The Hepatitis B Virus X Protein Functionally Interacts with CREB-binding Protein/p300 in the Regulation of CREB-mediated Transcription*

The hepatitis B virus infects more than 350 million people worldwide and is a leading cause of liver cancer. The virus encodes a multifunctional regulator, the hepatitis B virus X protein (HBx), that is essential for virus replication. HBx is involved in modulating signal transduction pathways and transcription mediated by various factors, notably CREB that requires the recruitment of the co-activators CREB-binding protein (CBP)/p300. Here we investigated the role of HBx and its potential interaction with CBP/p300 in regulating CREB transcriptional activity. We show that HBx and CBP/p300 synergistically enhanced CREB activity and that CREB phosphorylation by protein kinase A was a prerequisite for the cooperative action of HBx and CBP/p300. We further show that HBx interacted directly with CBP/p300 in vitro and in vivo. Using chromatin immunoprecipitation, we provide evidence that HBx physically occupied the CREB-binding domain of CREB-responsive promoters of endogenous cellular genes such as interleukin 8 and proliferating cell nuclear antigen. Moreover expression of HBx increased the recruitment of p300 to the interleukin 8 and proliferating cell nuclear antigen promoters in cells, and this is associated with increased gene expression. As recruitment of CBP/p300 is known to represent the limiting event for activating CREB target genes, HBx may disrupt this cellular regulation, thus predisposing cells to transformation.


Hepatitis B virus (HBV)
infection is a worldwide health problem and one of the major causes of hepatocellular carci-noma (HCC) (1). Despite epidemiological evidence linking HBV infection to HCC, the mechanisms underlying HBV-associated carcinogenesis are poorly understood. Among the proteins encoded by HBV, the hepatitis B virus X protein (HBx) is essential for virus replication in vivo (2,3) and is thought to contribute to hepatocarcinogenesis. HBx can transform SV40immortalized murine hepatocytes (4) and induce liver cancer in some transgenic mouse models (5). In other mouse models, HBx acts as a cofactor and may accelerate cancer development (6 -8). HBx may participate in cell transformation in several ways. On the one hand, it may act indirectly by activating virus replication (9 -12), which in turn activates immune responses and causes chronic liver inflammation. This leads to continuous destruction and regeneration of the hepatocytes, increasing the chances of genetic alterations (13). On the other hand, by providing a favorable environment for virus replication, HBx may directly disrupt cellular functions such as cell signaling in part through modulation of cytoplasmic calcium, transcription, cell proliferation, and apoptosis (14 -21). Thus, dysfunctions and alterations may accumulate, leading ultimately to cell transformation.
HBx exerts most of its activities through interaction with a large array of cellular partners, such as p53, DDB1, and CRM1 (18,(22)(23)(24)(25). Furthermore HBx acts as a transcriptional co-activator through direct interaction with various proteins, such as the RPB5 subunit of RNA polymerase II, TF IIB, TF IIH, TATAbinding protein, and the basic domain-leucine zipper (bZIP) family of proteins that include the cyclic AMP-response element (CRE)-binding protein (CREB) (14).
The CREB/bZIP family of proteins plays an essential role in the liver by regulating gene expression and different processes such as gluconeogenesis, lipid metabolism, and cell proliferation (26). Recently CREB has also been implicated in leukemogenesis and, interestingly, in hepatocarcinogenesis (27,28). Activation of proteins of the CREB/bZIp family is mediated through phosphorylation of a key acceptor site (serine 133 in CREB) induced by a variety of growth factors and stress signals that activate various signaling pathways. Among other signals, cAMP is known to induce CREB phosphorylation through the activation of cyclic AMP-dependent kinase (PKA). This phosphorylation allows the CREB-binding protein (CBP) and p300, co-activator paralogs, to be recruited (29). In turn, CBP/p300 mediate the transcriptional activation of target genes by allowing a multiprotein activation complex to be formed that links the transcription factors to the basal transcriptional machinery. CBP/p300 also possess intrinsic acetyltransferase activity, which is responsible for the acetylation of histones thought to relieve chromatin-dependent repression (30). Finally CBP/ p300 also modify transcriptional activity through the acetylation of non-histone proteins (31). The CREB/bZIP factors can form homo-or heterodimers that bind to a DNA sequence known as the CRE. This is found in the regulatory region of various cellular genes and also as a variant in the HBV enhancer I (32).
HBx interacts directly in vivo and in vitro with the CREB/ bZIP family of proteins via their bZIP domain and increases their DNA binding affinity and transcriptional activity (33,34). It is still unknown whether HBx stabilizes CREB binding to DNA or increases CREB dimerization (35)(36)(37). However, studies have clearly shown that an increase in CREB DNA binding affinity by HBx is necessary but not sufficient to explain the co-activation of CREB by HBx (38,39). In particular, it has been suggested that HBx might recruit CBP/p300 to CREB/CRE independently of CREB phosphorylation at Ser-133 (39). This mechanism has already been proposed for the Tax protein of the human T-cell leukemia virus type I (HTLV-I) (40).
