The hepatitis B virus X protein enhances the DNA binding potential and transcription efficacy of bZip transcription factors.

The hepatitis B virus X protein interacts with the basic-region, leucine zipper protein (bZip) domain of cAMP response element-binding protein increasing its affinity for the cAMP response element site in vitro and its transcriptional efficacy in vivo (Williams, J. S., and Andrisani, O. M. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 3819-3823). Here we examine pX interactions with bZip transcription factors ATF3, gadd153/Chop10, ICER IIgamma, and NF-IL6. We demonstrate direct interactions in vitro between pX and the bZip proteins tested. In contrast MyoD and Gal4(1-147) fail to interact with pX. We also demonstrate by the mammalian two-hybrid assay the direct interaction of pX with cAMP response element- binding protein, ICER IIgamma, ATF3, and NF-IL6 in hepatocytes. In addition, pX increases the DNA binding potential of bZip proteins for their cognate DNA-binding site in vitro. In transient transfections in hepatocytes (AML12 cell line), pX increases the transcriptional efficacy of the bZip transcription factors. NF-IL6-mediated transcriptional activation is enhanced 3-fold by pX. Most interestingly, pX augments the repression mediated by bZip repressors ATF3 and ICER IIgamma, by 6- and 7-fold, respectively, demonstrating for the first time the involvement of pX in gene repression. We conclude that pX is an enhancer of the DNA binding potential of bZip transcription factors, thereby increasing the transactivation or repression efficacy of bZip-responsive genes.

The hepatitis B virus genome encodes a 16.5-kDa protein, termed X antigen (2), expressed during viral infection (3)(4)(5) and required for the viral life cycle (6,7). Expression of pX in liver of transgenic animals has been shown in some cases to induce liver cancer (8), implicating pX in the development of hepatocarcinogenesis in humans chronically infected with HBV. 1 However, the mechanism of hepatocarcinogenesis remains unknown.
Many studies have examined the role of pX in cellular sig-naling, but its mechanism of action remains obscure. HBV X is a multifunctional protein; it interacts in vitro and in vivo with the tumor suppressor p53 protein (9 -11), with the putative DNA repair protein XAP-1 (12), and acts as a promiscuous transactivator (13). pX transactivates cis-acting elements, such as those present within the HBV (14,15) and SV40 enhancers (16,17), and the HIV-LTR (18 -21). X-responsive elements include the NF-B (18,20,(21)(22)(23), AP-1 (24 -28), AP-2 (24), and CRE (1,29) sites. Interestingly, pX does not appear to bind double-stranded DNA. These multiple X-responsive, cis-acting elements suggest that the mechanism of pX action is pleiotropic. Studies to date support pX transactivation by a dual mechanism, i.e. both the activation of cytoplasmic signaling pathways and direct interactions with cellular transcription factors and the transcriptional machinery. Regarding this dual mechanism, pX activates in vivo the Ras, Raf, mitogen-activated protein kinase, and JNK signaling pathways, leading to transactivation of AP-1 and NF-B sites (27,30,31). pX interacts with several members of the basal transcriptional apparatus, such as the RPB5 subunit of eukaryotic RNA polymerases (32), TBP (33), and with components of TFIIH (34). Furthermore, it has been proposed that pX is a general viral transactivator, acting by influencing the coactivator process (35).
Direct interactions between pX and cellular transcription factors have been uniquely demonstrated for the CREB/ATF family of proteins (1,29). CREB, which mediates the transcriptional induction of the cAMP-transduction pathway (36,37), is also the interaction target of the HTLVI Tax oncoprotein (38,39). Our studies demonstrated that the interaction of pX with CREB increases by 1 order of magnitude its affinity for the CRE site, thereby increasing its in vivo transcriptional efficacy by 13-fold. In contrast to the CREB/Tax interactions (40), pX does not enhance the dimerization rate of CREB; it seems to target the basic, DNA-binding region of CREB (1).
