (Kodak, New Haven, Conn.). The resulting plasmid, pf:E4BP4, was digested with ApaI and SalI and then recircularized to generate pf:E4BP4ΔApa and pf:E4BP4ΔSal, respectively. The plasmid pf:E4BP4Pvu was generated by cloning the BamHI-PvuII fragment from pf:E4BP4 into BglII- and SmaI-digested pFLAG-CMV2. To remove the repression domain of E4BP4, pCMVP4 was first digested with BstBI to derive a 5.6-kb BstBI fragment, which was subsequently digested with BamHI, followed by filling in of all 3′-recessed ends with the Klenow fragment of Escherichia coli DNA polymerase. Two BamHI-BstBI fragments of 4,513 and 1087 bp, were generated. An 885-bp BamHI-HaeIII fragment was derived from the digestion of the 1,087-bp BamHI-BstBI fragment with HaeIII. The 4,513-bp BamHI-BstBI and 885-bp BamHI-HaeIII fragments were ligated together to generate pCMVP4ΔHae/BstB. The BamHI-KpnI fragment of pCMVP4ΔHae/BstB was cloned into the BglII and KpnI sites of the pFLAG-CMV2 expression vector to generate the plasmid pf:E4BP4ΔHae/BstB. All of these constructs were confirmed by DNA sequencing.

The E4BP4 antisense plasmid pCMV4Ns/s was generated by cloning of the 378-bp SalI-SspI fragment, which corresponds to the 5′-end region of E4BP4 from nt 117 to 494, into the SalI-SmaI sites of the pCMVIE expression vector.

Bacterial fusion proteins.

The E4BP4 expression construct pXa-2-P4 was used to express biotinylated fusion proteins in strain JM109. JM109 cells harboring E4BP4 vectors were grown to log phase and induced with 100 μM isopropyl-β-d-thiogalactopyranoside (IPTG) (Sigma). Six hours following induction, the bacteria were centrifuged and lysed in 1 mg of lysozyme per ml–0.1% Triton X-100–200 U of DNase. The lysates were then clarified by centrifugation at 10,000 × g for 15 min at 4°C and mixed with avidin resin (Promega). Following a 6-h incubation, the resin was washed three times in cold buffer (50 mM Tris-HCl, 4 mM dithiothreitol [DTT], 2 mM EDTA, 10% glycerol). If the fusion protein was to be eluted, the pelleted resin was washed with buffer containing 5 mM biotin. The eluent was aliquoted, quickly frozen under liquid nitrogen, and kept frozen at −70°C.

Preparation of anti-E4BP4 polyclonal antibody.

Rabbits were immunized with a 16-amino-acid peptide corresponding to amino acids 446 to 461 of E4BP4. After three booster injections with conjugated peptide (the carrier protein was keyhole limpet hemocyanin or bovine serum albumin), rabbit sera were tested for reactivity with E4BP4 by immunoprecipitation and Western blotting.

Cell lines, transfection, and CAT assay.

The culture and transfection of human hepatoma cell lines HepG2 and HuH-7 were performed as previously described (7). All plasmids used in one set of experiments were simultaneously prepared, checked for supercoiled forms, aliquoted in small amounts, and stored in 70% ethanol. Each set of experiments was performed with two different preparations of plasmids and repeated two to three times for each preparation. The chloramphenicol acetyltransferase (CAT) activity was normalized against the CAT activity exhibited by a control plasmid, pSV2CAT, which was taken as 100%. In pSV2CAT, the expression of the CAT gene is driven by the SV40 early promoter and 72-bp enhancer. When the CAT activity was high, assays were performed on serially diluted cell lysates to ensure that CAT activity fell in a linear range for all assays.

Preparation and heparin-Sepharose fractionation of nuclear extracts.

Fractionated nuclear extracts from differentiated human hepatoma cell lines HepG2 and HuH-7 were prepared as previously described (8, 68). The crude and fractionated nuclear extracts were aliquoted, quickly frozen under liquid nitrogen, and kept frozen at −70°C.

Preparation of mini-nuclear extracts from transfected cells.

Mini-nuclear extracts were prepared by the method of Schreiber et al. (56). HuH-7 cells were transiently transfected with pFLAG-CMV2, pf:E4BP4, and expression plasmids containing deletion mutants of f:E4BP4 by the calcium phosphate precipitation method. Forty-eight hours later, transfected HuH-7 cells were collected, washed with Tris-buffered saline (TBS) (10 mM Tris-HCl [pH 7.45] and 150 mM NaCl), and pelleted by centrifugation at 1,500 × g for 5 min. The cell pellet was resuspended in TBS, transferred into an Eppendorf tube, and pelleted again by being spun for 20 s in a microcentrifuge. TBS was removed, and the cell pellet was resuspended in cold buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, and 0.5 mM phenylmethylsulfonyl fluoride) by gentle pipetting. The cells were allowed to swell on ice for 15 min, after which a 10% solution of Nonidet P-40 was added and the tube was vigorously vortexed for 10 s. The homogenate was centrifuged for 30 s in a microcentrifuge. The nuclear pellet was resuspended in ice-cold buffer C (20 mM HEPES [pH 7.9], 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, and 1 mM phenylmethylsulfonyl fluoride), and the tube was vigorously rocked at 4°C for 15 min on a shaking plate. The nuclear extract was centrifuged for 5 min in a microcentrifuge at 4°C, and the supernatant was aliquoted, quickly frozen under liquid nitrogen, and kept frozen at −70°C.

Gel shift analysis.

The probe was prepared with annealed double-stranded oligonucleotide (100 ng) corresponding to the box α sequence of HBV (Fig. 1A) and end labeled with [γ-³²P]ATP and T4 polynucleotide kinase. Gel shifting and competition experiments were done as previously described (68) except that 5 instead of 10 μg of nuclear extract was used (68). Supershifts were generated with anti-E4BP4 antiserum, anti-FLAG M2 monoclonal antibody (Kodak), or anti-C/EBP polyclonal antibody for C/EBPα, C/EBPβ, or C/EBPδ (Santa Cruz Biotechnology, Santa Cruz, Calif.).

