The Hepatitis B Virus X Protein Is a Co-activator of Activated Transcription That Modulates the Transcription Machinery and Distal Binding Activators*

Hepatitis B virus X protein (HBx) transactivates viral and cellular genes through a wide variety ofcis-elements, but the mechanism has not been well elucidated. Evidence for nuclear events in HBx transactivation has been reported. Here we examine the role of HBx in modulation of transcription with a transient transfection system and an in vitro transcription assay. Reporters bearing Gal4-binding sites were applied to avoid the effects of endogenous transcription factors with or without signaling processes. The Gal4-DNA binding domain fused form of HBx exhibited no effect on Gal4-responsive reporters. However, HBx augmented activated transcription by transcriptional activators, suggesting HBx retains a co-activator but not a transcriptional activator function. The functional domain for co-activation was the same as that for HBx transactivation, and the transcription factor IIB- and RNA polymerase II subunit 5-interacting sites of HBx, which were critical for HBx transactivation, were shown to be crucial for the co-activation function. Importantly, HBx stimulated transcription on templates bearing the X responsive elements in vitro with endogenous activators. These results imply that HBx acts as a co-activator that modulates transcriptional machinery and distal-binding activators, which may explain one of the mechanisms of transactivation by HBx when localized in nuclei.

Epidemiological studies have established a close association between chronic infection with human hepatitis B virus (HBV) 1 and the development of primary hepatocellular carcinoma. However, the nature of this association and the mechanism by which HBV infection leads to tumor formation are not well understood (1,2). Studies of HBV-related hepadnaviruses have shed some light on the positive role of the X gene product in hepatocarcinogenesis. The woodchuck and ground squirrel hepatitis viruses, the genomes of which harbor the X gene, are associated with development of hepatocellular carcinoma. In contrast, the oncogenic potential of avian hepadnaviruses, that are devoid of an X open reading frame, has not been conclusively established (3). HBx, the smallest protein encoded by the HBV genome, is essential for the viral life cycle in vivo (4,5). HBx is a transactivator that regulates a wide range of cellular and viral genes transcribed by RNA polymerase II (6 -11). RNA polymerase III transcribed genes are also transactivated by HBx (12,13). The known X responsive elements (XRE) include activated protein-2, activated transcription factor, CCAAT and enhancer-binding protein, nuclear factor-B sites, and the serum responsive element. Several host genes important for cell proliferation and acute inflammatory responses, such as c-fos, c-jun, tumor necrosis factor-␣, endothelial cell adhesion molecule 1 (ECAM-1), and human interleukin-8, are activated by HBx (6, 14 -16). This broad gene regulation function suggests that HBx not only up-regulates the expression of HBV genes by transactivating the HBV enhancer but also modifies the environment by transactivating cellular genes in infected cells to facilitate viral replication. It also implies that HBx plays a positive role in hepatocellular carcinogenesis (17)(18)(19).
The mechanisms of HBx transactivation appear to be complex. Since HBx cannot bind double strand DNA directly, protein-protein interaction is crucial for HBx transactivation (20). The reported HBx binding proteins include a variety of factors for transcription (21)(22)(23)(24)(25)(26), a probable DNA repair enzyme (27), a human homologue of Drosophila 20 S proteasome subunit (28,29), and the tumor suppressor p53 (30 -32). These findings are so diverse that the mechanism of HBx transactivation remains unclear. It has also been reported that HBx mediates transcriptional activation through modification of signal transduction pathways in the cytoplasm (7,(33)(34)(35)(36). Although it is not well defined, the subcellular localization of HBx seems to be mainly cytoplasmic and some portion nuclear (37,39), 2 HBx thus may have a dual role in transcriptional regulation; cytoplasmic HBx influences the regulation of second messenger systems while nuclear HBx may function at the promoter level. Nuclear HBx may directly interact with the transcription machinery to facilitate transcription (22,24,38). Recently, we found that HBx specifically binds to RPB5, a common subunit shared by eukaryotic nuclear RNA polymerases I, II, and III and transcription factor TFIIB and that the trimeric interactions may be involved in HBx transactivation (22,24).
