Interaction of Hepatitis B Viral X Protein and CCAAT/ Enhancer-binding Protein α Synergistically Activates the Hepatitis B Viral Enhancer II/Pregenomic Promoter*

The hepatitis B viral X protein (HBx) is known to exert its transactivation activity by the interaction with several cellular transcription factors. Here we report the interaction of HBx and CCAAT/enhancer-binding protein α (C/EBPα) and their effects on the enhancer/promoters of hepatitis B virus (HBV). A chloramphenicol acetyltransferase assay showed that the cotransfection of HBx and C/EBPα strongly activated the enhancer II/pregenomic promoter of HBV in a synergistic manner. This effect was also observed in the heterologous expression system with promoters of SV40 and herpes simplex virus thymidine kinase genes. Serial deletion analysis of the enhancer II/pregenomic promoter identified the responsible region (nucleotides 1639–1679), in which two C/EBP-binding sites are located. An in vitro interaction assay and electrophoretic mobility shift assay showed that HBx augmented the DNA binding activity of C/EBPα by direct interaction with it, and its basic leucine zipper domain was responsible for the interaction with HBx. Domain analysis of HBx showed that the central region (amino acids 78–103) was necessary for direct interaction with C/EBPα. However, the complete form of HBx was necessary for the synergistic activation of the HBV pregenomic promoter. These results suggest that the interaction of HBx and C/EBPα enhances the transcription of the HBV pregenomic promoter for the effective life cycle of HBV in hepatocytes.

The hepatitis B viral X protein (HBx) is known to exert its transactivation activity by the interaction with several cellular transcription factors. Here we report the interaction of HBx and CCAAT/enhancer-binding protein ␣ (C/EBP␣) and their effects on the enhancer/promoters of hepatitis B virus (HBV). A chloramphenicol acetyltransferase assay showed that the cotransfection of HBx and C/EBP␣ strongly activated the enhancer II/ pregenomic promoter of HBV in a synergistic manner. This effect was also observed in the heterologous expression system with promoters of SV40 and herpes simplex virus thymidine kinase genes. Serial deletion analysis of the enhancer II/pregenomic promoter identified the responsible region (nucleotides 1639 -1679), in which two C/EBP-binding sites are located. An in vitro interaction assay and electrophoretic mobility shift assay showed that HBx augmented the DNA binding activity of C/EBP␣ by direct interaction with it, and its basic leucine zipper domain was responsible for the interaction with HBx. Domain analysis of HBx showed that the central region (amino acids 78 -103) was necessary for direct interaction with C/EBP␣. However, the complete form of HBx was necessary for the synergistic activation of the HBV pregenomic promoter. These results suggest that the interaction of HBx and C/EBP␣ enhances the transcription of the HBV pregenomic promoter for the effective life cycle of HBV in hepatocytes.
Hepatitis B virus (HBV) 1 is a partially double-stranded DNA virus that replicates through the reverse transcription of pregenomic RNA (1). HBV is a causative agent of chronic and acute hepatitis and is associated with the development of hepatocellular carcinoma (2). The X protein of HBV (HBx) has been implicated in HBV-mediated hepatocellular carcinoma by its abilities to induce liver cancer in some transgenic mice (3) and to transactivate a variety of viral and cellular promoters (re-viewed in Ref. 4). Being unable to bind DNA directly, the activity of HBx is known to be mainly mediated through the binding sites for other transcription factors such as AP-1 (5), NF-B (6), ATF/CREB (7,8), and acidic activators (9). HBx has been shown to activate AP-1 and NF-B by means of Rasmediated signaling pathways (5,6,10). Activation of Ras by HBx was also shown to increase the cellular level of TATAbinding protein, which induces RNA polymerase III-dependent transcription (11). Recently, it was reported that activation of Src family kinases by HBx is coupled with the activation of Ras (12).