FIG. 1.

Binding specificity of recombinant E4BP4 protein. (A) Summary of the oligonucleotide sequences of wild-type box α nt 1636 to 1668 [WT(36)] and 1646 to 1668 [WT(46)] and mutants AB, CD, EF, GH, IJ, and YZ and their binding activities toward recombinant E4BP4. In the WT(36) oligonucleotide, the lowercase letters represent the sequence which is different from the HBV sequence. In mutants, the lowercase letters represent the mutated nucleotides. (B) Competition of binding of recombinant E4BP4 protein to wild-type box α sequence [WT(36)] by wild-type box α sequence [either WT(36) or WT(46)] or six box α mutants in gel shift assays. Recombinant E4BP4 protein was first incubated with competing cold oligonucleotides in molar excess and then tested for its binding to labeled WT36 probe. Lane 1, labeled WT(36) probe with no protein; lane 2: labeled WT(36) probe with recombinant E4BP4 protein but no competitor; lanes 3 to 20, labeled WT(36) probe with recombinant E4BP4 protein and different unlabeled competitors at various molar excesses as indicated. HNF1 is a nonspecific competitor.

For Scatchard plot analysis, 7.5 μg of HuH-7 nuclear extract (0.5 M NaCl fraction) and 7 μl of recombinant E4BP4 protein were incubated with different amounts of ³²P-end-labeled box α oligonucleotides ranging from 0.032 to 1.411 ng. The resulting protein-DNA complexes were separated in a 4% polyacrylamide gel, and the bands representing free and bound ligands were identified; this was followed by drying and quantification with a Molecular Dynamics PhosphorImager. Standard Scatchard plot analysis allowed determination of the appropriate K_d values (4, 55).

Northern blotting.

Total cellular RNA was prepared from HepG2 cells, HuH-7 cells, or transfected HuH-7 cells with the RNAzol B kit (Cinna/Tiotecx Laboratories, Inc., Houston, Tex.). Twenty or 40 μg of total cellular RNA was electrophoretically separated on a 1% formaldehyde-agarose gel and transferred to a Hybond nylon membrane (Amersham). In addition, nitrocellulose filters containing approximately 2 μg of poly(A)⁺ RNAs from 16 different adult human tissues (Clontech, Palo Alto, Calif.) were used for Northern analysis. These membranes were hybridized with a ³²P-random-prime-labeled 554-bp EcoRI fragment of the E4BP4 cDNA probe or 1,960-bp PstI fragment of the HBV DNA probe and washed at a high stringency under standard conditions. These blots were exposed to Fuji X-ray film at −70°C with an intensifying screen. The signals were normalized by hybridization with a probe for the β-actin or glyceraldehyde-3-phosphate dehydrogenase (G3PDH) gene followed by quantitation with a Molecular Dynamics PhosphorImager.

Assay for endogenous DNA polymerase activity.

To assay for endogenous DNA polymerase activity, the culture supernatant was collected 3 days after transient transfection, treated with 1% Nonidet P-40 for 4 h at room temperature, and centrifuged at 17,000 × g for 30 min at 4°C. The supernatant was then centrifuged at 227,000 × g for 1 h at 4°C. The pellet from the second centrifugation, which contains HBV viral core particles, was resuspended in TNE buffer (10 mM Tris-HCl [pH 7.5], 50 mM NaCl, and 0.1 mM EDTA) and assayed for endogenous polymerase activity as previously described (70).

Western blotting.

HuH-7 cells were transfected with pFLAG-CMV2, pf:E4BP4, and plasmids containing deletion mutants of f:E4BP4. After 48 h, cells were collected and lysed in Laemmli sample buffer at 95°C for 10 min. Proteins were separated by electrophoresis through a 10 or 12.5% polyacrylamide gel, transferred onto a Hybond (Amersham) enhanced chemiluminescence (ECL) nitrocellulose membrane, and probed with 6 μg of monoclonal anti-FLAG antibody (Kodak) per ml or with a 1,000× dilution of polyclonal anti-E4BP4 antibody. Blots were incubated with anti-rabbit or anti-mouse immunoglobulin G (IgG)–horseradish peroxidase conjugate (Promega), and immunoreactive proteins were visualized with 4-chloro-1-naphthol (Sigma) or by using the ECL system (Amersham).

Immunofluorescence.

HuH-7 cells cultured on Chamber Slides (Nunc) were transfected with pFLAG-CMV2, pf:E4BP4, and plasmids containing deletion mutants of f:E4BP4. Following washing in phosphate-buffered saline (PBS), the cells were fixed with 2% formaldehyde in PBS for 20 min at room temperature, permeabilized by expoure to cold acetone for 3 min, and washed once with PBS. The permeabilized cells were detected with 6 μg of monoclonal anti-FLAG antibody (Kodak) per ml for 1 h and then with fluorescein isothiocyanate-conjugated goat anti-mouse IgG (Cappel) and 0.1 μg of Hoechst 33258 per ml in 1% bovine serum albumin–PBS for 1 h at room temperature. Finally, the cells were washed three times with PBS and examined by fluorescence microscopy.

RESULTS

E4BP4 as a box α binding protein.

To search for the box α binding protein(s), we screened a cDNA library made from a differentiated human hepatoma cell line, HepG2, with labeled concatemers of the box α sequence. A cDNA which was identical to that of a previously described transcription factor, E4BP4 (also named NF-IL3A), was obtained. A member of the bZIP family, E4BP4 has been identified as a binding protein for a variety of promoters, including the ATF site in the E4 promoter of adenovirus, the CRE/ATF-like site of the interleukin-1β (IL-1β) promoter, the gamma interferon promoter, and the A site of the IL-3 promoter (11, 12, 27, 71). E4BP4 contains 462 amino acids. The bZIP domain of E4BP4 is located in the N-terminal region of the protein (see Fig. 6A). E4BP4 has been shown to function as a dimer (12).