In this study, we investigated potential interactions between HBx and CBP/p300. We first addressed the question of whether HBx cooperates with CBP/p300 to activate CREB-dependent transcription in human hepatoma cells by reporter assays. We also examined the requirement of PKA-induced phosphorylation of CREB in HBx transactivation capacity. Then direct binding of HBx to CBP/p300 was assessed in vitro and in vivo. Moreover we used chromatin immunoprecipitation (ChIP) assay to analyze the mechanisms underlying the cooperation between HBx and CBP/p300. Our data show that activated transcription of endogenous HBx target genes is associated with increased p300 occupancy at CREB binding sites in the promoters of these genes. Thus, interaction between HBx and p300 could play a predominant role in HBx-dependent regulation of gene transcription.

EXPERIMENTAL PROCEDURES
Cell Culture, Infection, and Transfection-HeLa, HEK293, and HEK293 T cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum. HepG2 and HepAD38 cells were maintained in Dulbecco's modified Eagle's medium/F-12 complemented with 10% fetal calf serum, 3.5 ϫ 10 Ϫ7 M hydrocortisone, and 5 g/ml insulin. Cells were transfected with different vectors as indicated in the figure legends using the Exgen reagent (Euromedex) or Lipofectamine reagent (Invitrogen). Total amounts of transfected DNA were kept constant by adding corresponding empty vectors. All transfection experiments were repeated at least three times in duplicate. For luciferase assays, cells were lysed and assayed 48 h after transfection. HBx was found to activate transcription of the thymi-dine kinase-␤-galactosidase plasmid used to normalize luciferase activity for transfection efficiency. We therefore confirmed the results by multiple independent assays. Human primary hepatocytes were prepared from resected normal human livers as described previously (41). Briefly cells were seeded on collagen I-coated plates and maintained in William's medium supplemented with 10 nM insulin, 100 mM triiodothyronine, and 1 mg/ml bovine serum albumin. All experimental procedures were in conformity with French laws and regulations and with informed consent of the patients. For infection, virions were produced in HEK293 T cells as described previously (42). Hepatocytes were incubated for 2 h with virion preparations normalized to 800 ng/ml viral p24. Fresh medium was then added, and cells were incubated further for 48 h.
Recombinant Proteins and in Vitro Binding Assays-GST-CBP fusion proteins were expressed in E. coli BL21 and purified by glutathione-Sepharose affinity (Sigma) according to standard protocols. HBx proteins were translated in vitro in the presence of [ 35 S]methionine using the TNT coupled reticulocyte lysate system (Promega). For in vitro binding assay, 35 S-labeled HBx proteins were mixed with Sepharose beads carrying either GST or GST-CBP and incubated for 2 h at 4°C in binding buffer (20 mM Hepes, pH 7.9, 300 mM NaCl, 1 mM MgCl 2 , 0.8% Nonidet P-40, 1 mM dithiothreitol, and 0.02% bovine serum albumin). The beads were then washed three times with binding buffer, and bound proteins were analyzed by SDS-PAGE and autoradiography.
Immunoprecipitation and Western Blotting-Cells in 10-cm dishes were lysed in 800 l of lysis buffer (150 mM NaCl, 50 mM Tris/HCl (pH 7.4), 0.5% (v/v) Nonidet P-40, and protease inhibitor mixture (Roche Applied Science)) and quickly frozen on dry ice. Lysates were thawed and centrifuged at 14,000 rpm for 10 min, and 700 l of the supernatant were cleared by incubation with 30 l of protein A/protein G mixture. Antibodies (2.5 g) were initially incubated with 50 l of 50% (v/v) protein A/protein G-Sepharose for 2 h and then added to the lysates. After overnight incubation, the beads were extensively washed with the same buffer, and bound proteins were resolved by SDS-PAGE.
For Western blot analysis, protein lysates were transferred to nitrocellulose membrane and incubated with the indicated antibodies for 1 h. Reactive proteins were developed with secondary antibodies conjugated to alkaline phosphatase and visualized using chemiluminescence according to the manufacturer's protocol (Tropix).
