Hepatitis B virus X protein is a transcriptional modulator that communicates with transcription factor IIB and the RNA polymerase II subunit 5.

Hepatitis B virus X protein (HBx) transactivates viral and cellular genes through a wide variety of cis-elements. However, the mechanism is still obscure. Our finding that HBx directly interacts with RNA polymerase II subunit 5 (RPB5), a common subunit of RNA polymerases, implies that HBx directly modulates the function of RNA polymerase (Cheong, J. H., Yi, M., Lin, Y., and Murakami, S. (1995) EMBO J. 14, 142-150). In this context, we examined the possibility that HBx and RPB5 interact with other general transcription factors. HBx and RPB5 specifically bound to transcription factor IIB (TFIIB) in vitro, both of which were detected by either far-Western blotting or the glutathione S-transferase-resin pull-down assay. Delineation of the binding regions of these three proteins revealed that HBx, RPB5, and TFIIB each has two binding regions for the other two proteins. Co-immunoprecipitation using HepG2 cell lysates that express HBx demonstrated trimeric interaction in vivo Some HBx substitution mutants, which had severely impaired transacting activity, exhibited reduced binding affinity with either TFIIB or RPB5 in a mutually exclusive manner, suggesting that HBx transactivation requires the interactions of both RPB5 and TFIIB. These results indicated that HBx is a novel virus modulator that facilitates transcriptional initiation by stabilizing the association between RNA polymerase and TFIIB through communication with RPB5 and TFIIB.


Hepatitis B virus X protein (HBx) transactivates viral
and cellular genes through a wide variety of cis-elements. However, the mechanism is still obscure. Our finding that HBx directly interacts with RNA polymerase II subunit 5 (RPB5), a common subunit of RNA polymerases, implies that HBx directly modulates the function of RNA polymerase (Cheong, J. H., Yi, M., Lin, Y., and Murakami, S. (1995) EMBO J. 14, 142-150). In this context, we examined the possibility that HBx and RPB5 interact with other general transcription factors. HBx and RPB5 specifically bound to transcription factor IIB (TFIIB) in vitro, both of which were detected by either far-Western blotting or the glutathione S-transferaseresin pull-down assay. Delineation of the binding regions of these three proteins revealed that HBx, RPB5, and TFIIB each has two binding regions for the other two proteins. Co-immunoprecipitation using HepG2 cell lysates that express HBx demonstrated trimeric interaction in vivo. Some HBx substitution mutants, which had severely impaired transacting activity, exhibited reduced binding affinity with either TFIIB or RPB5 in a mutually exclusive manner, suggesting that HBx transactivation requires the interactions of both RPB5 and TFIIB. These results indicated that HBx is a novel virus modulator that facilitates transcriptional initiation by stabilizing the association between RNA polymerase and TFIIB through communication with RPB5 and TFIIB.
Human hepatitis B virus (HBV) 1 is one major risk factor associated with primary hepatocellular carcinoma. HBx, the product of the smallest open reading frame of the HBV genome, is required for virus multiplication in vivo (1,2). Transient transfection has demonstrated that HBx transactivates a wide variety of viral and cellular promoters. The HBx responsible cis-elements present in RNA polymerase II (RNAPII) promot-ers include AP-1, AP-2, ATF, c/EBP, NF-B sites, and serumresponsive element (3)(4)(5)(6)(7)(8)(9)(10)(11). One of the RNA polymerase III promoters is also transactivated by HBx (12). Several endogenous genes important for cell proliferation and acute inflammatory response, such as c-fos, c-jun, and human interleukin-8, are activated by HBx (3,7,13). This broad gene-regulating 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 (14,15).