FIG. 6.

Minimal essential region of E4BP4 required for transcription repression. HuH-7 cells were cotransfected with either pα/BCP-CAT or p(1687-1851)CAT plus either pf:E4BP4 or one of the four E4BP4 deletion constructs pf:E4BP4ΔHae/BstB (ΔH/B), pf:E4BP4ΔApa (ΔApa), pf:E4BP4Pvu (Pvu), and pf:E4BP4ΔSal (ΔSal). Cotransfection with an insertless pFLAG-CMV2 was also included as a negative control. (A) Repression functions of different E4BP4 deletion mutants. The diagram displays the repression of the CAT activity exhibited by pα/BCP-CAT or p(1687-1851)CAT by a cotransfected wild-type E4BP4 or deletion mutants of E4BP4. The resulting CAT activities were normalized against those obtained with a pα/BCP-CAT reporter alone. (B) Expression of E4BP4 protein by transfectants. Protein expression by untransfected cells (un) (lanes 1 and 8) and cells transfected with either wild-type E4BP4 (f.l) (lanes 2 and 9) or its deletion mutants (with a whole cell lysate of 10⁵ cells/lane) was analyzed by Western blotting with anti-FLAG monoclonal antibody and the ECL system as described in Materials and Methods. Numbers on the left are kilodaltons. (C) Localization of f:E4BP4Pvu and f:E4BP4ΔSal mutant proteins. The localization of f:E4BP4Pvu (left panels) and f:E4BP4ΔSal (right panels) mutant proteins was detected with anti-FLAG monoclonal antibody and a fluorescein isothiocyanate-conjugated goat anti-mouse IgG (bottom panels). The nuclear DNA was stained with Hoechst 33258 (top panels). (D) DNA binding activity of wild-type E4BP4 and its various deletion mutants. Gel shift experiments were performed with end-labeled box α oligonucleotide in the presence of 5 μg of mini-nuclear extracts obtained from untransfected cells (lanes 14 and 15) and cells transfected with either wild-type E4BP4 (lanes 2 and 3) or different deletion mutants of E4BP4 (lanes 4 to 11). For lanes 3, 5, 7, 9, 11, 13, and 15, 3 μg of anti-FLAG monoclonal antibody (α flag) was added for supershifting. No antibody was added in lanes 2, 4, 6, 8, 10, 12, and 14.

To examine the binding specificity of E4BP4, recombinant E4BP4 protein was obtained by fusion of the E4BP4 open reading frame with that of the 1.3S subunit of the Proprionibacterium shermanii transcarboxylase. Labeled double-stranded oligonucleotides (nt 1636 to 1668) corresponding to the wild-type box α sequence were incubated with recombinant E4BP4 protein in gel shift assays. Oligonucleotides containing different mutated box α sequences were added in molar excess as competitors in gel shift assays. As shown in Fig. 1B, mutants IJ, GH, EF, and YZ competed efficiently for binding, while mutants AB and CD did not. These results, summarized in Fig. 1A, indicated that the sequence from nt 1650 to 1662 in box α was the binding site for E4BP4. This segment contains the sequence cTTACaTAAg (the lowercase letters represent the sequence which is different from the consensus E4BP4 binding sequence), which resembled the consensus E4BP4 binding sequence (A/G)T(G/T)A(T/C)GTAA(T/C) (12).

We then examined the binding specificity of overexpressed E4BP4 protein in cells. E4BP4 was first epitope tagged with the FLAG sequence at its N terminus. The coding sequence of FLAG-E4BP4 was then cloned downstream of the CMVIE in the expression plasmid pf:E4BP4. An empty vector, pFLAG-CMV2, was included as a negative control. The expression of E4BP4 was detected with both anti-FLAG monoclonal antibody and anti-E4BP4 antiserum. The anti-E4BP4 antiserum was raised against a peptide derived from the C-terminal region of E4BP4 (see Materials and Methods). Antihemagglutinin (anti-HA) monoclonal antibody and/or preimmune serum was used as a negative control.

HuH-7 cells were transfected with pf:E4BP4 or empty vector in transient-transfection assays. Total cell lysates from untransfected and transfected cells were obtained for Western and gel shift experiments. As shown in Fig. 2A, two forms of E4BP4 were detected with anti-E4BP4 antiserum in pf:E4BP4-transfected cells. Only the larger form of E4BP4 was detected with anti-FLAG monoclonal antibody. The small form of E4BP4 is most likely the translation product from the AUG initiation codon of E4BP4 in the FLAG-tagged E4BP4 from a downstream position. The apparent molecular masses of these two forms of E4BP4 are approximately 64 and 61 kDa instead of the estimated 51 kDa. The reason for this discrepancy is not clear, although an earlier report suggests that phosphorylation may play a role (11). No signal was observed in cells transfected with empty vectors or in untransfected cells. No signal was observed with preimmune serum or anti-HA antibody.

FIG. 2.

DNA binding specificity of overexpressed E4BP4. HuH-7 cells at a density of 1.1 × 10⁷ cells per 15-cm-diameter plate were untransfected (un) or transfected with 62.5 μg of pFLAG-CMV2 (flag) or pf:E4BP4 as indicated. Mini-nuclear extracts were prepared as described in Materials and Methods. (A) E4BP4 expression in transfectants determined by Western blot analysis. Mini-nuclear extracts (20 μg per lane) were used for Western blot analysis, with 1,000× dilutions of anti-E4BP4 (lanes 1 to 3) and preimmune serum (lanes 4 to 6), 6 μg of anti-FLAG (α flag) antibody per ml (lanes 7–9), and 10 μg of anti-HA antibody per ml (lanes 10 to 12). Immunoreactive proteins were visualized with 4-chloro-1-naphthol. (B) Binding of E4BP4 to the box α sequence. Five micrograms of mini-nuclear extracts was incubated with 10⁵ cpm of labeled box α probe [WT(36)] in gel shift experiments. The sources of nuclear extracts are indicated. For supershift experiments, 1 μl of preimmune serum (lane 5) or anti-E4BP4 (lane 6) or 3 μg of anti-FLAG antibody (lane 11) was added. (C) DNA binding specificity of E4BP4. Five micrograms of mini-nuclear extracts was incubated with the labeled wild-type box α probe [WT(36)] in the presence of unlabeled WT(36) or different mutant competitors at various molar excesses as indicated.