To gain insight into the transactivation mechanism of HBx, we examined the role of HBx in modulation of transcription with a transient transfection in vivo system and an in vitro * This work was supported in part by a Grant-in-aid from the Ministry of Education and Culture of Japan. 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.
¶ To whom correspondence should be addressed: transcription assay using nuclear extract and recombinant proteins. We utilized artificial reporters bearing Gal4 binding sites since in this system the effects of endogenous factors with or without signaling may be largely avoided. Here we present results showing that HBx cannot act as a transcriptional activator but that it co-activates activated transcription. The TFIIB-and RPB5-interacting sites of HBx were also shown to be important for HBx co-activation by using substituted HBx mutants. HBx stimulated transcription from templates bearing the X responsive element (XRE) in an in vitro transcription assay with endogenous transcriptional activators in nuclear lysates. These results suggested that HBx acts as a co-activator in transcription through modulation of the transcriptional machinery and distal binding activators, and this co-activation could explain the transactivation mechanism of nuclearly localized HBx in vivo.

EXPERIMENTAL PROCEDURES
Plasmids-The mammalian expression plasmid pSG5UTPL has been described previously (40). HBx, HBx-5D1, and HBx-3D5 encode HBx 1-154 amino acids (full size), 51-154 amino acids, and 1-50 amino acids, respectively (22,24,40). In substitution HBx mutant HBx-m120, WEE at 120 -122 positions is substituted by AAA. In HBx-m76 and HBx-m93, four amino acid residues were substituted by GAGA, ARRM at positions 76 -79 of HBx-m76 and LHKR at positions 93-96 of HBx-m93, respectively (24). The Escherichia coli histidine-tagged protein expression plasmid pLHis was derived from pET11d, by replacing the NdeI-BamHI fragment with the annealed complementary oligonucleotides, TATGAATTCCATGAAGCTTGGATC and GATCCAAGCTTCAT-GGAATTCATA, to introduce the EcoRI, HindIII, and BamHI digestion sites. Plasmids pLHis-HBx, pLHis-Xm76, pLHis-Xm93, and pLHis-Xm120 were constructed by inserting the coding sequence of wild-type and each mutant HBx into the EcoRI and BamHI sites of pLHis, respectively. The E. coli expression plasmids His6-Gal4-VP16, pFlag-Gal4-E1a, and pFlag-Gal4-p53 encode activation domains of the each transcriptional activators fused to the Gal4 DNA binding domain. The in vitro transcription template plasmids pG5MLP-G-less and p⌬53-MLP-G-less contain an adenovirus major late promoter (AMLP) and a G-less cassette as a reporter sequence, and the former has five Gal4 binding sites upstream of the G-less cassette as a regulatory element (41,42). Plasmid pHis6-Gal4 was derived from pHis6-Gal4-VP16 by deleting the VP16 sequence by HindIII and BamHI digestion. pHis6-Gal4-HBx was constructed by inserting the HBx sequence into the HindIII and BamHI sites of pHis6-Gal4. The sequence of Gal4-VP16 was amplified by a polymerase chain reaction from pHis6-Gal4-VP16 and inserted into the EcoRI and BamHI sites of pSG5UTPL to construct the mammalian expression vector pSG5UTPL-Gal-VP16. Plasmids pSG5UTPL-Gal and pSG5UTPL-Gal4-HBx were derived by deletion or replacement with the HBx coding sequence of the VP16 sequence. The HBx-responsive CAT reporter pHECx2CAT has two tandems of the HBV enhancer I core, an X-responsive element upstream of a SV40 promoter (43). The pGal-CAT reporter was constructed by insertion of five Gal4 binding sites into pSV2CAT. The pGal-Luc reporter was constructed by insertion of five Gal4 binding sites and the AMLP promoter sequence into pGL3 basic vector (Promega). The Gal4 binding sites in pG5MLP-G-less were replaced with the following annealed complementary oligonucleotides, TGACTCATGACTCA and AAGGTA-CCTTGCTGACGCAACCCCCACTGGCTTGCTGACGCAACCCCCAC-TGGCGAGCTC to create plasmids pAP1MLP-G-less and pENHIC-MLP-G-less, which harbor two copies of AP1 binding site HBV enhancer I core, respectively. All the constructs were sequenced using Taq sequencing kits and a DNA sequencer (370A; Applied Biosystems).