HBx has been shown to interact with some basal transcription factors (13)(14)(15)(16)(17), the p53 tumor suppressor (18,19), and basic leucine zipper (bZip) proteins (8,20). Stimulation of RNA polymerase II transcription by HBx in vitro or in cultured cells seems to require a specific cis-element or other activator proteins (9,14,16), which suggests a role for HBx as a coactivator. These two distinct roles for HBx, as a cellular signaling molecule or as a coactivator of transcription, have been suggested to be dependent on its intracellular distribution (21). This is also implied for the interaction of HBx and the p53 tumor suppressor, which modulates p53-mediated gene transcription and apoptosis (18,19,22). Other functions of HBx suggested so far include the interactions of HBx with UV-damaged DNA-binding protein (23), proteasome complexes (24), and protease tryptase TL 2 (25). Recent reports suggesting a role of HBx in DNA repair are noteworthy (17,26). Two distinct functions and these alternatives of HBx action may explain the multifunctional role of HBx in vivo.
Although the role of HBx during the infection of HBV is still unclear, it seems to play an essential role in infection by woodchuck hepatitis B virus (27,28). In addition, transactivation of HBV enhancer/promoters by HBx (29,30) increases its importance in the HBV life cycle. A binding site for AP-1, a ubiquitous transcription factor, in enhancer I is the only cis-element of HBV yet known to respond functionally to HBx (7).
The CCAAT/enhancer-binding protein (C/EBP) was first identified in rat liver (31) and is expressed mainly in highly differentiated cells such as liver and fat cells (32), where it plays a key role in cell differentiation (33,34). C/EBP belongs to the bZip family of transcription factors and activates transcription of several genes through its binding sites (a consensus site: 5Ј-RTTGCGYAAY-3Ј) (35) in liver and fat cells. C/EBP has been shown to bind and modulate enhancer I (36) and the enhancer II/core promoter (37,38) of HBV. A possible role of C/EBP in the HBx-stimulated expression of promoters other than HBV has been previously suggested by indirect evidence (39,40). In the course of our study, a physical interaction of HBx and C/EBP␤ without any functional implications in vivo was reported (20). We investigated a role for HBx in C/EBP␣ function and their effects on the HBV enhancer/promoters in * This work was supported in part by the Korea Ministry of Education and the Korea Science and Engineering Foundation through the Research Center for Cell Differentiation at Seoul National University. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
HepG2 cells. We report here that the direct interaction of HBx and C/EBP␣ strongly activated the enhancer II/pregenomic promoter (EnII/Cp) of HBV in a synergistic manner.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The pCENCAT reporter plasmid was constructed by inserting the XhoI-AluI fragment (nt 127-1873) of HBV subtype adr-k (41), which contains the upstream enhancer, the enhancer I/X promoter (EnI/Xp), and EnII/Cp, in front of the CAT reporter gene (see Fig. 1A). Plasmids pUEIXp and pCpBm were constructed by dividing the HBV enhancer/promoters using the BamHI site (nt 1937) into the regions containing the upstream enhancer and EnI/Xp, and EnII/Cp, respectively. The pEIXp plasmid was constructed by deleting the XhoI-AccI fragment (nt 127-1070) containing the upstream enhancer from pUEIXp. The derivatives of pCpBm were constructed by serially digesting the plasmid with SacII (pCpSc), ApaLI (pCpAp), StyI (pCpSy), HincII (pCpHc), and DraI (pCpDa) as described previously (42). The position of each deletion site is indicated in Fig. 2A. The pC/EBP-CAT reporter plasmid was constructed by inserting eight copies of the consensus C/EBP-binding site (5Ј-TGCAGATTGCGCAATCT-GCA-3Ј) into the BglII site of the pCATpromoter (Promega) containing the minimal SV40 early promoter. The pBLCAT2 plasmid has a minimal promoter (Ϫ150 to ϩ50) of the herpes simplex virus thymidine kinase gene that contains its own C/EBP-binding site (43).