Nuclear extracts derived from the transfected cells described above were used in gel shift assays. As shown in Fig. 2B, box α binding activity was present in both untransfected cells and cells transfected with pf:E4BP4 or empty vector. In the nuclear extracts obtained from pf:E4BP4 transfectants, a supershift of box α binding activity by anti-FLAG and anti-E4BP4 antibodies was observed. This supershift was not seen with a control preimmune serum. These results indicated that E4BP4 could indeed bind to the box α sequence. They also showed that E4BP4 is not a major component of endogenous box α binding proteins (Fig. 2C). Moreover, inclusion of unlabeled competing oligonucleotides in molar excess showed that wild-type box α and the EF sequences could effectively abolish the box α binding activity that was supershifted with anti-FLAG antibody. FLAG-tagged E4BP4 expressed in transient transfection, therefore, exhibited the same binding specificity as its bacterially expressed counterpart. Identical results were obtained with another E4BP4-transfected differentiated human hepatoma cell line, HepG2 (data not shown).

To determine the intracellular localization of E4BP4, immunofluorescence of E4BP4-transfected and untransfected HuH-7 cells with anti-FLAG antibody was performed. E4BP4 was detected as a nuclear protein. Identical results were obtained with anti-E4BP4 antiserum (data not shown).

Repression of the transcription stimulation effect of box α by E4BP4.

We have previously shown that a single copy of the box α sequence in an upstream position stimulates the transcription of the HBV basal core promoter in both HepG2 and HuH-7 cells (68). To examine the effect of E4BP4 on box α, the reporter plasmid pα/BCP-CAT was cotransfected with pf:E4BP4 or pFLAG-CMV2 vector in HepG2 and HuH-7 cells. pα/BCP-CAT contains a CAT reporter gene, driven by the HBV basal core promoter, which is preceded by an upstream box α sequence. Increasing amounts of E4BP4 expression plasmids were cotransfected with the reporter plasmid. The expression level of f:E4BP4 and the transcriptional activity of the basal core promoter as measured by CAT assays were determined (Fig. 3). Coexpression of E4BP4 reduced the CAT activity by 35- and 14-fold in HepG2 and HuH-7 cells, respectively (Fig. 3). Cotransfection with pFLAG-CMV2, which had no insert, had no significant effect. It is worth noting that the suppression by E4BP4 was observed with the expression of E4BP4 at a very low level (data not shown). Since the promoter activity of the basal core promoter (pBCP-CAT) is already very low, it is difficult to examine the effect of E4BP4 on the basal core promoter directly. To circumvent this problem, we placed another positive element from the CURS upstream of the basal core promoter instead of box α. This reporter plasmid, p(1687-1851)CAT, contained an extra sequence from nt 1687 to 1743 in addition to the basal core promoter. E4BP4 decreased the activity of p(1687-1851)CAT by threefold (Fig. 4). Weak suppression of the basal promoter by E4BP4 has been previously noted (12). Taken together, these data indicated that E4BP4 suppressed the transcription-stimulatory activity of the basal core promoter mediated by box α.

FIG. 3.

Repression of the box α activity by E4BP4. HuH-7 (lanes 1 to 14) and HepG2 (lanes 15 to 28) cells were transfected with 8 μg of pBCP-CAT only (lanes 1 and 15), 8 μg of pα/BCP-CAT only (lanes 2 and 16), 8 μg of pα/BCP-CAT plus 31.2, 62.5, 125, 250, 500, or 1,000 ng of pf:E4BP4 (lanes 3 to 8 and 17 to 22, respectively), or 8 μg of pα/BCP-CAT plus 31.2, 62.5, 125, 250, 500 or 1,000 ng of pFLAG-CMV2 (lanes 9 to 14 and 23 to 28, respectively). The cell densities for HuH-7 and HepG2 cells were 1.5 × 10⁶ and 2.8 × 10⁶ per 5-cm-diameter plate, respectively. The ability of a cotransfected expression vector to modulate the box α activity was determined by CAT assay. (A) Autoradiogram of CAT activities in a representative assay. (B) Diagram showing the suppression of CAT activity produced by pα/BCP-CAT in the presence of an increasing amount of cotransfected pf:E4BP4. The diagram shows the CAT activity exhibited by pα/BCP-CAT relative to that of pBCP-CAT. Results were quantitated by PhosphorImager counting as described in Materials and Methods.

FIG. 4.

Repression of the activities of the CURS and second enhancer by E4BP4. HepG2 cells at a density of 1.4 × 10⁶ per well in a six-well plate were either transfected with 4 μg of pBCP-CAT or cotransfected with pα/BCP-CAT, pCURS/BCP-CAT, pNRE-CURS/BCP-CAT, pSVpCAT/ENII, pSVpCAT/αβ, or a p(1687-1851)CAT control plasmid in the presence of 500 ng of either pCMVP4 (CMV-E4BP4) or pCMVIE (CMV). The diagram shows the CAT activity exhibited by pα/BCP-CAT relative to that of pBCP-CAT. Results were quantitated with a PhosphorImager as described in Materials and Methods. The data represent the mean results obtained from at least four experiments. Error bars represent the standard errors of the mean values obtained.

Since box α is a functional element of the CURS (nt 1636 to 1741), the effect of E4BP4 on the entire CURS was determined. The positive regulatory activity of the CURS was reduced sevenfold in HepG2 cells (Fig. 4).