Histidine-tagged proteins were purified by incubating the sonication supernatant with nickel-resin, followed by extensive washing with denaturation binding buffer and by modified denaturation washing buffer (6 M guanidine chloride, 20 mM sodium phosphate, 500 mM NaCl, pH 6.0, 10 mM imidazole) and elution with denaturation elution buffer (20 mM sodium phosphate, 500 mM NaCl, pH 4.0). The eluted proteins were renatured by dialysis with sequentially reducing concentrations of urea in dialysis buffer (20 mM sodium phosphate, 500 mM NaCl, pH 7.8).
Flag-tagged Gal4-E1A and Gal-p53 were expressed in E. coli by induction for 3 h at 30°C with 0.4 mM IPTG. Cells were harvested and sonicated in TBST buffer (50 mM Tris-HCl, pH 7.5, 0.2 M NaCl, 0.05% Tween 20). After centrifugation, the extracts were incubated with monoclonal anti-FLAG antibody (M2)-conjugated Sepharose 4B resin (Eastman Kodak Co.) for 6 h at 4°C. The beads were collected and washed four times with TBST buffer followed by three elutions with 0.4 mM FLAG peptide in TBST. The eluted proteins were dialyzed in BC100, divided into aliquots, and stored at Ϫ80°C.
GST and GST-fused proteins were prepared as described previously (29). Nonfused HBx was prepared by digesting GST-HBx with thrombin (29). HBx protein was separated by SDS-polyacrylamide gel electrophoresis and eluted from the gels. After renaturation and dialysis in BC100 buffer, HBx protein was divided into aliquots and stored at Ϫ80°C.
In Vitro Transcription-HeLa cell nuclear extracts were prepared as described previously (44) and dialyzed in BC100 buffer. In vitro transcription was carried out as described elsewhere (41). Briefly, 25 l of reaction mixtures contained 50 ng of pG5MLP-G-less and p⌬53-MLP-G-less supercoiled plasmid templates, 40 g of HeLa cell nuclear extract, 0.5 mM UTP and ATP, 0.025 mM CTP, 5 Ci of [ 32 P]CTP, 4 units of RNase T1, 0.8 unit of RNase inhibitor, 0.1 mM O-Me-GTP, 0.1 mg/ml bovine serum albumin, 12 mM Tris-HCl, pH 7.9, 4 mM MgCl 2, 0.06 mM EDTA, and 12% glycerol. The reaction proceeded by incubation at 30°C for 60 min and was stopped by addition of 175 l of stop buffer (1% SDS, 10 mM EDTA, 0.1 mg/ml glycogen, 100 units/ml proteinase K). After incubation at 37°C for a further 30 min, RNA product was purified by organic extraction followed by ethanol precipitation, resolved in denaturing 4% polyacrylamide gel, and visualized by autoradiography. Signals were quantified using a Bioimage analyzer (BAS1000; Fuji).
Western Blot Analysis-Transfection of HepG2 cells and preparation of cell lysates were described previously (40). Fifty micrograms of proteins of cell lysates were fractionated in SDS-polyacrylamide gels and Western blotted as described previously (24). Anti-HBx antibody has been described elsewhere (24). Anti-Gal4 DNA binding domain (DBD) antibody was purchased from Santa Cruz Biotechnology.