The C/EBP␣ expression vector pMSV-C/EBP␣ is a kind gift of Dr. S. L. McKnight. The bacterial expression vectors for full-length C/EBP␣ and the bZip domain of C/EBP␣ (pGST-C/EBP␣ and pGST-bZip, respectively) were constructed by inserting the NcoI-NcoI fragment (amino acids 1-358) and the SmaI-NcoI fragment (amino acids 245-358) of C/EBP␣ cDNA into the SmaI site of pGEX-4T1 and pGEX-4T2 (Amersham Pharmacia Biotech), respectively. The eukaryotic expression vector for HBx (pSVX2) was constructed by placing its cDNA (nt 1371-1833) under the control of the SV40 enhancer and early promoter. The derivatives of pSVX2 (pSVX⌬BA, pSVX⌬BH, and pSVX⌬SH) were con-  35 and 49. b Known C/EBP sites in the upstream enhancer (UE) and enhancer I (EnI) of HBV (36, 44) (data not shown). c The results are shown in Fig. 2B. Numbers in parentheses correspond to the known C/EBP-binding sites in Fig. 2A (37). Other sequences were shown to be important in the function of EnII/Cp and have homologies to the consensus C/EBP site (47,48).

FIG. 1. Synergistic effect of HBx and C/EBP␣ on the HBV enhancer/ promoters. A, schematic presentation of HBV
enhancer/promoter constructs. Known factor-binding sites are shown at the top. As indicated, plasmid pCENCAT has a whole region from the upstream enhancer (UE) to enhancer II (EnII) and the pregenomic promoter (Cp) (nt 127-1873 of HBV subtype adr-k). Plasmids pCpBm and pUEIXp have the regions encompassing the pregenomic promoter and the X promoter (Xp), respectively, of pCENCAT. Plasmid pEIXp has EnI/Xp alone without the upstream enhancer. Ac, AccI; Bm, BamHI. B, 3 g of pUEIXp and 1 g each of the other reporters were transfected into HepG2 cells with the HBx (pSVX2) and C/EBP␣ (pMSV-C/ EBP␣) expression vectors as indicated. pMSV-C/EBP␣ (0.5 g) and pSVX2 (4 g) were used for each reporter plasmid. structed using the BamHI, ApaLI, and HincII sites, respectively, within the HBx cDNA as described for Fig. 5A. The bacterial expression vectors for HBx and its derivatives (pMBP-X, pMBP-Mx, and pMBP-Sx) were described previously (7).
Transient Transfection and CAT Assay-HepG2 cells were transfected with reporter and activator plasmids as indicated in the figure legends using the calcium phosphate co-precipitation method with BES as described previously (7). The total amount of transfected DNA for each reaction was adjusted to 9 g with pUC19. The CAT assay was performed with cell extracts normalized for the total amount of protein using the Bradford assay (Bio-Rad). CAT activity was quantified by measuring the conversion of [ 14 C]chloramphenicol to its acetylated forms using a Fuji BAS bioimaging analyzer.
Preparation of Recombinant Proteins and Electrophoretic Mobility Shift Assay (EMSA)-The GST-fused C/EBP␣ and MBP-fused X proteins were affinity-purified, and the amount of purified proteins was determined by the Bradford assay. DNA binding reactions were carried out with the indicated amount of proteins as described previously (7), except that the proteins were preincubated in a reaction mixture without a probe for 10 min at room temperature and then incubated with labeled probes for 5 min. Samples were loaded on a 5% native polyacrylamide gel (40:1 acrylamide/bisacrylamide) in 0.5ϫ Tris borate/ EDTA, and the gels were dried and exposed to x-ray film. In Vitro Interaction Assay-One milliliter of bacterial extracts containing GST-C/EBP␣ or GST-bZip were incubated with 20 l of amylose resin bound to MBP or MBP-fused HBx proteins for 12 h at 4°C. After extensive washing (10 ϫ 10-min incubation with 1 ml of column buffer (10 mM sodium phosphate (pH 7.2), 0.5 M NaCl, 1 mM sodium azide, 1 FIG. 2. Identification of the region in EnII/Cp responsible for the synergistic effect of HBx and C/EBP␣. A, serial deletion analysis of EnII/Cp. Serial deletion clones of pCpBm were made using the restriction enzymes indicated at the top and were examined for the effect of HBx and C/EBP␣ using a CAT assay. Known C/EBP-binding sites of EnII/Cp are indicated by