In this study we investigate the potential interaction of pX with other bZip transcription factors normally expressed in the hepatocyte. Our hypothesis is that pX interacts not only with CREB but also with other bZip proteins, by recognizing common structural features shared by bZip proteins (41). To test this hypothesis, we selected for analysis bZip proteins that play a role in hepatocyte physiology such as, NF-IL6 or C/EBP␤ (42), gadd153/Chop10 (43), ATF 3 (44), and ICER II␥ (45,80) which may play such a role.
NF-IL6, a member of the C/EBP leucine zipper family (42), is selected in this study, since it is expressed in liver (46) where it is implicated as the master regulator of the acute phase response genes (42). Expression of NF-IL6 is induced by IL-6 (42, 47), IL-1 (48), and other inflammatory mediators. NF-IL6 forms heterodimers with other bZip proteins in the liver (43), including gadd153/Chop10 (43) which is induced by cellular stress conditions (49,50), acute phase (inflammatory) response (51), and exposure to toxins (52,53). While gadd153/Chop10 has strong sequence similarity to C/EBP-like proteins within the bZip region (43), it contains substitutions of three conserved residues in the basic region critical for DNA binding. Therefore, heterodimers of gadd153/Chop10 and C/EBP-like proteins are unable to bind to their cognate DNA binding site (43). Accordingly, it has been proposed that gadd153/Chop10 functions as a stress-inducible transcriptional inhibitor (43). However, recent studies indicate that in certain cases gadd153/ Chop10 may function as a direct transcriptional activator (54). ATF 3 , is also of interest in this study since it is a member of the CREB/ATF family of transcription factors, sharing sequence similarity with the bZip domain of CREB and binding to the CRE site (44). However, while CREB is a positive regulator of transcription, ATF 3 is a transcriptional repressor (56). Recent studies indicate that ATF 3 is induced by a variety of physiological stress conditions, such as mechanically injured and toxin-injured liver (55,57). ATF 3 and gadd153/Chop10 form non-DNA-binding heterodimers (55). Interestingly, during carbon tetrachloride injury of liver, gadd153/Chop10 and ATF 3 mRNAs are inversely induced but in an overlapping manner; gadd153/Chop10 mRNA, high in normal liver, decreases upon CCl 4 exposure; ATF 3 mRNA, low in normal liver, increases upon CCl 4 exposure (55).
ICER II␥ (inducible cAMP early repressor) (45), a member of the cAMP-responsive element modulator subfamily of bZip proteins, is of interest since it is an inducible repressor, displaying high degree of amino acid sequence identity with the bZip of CREB and is devoid of a transactivation domain. Thus, ICER II␥ provides an ideal model system to demonstrate the effect of pX in repression of bZip transcription. Recent studies demonstrated ICER II␥ expression in liver regeneration (80).
In this study, we employ in vitro and in vivo (cellular) assays to examine the interaction of pX with the aforementioned bZip transcription factors.

MATERIALS AND METHODS
In Vitro Protein-Protein Interaction Assays-Transcription factors ICER II␥ (45), the bZip domain of NF-IL6, amino acid residues 259 -335, and CREB 327 (59) were cloned in T 7-7 vector. ATF 3 (55) and gadd153/Chop10 (55) were in a derivative of pTM1 (60) vector. 35 S-Labeled proteins were synthesized by the TNT (Promega) in vitro transcription/translation system. Plasmids GST-X 1-154 and GST-X 49 -154 were constructed by polymerase chain reaction of pX DNA fragments corresponding either to amino acid residues 1-154 or 49 -154 cloned into the EcoRI-HindIII sites of plasmid pGEX-KG (61). GST-X fusion proteins were expressed in Escherichia coli and purified on glutathione-Sepharose 4B resin (Pharmacia Biotech Inc.), as described previously (62). Bacterial extract obtained from 1 liter of bacterial culture was bound to 500 l of resin for 30 min and washed (62). Protein concentration of GST and GST-X 1-154 bound to the resin was estimated by Coomassie Blue staining of SDS-PAGE and by comparison to known amounts of bovine serum albumin run on the same gel. In vitro proteinprotein interaction assays were carried out as follows: 10 g of GST and 2 g of GST-X 1-154 proteins immobilized onto 20 l of glutathione-Sepharose 4B resin were incubated with 4 l of TNT lysate for 3 h at 4°C, in buffer containing 25 mM Hepes, pH 7.5, 100 mM KCl, 5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 40 g/ml bovine serum albumin, 0.1% Triton X-100, and a protease inhibitor mixture. After incubation, the beads were washed six times in the above buffer. Analysis of the bound protein was by SDS-PAGE and fluorography.