Cell Culture, Transfection, CAT, and Luciferase Assays-HepG2 cells were tansfected with different combinations of plasmid DNA as indicated in the figures. Total cell lysates were prepared from cells harvested 48 h after transfection. The CAT assay reactions proceeded for 60 min at 37°C using 20 g of protein from the transfected cell lysates (40). The fractionated TLC plates were exposed to imaging plates, and CAT activities were measured as percentages of conversion to the acetylated forms of [ 14 C]chloramphenicol (Amersham Pharmacia Biotech) (percent acetylation) using a Bioimage analyzer (BAS1000; Fuji). Transfection and CAT assays were performed at least three times with each combination of transactivator and CAT reporter constructs. Luciferase assay was done using 5 l of protein from the transfected cell lysates with the luciferase assay system (Promega). Representative data are shown.

RESULTS
HBx Co-activates Transcription by Gal-VP16 whereas Gal-HBx Is Inactive upon Gal-CAT in Vivo-Having established that HBx binds to TFIIB and RPB5, and the trimeric interaction of these three factors is involved in HBx transactivation, we were interested in the role of HBx in transcriptional modulation. First, we addressed whether HBx can act as a transcriptional activator when tethered to the distal region of reporter genes. For this purpose, Gal-HBx, the Gal4 DNA binding domain fused to the N terminus of HBx, was constructed. The effect of Gal-HBx on the expression of Gal-CAT, a CAT reporter with a SV40 promoter driven by five Gal4-binding sites, was determined in transiently transfected HepG2 cells by CAT assay. Gal4 DNA binding domain, Gal, and the Gal-fused VP16 activation domain, Gal-VP16, served as negative and positive controls, respectively. Gal-VP16 markedly activated the expression of Gal-CAT (Fig. 1A, lanes 2-5), whereas Gal has no effect on reporter gene expression (lanes 6 -9), indicating that the transient transfection and CAT assay are suitable for this functional assay. Unexpectedly, Gal-HBx also did not activate the reporter (lanes 10 -13). The inability of Gal-HBx to activate the expression of Gal-CAT was not because of low levels of expression or instability of Gal-HBx protein since Gal-HBx and HBx were expressed at similar levels in the transfected cells (Fig. 1B, compare lanes 1 and 2). As we reproducibly detected no activity of Gal-HBx with various amounts of transfected plasmid DNA (data not shown), the possibility of self-squelching as seen with other transcriptional factors was unlikely (45,46). Importantly, Gal-HBx exhibited transactivation activity upon reporters harboring XRE, indicating that Gal-HBx was correctly folded and functionally active in the transfected cells (Fig. 1C, lanes 11-13). It was less likely that Gal-HBx bound to the reporter DNA inefficiently since Gal-HBx repressed the activation by Gal-VP16 in a dose-dependent manner, similarly to Gal-DB competing the Gal4-responsive element (Fig. 1D,  lanes 10 -12, and data not shown). These results indicated that HBx cannot act as a transcriptional activator when artificially tethered to the promoter.
Gal-HBx Is Inactive while HBx Functions as a Co-activater in Vitro-Since the in vivo experiment was somewhat indirect, it is important to confirm these observations in a cell-free system. We used an in vitro transcription with supercoiled plasmid pG5MLP-G-less, containing five Gal4 binding sites as a template (41). pMLP⌬53-G-less plasmid DNA, with no Gal4 binding sites, was included as a control of basal transcription. All of the basal transcription factors were provided by HeLa cell nuclear extract. Histidine-tagged Gal-HBx, Gal-VP16, and Gal4-DNA binding domain (Gal) were expressed in and purified from E. coli ( Fig. 2A). As shown in Fig. 2B, under our experimental conditions, 15 and 30 ng of Gal-VP16 activated transcription from pG5MLP-G-less by 2-and 4-fold, respectively. However, similar to Gal alone, Gal-HBx did not stimulate transcription. Similar DNA binding abilities of Gal-HBx and Gal-VP16 were detected by electrophoretic mobility shift assay using a probe containing two Gal4 binding sites, suggest-ing that the bacterially expressed proteins were correctly folded and functionally active (Fig. 2C). These results confirmed the in vivo observation that Gal-HBx could not act as a transcriptional activator when tethered to DNA.