ChIP Assay-HeLa cell extracts were prepared 24 h after transfection, and ChIP assays were carried out as described previously (52) with minor modifications. In brief, cells were crosslinked by incubation with 1% formaldehyde for 5 min at 37°C, and the nuclear extracts were prepared and sonicated. The bound protein-DNA complexes were incubated overnight at 4°C with or without 2 g of rabbit polyclonal anti-p300 antibodies. Immune complexes were incubated (2 h at 4°C with rotation) with 50 l of a protein A/protein G-agarose mixture. The immunoprecipitates were washed five times in radioimmunoprecipitation assay buffer, once in LiCl buffer, and twice in Tris-EDTA buffer and then eluted in elution buffer (1% SDS, 0.1% NaHCO 3 , 0.5 mM Pefabloc, protease inhibitors (Sigma P8340; dilution, 1:1000), and phosphatase inhibitors (Sigma P5726; dilution, 1:1000). After purification of the immunoprecipitated DNA, a 180-bp region encompassing the Gal4 site was amplified using semiquantitative stepdown PCR (see below) using the following primers: pG5gal4-5Ј, 5Ј-TACCCTCTAG-AGTCGACGGAT-3Ј, and pG5gal4-3Ј, 5Ј-AAGCTAATTCC-CGGGATCCGC-3Ј (T m ϭ 67°C).
As negative control, a sequence within the ␤-lactamase gene was amplified with the following primers: Ampr-c5Ј, 5Ј-CGG-CATCAGAGCAGATTGTA-3Ј, and Ampr-c3Ј, 5Ј-CTGGCG-TAATAGCGAAGAGG-3Ј. For analysis of endogenous promoters, 6 ϫ 10 6 HepG2 or HepAD38 cells were seeded in 75-cm 2 flasks and treated or not with 0.3 g/ml tetracycline. After 5 days, the cells were treated for 1 h with forskolin (10 M), and ChIP assays were carried out using 2 g of anti-p300 antibody or 5 g of mouse monoclonal anti-HBx for the immunoprecipitation step.
RT and Quantitative PCR-Total RNA was extracted from primary human hepatocyte cultures or HepAD38 cells using the RNeasy kit (Qiagen) according to the manufacturer's instructions. RNA (1 g) was reverse transcribed using random primers and the Superscript reverse transcriptase (Invitrogen). Real time quantitative PCR was carried out on the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems) using the standard PCR protocol (denaturation at 95°C and annealing/extension at 60°C) with the addition of a final dissociation step to ensure amplicon-specific detection by SYBR Green. Samples were prepared by adding cDNA to SYBR Green PCR Master Mix (Applied) using the following primers: IL-8.s, 5Ј-GCCTTCCTGATTTCTGCAGC-3Ј, and IL-8.a, 5Ј-CGCA-GTGTGGTCCACTCTCA-3Ј; PCNA.s, 5Ј-GGTCAGCCTTC-CCTAGCC-3Ј, and PCNA.a, 5Ј-CGCCTCTCGACTCTG-CTC-3Ј; CCNA2.s, 5Ј-TTGCTGGAGCTGCCTTTCAT-3Ј, and CCNA2.a, 5Ј-GCATGCTGTGGTGCTTTGAG-3Ј. We chose PNN as a reference gene because it has a very low variation coefficient in arrays of human liver tumors and in liver cell lines. 8 The primers were as follows: PNN.s, 5Ј-CCTTTCTGG-TCCTGGTGGAG-3Ј, and PNN.a, 5Ј-TGATTCTCTTCTGG-TCCGACG-3Ј.

RESULTS
HBx Cooperates with CBP/p300 in CREB/ATF Family-dependent Transcriptional Activation-We used reporter assays to investigate whether HBx cooperates with CBP/p300 in the activation of CREB/ATF-dependent transcription. HeLa cells were transfected with pCRE-Luc, a synthetic luciferase reporter containing four CRE sites either alone or together with HBx, PKA, and p300. Although HBx alone could not efficiently activate the pCRE-Luc reporter, we observed a 15.4-fold increase in pCRE-Luc activity when HBx was co-transfected with PKA ( Fig. 1). Consistent with previous reports (29), p300 increased pCRE-Luc reporter activation about 16-fold after the induction of CREB phosphorylation by PKA. CRE reporter activity was 53.7 times higher after co-transfection with HBx and p300 than with the reporter alone, suggesting cooperative action of the two factors (Fig. 1A). We also observed a similar cooperation between HBx and CBP/p300 in activating CREB/ATF-dependent transcription in HEK293 cells (data not shown).
We then investigated whether HBx cooperates with p300 in activating CREB/ATF-dependent transcription in hepatoma cells. HepG2 cells were transfected with pCRE-Luc together with different combinations of HBx, PKA, and p300 expression vectors. In HepG2 cells transfected with PKA, co-transfection with HBx and p300 led to a 28-fold induction, whereas cotransfection with HBx or p300 alone led to 3.5-and 15-fold induction (Fig. 1B). In cells that were not transfected with PKA, HBx did not cooperate with p300. Our results show that HBx and p300 cooperatively activate CREB/ATF-mediated transcription when the cAMP pathway is activated.