closed circles and are numbered consecutively. One microgram each of the deletion clones was transfected into HepG2 cells in combination with 4 g of pSVX2 and 0.5 g of pMSV-C/EBP␣ as described in the legend to Fig. 1. Relative CAT activities are listed on the right. Bm, BamHI; Sa, SacII; Ap, ApaLI; Sy, StyI; Hc, HincII; Da, DraI. B, identification of the binding site of C/EBP␣ in HBV EnII/Cp. Oligonucleotides covering sequences downstream of the StyI site (nt 1639) were used as probes for EMSA as indicated, and their sequences are described under "Experimental Procedures." A consensus C/EBP-binding site was used as a positive control. EMSA was accomplished with 50 ng of GST-C/EBP␣ as described under "Experimental Procedures." The DNA-protein complex is indicated by the arrow. C, HBx-enhanced binding of C/EBP␣ to HBV EnII/Cp and competition analysis. EMSA was carried out on the oligonucleotides at nt 1639 -1673 (lanes 1-5) and nt 1666 -1690 (lanes 6 -10) with 50 ng of GST-C/EBP␣ and 100 ng of MBP-X protein as indicated. Purified MBP was added to the reaction mixtures without MBP-X protein to adjust the amount of total proteins (lanes 1 and 6). A 50-fold molar excess of the unlabeled probes (self) and the consensus (C/EBPwt) and mutant C/EBP (C/EBPmt) oligonucleotides were added as competitors (lanes 3 -5 and 8 -10). The sequences of C/EBPwt and C/EBPmt are also described under "Experimental Procedures." mM EGTA, and 10% glycerol)), bound proteins were eluted with 50 l of column buffer containing 10 mM maltose. Ten microliters of eluted proteins were assayed by SDS-polyacrylamide gel electrophoresis and electrophoretically transferred to nitrocellulose membranes at 4°C. Membranes were blocked with 5% nonfat dry milk in phosphate buffer (80 mM Na 2 HPO 4 , 20 mM NaH 2 PO 4 (pH 7.5), 100 mM NaCl, and 0.1% Triton X-100), briefly washed, and incubated with anti-C/EBP␣ antibody (1:5000) for 1 h at 25°C. After extensive washing, membranes were incubated with a horseradish peroxidase-linked anti-rabbit mouse antibody (1:10,000) for 1 h at 25°C. Protein-antibody complexes were visualized by the ECL Western blotting detection system (Amersham Pharmacia Biotech) according to the manufacturer's instructions.

HBx and C/EBP␣ Synergistically Activate the Pregenomic
Promoter of HBV-HBV has been previously shown to have C/EBP-binding sites in the upstream enhancer, enhancer I , and EnII/Cp (37,44). We first examined the effect of HBx and C/EBP␣ on the whole HBV enhancer/promoters (pCENCAT) in HepG2 cells, which contain low levels of C/EBP␣ compared with normal liver (45). The effects of both proteins on each enhancer/promoter (pCpBm, pUEIXp and pEIXp) were then examined. As shown in Fig. 1B (lanes 3, 7, 11, and 15), HBx alone activated all four reporters. In contrast to previous reports (36,38), C/EBP␣ alone neither activated nor repressed notably the HBV enhancer/promoters in our experiments (Fig.  1B, lanes 2, 6, 10, and 14). However, cotransfection of HBx and C/EBP␣ significantly elevated the activities of pCENCAT and pCpBm plasmids (Fig. 1B, lanes 4 and 8), suggesting a synergistic effect of HBx and C/EBP␣ on HBV EnII/Cp. Contrary to our expectation, the cotransfection of HBx and C/EBP␣ had only a small effect on pUEIXp (Fig. 1B, lane 12) and inhibited the activation of pEIXp by HBx (lane 16), although they have C/EBP-binding sites. By EMSA, purified C/EBP␣ efficiently bound to the known C/EBP site of enhancer I, but weakly bound and did not bind to the downstream and upstream C/EBP sites of the upstream enhancer, respectively (data not shown). When compared with the consensus sequence of C/EBP, the C/EBP site at the upstream site of the upstream enhancer (nt 974 -995) does not have any homology, whereas the one at the downstream site (nt 1024 -1046) and the other one in enhancer I (nt 1170 -1198) have 80% homology (Table I). The discrepancy of the results from the CAT assay and EMSA indicates that C/EBP␣ did not act as an influential activator in EnI/Xp.