Since HBx cannot bind DNA directly, protein-protein interaction is crucial for HBx transactivation (16). The reported HBx-binding proteins include transcription factors such as TBP (17), RPB5 (18), CREB/ATF2 (19), and Oct-1 (20), a probable DNA repair enzyme (21), a human homologue of Drosophila 20 S proteasome subunit (22,23), and the tumor suppressor p53 (24,25). These findings are so diverse that the mechanism of HBx transactivation remains obscure. It has been reported that HBx mediates transcription activation through the modifying signal transduction pathways in cytoplasm such as the activation of AP-1 via protein kinase C and the signaling pathway of Ras-Raf-mitogen-activated protein kinase (4, 5, 26 -29). However, it is difficult to reconcile the signal transduction models with the observations that HBx acts in in vitro transcription (30). Since the subcellular localization of HBx is both cytoplasmic and nuclear (16,31), 2 HBx may have a dual role in transcriptional regulation, such that nuclear HBx functions at the promoter level as cytoplasmic HBx influences the regulation of second messenger systems (31). Another model postulates that HBx directly interacts with the transcription machinery. This model is supported by several lines of evidence. HBx can activate transcription in vitro using whole cell extracts (30). HBx fused to LexA or c/EBP DNA-binding domains can transactivate the reporters harboring the responsive ciselements (9,32). HBx co-activates potent activators, including acidic activators such as VP16 and p53 as well as non-acidic activators such as E1a (33). Finally, we showed that HBx can specifically bind to RPB5, a common subunit shared by eukaryotic nuclear RNA polymerase I, II, and III, both in vitro and in vivo. RPB5 overexpression can stimulate the transcription of HBx responsible reporters (18) supporting the notion that HBx transactivation acts through direct interaction with the transcription machinery.
To elucidate the role of HBx-RPB5 interaction in transcription modification, we investigated whether or not HBx and * This work was supported in part by grants-in-aid from the Ministry of Science, 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. RPB5 interact with other transcription factors. Here we report that both HBx and RPB5 interact with TFIIB through different sites and that they may form a ternary complex that facilitates transcription.
Preparation of Recombinant Proteins-GST-fused proteins were expressed in E. coli by induction for 3 h at 30°C with 0.4 mM isopropyl-␤-D-thiogalactopyranoside. Cells were harvested and sonicated in PBST buffer (phosphate-buffered saline containing 1% Triton X-100). After centrifugation, the extracts (supernatants) were collected and stored at Ϫ80°C. For purification, the extracts were incubated with glutathione-Sepharose 4B resin (Pharmacia Biotech Inc.) for 1 h at room temperature. The beads were precipitated and washed four times with a 50-fold volume of PBST buffer followed by elution with 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0). The eluted proteins were divided into aliquots and stored at Ϫ80°C.
Recombinant HBx was prepared by digesting the GST-HBx-fused protein with thrombin as described (36). The HBx protein was purified by separation in and elution from SDS-PAGE with acetic buffer (50 mM NaOAc, pH 5.2, 0.1% ␤-mercaptoethanol), then stored at Ϫ80°C.
Far-Western Blotting-Far-Western blotting proceeded as reported with some modification. Purified GST-fused proteins were labeled in vitro as probes with ␥-32 P, using the catalytic subunit of cAMP-dependent protein kinase (Sigma). The target proteins (200 ng each) were fractionated in SDS-PAGE gels and then electrotransferred onto nitrocellulose membranes, which were denatured, renatured, blocked with 5% skim milk in modified GBT buffer (10% glycerol, 50 mM Hepes-NaOH, pH 7.5, 170 mM KCl, 7.5 mM MgCl 2 , 0.1 mM EDTA, 0.1 mM dithiotreitol, 1% Triton X-100) and far-Western blotted (18). The binding reactions proceeded in modified GBT buffer containing 32 P-labeled probe (40 -100 ng/ml protein, 2 ϫ 10 6 cpm/g of protein), 1% bovine serum albumin, 2 mM unlabeled ATP, and the sonication supernatant of E. coli JM109 transformed with pGENK1 containing a final protein concentration of 1 mg/ml GST. The membranes were rotated in the reaction mixture for 1 h at room temperature, washed five times with modified GBT buffer, then exposed to x-ray films (X-Omat AR, Eastman Kodak Co.) for 1-3 days or to imaging plates (Fuji) overnight.