Interaction of HBx with CBP/ p300-We next investigated whether HBx directly interacts with CBP and p300. HeLa cells were transfected with a plasmid expressing HA-tagged HBx, and cell lysates were immunoprecipitated with anti-CBP or anti-p300 antibodies followed by Western blotting with anti-HA antibodies. HBx co-immunoprecipitated with endogenous CBP and p300 (Fig. 2), whereas no signal was detected after control immunoprecipitation with irrelevant anti-FLAG antibody. To confirm the specificity of the HBx/CBP and HBx/p300 interactions, we used cells expressing HA-tagged PIASy. Indeed it has been shown previously that PIASy does not interact with CBP/p300 (53). In our conditions, HA-PIASy did not co-immunoprecipitate with p300 or CBP (Fig. 2).
We then investigated the CBP/p300 domains involved in HBx binding using pulldown assays. Different CBP fragments expressed as GST fusion proteins (depicted in Fig. 3A, upper panel, left) were incubated with in vitro translated, radiolabeled full-length HBx. HBx bound to the N-terminal domain of CBP as shown by the interaction with CBP-(1-1099) and CBP-(1-495), both of which encompass the CH1 domain (Fig. 3A, upper panel, right). We also observed HBx binding to CBP-(1099 -1877) and CBP-(1620 -1877), both of which contain the CH3 domain, and a weaker interaction with CBP-(1099 -1758), which may reflect that HBx binds to the small part of the CH3 domain present in this fragment. We observed no interaction between HBx and CBP-(409 -683), which contains the KIX domain (amino acids 588 -683).
We repeated the binding assays using GST-CBP-(1-495) and GST-CBP-(1099 -1877) fusion proteins together with different in vitro translated, in-frame deletion mutants of HBx to identify the domain that interacts with CBP in HBx. We observed no binding using the X7 mutant, indicating that the HBx region spanning the amino acids 58 -84 was needed for binding to CBP-(1-495) and CBP-(1099 -1877) (Fig. 3B). However, amino acids 120 -140 might also have been involved because deletion of this region (X10 mutant) resulted in very weak binding to CBP-(1-495) and CBP-(1099 -1877) (Fig. 3B). Finally the X9 mutant, which contains a deletion of amino acids 85-119, had a minimal residual binding with both CBP-(1-495) and CBP-(1099 -1877). Thus, the central part of HBx from amino acids 61 to 140 is necessary for maximal binding to CBP. It has been reported that the minimum HBx domain required for a direct CREB interaction is the region from amino acids 49 to 115 (38), and thus it overlaps with the domains needed for CBP binding.
We confirmed that these domains are required for HBx and CBP cooperation in the co-activation of CREB/ATF-mediated  transcription by investigating the activity of HBx deletion mutants by reporter assays. HeLa cells were transfected with the pCRE-Luc reporter and PKA expression vector together with wild type or mutated forms of HBx in the presence or absence of p300. In these experiments, fold activations were expressed relative to the activity in cells co-transfected with pCRE-Luc reporter and PKA. The X7 (⌬58 -84) and X9 (⌬85-119) mutants that bind neither CREB nor CBP/p300 failed to stimulate CREB/ATF transcriptional activity (Fig. 3C, compare  lanes 8, 10, and 13). Moreover the X10 (⌬120 -140) mutant, which binds CREB but binds CBP/p300 very weakly, did not activate CREB or cooperate with CBP/p300 in CREB/ATF co-activation (Fig. 3C, lanes 11 and 12). These results indicate that the physical interaction of HBx with CBP/p300 is required for HBx co-activation activity in CREB/ATF-dependent transcription.
HBx Cooperates with CBP/p300 in the Activation of Gal4-CREB-dependent Transcription and Requires the Phosphorylation of CREB-Previous studies have suggested that the increased affinity of CREB DNA binding by HBx contributes to, but does not fully explain, HBx activity on CREB (38,39). Moreover our results suggested that an interaction between HBx and CBP/p300 is needed for CREB transcriptional activation by HBx. Therefore, we tested whether HBx still cooperates with p300 independently from CREB DNA binding. We used a Gal4-CREB chimera containing full-length CREB fused to the Gal4 DNA-binding domain (47). It has been reported that HBx does not affect the binding of Gal4-CREB to Gal4 response elements (34). We transfected 293 cells with the G5-E1B-Luc reporter, which contains five Gal4 binding sites, alone or in combination with HBx, PKA, and p300. PKA activated Gal4-CREB 2.7-fold compared with the level in cells co-transfected with G5-E1B-Luc and Gal4-CREB plasmids (Fig. 4, lanes 2 and 7). HBx cotransfected with PKA further increased Gal4-CREB activity 4.4-fold (Fig. 4, compare lanes 4 and 7). Moreover G5-E1B-Luc activity was increased 3.3-fold in cells co-transfected with p300 and PKA compared with PKA alone (Fig. 4, compare lanes 6  and 7), and co-transfection of HBx and p300 with PKA further increased CREB activity 2.7-fold (Fig. 4, compare lanes 5 and 6).