Identification of the Region in HBV EnII/Cp Responsible for the Synergism between HBx and C/EBP␣-We analyzed fur-
ther the EnII/Cp region that was highly activated by the cotransfection of HBx and C/EBP␣ (Fig. 1B). Previously, HBV EnII/Cp has been shown to contain several C/EBP-binding sites (37). To identify the responsible one in EnII/Cp, the effect of HBx and C/EBP␣ on the serial deletions of pCpBm was examined by CAT assay (Fig. 2A). The basal activities of plasmids pCpAp and pCpDa were lower than those of the other reporter plasmids, which is consistent with previous reports (42,46). The synergistic effect of HBx and C/EBP␣ was maintained on plasmid pCpSy, but not on plasmids pCpHc and pCpDa. In contrast, HBx alone slightly activated pCpHc. These results show that the StyI-HincII fragment (nt 1639 -1679) in the enhancer II region is responsible for the synergistic effect of HBx and C/EBP␣ on the HBV pregenomic promoter.
To find the binding site of C/EBP␣ in HBV EnII/Cp, we performed EMSA with oligonucleotides synthesized based on the sequences of the probable C/EBP sites and the previous data on this region (37,47,48). Purified GST-C/EBP␣ bound to the regions spanning nt 1639 -1673 and 1666 -1690 (Fig. 2B,  lanes 1 and 2; known C/EBP sites depicted as sites 1 and 2 in Fig. 2A), but not to the other regions (Fig. 2B, lanes 3-8; downstream of the HincII site). The consensus C/EBP site (C/EBPwt), used as a positive control, also bound to GST-C/ EBP␣ (Fig. 2B, lane 9). Siddiqui and co-workers (37) have identified the multiple C/EBP-binding sites in this region and the downstream sequences near the translation start site of core gene expression by DNase I protection analysis. However, downstream C/EBP sites (nt 1740 -1761, 1798 -1822, and 1826 -1846; known C/EBP sites depicted as sites 3-5 in Fig. 2A) were not bound by purified C/EBP␣ in our experiment (Fig. 2B,  lanes 6 -8), although all these sequences have 60 -70% homologies to the consensus C/EBP site (Table I). The discrepancy between our results and theirs may be due to the assay methods, and C/EBP␣ binding to DNA seems to require conservation of particular bases or secondary structure more than a consensus sequence (35,49). The binding of GST-C/EBP␣ to these two regions (nt 1639 -1673 and 1666 -1690) was augmented by the addition of purified MBP-X protein (Fig. 2C, lanes 2 and 7) without a supershifted band, which is consistent with previous reports (8,20). This binding was abolished by the unlabeled FIG. 3. Synergistic effect of HBx and C/EBP␣ on the minimal promoters of SV40 and thymidine kinase genes. A, schematic presentation of reporters. The plasmid pCATpromoter has a minimal SV40 early promoter (SV40-p) without a C/EBP-binding site. The plasmid pC/EBP-CAT has eight copies of the consensus C/EBP-binding site before the SV40 promoter of the pCATpromoter. The plasmid pBLCAT2 has a minimal thymidine kinase gene promoter (tk-p) with its own C/EBP-binding site. B, the role of the C/EBP-binding site in the activation of the minimal SV40 promoter by HBx and C/EBP␣. Three micrograms of each reporter, 4 g of pSVX2 (HBx), and 0.5 g of pMSV-C/EBP␣ (C/EBP␣) were transfected into HepG2 cells as indicated. C, effect of HBx and C/EBP␣ on the minimal promoter of the thymidine kinase gene. Two micrograms of pBLCAT2 were transfected into HepG2 cells with 2 g of pSVX2 (HBx) and 0.5 g of pMSV-C/ EBP␣ (C/EBP␣).
probe and consensus C/EBP oligonucleotide (Fig. 2C, lanes 3, 4,  8, and 9), but not by the mutant C/EBP oligonucleotide (lanes 5 and 10). These results show that HBx enhances the binding of purified C/EBP␣ to the two C/EBP-binding sites in nt 1639 -1690 of HBV enhancer II, which is consistent with the results of the CAT assay ( Fig. 2A).