We next tested the co-activator activity of HBx using the same in vitro transcription system. When histidine-tagged HBx was added to the reaction together with Gal-VP16, augmented transcription was observed. The co-activation by HBx was dosedependent (Fig. 3A, lanes 4 and 5), whereas the negative controls, GST and bovine serum albumin, had no effect on transcription (lanes 6 and 7 , and data not shown). To confirm this result, nonfused HBx protein derived from digestion of GST-HBx was also tested in vitro. Similar to histidine-tagged HBx, the nonfused HBx co-activated transcriptional activation by Gal-VP16 (Fig. 3B). GST-HBx also co-activated transcriptional activation by Gal-VP16, although to a lesser extent (data not shown). These results clearly demonstrated that HBx acts as a co-activator facilitating activated transcription by Gal-VP16.
To examine whether the co-activation by HBx is a general effect, we tested the effects of HBx on transcriptional activation by other transcriptional activators. Gal-p53, a protein consisting of Gal4-DNA binding domain fused to two tandem repeats of the p53 activation domain, and Gal-E1a, a Gal4-DNA binding domain fused to adenovirus E1a activation domain, were examined. The p53 activation domain, similar to VP16, is an acidic activator, whereas E1a is a nonacidic activator (48). As shown in Fig. 4, HBx stimulated both Gal-p53-and Gal-E1aactivated transcription in vitro. Similar co-activation was also detected in vivo (data not shown). Since HBx co-activated transcriptional activation by different groups of activators, this protein seems to be a general co-activator at least of acidic and nonacidic transcriptional activators.
RPB5 and TFIIB Bindings of HBx Are Involved in HBx Co-activation-We previously reported that HBx consists of a functional transactivation (51-148 amino acids) and a regulatory (1-50 amino acids) domains (40). Luciferase assay showed that the HBx co-activation activity resided in the transactivation (HBx-5D1) but not the regulatory domain (HBx-3D5) of HBx (Fig. 5A). This suggested that the HBx transactivation observed in vivo may have been due to a co-transactivator function. To examine this possibility, we investigated whether the RPB5 and TFIIB interactions, which are necessary for HBx transactivation, are crucial in HBx co-activation by Gal-VP16. To this end, we chose Xm93 and Xm120, which have reduced binding abilities with RPB5 and TFIIB, respectively (24). Xm76, which has RPB5 and TFIIB binding activity similar to that of wild type HBx (24), was used as a control. Similar to wild type HBx, Xm76 co-activated transcriptional activation by Gal-VP16, whereas both Xm93 and Xm120 lost co-activation activity (Fig. 5A). The inability of Xm93 and Xm120 to stimulate transcription was not due to low levels of expression or instability of mutated HBx proteins, since Xm76, Xm93, Xm120, and wild type HBx were expressed at similar levels in the transfected cells (24) (data not shown). Consistent with the results of the in vivo experiment, Xm76 co-activated transcriptional activation by Gal-VP16 in vitro, whereas both Xm93 and Xm120 failed to show co-activation (Fig. 5B, compare lanes 5-8  and 4). These results indicated that both of RPB5 and TFIIB interacting sites are important not only for the HBx transactivation but also for HBx co-activation.