In Vitro DNA-Protein Binding Assays-DNA protein binding assays were carried out as described previously (1). The somatostatin CRE (62,63) was used as the oligonucleotide probe for CREB, ICER II␥, ATF 3 , and gadd153/Chop10. The nucleotide sequence spanning positions Ϫ165 to Ϫ173 of the HIV-LTR (58) was used as the DNA-binding probe for NF-IL6. The MyoD/E 47 E-box DNA-binding site (64) was kindly provided by Dr. S. Konieczny. Proteins for ICER II␥, ATF 3 , gadd153/ Chop10, ATF 3 /gadd153/Chop10, and MyoD/E 47 were obtained by in vitro translation (TNT, Promega); 1-3 l of in vitro translation mixture was used for each reaction. Bacterially produced CREB 327 (70) and bZip NF-IL6, 15 ng each, were employed in the binding reactions. Recombi-nant GST-X 49 -154 was added to binding reactions; control lanes contained equal amounts of GST protein. The reaction mixtures were analyzed by native acrylamide gel electrophoresis, as described (62,63).
In functional assays, AML12 cells were transfected by the calcium phosphate coprecipitation method, using the Life Technologies, Inc. transfection kit. Briefly, 10 -20% subconfluent cultures were transfected with 5 g of CAT reporter plasmid and the indicated amount of expressor plasmid. Cells were harvested 48 h after transfection. CAT activity was determined as described (66) using equal amounts of protein extract per assay. Protein concentration of cellular extracts was determined by the Bio-Rad protein assay. Each experiment was repeated a minimum of three times. The CAT reporter vector pC15⌬XE, kindly provided by Dr. M. Bina, contains only the NF-IL6-binding site I, at position Ϫ158 to Ϫ178 (58).
The mammalian two-hybrid assay was carried out using the 5 ϫ Gal4-E 1 b TATAA-luciferase reporter plasmid (67). The RSV-CREB-VP16 and RSV-Gal4 vectors were kindly provided by Dr. R. Gaynor (68). The CMV4-ICER II␥ (45) was kindly provided by Dr. C. Molina. The ATF 3 , gadd153/Chop10, ICER II␥, and bZip of NF-IL6 (amino acid residues 259 -335) were cloned by polymerase chain reaction amplification of the respective fragments into the RSV-VP16 vector (68), at the NcoI site, resulting in the construction of bZip-VP16 fusion proteins, as described by Yin et al. (68). Similarly, RSV-X-Gal4 constructs were prepared by inserting the coding region of pX at the NcoI site, resulting in fusion proteins with the Gal4 DNA-binding domain (amino acids 1-147) at its C terminus. 5 g of reporter plasmid, 5 ϫ Gal4-E1b TATAA-luciferase, were cotransfected with 5-10 g of each of the expressor plasmids. Transfections were performed in duplicates (60-mm plates) and repeated a minimum of three times.