These results indicate that, in addition to its known effect on CREB DNA binding, HBx cooperates with CBP/p300 in activation of CREB independently of CREB DNA binding affinity.
A previous study had suggested that the activation of CREBdependent transcription by HBx relies on its ability to recruit CBP/p300 to non-phosphorylated CREB (39). However, our results showed that HBx could only activate CREB-dependent transcription after activation of CREB by PKA, suggesting that phosphorylation of CREB is required for transcriptional activation by HBx. We tested whether HBx can cooperate with CBP/ p300 in the context of non-phosphorylated CREB in a reporter assay using a Gal4-CREB-M1 mutant (S133A) that cannot be phosphorylated by PKA. HBx cannot activate the luciferase reporter when the cAMP signaling pathway is not activated (Fig. 4, compare lanes 10 and 11). However, co-transfection with HBx and PKA slightly increased the transcriptional activity of CREB-M1 (2.9-fold), but this activation was lower than the activation obtained with the wild type CREB (12-fold) (Fig.  4, compare lanes 11 and 12 with lanes 3 and 4). There was no cooperation between HBx and p300 when using CREB-M1 (Fig.  4, compare lanes 12, 13, and 14 with lanes 4, 5, and 6). These data show that HBx activates CREB and cooperates with CBP/ p300 only when CREB is phosphorylated at Ser-133.
Endogenous CBP/p300 Are Required for HBx Activation of CREB-mediated Transcription-We confirmed the role of CBP/p300 in HBx-induced activation of CREB using siRNAs to inhibit the endogenous expression of CBP and p300 in HeLa cells. The transfection of HeLa cells with an siRNA specific for p300 greatly reduced p300 expression without affecting CBP expression (Fig. 5, upper panel). Similarly transfection with the CBP-specific siRNA greatly reduced CBP expression without affecting p300 expression. By contrast, the control siRNA did not affect expression of either CBP or p300 as compared with endogenous levels in non-transfected cells (Fig. 5). Therefore, we tested whether HBx could activate CREB-dependent transcription in the presence of siRNAs against p300 and CBP using the Gal4-CREB chimera and the G5-E1B-Luc reporter. Depletion of p300 and CBP by siRNAs greatly reduced PKA-mediated stimulation of CREB-dependent transcription consistent with the crucial role of CBP/p300 recruitment by phospho-CREB for transactivation (Fig. 5, lower panel). Transcriptional activation by HBx was abolished in cells transfected with siR-NAs directed against CBP and p300. Taken together, these data suggest that HBx cooperates with CBP/p300 in the activation of

CREB-mediated transcription and that endogenous CBP/p300 is necessary for HBx transcriptional activity.
HBx Increases the Binding of CBP/p300 on DNA-bound Gal4-CREB in Vivo-As HBx interacts and cooperates with CBP/ p300 but cannot recruit CBP/p300 to non-phosphorylated CREB, we wondered whether HBx could increase the recruitment of CBP/p300 onto DNA-bound CREB. Therefore, we transfected HeLa cells with the G5-E1B-Luc reporter and Gal4-CREB construct alone or in combination with PKA and HBx and then assessed the occupancy of CRE sites by endogenous CBP/p300 using a ChIP assay. Equivalent amounts of crosslinked chromatin were immunoprecipitated or not with p300 antibodies. The precipitated DNA was then subjected to semiquantitative PCR amplification using primers encompassing the Gal4 binding sites. We used primers targeting plasmid sequences outside the Gal4 binding sites in the ␤-lactamase gene as negative controls. Although PKA efficiently induced the recruitment of p300 to Gal4 binding sites (Fig. 6, lane 7), HBx in the absence of PKA was unable to induce significant recruitment of p300 (Fig. 6, lane 9). This observation correlates with our previous finding that HBx was unable to activate CREB-dependent transcription in the absence of activation of the cAMP pathway. Importantly the amount of p300 on Gal4 binding sites was higher in the presence of HBx and PKA than in cells transfected with PKA alone (Fig. 6, compare lanes 11 and 7). Recruitment of p300 to Gal4 binding sites was specific because we did not observe any amplification of the ␤-lactamase gene (Fig. 6). These data showing that HBx increased CBP/p300 occupancy of a synthetic promoter suggested that HBx might reinforce the recruitment of p300 on phospho-CREB.