GST-Resin Pull-down Assay-Equal amounts (approximately 1 g) of GST, GST-TBP, GST-RPB5, and GST-HBx immobilized on 10 l of glutathione resin were incubated with 100-ng target proteins in modified GBT buffer containing 1% bovine serum albumin for 2 h at 4°C. After 4 washes with modified GBT buffer, the bound proteins were eluted, then Western blotted.
Immunoprecipitation and Western Blotting-Antibodies against HBx, RPB5, and TFIIB were generated by immunizing rabbits with purified GST-fused proteins. IgG fractions were purified by affinity chromatography using protein A followed by GST (18). HepG2 cells were transfected under the same conditions as described for CAT assays except that they were incubated in 100-mm culture dishes (18). The cells were sonicated in LAC buffer (10% glycerol, 20 mM HEPES, pH 7.9, 50 mM KCl, 0.4 mM NaCl, 10 mM MgCl 2 , 0.1 mM dithiothreitol, 0.1 mM EDTA, 9 mM CHAPS, 0.5 mM phenylmethylsulfonyl fluoride, and 10 g/ml aprotinin and leupeptin) followed by centrifugation. The supernatants were collected and stored at Ϫ80°C. The total cell lysate in 500 l of TBST buffer (50 mM Tris-HCl, pH 7.5, 0.2 M NaCl, 0.05% Tween 20) was clarified by an incubation with 20 l of swollen protein A-Sepharose resin (Pharmacia) for 30 min followed by centrifugation. The supernatants were then immunoprecipitated. After reaction with the first antibody (3 g of IgG) for 1 h at 4°C, 10 l of 50% swollen protein A-Sepharose resin was added. Samples were rotated for another hour followed by 3 washes with washing buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 1 mM EDTA). The bound proteins were eluted with elution buffer (10 mM Tris-HCl, pH 7.4, 2% SDS, 0.3 M NaCl, and 1 mM EDTA), separated by SDS-PAGE, then transferred onto nitrocellulose membranes, and Western blotted using second antibodies to detect HBx, TFIIB, or RPB5. The proteins were visualized by enhanced chemiluminescence (ECL) according to the manufacturer's (Amersham) instructions (18).
Transfection and CAT Assay-Transient transfection and CAT assays proceeded as described (7,28). HepG2 cells were cultured in 60-mm plates and transfected. Total cell lysates were prepared from the cells 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. The fractionated TLC plates were exposed to imaging plates, and CAT activities were measured as the percentage of conversion to acetylated forms of [ 14 C]chloramphenicol (% acetylation, Amersham) 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. Representative data are shown.

HBx, RPB5, and TFIIB Associate with Each Other in Vitro and in Vivo-Specific HBx-RPB5 interaction is involved in
HBx transactivation, suggesting that RPB5, as a communicating subunit of RNA polymerase, interacts with general transcription factors (18). We investigated whether or not RPB5 interacts with other general transcription factors by means of far-Western blotting using bacterially expressed GST fusion forms of TBP, TFIIB, and CTD (Fig. 1A). Comparable amounts of proteins were applied and fractionated by SDS-PAGE as shown by Coomassie Blue staining (lanes [25][26][27][28][29][30]. The RPB5 probe bound to TFIIB with high affinity (lane 2) as well as to HBx and to itself as reported (lanes 4 and 6) (18). Similarly, the TFIIB probe also associated with RPB5 (lane 10), TBP (lane 7), and with itself. The self-binding of TFIIB agrees with the reported findings (38). TFIIB and HBx associated with each other (lanes 12 and 14). None of the probes bound CTD or GST. The TBP and GST probes did not bind either HBx or RPB5 (lanes 19 -24 and data not shown), indicating that the binding among HBx, RPB5, and TFIIB is specific. These results suggested that TFIIB is another interaction target of HBx. To confirm the results of the far-Western blotting, we performed in vitro GST-resin pull-down assays. As shown in panel a of Fig. 1B, TFIIB-His6 was efficiently retained by GST-TBP, GST-RPB5, and GST-TFIIB (lanes 3-5) but barely recovered by GST (lane 2). The in vitro GST-resin pull-down assay was specific, since TBP-His6 was only retained by GST-TFIIB but not GST-HBx and GST-RPB5 (Fig. 1B, panel b). These results provided further evidence of the specific binding of TFIIB to RPB5 and HBx.