HBx Activates XRE in in Vitro Transcription-If the HBx transactivation previously observed in vivo was due to the co-activator function of this protein, HBx might augment activated transcription by transactivators that bind X-responsive cis-elements. As in above experiments we utilized artificial promoters and artificial Gal-fused activators; it is important to see whether the results reflect the mechanism of HBx transactivation. Therefore, we constructed several reporters containing the HBV enhancer I core or AP1-binding site, two well known XREs, upstream of the AMLP promoter and G-less cassette to serve as templates of in vitro transcription. HBx activated transcription of pENHICMLP-G-less and pAP1MLP-G-less templates in a dose-dependent fashion (Fig. 6). Since HBx has little effect on basal transcription (see ⌬p53 G-less bands), these observations appeared to be the result of co-activation by HBx of transcription activated by endogenous activators.  5), pSG5UTPL-HBx-m93 (lane 6), and pSG5UTPL-HBxm120 (lane 7), respectively. B, in vitro transcription. The in vitro transcription assay proceeded as described in Fig. 2. The amounts of recombinant proteins tested are: lane 1, Gal-VP16, 15 ng; lanes 2-7, Gal-VP16, 30 ng, plus 30 ng of GST, wild-type HBx, HBx-m120, HBx-m93, HBx-m76, respectively. The relative transcription activates are shown at the bottom.

DISCUSSION
Transactivation has been well documented as one of the discrete functions of HBx, although the mechanism remains controversial. HBx transactivates a wide range of cellular and viral genes transcribed by polymerases II and III (6, 7, 9 -12, 25, 26, 33, 47, 49), and a variety of cis-elements have been determined as XREs. As HBx is unable to bind double strand DNA, HBx may directly modulate transcription through interaction with a variety of transactivators (21,23,25,31) or factor(s) involved in transcription (22)(23)(24)26). They may also indirectly modulate transcription by interacting with factors such as those involved in signaling. These possible mechanisms of transactivation by HBx may not be mutually exclusive, since HBx seems to distribute in both the cytoplasm and nucleus, which may have resulted in the different mechanisms proposed previously. Several lines of evidence from independent groups indicate the direct interaction of HBx with TFIIB and RPB5 or polymerase II (23,24) and the direct effects of HBx on activated transcription in vivo and in vitro. This strongly suggests that the transactivation mechanism of nuclearly localized HBx involves the direct interaction of HBx and transcription machinery.
Here, we present results supporting the hypothesis that the HBx transactivation is the result of the co-activator function of HBx. For focusing on the effect of HBx on transcriptional modulation, we utilized artificial reporters bearing Gal4 binding sites. With this system, the effects of endogenous factors with or without signaling might be avoided. HBx could not act as a transcriptional activator when artificially tethered to promoters either in vivo and in vitro, but it can co-activate transcription activated by various activators including a variety of artificial activators. The co-activation by HBx is general at least to acidic and nonacidic activators. The requirements of the functional domain of HBx dissected by mutations for the transactivation were the same as for co-activation. Importantly, HBx co-activated the transcription activated by endogenous activators in vitro. The co-activation property could well explain the transactivation mechanism of nuclearly localized HBx in vivo.
Although Gal-HBx retains DNA binding ability comparable to those of Gal or Gal-VP16, our results clearly show that Gal-HBx could not transactivate pGal-CAT. The distance between the Gal4 DNA-binding elements and the TATA element may not explain the inability of transactivation function of HBx in our system, since similar results were obtained using pGal-CATϩ4, in which an additional 4 base pairs were inserted between the TATA box and Gal4 DNA-binding site (data not shown). Our results are in contrast to those of previous reports in which HBx was shown to act as a transcriptional activator when fused to LexA or CCAAT and enhancer-binding protein DNA binding domains (47,50). The reason for this discrepancy is not clear at present, although there were several differences between the experimental systems. First, the cell lines used were different. Unger and Shaul (50) used the human hepatoma Alexander cell line that has endogenous CCAAT and enhancer-binding protein. This cell line contains HBV proteins including HBx expressed from an integrated HBV subgenome (51,52). Whereas Seto et al. (47) performed their experiments in CV-1 monkey kidney cells. In this study, we used HepG2, a human hepatoma cell line free of HBV infection, and similar results were obtained by