Box α is an essential component of enhancer II of HBV. Enhancer II has a unique bipartite structure, and the cooperation of two noncontiguous sequence motifs, box α and box β, is required for its function (68). As shown in Fig. 4, E4BP4 could also suppress the stimulating activity of enhancer II. This suppressive effect was seen with the enhancer in its entirety (from nt 1636 to 1741) as well as with its minimal essential elements (box α and box β).

We have previously identified a negative regulatory element designated NRE. The sequence from nt 1613 to 1621 is essential for NRE activity. Located upstream of the CURS, NRE represses the activity of CURS and enhancer II (reference 44 and our unpublished results). We then examined whether E4BP4 could suppress the transcription-stimulatory activity of the CURS in the presence of NRE. A 25-fold reduction in CAT activity was observed in HepG2 cells (Fig. 4). E4BP4 therefore suppressed the stimulating activity of the CURS in the absence or presence of NRE.

E4BP4 suppresses HBV replication.

So far, we have shown that E4BP4 can suppress the activity of box α, which is a major component of the CURS and enhancer II. Enhancer II activates the transcription of both the large and middle/major surface promoters, while the CURS activates that of the core promoter (67, 70). The negative regulatory effects seen with E4BP4, therefore, may have a significant impact on viral gene expression and replication. To test this, we resorted to transient transfection with a more-than-unit-length HBV genome, pHBV3.6, into differentiated human hepatoma HuH-7 cells. Viral gene expression and production of mature virions that closely mimic viral infection in vivo have been seen after transfection (70). The effect of E4BP4 on the transcription and replication of HBV was tested by cotransfecting an E4BP4 expression plasmid, pf:E4BP4, or an empty vector, pFLAG-CMV2, with pHBV3.6. Three days after transfection, the amounts of the 2.4-kb large surface, 2.1-kb middle and major surface, and 3.5-kb precore and pregenomic transcripts were measured by Northern hybridization. The expression of G3PDH was used as an internal control. As shown in Fig. 5B, expression of E4BP4 reduced the expression of viral RNAs by five- to sixfold. The production of mature virions was quantified by an endogenous DNA polymerase activity assay. As shown in Fig. 5A, E4BP4 reduced the production of virions 20-fold. This was seen with pf:E4BP4 but not pFLAG-CMV2. These results clearly demonstrated that E4BP4 suppressed the gene expression and replication of HBV.

FIG. 5.

Repression of the production of HBV virions and gene expression by E4BP4. Forty-five micrograms of plasmid HBV3.6, which contains a more-than-unit-length HBV viral genome, was transfected into HuH-7 cells at a density of 1.1 × 10⁷ cells per 15-cm-diameter plate alone (lanes 1) or with 30 μg of pFLAG-CMV2 (flag) (lanes 3) or pf:E4BP4 (lanes 2). A pSV2CAT vector, which contains a CAT reporter gene driven by the SV40 early promoter and 72-bp enhancer, was included in all transfections as an internal control for transfection efficiency. (A) Production of virions. Media from the transfectants were collected 3 days after transfection to assay for the production of HBV virions and core particles. The amounts of virions and core particles produced were quantified by the endogenous DNA polymerase activity assay. L and NC, linear and nicked circular forms, respectively, of HBV DNA. Numbers on the left are kilobases. (B) Northern blot analysis of HBV transcripts. The intracellular RNAs were collected at day 3 posttransfection. Twenty micrograms of total RNA of each sample was analyzed by Northern hybridization with the entire HBV DNA as the probe. The same blot was reprobed with G3PDH to ensure equal loading of RNA samples.

Identification of E4BP4 regions essential for the suppression of HBV gene expression.

Earlier studies have identified a repression domain at the C-terminal portion of E4BP4 (12, 13). To test the involvement of this repression domain in the suppression of HBV gene expression, we tested several E4BP4 mutants, which included a series of C-terminal deletion mutants (pf:E4BP4ΔApa, pf:E4BP4Pvu, and pf:E4BP4ΔSal) and an internal deletion mutant lacking the repression domain (pf:E4BP4ΔHae/BstB) (Fig. 6A). All of these mutants were tagged with the FLAG epitope. These mutants were individually cotransfected into HuH-7 cells with a reporter plasmid, pα/BCP-CAT, or a control plasmid, p(1687-1851)CAT, which lacked the box α sequence. The levels of expression and the intracellular localizations of E4BP4 proteins were determined by Western blotting and immunofluorescence. The box α binding activities and the effects of transcription suppression of these mutants were examined with gel shifting experiments and CAT assays. All of these mutants were expressed at roughly comparable levels (Fig. 6B). All of these mutant proteins except f:E4BP4ΔSal were localized only in the nucleus. In Fig. 6C, the localization of a representative f:E4BP4Pvu mutant is shown (left panel). The f:E4BP4ΔSal was present mainly in the nucleus, but a faint signal could be detected in the cytoplasm (right panel).

Nuclear extracts of transfected HuH-7 cells were prepared to examine the box α binding activities of these different E4BP4 mutant proteins. Supershifting with anti-FLAG antibody was done to verify the specific binding of box α by E4BP4. As shown in Fig. 6D, most of these mutants could bind to box α. In contrast, the f:E4BP4ΔSal mutant did not bind to box α despite its predominant localization in the nucleus. It has been shown that E4BP4 dimers bind to DNA much better than monomers (12). The mutant f:E4BP4ΔSal lacks the leucine zipper dimerization domain but retains a basic region. The fact that f:E4BP4ΔSal could not bind to box α most likely resulted from its inability to form dimers. Mutants containing the DNA binding domain suppressed the activity of the reporter plasmid, pα/BCP-CAT. The f:E4BP4Pvu mutant, which contained an intact DNA binding domain, suppressed the activity of box α with the same efficiency as the wild-type f:E4BP4 (Fig. 6A). These results showed that the ability to bind to box α, but not the presence of the repression domain, was essential for the suppression of box α activity by E4BP4. Interestingly, only wild-type f:E4BP4 repressed the activity of p(1687-1851)CAT. These results indicated that suppression of box α or basal promoter activity is mediated by different domains. DNA binding activity is required for the suppression of the transcription-stimulatory effect of box α. The repression domain, in contrast, is essential for the repression of basal promoter activity.