Next we performed immunoprecipitation and Western blotting studies to examine whether these interactions proceed in vivo. HepG2 cell lysates transiently co-transfected with HBx, and RPB5 and TFIIB were precipitated with anti-RPB5 and anti-TFIIB. Antibodies against TBP, tubulin, and normal rabbit serum were used as controls. The antibody-bound proteins were precipitated with Sepharose-bound protein A, then Western blotted with anti-HBx antibody. HBx was co-precipitated with RPB5 and TFIIB ( Fig. 2A, lanes 3 and 4). Although only a small amount of HBx was co-precipitated, this phenomenon might be specific since HBx was not detectably precipitated by the control antibodies (lanes 2, 5, and 6). These results indicate that HBx associates with RPB5 and TFIIB in vivo when RPB5 and TFIIB are overexpressed. To determine whether these interactions occur when RPB5 and TFIIB are expressed at physiological levels, HepG2 cell lysate transiently transfected with HBx was precipitated with anti-TFIIB. The antibodybound proteins were Western blotted for HBx, RPB5, and TFIIB. HBx and endogenous RPB5 were co-precipitated with TFIIB by anti-TFIIB (Fig. 2B, lane 6) whereas none of these proteins were recovered by normal rabbit antibody (lane 5). Similarly, HBx and endogenous TFIIB were also recovered together with RPB5 in the anti-RPB5 precipitates (data not shown). When non-transfected HepG2 cell lysate was precipitated with anti-TFIIB, RPB5 was co-precipitated whereas no HBx band was detected (Fig. 2B, lane 3). These results indicate that HBx, RPB5, and TFIIB associate with each other under physiological conditions. The concurrent precipitation of HBx, RPB5, and TFIIB implies trimeric interaction or tripartite complex formation (see below).
HBx, TFIIB, and RPB5 Have Distinct and Separable Binding Sites for Each of the Other Two Proteins-Having established the interactions among HBx, RPB5, and TFIIB, we mapped the interaction regions of these proteins by far-Western blotting. As we reported, the HBx-binding site is located in the middle of RPB5 (10); HBx probe bound to the constructs harboring this region (73-120 aa) of RPB5, namely -d1 (47-210 aa), -d13 (47-120 aa), and -d83 (73-120 aa) (Fig. 3A). The TFIIB probe efficiently associated with the truncated forms of RPB5 proteins harboring the N-terminal region, namely -d4 (1-46 aa) and -d27 (21-47 aa), but interacted weakly or barely with constructs having the middle region (-d13, -d1, and -d83), indicating that the N-terminal portion is necessary for TFIIB binding (Fig. 3B). These results indicate that in RPB5, the binding sites of HBx and TFIIB are discrete and separable.
TFIIB contains two imperfect direct repeats in the C-terminal two-thirds and one putative zinc finger in the N-terminal portion (39,40). Far-Western blotting with the HBx probe showed that the HBx-binding site is located in the two direct repeats, since the probe bound to the constructs containing the repeats, including TFIIB-full (1-316 aa), -d9 (1-285 aa), and -d10 (124 -285 aa) (Fig. 4A). In contrast, RPB5 binding does not require the presence of the direct repeats, since RPB5 did not bind to -d10 (124 -285 aa). The RPB5-binding region is within the N-terminal one-fifth because the RPB5 probe bound to constructs containing this region, namely TFIIB-d1 (1-60 aa), -d3 (1-124 aa), -d9 (1-285 aa), and full-length TFIIB (1-316 aa) (Fig. 4B). Since this region contains a zinc finger, further delineation was conducted to see whether this motif contributes to RPB5 binding. The results showed that the zinc finger motif is needed for the binding, since destruction of the zinc finger motif in -d21 (21-316 aa) reduced the binding ability to RPB5 probe. Supporting this conclusion, the smallest construct TFIIB-d16 (21-60 aa) retained only reduced RPB5 binding ability, whereas the internal deletion mutant proteins, -I37 (17-20 and 37-316 aa) and -I61 (45-49 and 61-316 aa) could not bind to the RPB5 probe (Fig. 4B). Therefore, the RPB5binding region is located within 1-60 aa of TFIIB, and the zinc finger motif is necessary for the interaction. These results indicated that the binding regions for HBx and RPB5 in TFIIB are different.