Synergistic Effect of HBx and C/EBP␣ on the Heterologous Promoters-The synergistic effect of HBx and C/EBP␣ was tested first on the minimal SV40 promoter with or without C/EBP-binding sites (Fig. 3A). As shown in Fig. 3B, both reporter plasmids were activated by HBx (lanes 3 and 7), but the synergistic effect of HBx and C/EBP␣ was shown only with pC/EBP-CAT, which suggests a role of the C/EBP-binding site (compare lanes 4 and 8). C/EBP␣ alone slightly enhanced the activity of pC/EBP-CAT (Fig. 3B, lane 6). The effect of HBx and C/EBP␣ was then examined on the thymidine kinase gene promoter of herpes simplex virus, which has its own C/EBPbinding site located ϳ80 base pairs upstream of the transcription start site (43). As reported previously (50), C/EBP␣ alone highly activated the thymidine kinase promoter (Fig. 3C, lane  2). HBx alone slightly activated the thymidine kinase promoter (Fig. 3C, lane 3), but activated it Ͼ40-fold in the presence of C/EBP␣ (lane 4). The synergistic effect on the thymidine kinase promoter was stronger than that on HBV EnII/Cp and the SV40 promoter and was effective even with the low concentration of HBx and C/EBP␣ (data not shown). These results show that the cooperation of HBx and C/EBP␣ is also shown on the other promoters depending on the C/EBP-binding site.
HBx Binds to the bZip Domain of C/EBP␣-We analyzed the direct interaction of HBx and full-length C/EBP␣ or the bZip domain of C/EBP␣ using an in vitro interaction assay with MBP-X protein-coupled amylose resin (Fig. 4A). Both fulllength C/EBP␣ and the bZip domain of C/EBP␣ (GST-C/EBP␣ and GST-bZip, respectively) directly bound to MBP-X protein (Fig. 4A, lanes 2 and 4), but not to MBP alone (lanes 1 and 3).
Purified GST-C/EBP␣ and GST-bZip proteins were used as markers (Fig. 4A, lanes M1 and M2). MBP-X protein did not bind to GST protein (data not shown). The effect of HBx on the DNA binding affinity of C/EBP␣ was then examined by EMSA of the consensus C/EBP-binding site (the C/EBPwt probe). As shown in Fig. 4B (lanes 2 and 7), MBP-X protein increased the binding of both GST-C/EBP␣ and GST-bZip to the consensus C/EBP site. The addition of anti-C/EBP␣ antibody produced supershifted bands (Fig. 4B, lanes 3 and 8), and the complexes disappeared with the C/EBPwt competitor (lanes 4 and 9), but not with the mutant competitor (C/EBPmt) (lanes 5 and 10). These results show that HBx targets the bZip domain of C/EBP␣ and enhances its binding to DNA.
Central Region of HBx Is Necessary for the Direct Interaction with C/EBP␣-To assign the domains of HBx that are important for the interaction with C/EBP␣, derivatives of HBx were made (Fig. 5A). N-terminal deletions of HBx were made based on the two internal ATG codons, and the C-terminal deletion clone lacks its acidic transactivation domain. Smaller forms of HBx have been reported to be expressed from an intragenic promoter (51,52) and to transactivate differentially several class II promoters (53). Direct interaction of C/EBP␣ and the HBx derivatives (MBP-X protein, MBP-middle X protein, and MBP-small X protein) was examined by in vitro interaction assay as described above (Fig. 5B). Full-length and middle X proteins bound directly to both GST-C/EBP␣ (Fig. 5B, lanes 2  and 3) and GST-bZip (lanes 6 and 7), but small X protein (lanes 4 and 8) and MBP alone (lanes 1 and 5) did not. Purified GST-C/EBP␣ and GST-bZip proteins were used as markers (Fig. 5B, lanes M1 and M2). Coomassie Blue staining with an equal volume of bacterial extracts showed that the result was not due to the differences in the amounts of HBx proteins expressed in Escherichia coli (data not shown). These results suggest that the central region of HBx is necessary for the direct interaction with C/EBP␣. EMSA and the CAT assay using HBx derivatives showed somewhat different results. MBP-middle X protein did not efficiently enhance the binding of C/EBP␣ to the C/EBP site in EMSA (Fig. 5C, compare lanes 2 and 3). As expected, MBPsmall X protein showed no effect (Fig. 5C, lane 4). The results of the CAT assay using pC/EBP-CAT as a reporter also showed that smaller