transfecting Hep3B, another HBVnegative human hepatoma cell line (data not shown). Second, there were differences in DNA binding domains and responsive elements utilized, which may have affected the affinity and specificity of DNA binding. Third, the core promoters used were different. Unger and Shaul (50) used the HSV TK promoter, Seto et al. (47) used human metallothionein IIA promoter, while we used the SV40 promoter and AMLP. It has been reported that different activators require different basal elements of core promoters. For example, in the AMLP promoter, SP1 glutamine-rich activation domain preferentially activates through Inr, whereas VP16 activation is dependent on both FIG. 6. HBx co-activates endogenous activators. The in vitro transcription assay was made as described in Fig. 2, except that the test templates are pAP1MLP-G-less (lanes 1-4) and pENHICMLP-G-less (lanes 5 and 8). In lanes 3 and 7, 10 ng of HBx were added, and in lanes 4 and 8, 30 ng of HBx were added, respectively. Thirty nanograms of GST were included in lanes 2 and 6 as negative controls. The position of the 380-nucleotide (nt) test and 280-nucleotide control RNA products are indicated. The relative transcription activities are shown at the bottom. TATA and Inr (53). At present, the promoter basal element requirement for HBx is unknown, and different promoter contexts may affect the results of transcription.
Free HBx was capable of co-activating transcription, but when tethered to the promoter HBx did not stimulate, or even repress, transcriptional activation, perhaps by competing with activators for DNA binding. We previously reported that the interactions among HBx, TFIIB, and RPB5 are involved in HBx transactivation in vivo. In this study, we demonstrated that these interactions are also important for HBx co-activation both in vitro and in vivo, suggesting that co-activation through these interactions is involved in HBx transactivation in vivo. The close and fixed position of HBx relative to polymerase II and TFIIB, which might be crucial for HBx function, could be hindered when HBx is recruited to DNA through the fused DNA binding domain. Since both the associations of HBx with activators and the transcriptional machinery are indispensable, we propose a model for its function as shown in Fig. 7. As a co-activator, HBx may bridge the activators and the initiation complex through direct interactions with TFIIB and RPB5 and direct or indirect interactions with the distal element-bound activators, or modulate communications between activators and the transcription machinery (Fig. 7A). Alternatively, HBx may release co-repressors from the machinery to enable the communication between activators and the transcription machinery (Fig. 7B). The latter is consistent with the previous observation that overexpression of RPB5 and its HBx-interacting region could transactivate reporters harboring XRE (22). The recently identified novel RPB5-mediating protein, RMP, is a good candidate since it exhibits co-repressor activity and counteracts HBx transactivation. 3 Our model stresses that the communication between TFIIB and RPB5 is an important step in modulating activated transcription (38).
Several aspects of the mechanism of HBx co-activation remain to be elucidated, including how HBx affects activated transcription but not basal transcription, and how HBx augments transcription of only a fraction of genes defined by XRE (20). The role of TFIIB in activated transcription seems not to be general for different activators as reported in studies using TFIIB mutations (54). The ability of the transcription machinery, such as TFIIB and polymerase II, to have differential roles may be intrinsic or be achieved through communication with various co-activators and co-repressors. Alternatively, HBx may interact and modulate the activation process by transcriptional activators such as facilitating dimerization and binding the cis-element (25,55), then modulate basal machinery components and cooperatively induce the isomerization of the preinitiation complex to form an open complex. Such multiple roles of HBx in transcriptional modulation may partly explain the promoter selectivity in activated transcription.
The transcription process includes several steps including initiation, promoter clearance, elongation, and termination (56). HBx directly interacts with components of the initiation complex, making it possible to stimulate the initiation step. However, the involvement of HBx in regulation of other steps is also possible. HBx interacts with TFIIH, a basal transcription factor involved in both initiation and elongation. VP16, which also interacts with TFIIH, has been found to stimulate both initiation and elongation (48). Further studies are needed to determine which step (s) of transcription is modulated by HBx and how HBx exerts its effect on transcription.