HBx consists of functional transactivation (51-154 aa) and negative regulatory domains (1-50 aa) (37). Both RPB5-and TFIIB-binding sites are located within the transactivation domain of HBx (-5D1, 51-154 aa) but not in the regulation domain (-3D5, 1-50 aa) (Fig. 5). Further truncation of HBx showed that the RPB5-and TFIIB-binding sites might be dif- FIG. 2. HBx, RPB5, and TFIIB are associated in vivo. A, HepG2 cells were transfected with mammalian vectors expressing HBx, RPB5, and TFIIB (pSG5-HBx, -RPB5, and TFIIB). Total cell lysates were prepared as described under "Experimental Procedures." Cell lysates (1.5 mg) were immunoprecipitated with anti-RPB5 or anti-TFIIB antibody. Anti-TBP, anti-tubulin, and normal rabbit serum (NRS) were used as controls. The precipitated proteins were eluted, then fractionated by SDS-PAGE, and Western blotted for HBx by anti-HBx. Lane 1, 5% of the cell lysate used in immunoprecipitation. B, cell lysates were derived from HepG2 cells with or without transfection of pSG5-HBx. Total cell lysates (500 g) were immunoprecipitated with anti-TFIIB or NRS. The precipitated proteins were eluted, separated by SDS-PAGE, and transferred onto nitrocellulose membranes, which were cut into three pieces according to the molecular mass markers and Western blotted to detect HBx, RPB5, or TFIIB, respectively.  (Fig. 5A). The RPB5 probe bound neither -5D4 nor -D26 (Fig. 5B, lanes 4, 5, and 9) but bound the larger constructs such as D16 (51-136 aa) (lane 8). These results suggested that RPB5 and TFIIB binding requires RPB5 and TFIIB Binding Are Essential for the Transactivation Activity of HBx-HBx, RPB5, and TFIIB associated with each other through distinct binding sites, which may contribute to the transactivation function of HBx. To investigate this notion, we attempted to introduce substitution mutations into the HBx transactivation domain. Since we could not find a single mutation within the HBx transactivation domain that markedly affects the transactivation function as reported by others (41), we constructed mutants with multiple mutations (three or four substitutions in a row as described under "Experimental Procedures"). The HBx mutant proteins were examined for transactivation activities and binding affinity for RPB5 and TFIIB. All the mutants except Xm76 had severely impaired transactivation activity (Fig. 6A, left panel). The difference in the transactivation activities was not because of a different expression level or stability of the HBx mutants, since comparable HBx mutant proteins were detected by Western blotting (Fig. 6A, right panel). RPB5 binding of Xm89 and Xm93 was reduced whereas the TFIIB binding was similar to that of wild HBx (Fig. 6B, compare lanes 2 and 3 with lane 7). In contrast, mutation Xm120 had reduced binding to TFIIB, but there was no effect upon RPB5 binding (compare lane 4 to lane 7). The double mutant, Xm93 m120, exhibited reduced binding affinity for TFIIB and RPB5 (data not shown). The mutant Xm76 with normal transactivation activity retained binding affinity with both the TFIIB and RPB5 (lane 1). The results showed that RPB5 and TFIIB bind to different regions in the HBx transactivation domain and suggested that both the RPB5-and TFIIB-binding regions are essential for HBx transactivation function.