forms of HBx did not synergize with C/EBP␣ as FIG. 4. HBx directly interacts with the bZip domain of C/EBP␣ and enhances its binding to the consensus C/EBP site. A, in vitro interaction assay of the direct interaction of HBx and C/EBP␣ as described under "Experimental Procedures." MBP or MBP-X protein was used as a bait for full-length C/EBP␣ (lanes 1 and 2) or the bZip domain of C/EBP␣ (lanes 3 and 4). Eluted proteins were subjected to Western blotting using anti-C/EBP␣ antibody. Purified GST-C/EBP␣ and GST-bZip were used as markers (lanes M1 and M2, respectively). B, HBxenhanced binding of C/EBP␣ to the consensus C/EBP site. Fifty nanograms of full-length C/EBP␣ (GST-C/EBP␣) or the bZip domain of C/EBP␣ (GST-bZip) were incubated with the 32 P-labeled consensus C/EBP oligonucleotide (C/EBPwt) in the presence (lanes 2-5 and 7-10, respectively) or absence (lanes 1 and 6) of 100 ng of MBP-X protein.
Anti-C/EBP␣ antibody (Ab) was added just before the addition of the probe (lanes 3 and 8). A 50-fold molar excess of unlabeled C/EBPwt (lanes 4 and 9) or C/EBPmt (lanes 5 and 10) was used as a competitor. efficiently as full-length X protein (Fig. 5D, compare lane 2 with lanes [3][4][5]. In contrast to full-length HBx (Fig. 3B, lane 7), smaller forms of HBx by themselves did not activate pC/EBP-CAT (data not shown). These results suggest that the functional synergism of HBx and C/EBP␣ requires more than direct interaction of both proteins. The whole region of HBx is likely to be needed. DISCUSSION We have demonstrated here that HBx enhanced the binding activity of C/EBP␣ on the DNA by directly interacting with its bZip domain and that the interaction of both proteins synergistically transactivated the pregenomic promoter of HBV. These results revealed the C/EBP␣ as a cellular target of HBx for the transactivation of the HBV pregenomic promoter and can accordingly explain its liver specificity. The role of C/EBP family proteins in the function of HBx has been suggested previously by indirect evidence (39,40), and recently, the ␤-iso-form of C/EBP has been reported to interact with HBx without any effect on the functional implications in vivo or in cultured cells (20). However, our results clearly showed the functional synergism and direct interaction of HBx and C/EBP␣ both in HepG2 cells and in vitro.
The synergistic effect of HBx and C/EBP␣ differentially regulated EnI/Xp and EnII/Cp of HBV, although both promoters have C/EBP-binding sites (Fig. 1). This result is also implied by the effect on the heterologous promoters of SV40 and thymidine kinase genes (Fig. 3) and is relevant to a previous report that showed the differential regulation by C/EBP␣ of viral enhancer/promoters including HBV enhancer I depending on the promoter context (50). Activation of its own promoter (X promoter) by HBx seems to be mainly mediated by the AP-1-binding site in HBV enhancer I (7).
Serial deletion analysis of the HBV pregenomic promoter and EMSA revealed the region (nt 1639 -1679) with two C/EBP-binding sites functionally responsible for the effect of shown is a schematic presentation of fulllength HBx and its deletion mutants. The HBx derivatives used for the mammalian and bacterial expression are described under "Experimental Procedures." HBx, full-length X protein; Mx, middle X protein; Sx, small X protein; ⌬C, C-terminal deletion mutant of HBx. B, the direct interaction of C/EBP␣ and HBx derivatives was subjected to in vitro interaction assay. Full-length C/EBP␣ (lanes 1-4) or the bZip domain of C/EBP␣ (lanes 5-8) was assayed as described in the legend to Fig. 4. Purified GST-C/EBP␣ and GST-bZip were used as markers (lanes M1 and M2, respectively). C, 50 ng of GST-C/ EBP␣ were incubated with 100 ng each of the purified HBx derivatives (MBP-X protein (MBP-X), MBP-middle X protein (MBP-Mx), and MBP-small X protein (MBP-Sx)). EMSA was accomplished as described under "Experimental Procedures." D, 4 g each of the HBx derivatives (pSVX2, pSVX⌬BA, pSVX⌬BH, and pSVX⌬SH) and 0.5 g of pMSV-C/EBP␣ were transfected into HepG2 cells as indicated, and the CAT assay was performed.