HBx Increases the Recruitment of CBP/p300 to Endogenous IL-8 and PCNA Promoters in Vivo-We next investigated whether HBx could also activate p300 recruitment on endogenous cellular promoters. For these studies, we took advantage of our recent microarray-based screen for HBx target genes in primary human hepatocytes, 9 and we selected three CREB-responsive genes: IL-8 and PCNA that we found to be activated by HBx (54 -56) and the cyclin A2 gene (CCNA2) for which no activation was associated to HBx expression (57,58). First the ability of HBx to up-regulate PCNA and IL-8 was verified in primary human hepatocytes using real time PCR. Hepatocytes at 24 h postplating were infected with the lentiviral recombinant vector pTRIP-HBx expressing HBx under control of the cytomegalovirus promoter. Control hepatocytes were infected with the vector pTRIP⌬U3-GFP (51). As reported previously, the transduced protein was expressed in 90% of cells (41), and transgene expression was ascertained by Northern blotting (data not shown). RNA was extracted at 48 h after infection, and expression of IL-8 and PCNA was analyzed by quantitative RT-PCR. As shown is Fig. 7A, expression of IL-8 was activated by 2-fold, and PCNA expression was activated by 3-fold in hepatocytes expressing HBx compared with the green fluorescent protein control. By contrast, the CCNA2 gene was not induced by HBx in our experiments (Fig. 7A).
We then analyzed the expression of PCNA, IL-8, and CCNA2 in HepAD38, a HepG2-derived cell line that expresses the HBV genome under tetracycline control (59). This cell line is useful for studying the regulation of cellular genes by HBx as HBx expression is significantly increased by tetracycline removal (Fig. 7B). RNA was extracted from HepAD38 cells grown with or without tetracycline and treated or not with the cAMP agonist forskolin that activates PKA. Quantitative RT-PCR was used to investigate expression of PCNA, IL-8, and cyclin A2 genes in these different conditions. We observed a significant increase in IL-8 and PCNA expression in cells treated with forskolin and grown in the absence of tetracycline, whereas cyclin A2 expression remained unchanged (Fig. 7C). Our results suggest that, in HepAD38 cells, the expression of HBx can up-regulate the expression of IL-8 and PCNA only when the cAMP pathway is activated.
Next we studied the recruitment of HBx to PCNA, IL-8, and CCNA2 promoters in tetracycline-treated or untreated HepAD38 cells. ChIP assays were carried out in HepAD38 cells 9 L. Lé vy, C.-A. Renard, M. A. Buendia, and C. Neuveut, unpublished data. FIGURE 6. HBx increases the recruitment of p300 to a synthetic promoter. HeLa cells were transfected with 3 g of G5-E1B-Luc reporter vector in combination with Gal4-CREB (5 g), RSV-PKA (2.5 g), and HA-HBx (5 g). At 24 h after transfection, ChIP assays were carried out using specific antibodies against p300 (2 g). Immunoprecipitated DNA or control DNA was amplified using specific primers for either the Gal4 binding site region or the ␤-lactamase gene. Densitometry of ChIP PCR normalized to input is shown in the bottom panel. The value of amplification products from cells transfected with the G5-E1B-Luc and Gal4-CREB plasmids was taken as 1 with other values being calculated accordingly. The graph is the average of three separate experiments. FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 treated with forskolin using primers designed to amplify regions containing the CRE sites. In the PCNA promoter, the CRE site has been located from Ϫ52 to Ϫ45 from the start site. For the IL-8 gene, we used the rVISTA algorithm and comparative analysis of human and mouse promoters to define a putative CRE sequence located from Ϫ76 to Ϫ56 from the start site, consistent with previous studies (55,60). The CCNA2 promoter contains CRE sites located from Ϫ67 to Ϫ44. We found that in tetracycline-treated HepAD38 cells, which expressed very low levels of HBx, the PCNA and IL-8 promoters were occupied by an almost undetectable level of HBx (Fig. 8A, lane 3). By contrast, HBx recruitment greatly increased in the absence of tetracycline, which correlates with activated HBx expression (Figs. 8A, lane 5, and 7B). As a negative control, we observed no PCR amplification of IL-8 and PCNA in forskolin-treated HepG2 cells. HBx occupancy at the CCNA2 promoter could not be detected in forskolin-treated HepAD38 cells in the absence of tetracycline.

Functional Interaction between HBx and CBP/p300
We then investigated the role of HBx on the recruitment of CBP/p300 to endogenous promoters in HepAD38 cells treated with forskolin. High levels of HBx increased p300 recruitment to the IL-8 promoter 3.1-fold and to the PCNA promoter 17.3fold versus cells treated with forskolin and tetracycline (Fig. 8B, compare lane 3 with lane 1). HBx did not affect the recruitment of p300 to the CCNA2 promoter. These data showed a good correlation between the increase in CBP/p300 recruitment by HBx to the CRE binding site of the IL-8 and PCNA promoters and the induction of the corresponding genes by HBx (Figs. 8B  and 7C). Taken together, our results show that HBx activates CREB transcriptional activity by increasing the recruitment of CBP/p300 to the promoter of CREB endogenous cellular target genes.