HBx, RPB5, and TFIIB May Form a Ternary Complex-
Since HBx, RPB5, and TFIIB have two different binding sites each for the other two partners, these three proteins may form a tripartite complex. We examined this possibility by means of a GST-resin pull-down assay. The GST-fusion form of the full size RPB5 protein immobilized on glutathione-Sepharose resin was incubated in a mixture containing HBx and TFIIB-His6. After extensive washing, bound proteins were separated by SDS-PAGE and visualized by Western blotting (Fig. 7A). Both HBx and TFIIB were retained by GST-RPB5 resin (lane 2) but not by GST resin (lane 3). This result suggests that RPB5 simultaneously binds HBx and TFIIB. We then examined the possibility that these proteins form a trimeric complex using RPB5 mutants. The RPB5-full and -d1, harboring the HBxbinding site, bound HBx independently of the presence of TFIIB (Fig. 7B, lanes 7-10). However, RPB5-d4, which has the TFIIB-but not HBx-binding site, efficiently pulled down HBx only when TFIIB was present in the binding reaction (lane 12). As negative controls, neither GST nor GST-TBP bound HBx regardless of the presence of TFIIB (lanes 3-6). These results showed that RPB5-d4 associates with HBx indirectly through the bridging of TFIIB and that HBx, RPB5, and TFIIB simultaneously interact with one another to form a trimeric complex in vitro. DISCUSSION We present evidence of specific interactions among TFIIB, RPB5, and HBx, which were detected in vitro by far-Western blotting and the GST-resin pull-down assay as well as in vivo by co-immunoprecipitation. Since each of these three proteins has two distinct and separable binding sites for the other two, they may form a trimeric complex. The simultaneous binding of HBx and TFIIB to RPB5 detected by GST-resin pull-down assay, as well as the finding that the HBx-binding-defective deletion mutant RPB5-d4 (1-47 aa) bound HBx through the bridging of TFIIB, indicated trimeric complex formation at least in vitro. Our results imply that trimeric interaction or complex formation is required for HBx transactivation, since the HBx mutations, which reduced the binding ability either to TFIIB or RPB5, markedly impaired transactivation activity.
TFIIB is conserved from archaebacteria to humans (42,43). In combination with RNA polymerase II, TFIIB determines the transcription start site (44,45). The importance of TFIIB in activated transcription has been revealed by its interactions with a variety of gene-specific transcriptional regulator proteins (see Refs. 46 and 47 and references therein), cofactors (48 -50), and general transcription factors such as TFIID, TFIIF, and RNA polymerase II (37,38,51,52). The specific interaction of HBx and TFIIB implies that TFIIB serves as a regulation target of HBx. Several virus regulators interact with TFIIB, such as VP16, ICP4 of herpes simplex virus, EBNA-2 of Epstein-Barr virus, IE2 of human cytomegalovirus, E2TA of bovine papillomavirus, and Tat of human T-cell lymphotrophic virus (37,53). HBx can be categorized as a novel type of transcriptional modulator that communicates with RNA polymerase and the general transcription factor TFIIB, since it is distinct from the precedents that act through binding distal cis-elements and communicating with TFIIB to activate transcription. The HBx-binding region was confined within the two direct repeat regions of TFIIB, which is necessary for the binding of TBP, VP16, and some cofactors (37,39,54,55). HBx may modify TFIIB function by inducing a conformational change in TFIIB through interaction with the two direct repeats, like VP16 (37).
The RPB5-binding site was mapped within the N-terminal part of TFIIB (1-60 aa), and we found that the zinc finger is necessary for the binding. The site is mapped within the region (1-106 aa) important for the recruitment of RNA polymerase II to promoters (39,55,56), but seems to be distinct from the putative TFIIF-binding region, because the yeast TFIIB mutation E62K, outside of the RPB5-binding region, can be suppressed by the largest subunit of TFIIF (56). Therefore, the N-terminal part of TFIIB may play an important role in transcription initiation through interacting with RPB5, the exposed subunit of RNA polymerase. The N-terminal part of TFIIB including the zinc finger and its adjacent regions is involved in the transcriptional regulation of the glutamine-rich activator, ftz Q (47).