Direct Protein-Protein Interactions of HBV pX with bZip Proteins-Interactions
between pX and CREB were demonstrated by enhanced CREB binding to the CRE site in the presence of pX (1,29) and by altered methylation interference assays (1). In the present study we employed the CREB/pX-interacting proteins as the model system for developing an assay to detect direct protein-protein interactions. For this analysis the fulllength X protein was produced as a fusion with glutathione S-transferase (61). The GST-X 1-154 fusion protein is selectively retained by glutathione-Sepharose 4B resin and thus the complex provides a suitable affinity resin.
Recombinant CREB 327 (69, 70), 32 P-radiolabeled (66), was employed to establish the in vitro conditions for detecting specific binding of CREB 327 to the resin-immobilized GST-X 1-154 . We observe selective binding of 32 P-CREB 327 to GST-X 1-154 but no binding to the control GST-resin (Fig. 1A). Additional control experiments include the following: first, the demonstration that under the conditions detecting CREB/pX interactions (Fig.  1A), transcription factors of a different class, namely Gal4 1-147 (71) and MyoD (64), did not display detectable binding to GST-X 1-154 (Fig. 1, B and C); and second, the demonstration that the interaction of CREB 327 with GST-X 1-154 involved the bZip domain of CREB 327 . For this analysis, CREB 327 and the N-terminal portion of CREB 327 , amino acid residues 1-198, were synthesized in vitro ( Fig. 2A). In comparative analysis, the N-terminal region of CREB 327 did not interact with GST-X 1-154 (Fig. 2B), whereas 35 S-CREB 327 displayed specific binding to GST-X 1-154 (Fig. 2C). The results confirm our earlier observations that the bZip of CREB is the interacting target of pX and show that under the established conditions ( Fig. 1) we detect specific binding of CREB 327 to pX in vitro.
We employed the in vitro protein-protein interaction assay described in Fig. 1A to examine the interaction of bZip transcription factors ATF 3 , gadd153/Chop10, NF-IL6, and ICER II␥ with pX. The binding reactions with bZip proteins were carried out exactly as described for CREB 327 (Fig. 1A and Fig. 2C). Fig.  3A shows the binding of the bZip of NF-IL6 to the immobilized GST-X 1-154 is much higher than the control GST-resin. Similarly, Fig. 3, C and D, shows the specific binding of ATF 3 , gadd153/Chop10, and ICER II␥ to GST-X 1-154 . These observations demonstrate that pX can interact directly with bZip transcription factors.
The Effect of HBV pX on the DNA Binding Potential of bZip Proteins-Gel retardation assays demonstrated that CREB/pX interactions increase the DNA binding affinity of CREB for the CRE site (1). Therefore, we carried out DNA-binding assays to examine the effect of pX on the DNA-binding activity of the bZip proteins (Fig. 3). We incubated the bZip proteins with their cognate radiolabeled DNA-binding site, as a function of recombinant pX addition. The pX used is a truncated form of the full-length protein, lacking the N-terminal 48-amino acid residues. We constructed this truncation based on observations by Murakami et al. (72) that the N-terminal deleted portion does not affect pX transactivation. In the protein-protein interaction assay GST-X 49 -154 behaves similarly to X 1-154 and interacts equally with all the bZip proteins factor tested. 2 Furthermore, similar to pX 1-154 , pX 49 -154 enhances equally the DNA binding activity of CREB (data not shown). We employed the GST-X 49 -154 because, in our hands, it displays consistent activity and does not precipitate as readily as pX  or GST-X 1-154 . Fig. 4 shows enhanced DNA binding in the presence of pX 49 -154 for CREB, in agreement with our earlier results (1), and for ICER II␥, ATF 3 , and NF-IL6. Interestingly, neither the non-DNA-binding gadd153/Chop10 nor the ATF 3 /gadd153/ Chop10 heterodimer displays enhanced DNA binding by pX. Likewise, the MyoD/E 47 heterodimer, used here as the negative control, displays only minimal increase in DNA binding in the presence of pX, demonstrating the preference of pX for interaction with bZip transcription factors.