Negative regulatory effects of E4BP4 in human hepatoma cells.

Northern hybridization was performed to analyze the expression pattern of E4BP4. A 1.9-kb E4BP4 transcript was detected in total RNAs extracted from HepG2 and HuH-7 cells, which are differentiated human hepatoma cell lines (Fig. 7A). As a control, the blot was reprobed with a radiolabeled G3PDH cDNA for equal loading of RNA samples. In addition, poly(A)⁺ RNAs from a variety of human tissues were subjected to Northern blot analysis. The 1.9-kb E4BP4 message could be detected in all tissues. Modest expression was seen in brain, liver, and thymus, while very strong expression was seen in peripheral blood leukocytes, skeletal muscle, and testis (Fig. 7B). These blots were subsequently reprobed with a radiolabeled human β-actin cDNA to ensure equal loading. The modest level of E4BP4 expression in liver was in line with the observation that there was very little supershifting by anti-E4BP4 antibody of box α binding activity in nuclear extracts derived from untransfected HepG2 and HuH-7 cells (Fig. 2B). To test the involvement of E4BP4 in the negative regulation of HBV promoters, we overexpressed an E4BP4 antisense transcript and measured the transcriptional activity of a cotransfected basal core promoter. The E4BP4 antisense transcript was driven by the CMV promoter, while an insertless vector containing only the CMV promoter was used as a negative control. As shown in Fig. 8, the activity of the basal core promoter in the presence of an upstream box α or the CURS was increased by threefold. The activation of the basal core promoter by another upstream element containing nt 1687 to 1743, was not significantly affected.

FIG. 7.

Expression of E4BP4 RNA. (A) Expression of E4BP4 mRNA in the human hepatoma cell lines HuH-7 and HepG2. Thirty-five micrograms each of total RNAs from HuH-7 (lane 1) and HepG2 (lane 2) cells was loaded for Northern blotting. The filter was probed with a 554-bp EcoRI fragment of E4BP4 cDNA (upper panel) or G3PDH cDNA (lower panel). Numbers on the left are kilobases. (B) Expression of E4BP4 mRNA in various human tissues. Approximately 2 μg of poly(A)⁺ RNAs from various human tissues (Clontech) was used for Northern blot analysis. The tissue origins of the RNA samples are indicated. The filters were probed with a 554-bp EcoRI fragment of E4BP4 cDNA (upper panel) or β-actin cDNA (lower panel). Probing with either G3PDH or β-actin is to ensure approximately equal loading of RNA samples. PB, peripheral blood.

FIG. 8.

Elevation of the activities of box α and the CURS by E4BP4 antisense RNA. Three micrograms of plasmid pα/BCP-CAT, pCURS/BCP-CAT, or p(1687-1851)CAT was cotransfected with either 3 μg of pCMV4Ns/s, which is an antisense E4BP4 expression construct, or 2.35 μg of a control vector, pCMVIE (CMV), into HuH-7 (top panel) and HepG2 (bottom panel) cells. The promoter activity of pα/BCP-CAT, pCURS/BCP-CAT, or p(1687-1851)CAT was measured by CAT assay. The CAT activity exhibited by each promoter in cells cotransfected with the antisense construct is normalized against that in cells cotransfected with the control vector pCMVIE. Error bars indicate standard errors of the means.

The results described above indicate that there is a low abundance of E4BP4 protein and mRNA in liver and differentiated human hepatoma cell lines and that E4BP4 is not a major component of box α binding proteins in untransfected HepG2 and HuH-7 cells. Despite this, the amount of E4BP4 appears to be sufficient to exert a negative regulatory effect on the HBV core promoter in these differentiated human hepatoma cell lines. In support of this argument, a small amount of E4BP4 can significantly suppress the activity of box α, as shown in Fig. 3.

E4BP4-mediated suppression of transcription.

Box α stimulated the basal core promoter activity in HuH-7 and HepG2 cells. Box α is also an essential component of enhancer II. E4BP4, however, suppressed the stimulatory function of box α in the context of the CURS and enhancer II. E4BP4 does not appear to be a major component of box α binding proteins in HepG2 and HuH-7 cells. For example, the box α sequence (CAAGGTCTTACATAAGAGGACTCTT [nt 1645 to 1669]) is similar to the consensus C/EBP binding site (A/G/C)T(T/G)NNG(T/C)AA(T/G). We previously reported that box α binding proteins were C/EBP-like proteins (68). To test whether members of the C/EBP family are present among box α binding proteins in HepG2 and HuH-7 cells, gel shift experiments were performed with the labeled box α sequence in the presence of, as unlabeled competing sequences, the C/EBP consensus sequence or the mutant box α sequence AB, CD, or EF. As shown in Fig. 9A, mutant EF, carrying mutations in box α and the C/EBP consensus sequence, but not mutants AB and CD, carrying mutations in box α, could compete as effectively for box α binding as the wild-type box α sequence. This shows that these endogenous C/EBP-like proteins have the same binding site specificity as E4BP4. We have noticed that EF could compete for box α binding, which was different from our previous observation (68). These gel shift and competition experiments were done as previously described except that 5 instead of 10 μg of nuclear extract was used. Whether this minor modification or other conditions that we could not control for led to this change in result is not clear. This result, however, does not change our conclusion that C/EBP-like proteins are major box α binding proteins in liver cells (see Fig. 9B and C and Discussion).

FIG. 9.