Two models of transcriptional initiation have been proposed. According to the stepwise model, TFIID first binds the promoter through its TBP subunit, perhaps assisted by TFIIA. The TFIID-promoter complex is subsequently recognized by TFIIB, which acts as a bridging factor by recruiting RNAPII/TFIIF (57,58). Since HBx can simultaneously bind RPB5 and TFIIB, it may facilitate recruiting RNAPII/TFIIF to the promoter and/or stabilize the pre-initiation complex through HBx-RPB5-TFIIB interactions. Another model is that the RNA polymerase holoenzyme, a large complex containing RNAPII and other factors such as TFIIF, TFIIB, TFIIH, and mediators, is directly recruited to the promoter (58 -60). Activators may enhance the initiation by interaction with a component of the holoenzyme (58,59,61). According to the latter model, HBx may promote holoenzyme assembly or stabilize it through communication with TFIIB and RPB5. Alternatively, HBx may facilitate the recycling of transcription factors, which is thought to be a regulating step during the transcription process (62). HBx expressed by HepG2 has been recovered together with RPB5 in the fast sedimenting fractions (over 1000 kDa) by glycerol gradient centrifugation (18), suggesting that HBx is associated with the holoenzyme and that it modulates the function of the holoenzyme.
Despite the direct interaction of HBx and basal transcription factors, HBx transactivation requires distinct cis-elements, such as AP1-, NF-B-, and SRF-binding sites, in a given reporter as reported by several groups. Therefore an obvious question is how the HBx-RPB5-TFIIB trimeric interaction enables promoter selectivity in HBx transactivation. HBx alone does not transactivate the pGal4-CAT reporter but facilitates the activation of Gal4-VP16, suggesting that HBx acts during activated, but not basal transcription (33,63). 3 In this regard, HBx may directly or indirectly communicate with some transcriptional activators (19,20). Another possibility is the presence of RPB5-binding negative factor(s) or a co-repressor that can be replaced by HBx. The latter was suggested from the observation that the overexpression of RPB5 or its HBx-binding region exhibits transacting ability that requires HBx-responsive cis-elements (18). These two possibilities are not mutually exclusive. Isolating and characterizing the host proteins that interact with RPB5 may clarify the mechanism of HBx transactivation.
Both the RPB5-and TFIIB-binding regions in the transactivation domain of HBx were indispensable for HBx transactivation, suggesting that HBx transactivates through direct interactions with general transcription factors. However, these results do not exclude the possibility that HBx transactivates through other routes such as the Ras signaling pathways in the cytoplasm. If HBx has several independent pathways for transactivation, some mutations of HBx that differentially affect them may be isolated, and host proteins that selectively affect one of these pathways may be identified, as shown for human T-cell lymphotrophic virus Tax-1 (64) .  FIG. 7. HBx, RPB5, and TFIIB may form a tripartite complex. A, GST-RPB5 simultaneously binds HBx and TFIIB in vitro. The GST-resin pull-down assay proceeded as described in the legend to Fig. 1. HBx protein was prepared by digesting GST-HBx with thrombin as described under "Experimental Procedures." Lane 1, 20% of input HBx and TFIIB-His6, respectively. B, GST-resin pull-down assay using glutathione-Sepharose resin pre-bound to GST-fused full-length or truncated RPB5 protein. The resin was incubated and washed as described in the legend to Fig. 1. Lanes 3,5,7,9,and 11, input is HBx protein alone (100 ng); lanes 4, 6, 8, 10, and 12, input is HBx (100 ng) together with TFIIB (100 ng). Lanes 1 and 2, 10% of input of HBx or HBx/TFIIB, respectively. TFIIB and TBP, to R. G. Roeder for TFIIB-His6 and TBP-His6 plasmids, to K. Matsushima, M. Seiki, and S. Kaneko for critical discussion, and to C. Matsushima and F. Momoshima for technical assistance.