Furthermore, our observations suggest that the bZip transcription factors, which are capable of both DNA binding and protein-protein interactions with pX (CREB, ICER II␥, ATF 3 , and NF-IL6) (Figs. 3 and 4), display enhanced DNA binding in vitro in the presence of pX. However, pX does not convert a non-DNA-binding protein, such as gadd153/Chop10, to become DNA binding. pX only potentiates the existing DNA binding activity of bZip proteins. Similarly, the non-bZip transcription factors GAL4   (1) and MyoD/E 47 , which fail to directly interact with pX (Fig. 3), do not acquire enhanced DNA binding by pX. We conclude that pX interacts directly and specifically with bZip transcription factors and increases their DNA binding potential.
Interaction of HBV pX with bZip Transcription Factors in Vivo-To further establish direct interactions between pX and bZip transcription factors in the hepatocyte, we employed the mammalian two-hybrid assay, in the AML12 cell line (65). AML12 cells, derived from transgenic mouse liver, over-express transforming growth factor-␣, but otherwise exhibit properties of normal hepatocytes (65). For this analysis we constructed an expression vector containing the activation domain of VP16 412-490 in fusion with the C terminus of CREB (68), ICER II␥, ATF 3 , and NF-IL6. We also constructed a mammalian expression vector encoding pX in fusion at its C terminus to the DNA-binding domain of Gal4 1-147. Transient transfections of RSV-X-Gal4 in AML12 cells demonstrated that the pX-Gal4 fusion protein is devoid of detectable transactivation potential (Fig. 5). Fig. 5 shows the relative amount of reporter expression following cotransfection of both RSV-X-Gal4 and RSV-bZip-VP16 expressors, when compared with control transfections. With cotransfection of either RSV-CREB-VP16 or RSV-ICER II␥-VP16 with RSV-X-Gal4 plasmid, we observe a 3-and 6-fold induction of luciferase expression, respectively. Cotransfection of RSV-NF-IL6-VP16 or RSV-ATF 3 -VP16 with RSV-X-Gal4 induces the expression of luciferase by 4-and 14-fold, respectively. To verify the specificity of our in vivo assay, we employed the pEM-MyoD-VP16 expressor as our control, kindly provided by Dr. S. Konieczny. 3 We observe no detectable interactions occurring in vivo between MyoD and pX by this assay. These observations agree with our in vitro results with MyoD ( Figs. 1 and 4). We conclude that in vivo pX interacts directly via protein-protein interactions with the bZip proteins CREB, ICER II␥, ATF 3 , and NF-IL6 (Fig. 5).
HBV pX Enhances the Activity of bZip Transcription Factors in Vivo-Having demonstrated that pX interacts directly with bZip proteins in the hepatocyte (Fig. 5), we performed transient transfection assays to assess the effect of these interactions on the transcriptional activity of the aforementioned bZip proteins. Among the bZip proteins, NF-IL6 is a well documented transcriptional activator (73). Earlier studies by Tesmer et al.

FIG. 1. 32 P-CREB 327 interacts directly with HBV X protein.