C/EBPα and C/EBPβ are box α binding proteins. (A) Binding site specificity of endogenous box α binding protein determined by gel shift experiments in the presence of competing oligonucleotides in excess. The relative efficiencies of wild-type box α (WT) (Lanes 3 to 5), three mutated forms of box α sequences (AB [lanes 6 and 7], CD [lanes 8 and 9], and EF [lanes 10 and 11]), a C/EBP consensus sequence (lanes 12 and 13), and nonspecific competitor HNF1 (lane 14) in abolishing box α DNA-protein complexes formed by endogenous box α binding proteins present in nuclear extracts are shown. Lane 1, no protein; lane 2, 5 μg of nuclear protein of the 0.5 M NaCl eluent from HepG2 cells and no competitor; lanes 3 to 14, 5 μg of nuclear protein of the 0.5 M NaCl eluent form HepG2 cells in the presence of different unlabeled competitors at various molar excesses as indicated. (B and C) supershifting of box α binding complexes formed by nuclear extracts by antibodies against different members of the C/EBP family. Lane 1, no protein; lane 2, 5 μg of nuclear protein of the 0.5 M NaCl fraction from HuH-7 (B) or HepG2 (panel C) cells; lanes 3 to 6, cold wild-type box α, AB, CD, or C/EBP consensus sequence was added at a 125-fold molar excess for competition experiments; lanes 7 to 13, anti-C/EBPα, anti-C/EBPβ, and anti-C/EBPδ antibodies (0.5 μg each) alone or in combinations were added for supershifting.

To identify which C/EBP family protein(s) could bind to the box α DNA sequence, anti-C/EBPα, -β, and -δ antibodies were used in supershift experiments. Anti-C/EBPβ antibody could significantly supershift the protein-DNA complexes formed by the box α sequence after incubation with nuclear extracts derived from HuH-7 (Fig. 9B) and HepG2 (Fig. 9C) cells. Anti-C/EBPα antibody caused modest supershifting. Anti-C/EBPδ antibody did not display any apparent activity. When combinations of anti-C/EBP antibodies were tested, a mixture of anti-C/EBPα and anti-C/EBPβ antibodies supershifted almost all of the box α-protein complexes. Identical results were obtained with nuclear extracts derived from HepG2 and HuH-7 cells. C/EBPα and C/EBPβ, therefore, can bind to the box α sequence in both HuH-7 and HepG2 cells, where box α displays stimulating activity.

E4BP4 and endogenous box α binding proteins apparently all bind to the same sequence in box α. We then tested whether binding site occlusion might be the mechanism of suppression mediated by E4BP4. In this scenario, occupancy of box α by E4BP4 might hinder the binding, and therefore the function, of positive regulatory factors such as endogenous box α binding proteins. To test this possibility, fractionated nuclear extracts of HuH-7 and HepG2 cells were incubated with labeled box α sequences in the presence of increasing amounts of recombinant E4BP4 protein. The DNA-protein complexes formed by box α and endogenous box α binding proteins migrated differently from that formed by box α and E4BP4. As shown in Fig. 10, recombinant E4BP4 could compete out the endogenous proteins present in nuclear extracts in binding to box α in a dose-dependent manner. We then determined the binding affinity of recombinant E4BP4 and endogenous box α binding proteins for the box α sequence by Scatchard plot analysis. For Scatchard plot analysis, known amounts of recombinant E4BP4 protein and a 0.5 M NaCl fraction of nuclear extracts obtained from HuH-7 cells were incubated with increasing amounts of end-labeled box α oligonucleotides ranging from 0.032 to 1.411 ng. The resulting protein-DNA complexes were separated on a 4% polyacrylamide gel. After drying of the gels, the bands representing the free and bound oligonucleotides were identified, and their relative intensities were quantified with a Molecular Dynamics PhosphorImager. Scatchard plots were done by plotting the ratio of the amount of the protein bound DNA to that of the free DNA against the amount of protein-bound DNA. Recombinant E4BP4 protein and a 0.5 M NaCl fraction of nuclear extracts obtained from HuH-7 cells had apparent dissociation constants of 0.08 and 0.26 nM, respectively (Fig. 11). Recombinant E4BP4, therefore, bound to the box α sequence with a stronger affinity than endogenous box α binding proteins (threefold stronger).

FIG. 10.

Competition for binding to the box α sequence by recombinant E4BP4 and the endogenous box α binding activity present in nuclear extracts. The ability of recombinant E4BP4 protein to outcompete endogenous box α binding activity present in nuclear extracts was determined. Gel shift competition experiments were performed with end-labeled wild-type box α oligonucleotide and 7.5 μg of protein prepared from the 0.5 M NaCl fraction of HuH-7 (lanes 2 to 5) or HepG2 (lanes 6 to 9) nuclear extracts. Mixtures were preincubated with 1 (lanes 3 and 7), 2 (lanes 4 and 8), or 4 (lanes 5 and 9) μl of recombinant E4BP4 protein before loading. Lane 1, box α alone; lane 10, mixture of box α and 4 μl of recombinant E4BP4 protein.

FIG. 11.

Binding affinities of recombinant E4BP4 and the endogenous box α binding activity towards the box α sequence (Scatchard plot analysis). A series of gel shift assays were performed with fixed amounts of the 0.5 M NaCl fraction of HuH-7 nuclear extracts (NE) or recombinant E4BP4 protein in the presence of increasing amounts of ³²P-end-labeled box α oligonucleotides. The bands representing free and bound oligonucleotides were identified and isolated. After drying, the radioactivities of these bands were quantified with a Molecular Dynamics PhosphorImager. The dissociation constants (Kd) were determined by the standard method of Scatchard plot analysis.