DISCUSSION
HBx is described as a multifunctional protein essential for virus replication. It can subvert many cellular processes such as signal transduction, transcription, and apoptosis through protein-protein interaction (14). In this study, we showed that HBx interacts in vitro and in vivo with the co-activators CBP/p300. We studied this interaction for CREB transcriptional activity and found that HBx cooperates with CBP/p300 in the CREBmediated activation of a synthetic promoter containing four CRE sites. We used a CREB mutant that cannot be phosphorylated on Ser-133 to show that HBx cannot recruit CBP/p300 to non-phosphorylated CREB. Moreover we used ChIP experiments to show clearly that HBx favors the recruitment of p300 to phosphorylated CREB bound to DNA. The ChIP experiments showed that HBx can also increase the recruitment of the CBP/p300 co-activators to endogenous cellular target genes. The favoring of CBP/p300 recruitment to particular genes by HBx may be very important in vivo as Zhang et al. (60) showed that tissue-specific activation of selected genes by cAMP is not due to differences in phospho-CREB occupancy but rather to the selective recruitment of CBP/p300.
In this study, we analyzed the recruitment of p300 to the promoters of two HBx target genes, IL-8 and PCNA, both of which contain a CRE sequence and are known to be CREB target genes (54,55). Although we observed p300 recruitment to the PCNA and IL-8 promoters after treatment with forskolin in HepAD38 cells, HBx overexpression strongly increased p300 occupancy levels. The induction of CREB target genes may play an important role in the development of HCC associated with HBV infection. Indeed CREB is known to participate in many functions, such as cell survival, proliferation, and glucose metabolism, and is now considered an important actor in cancer development. A translocation that fuses the bZIP domain of ATF1, a CREB family member, to the oncogene product of Ewing sarcoma is found in clear cell sarcoma of soft tissues, and this translocation is responsible for the constitutive activation of CREB target genes (61). Recently it has been shown that CREB plays an important role in hepatocellular carcinoma development by modulating tumor growth, angiogenesis, and apoptosis (27). Shankar et al. (28) revealed that CREB plays a role in leukemogenesis by showing that CREB overexpression causes the abnormal proliferation and survival of myeloid cells; this correlates with the up-regulation of cyclin A. Finally virally FIGURE 7. HBx increases the expression of IL-8 and PCNA genes. A, expression profiles of IL-8, PCNA, and CCNA2 in primary human hepatocytes expressing or not expressing HBx were analyzed by quantitative RT-PCR in three independent pools of HBx-expressing hepatocytes. The results are the average of three independent experiments. B, the expression of HBx in HepAD38 cells grown in the presence or absence of tetracycline was analyzed on 50 g of protein extract by Western blotting using anti-HBx antibodies. HepG2 cells were used as a negative control. Actin was used as a loading control. C, expression profiles of IL-8, PCNA, and CCNA2 in HepAD38 cells grown in the presence or absence of tetracycline and treated or not with forskolin (10 M) for 1 h. RNA levels were analyzed by quantitative RT-PCR. Tet, tetracycline; WB, Western blot. encoded oncoproteins act also through CREB signaling to promote cell transformation. The Tax oncoprotein of HTLV-I and E1A of adenovirus (Ads), both of which affect cellular events such as cell proliferation, apoptosis, and oncogenic transformation, can interact with CREB and CBP/p300 (19,62,63). This interaction is thought to activate virus replication and also modulate the expression of cellular genes. Both proteins can either activate or repress cellular genes via the CREB/CBP/p300 interaction. In this study, we analyzed two HBx target genes containing a CRE sequence: IL-8 and PCNA. Those two genes are very likely members of a large group of HBx target genes that may play an important role in HBV-associated cancer development. Indeed IL-8 is described as a leukocyte chemotactic molecule and is thought to be responsible for maintaining the inflammatory environment associated with HBV infection that may play a role in the development of cancer. Moreover IL-8 has been described as a mitogenic, motogenic, and angiogenic factor and is up-regulated in a variety of human cancers, implying that it plays an important role in tumorigenesis (41). PCNA is a component of cellular complexes involved in eukaryotic DNA replication and repair. The dysregulation of these two functions is a key step in neoplastic transformation. Moreover the modulation of PCNA expression has been described previously for HTLV-I and adenovirus types 2 and 5 (62,63). Finally the lack of activation by HBx of some CRE-containing genes such as cyclin A2 suggests that HBx may act via specific combinations of general transcription and regulatory factors (14,64).