A, protein-protein interaction assays with recombinant GST and GST-X 1-154 fusion proteins. 10 g of GST or 2 g of GST-X 1-154 immobilized onto 20 l of glutathione Sepharose-4B resin were incubated with 10 ng of 32 P-CREB 327 , purified as described previously (68), in combination with CRE-affinity chromatography (68,69), and phosphorylated in vitro with [␥-32 P]ATP, using the catalytic subunit of protein kinase A (65). Incubation was for 3 h at 4°C, in buffer containing 25 mM Hepes, pH 7.5, 100 mM KCl, 5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 40 g/ml bovine serum albumin, 1 mM NaF, 0.1% Triton X-100, and a protease inhibitor mixture. After incubation, the beads were washed 6 times. Input 32 P-CREB 327 (INPUT) and 32 P-CREB 327 bound to GST-or GST-X 1-154 resin, respectively, and eluted by boiling in SDS-PAGE loading buffer. Analysis is on 10% SDS-PAGE. B and C, protein-protein interaction assays using 4 l of in vitro translated 35 S-labeled GAL4 1-147 and MyoD proteins, respectively. Binding and washing conditions were as described for 32 P-CREB 327 . (58) identified the cis-acting element located between nucleotide position Ϫ158 to Ϫ178 on the LTR of HIV as an NF-IL6 binding site. We employed an HIV-LTR-driven reporter, pC15⌬XE (58), as the model system for analyzing NF-IL6mediated transcriptional induction. A representative experiment is shown in Fig. 6A. AML12 cells were transfected under low serum conditions in the presence of 10 nM IL-6 (73). Increasing amounts of CMV4-X expressor were transfected in the presence or absence of CMV4-NF-IL6 vector (58). We observe a 3-fold enhancement in HIV-LTR-CAT transcription by pX only in the presence of the NF-IL6 encoding vector, suggesting that pX action augments the transcriptional efficacy of NF-IL6. In contrast, the pX encoding vector failed to demonstrate enhanced transcriptional induction of a MyoD-dependent transcription system (74), indicating that pX enhances transactivation in a specific manner (Fig. 6B).
HBV pX Enhances the Transrepression Potential of bZip Repressors-We have likewise employed transient transfections in AML12 cells to examine the effect of pX on the bZip repressors ICER II␥ and ATF 3 . In our in vitro assays (Figs. 3  and 4) we included the bZip inhibitor gadd153/Chop10. Although pX interacted with gadd153/Chop10 in vitro (Fig. 3), we did not further analyze gadd153/Chop10 because pX did not alter or augment its DNA binding potential (Fig. 4).
The assay system used to monitor the effect of pX on bZip repressor activity includes the CRE 3 -CAT reporter plasmid and the ATF 3 (45) and ICER II␥ (55) expressor plasmids. The pX expression vector encodes the nuclear localization sequence (NLS) in fusion with pX (30) which targets pX exclusively in the nucleus. Titration experiments of increasing amounts of bZip repressor were carried out in the presence or absence of CMV-NLS-pX plasmid which was used interchangeably with CMV4-X, as in Fig. 6.
In Fig. 7, CRE 3 -CAT reporter activity was monitored in the presence of forskolin which activates the endogenous CREB (1,66). Transfection of increasing amounts of ICER II␥ expressor in the absence of pX expression results in a 2-8-fold inhibition of CRE-driven transcription. In agreement with our earlier observations (1), transfection of pX, in the presence of forskolin, induces the transcriptional efficacy of endogenous CREB, from 13.5 to 81%. In Fig. 7, we show that increasing amounts of transfected ICER II␥ DNA in the presence of pX expression resulted in 7-23-fold repression. Therefore, the repression is higher in the presence of pX (7-23-fold) than that in the absence (2-8-fold).
Similarly, for ATF 3 in the absence of pX (Fig. 8), 1 g of transfected ATF 3 expressor vector results in an approximately 3-fold repression (55); in the presence of pX, 1 g of transfected   FIG. 2. In vitro translated 35 S-labeled CREB variants. A, full-length CREB 327 synthesized in vitro via the T 7 -CREB 327 template (58); CREB  contains the N-terminal 198 amino acid residues, synthesized in vitro from T 7 -CREB 327 template digested with KpnI; the resulting truncated CREB peptide, lacks the bZip domain. 4 l of in vitro translated CREB 327 or CREB  proteins (TNT, Promega), analyzed by 12% SDS-PAGE and autoradiography. B, protein-protein interaction assay with 35 S-CREB  . Input (INPUT) is 4 l of 35 S-CREB 1-198 loaded onto GSF-or GST-X 1-154 resin, respectively. 35 S-CREB  bound to GST or GST-X 1-154 resin, respectively. Analysis is by 12% SDS-PAGE and autoradiography. C, protein-protein interaction assay with 35 S-CREB 327 , as described in Fig. 1.