DISCUSSION

In this paper, we describe the cloning of a bZIP type of transcription factor, E4BP4, that specifically binds to the box α sequence in HBV. Box α is an essential element of both the CURS and the second enhancer, which positively regulate the transcription of HBV genes (68, 70). E4BP4 appears to function as a negative transcription regulator at box α. Overexpression of E4BP4 represses the stimulating activity of box α in the CURS and the second enhancer in constructs containing CAT reporters. E4BP4 can also repress the transcription of HBV genes and the production of HBV virions in a transient-transfection system that mimics the viral infection in vivo. In addition, expression of an E4BP4 antisense construct can elevate the transcription of HBV genes.

The following results suggest that binding site occlusion is the mechanism whereby E4BP4 suppresses transcription in HBV: (i) E4BP4 and the endogenous box α binding protein(s) bind to the same sequence in box α, (ii) E4BP4 readily outcompetes the endogenous box α binding activity present in nuclear extracts in a dose-dependent manner, (iii) E4BP4 has a stronger binding affinity towards the box α sequence than the endogenous box α binding activity present in nuclear extracts, (iv) examination of the structure-function relationship of E4BP4 reveals that DNA binding activity is sufficient to confer the negative regulatory function of E4BP4, and (v) E4BP4 does not bind either C/EBPα or C/EBPβ to a significant extent when coexpressed in human hepatoma cells (data not shown).

In addition to box α, E4BP4 also binds to the E4 promoter of adenovirus, the IL-1β promoter, the gamma interferon promoter, and the A site of the IL-3 promoter. Among these promoters, overexpression of E4BP4 has been shown to suppress the transcriptional activity of the E4 promoter of adenovirus and the IL-1β promoter like that of box α. The exact repression mechanism of E4BP4 on the E4 and IL-1β promoters has not been completely elucidated. It is not clear, for example, if positive factors bind to overlapping or identical sites in these promoters and if E4BP4 functions by binding site occlusion as is the ease for box α of HBV (11, 12). An earlier report showed that E4BP4 could act as an active repressor to suppress the transcriptional activity of a basal promoter in the presence of an E4BP4 binding site, probably via an interaction with Dr1 (13, 14). Dr1 is a TBP binding protein, which can down regulate both basal and activated transcription at a variety of promoters (28, 66). In the same report, a repression domain that was required for suppression of transcription and interaction with Dr1 by E4BP4 was described (14). This repression domain, however, is not required for the repression of the activity of box α in HBV. E4BP4, therefore, represses HBV transcription by binding site occlusion rather than by acting as an active repressor. Interestingly, E4BP4 may also function as an activator of transcription. Overexpression of E4BP4 (NF-IL3A) activates the transcriptional activity of the A site of the IL-3 promoter in a binding-site-specific manner (27, 71). Furthermore, although E4BP4 can bind to the human gamma interferon promoter, it does not confer either an activating or a repressing function (71). It is possible that the different regulation by E4BP4 is influenced by the nucleotide sequences surrounding the specific binding sites, the presence or absence of other factors binding to overlapping or identical sites, and/or the presence or absence of proteins interacting with E4BP4.

Mutant EF can compete fairly efficiently with the wild-type sequence in binding to cellular factors present in a 0.5 M NaCl fraction. This result appears to be at odds with that of our prior study that the EF mutant will not bind to cellular factors in extracts prepared in similar ways (68). The reason for this discrepancy is unknown. The fact that anti-C/EBPα and -β antibodies can almost completely supershift the box α binding activity nevertheless strongly supports our conclusion that C/EBPα and -β are major box α binding factors.

Box α is an essential component of both the CURS and the second enhancer. In both cases, it has a stimulatory effect in the differentiated human hepatoma cell lines HuH-7 and HepG2 (68, 70). E4BP4 does not appear to be a major component of the box α binding activity, and E4BP4 is present in low abundance in HuH-7 or HepG2 cells. Given the strong binding affinity of E4BP4 towards box α and the ability of a small amount of E4BP4 to significantly suppress the activity of box α, E4BP4 appears to be at a sufficient level to exert a negative regulatory effect on HBV transcription and replication. C/EBPα and C/EBPβ, in contrast, are major components of the box α binding activity present in nuclear extracts as demonstrated by supershift experiments. Overexpression of C/EBPα or C/EBPβ can potentiate the stimulating activity of box α (data not shown). Moreover, recombinant C/EBPβ protein can bind to the same sequence in box α as recombinant E4BP4 and the endogenous box α binding activity present in nuclear extracts (data not shown). However, the DNA-protein complexes formed by endogenous box α binding activity derived from nuclear extracts migrated differently from those formed by homodimers of either C/EBPα or C/EBPβ (data not shown). These results suggested that both C/EBPα and C/EBPβ would heterodimerize with each other or with other binding proteins when binding to the box α sequence. In addition to the formation of homodimers, C/EBPs have been reported to form heterodimers with other C/EBP family members or other bZIP proteins such as CREB, C/ATF, and AP1 (18, 26, 33, 34, 37, 38, 48, 61–65). C/EBPs may also interact with proteins that do not contain leucine zippers, such as YY1, Rb, Rel, NF-κB, Sp1, TBP, TFIIB, or glucocorticoid receptor (3, 10, 16, 39–41, 45, 46, 58–60). The nature of the partners that heterodimerize with either C/EBPα or C/EBPβ to form the endogenous box α binding activity in nuclear extracts is yet to be determined.

Transcription of HBV genes displays a strong preference for liver and differentiated hepatoma cell lines (2, 7, 17, 22). This preference has been attributed to the circumscribed action of certain liver-enriched transcription activators, including HNF1, HNF3, and HNF4, binding to the cis-acting regulatory elements present in the HBV genome (for example, HNF3 and HNF4 on both the X promoter/enhancer I and core promoter/enhancer II and HNF1 and HNF4 on the large surface promoter) (8, 9, 19, 23, 30, 43, 47, 49, 50, 51, 72). On the other hand, E4BP4 is a negative transcription factor for HBV transcription. It is intriguing that E4BP4 transcripts are present in larger amounts in many tissues other than liver. The potent suppression of box α activity by E4BP4 may contribute to the preferential silencing of HBV gene expression in tissues other than liver.