In this study, we showed that HBx binds two separate domains of the CBP/p300 protein: an N-terminal domain from amino acids 1 to 495 that contains the nuclear receptor-binding domain and the CH1 domain and a region from amino acids 1620 to 1877 that contains the CH3 domain. CH1 and CH3 are two cysteine-and histidine-rich domains that bind Zn 2ϩ (also termed TAZ1 and TAZ2) as well as many cellular and viral proteins, including transcription factors or factors of the general transcriptional machinery (30). Among others, CH1 binds p53, mdm2, p65, HTLV-I Tax, and E1A, and CH3 binds p300/CBP-associated factor, p53, RNA helicase A, c-Fos, Myo-D, E1A, and human immunodeficiency virus type 1 Tat. We did not investigate whether the binding of HBx to different CBP/ p300 domains affected the different HBx activities such as transcriptional activation or p53 transcriptional repression (65). Nevertheless other viral proteins, such as HTLV-I Tax or adenovirus E1A, also bind to multiple p300 and CBP subdomains; this modulates their transcriptional and transforming activities. For example, Tax is known to activate CREB transcriptional activity partly through the recruitment of CBP/p300 via the KIX domain and to block the activity of transcription factors such as p53 and p73␤ by competing for the KIX and CH1 domains. Tax also inhibits nuclear receptor signaling possibly by interacting with the same SRC-1 domain on CBP/p300 (40). The E1A and CBP/p300 interaction also seems to be very complex as different E1A domains contact CBP/p300, and E1A binds different domains of CBP/p300. The functional consequences of these interactions are also very complex: E1A acts as either a coactivator or a co-repressor of transcription factors such as CREB, c-Fos, MyoD, and Sat1 (62,66).
We also used a phosphorylation-deficient CREB mutant in reporter assays to show that HBx cannot recruit CBP/p300 to a non-phosphorylated CREB bound to target DNA. A study by Pflum et al. (39) using in vitro transcription assays with either phosphorylated or non-phosphorylated CREB protein showed that HBx activates only the non-phosphorylated form of CREB. However, using only the wild type CREB protein, it cannot be completely excluded that a few CREB molecules may be phosphorylated.
HBx can induce the recruitment of CREB and ATF2 to the CRE-like sequence present in the HBV enhancer I (33). Pollicino et al. (67) showed that the recruitment of CBP/p300 to the HBV minichromosome correlates with virus replication. Further studies are needed to determine whether CREB is involved in virus replication and whether HBx participates in the recruitment of CBP/p300 to the minichromosome. Moreover CBP/p300 binds a large variety of cellular transcription factors activated by HBx, such as c-Jun, c-Fos, and p65-NF-B. It will be interesting to test whether HBx is involved in the recruitment of CBP/p300 to these factors. This mechanism has FIGURE 8. HBx increases the recruitment of p300 to endogenous target gene promoters. A, HBx is recruited to the IL-8 and PCNA promoters. HepG2 or HepAD38 cells were grown in the presence or in the absence of tetracycline and treated with forskolin. ChIP assays were carried out using 5 g of anti-HBx antibodies and specific primers to amplify the CRE sites in the IL-8 and PCNA promoters. The CCNA2 promoter was used as negative control. PCR products were quantified, and values are expressed relative to PCR fragment amplified from HepG2 cells that was taken as 1. The result shown is the average of three independent experiments. B, HBx increases the recruitment of p300 to the IL-8 and PCNA promoters. ChIP assays in HepAD38 cells were carried out with 2 g of anti-p300 antibodies. DNA fragments containing the CRE sites in IL-8, PCNA, or CCNA2 promoters were amplified by semiquantitative PCR. PCR products were quantified as in A except that the value from HepAD38 cells treated with tetracycline was taken as 1. Results are the average of three independent experiments. Tet, tetracycline. FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 already been proposed for the transcriptional activation of the hypoxia-inducible factor-1␣ and the early growth response factor by HBx (68,69).

Functional Interaction between HBx and CBP/p300
In conclusion, we demonstrated that HBx co-activates CREB partly through the recruitment of CBP/p300. Thus, HBx may be considered as a potentiator of the signal mediated by CREB, and this mechanism may be involved in HBV-mediated oncogenesis. Indeed HBV infection is associated with continuous liver regeneration; an important feature of liver regeneration is the increase in intracellular cAMP, which in turn meditates the finely tuned regulation of gene expression (26). In this cellular environment, HBx may facilitate the recruitment of CBP/p300 to phospho-CREB, leading to the deregulation of gene expression, and thus predispose cells to tumorigenesis. Finally as HBx has been described as a pleiotropic co-activator, the stabilization of CBP/p300 on transcription factors may be a more general mechanism of HBx activity.