FIG. 3. Protein-protein interaction assay with 35 S-NF-IL6 bZip (A); 35 S-ATF 3 (B); gadd153/Chop10 (C); and ICER II␥ (D). Input (INPUT) is 4-l aliquot of in vitro translated mix loaded onto
GST or GST-X 1-154 resin, as described in Fig. 1. ATF 3 plasmid DNA brings about a 6-fold enhancement of repression. The assay monitoring ATF 3 repression is carried out in the absence of forskolin stimulation; accordingly, pX expression does not enhance the transcriptional efficacy of endogenous CREB, as shown earlier in Fig. 7. Importantly, all the transfection assays contained equal amounts of total expressor plasmid DNA, to account for squelching. We observe at lower levels of transfected ATF 3 expressor (Fig. 8) increased repression efficacy by ATF 3 due to pX action.
We conclude that the viral X protein interacts directly in the hepatocyte with bZip transcription factors and effects increased DNA binding potential to their cognate DNA-binding site and, in doing so, brings about augmented transcription or repression efficacy. DISCUSSION In this study we employed in vitro and in vivo (cellular) approaches to examine the interactions of pX and several bZip transcription factors, especially those that play a role in hepatocyte function. We employed the CREB/pX system (1) to establish the in vitro conditions to detect direct and specific protein-protein interactions (Figs. 1 and 2). We show (Fig. 3) that the bZip proteins ATF 3 , ICER II␥, gadd153/Chop10, and NF-IL6 directly interact with pX. Furthermore, the mammalian two-hybrid assay demonstrated direct interactions between pX and the bZip proteins in vivo (Fig. 5), confirming our in vitro observations. This is the first demonstration of direct interactions between pX and bZip proteins in the cellular environment of the hepatocyte.
The bZip proteins, which are shown to directly interact with pX, are induced in response to environmental cues such as growth factors, cellular stress, and cytokines. ATF 3 and ICER II␥ are induced as immediate early genes, in response to conditions of cellular stress (55) and cAMP induction (45), respectively. ATF 3 mRNA expression is of particular interest for pX interactions due to its transcriptional induction in the hepatocyte following mechanical or toxin-induced liver injury (55). Importantly, the ATF 3 homodimer acts as a repressor of CREmediated transcription and forms heterodimers with gadd153/ Chop10 (55). Similarly, NF-IL6 is rapidly induced in the liver during the acute phase response (42).
We demonstrate here the functional significance of these bZip/pX interactions by in vitro (Fig. 4) and in vivo assays (Figs. [5][6][7][8]. As we demonstrated previously with CREB (1), we observe that pX significantly enhances the DNA-binding potential of these bZip proteins, in vitro. Importantly, pX enhances the DNA binding potential of bZip proteins that are capable of DNA binding; it does not convert non-DNA-binding bZip proteins, such as gadd153/Chop10 or ATF 3 /gadd153/Chop10 heterodimer, to DNA binding.
In the hepatocyte, pX expression enhances the transcriptional efficacy of each of the bZip proteins tested. Employing the HIV-LTR as our model system, we show that the transcriptional activation by NF-IL6 in the hepatocyte is enhanced 3-fold by pX. This is in agreement with our earlier observations (1). Based upon the results of Figs. 3, 4, and 6, we conclude that 1) in vivo, NF-IL6 and pX interact directly, 2) these interactions increase the DNA binding affinity of NF-IL6 for its binding site, and 3) this increased affinity increases the transcriptional efficacy of NF-IL6 by pX.