C/EBP␣ and HBx (Fig. 2). This region has also been shown to bind other transcription factors (54,55), of which HNF-4 and Sp1 were shown to synergize with C/EBP␣ (56,57). Other known C/EBP-binding sites of HBV EnII/Cp also have homologies to the consensus C/EBP site similar to these sites, but did not bind to purified C/EBP␣ in our experiment (Fig. 2B). Extensive analysis of the binding sites of C/EBP␣ has shown that bases at Ϯ3 and Ϯ4 in the 10-base pair consensus sequence are most important and that bases at Ϯ1 and Ϯ5 are least important (35,49). In addition, the consensus sequence of the halfsite 5Ј-RTTGC-3Ј appears to be more important for C/EBP␣ binding than the palindrome structure of DNA sequences. The C/EBP␣ sequences in the regions spanning nt 1639 -1673 and 1666 -1690 are more similar to the above requirements than other sites (Table I).
The C/EBP family proteins were reported to interact directly and/or functionally with various transcription factors, including HNF-1 (58), v-Myb (59), AML1 (60), glucocorticoid receptor (61), HNF-4 (57), Sp1 (56), and NF-B (62-64), which synergistically activates the responsive promoters through their overlapping binding sites. Among these factors, Sp1 synergizes with C/EBP␣, but not with C/EBP␤ (56), and the interactions of NF-B with C/EBPs are so complex that p50 and p65 subunits of NF-B seem to selectively interact with the different isoforms of C/EBP to activate or repress promoters depending on the cis-elements. From these facts, it is likely that the C/EBP family proteins can interact with HBx differentially to act on the different enhancer/promoters, although we did not test the other C/EBP isoforms except for C/EBP␣.
HBx is somewhat different from other C/EBP-interacting proteins in its inability to bind DNA. The role of HBx in the interaction with C/EBP␣ is similar to that of human T-cell lymphotrophic virus type 1 tax protein in its ability to stimulate the DNA binding of some bZip proteins such as the ATF/ CREB family protein and GCN4 (65,66). The two proteins also share the common feature of forming a complex with bZip proteins only in solution, but dissociate under native gel electrophoresis conditions (65). The human T-cell lymphotrophic virus tax protein has been shown to recognize a basic region of the bZip protein and to incorporate into the ternary complex as a dimer (66). The detailed mechanism of HBx action should be studied further. HBx has been shown to coactivate potent activation domains (9) and to interact with basal transcription factors (13)(14)(15)(16). In this respect, the functional role of HBx in cooperation with C/EBP␣ may include its interaction with other activators or basal transcription machinery, not merely enhancing the binding of C/EBP␣. This hypothesis may be supported by the synergism between C/EBP␣ and the acidic activation domain of VP16 fused to the DNA-binding domain of HNF-1 (58) and also by the diverse interactions of C/EBPs with other transcription factors to activate responsive promoters.
Domain analysis tentatively revealed that the C-terminal two-thirds of HBx are important in the direct interaction with the bZip domain of C/EBP␣ (Fig. 5B), but full-length HBx seems necessary for functional cooperation with C/EBP␣ (Fig.  5D). The importance of the C-terminal activation domain of HBx in the interaction with basal transcription factors (13)(14)(15)(16) is relevant to this result. The N-terminal regulatory domain of HBx was not necessary for the direct interaction with C/EBP␣, and the C-terminal acidic activation domain of HBx alone was not sufficient. The importance of the bZip domain of C/EBP␣ in the interaction with HBx is consistent with the previous result on the interaction of HBx and CREB. It also suggests a role of the bZip domain that is more than just dimerization and DNA binding, which is also the case in the interaction of C/EBP␣ with other transcription factors.