Regarding the transactivating effect of pX on the HIV-LTR, it has been demonstrated (18,21,75) that pX requires multiple cis-acting elements (18) for full transactivation, including the sequences containing the NF-IL6-binding site I at positions Ϫ158 to Ϫ178 of the HIV-LTR (18). Although there is no direct evidence demonstrating HIV infection in hepatocytes in vivo, HBV DNA has been detected in CD4 ϩ T lymphocytes and monocytes (76). Our results are consistent with the possibility that the NF-IL6/pX interactions in T lymphocytes lead to higher HIV expression, contributing to the observed synergy between HIV and HBV infections (76). Recent studies have explored the mechanisms of NF-IL6 activation of HIV-LTR in monocytic U937 cells (77,78). However, the effect of NF-IL6/pX interactions in lymphocytic or monocytic cell culture systems is not yet understood.
Importantly, pX also enhances the repression efficacy of bZip repressors ATF 3 and ICER II␥. We interpret the results of the Binding reactions were carried out as described previously (2), employing 15 ng of recombinant CREB 327 (68) and 15 ng of recombinant bZip NF-IL6, or 1 l of in vitro translated reaction mixture for ICER II␥, ATF 3 , gadd153/Chop 10 heterodimer, control TNT lysate, and 3 l of MyoD/E 47 heterodimer. The respective radioactive probes (10,000 cpm/reaction) were included in binding reaction, in the presence of 1 g poly(dI-dC). Ϫ lanes, without HBV pX, containing equivalent amount of recombinant GST protein; ϩ lanes, DNA-protein binding reactions carried out with GST-X 49 -154 protein. in vitro (Figs. 3 and 4) and in vivo (Figs. 7 and 8) assays to mean that direct interactions between pX and bZip repressors, ICER II␥ and ATF 3 , increase their DNA binding affinity, thus resulting in increased repression efficacy by pX. This is the first report describing the involvement of pX in enhancing transcriptional repression of genes responsive to bZip repressors ATF 3 and ICER II␥. The implication of this observation is that pX, by targeting bZip transcription factors, can potentiate not only specific transcriptional activation but also transcriptional repression of cellular genes. In light of the recent observation that ATF 3 is transcriptionally induced in injured liver (55), our observation is of importance and significance for hepatocyte physiology.
It is well established that CREB/ATF and possibly C/EBP proteins are required for the activation of transcription from the viral HBV enhancer I (79). The effect of the aforementioned inducible bZip proteins on the interaction of CREB with pX in the activation of HBV enhancer I-mediated transcription is unknown. Competition of the inducible bZip proteins with the constitutive CREB for interaction with pX may have direct implications in the progression of the viral infection. We are currently investigating the role of pX in the nucleus in mediating CREB/pX and/or bZip/pX interactions during hepatocyte growth and hepatocarcinogenesis.
In conclusion, the results presented in this study support the hypothesis that pX acts as an enhancer of the DNA binding potential of bZip transcription factors, thereby increasing their transcriptional activation or repression of cellular and/or viral FIG. 5. Mammalian two-hybrid assay of HBV pX and bZip transcription factors. 60-mm culture dishes of subconfluent AML12 cells transfected with 5 g of reporter plasmid 5xGAL4-E 1 b TATAA-luciferase (66) and the following amounts of expressors: 2 g of pEM-MyoD and 10 g of RSV-X-Gal4; 5 g RSV-CREB-VP16 in the presence of 5 g of RSV-X-Gal4; 5 g of RSV-ICER II␥-VP16 and 10 g of RSV-X-Gal4; 5 g of RSV-ATF 3 -VP16 and 10 g of RSV-X-Gal4; 10 g of RSV-NF-IL6-VP16 and 10 g of RSV-X-Gal4. Control transfections contained instead of the RSV-X-Gal4 DNA equal amounts of PUC19 DNA or RSV-empty vector. Cellular extracts were prepared 16 h later and assayed for luciferase activity. Luciferase activity was quantitated per g of cellular extract. Relative luciferase activity is the ratio of reporter activity obtained from cotransfection of both RSV-X-Gal4 and RSV-bZip-VP16 plasmids, in comparison to the control transfection.