JOURNAL OF VIROLOGY, Feb. 1998, p. 1013–1019 0022-538X/98/$04.0010 Copyright © 1998, American Society for Microbiology
Vol. 72, No. 2
Functional Interaction of the Bovine Papillomavirus E2 Transactivation Domain with TFIIB JUN-MEI YAO,1 DAVID E. BREIDING,1
AND
ELLIOT J. ANDROPHY1,2*
Department of Dermatology, New England Medical Center and Tufts University School of Medicine,1 and Department of Molecular Biology and Microbiology, Tufts University School of Medicine,2 Boston, Massachusetts 02111 Received 13 August 1997/Accepted 5 November 1997
Induction of gene expression by the papillomavirus E2 protein requires its ;220-amino-acid amino-terminal transactivation domain (TAD) to interact with cellular factors that lead to formation of an activated RNA polymerase complex. These interaction partners have yet to be identified and characterized. The E2 protein localizes the transcription complex to the target promoter through its carboxy-terminal sequence-specific DNA binding domain. This domain has been reported to bind the basal transcription factors TATA-binding protein and TFIIB. We present evidence establishing a direct interaction between amino acids 74 to 134 of the E2 TAD and TFIIB. Within this region, the E2 point mutant N127Y was partially defective and W99C was completely defective for TFIIB binding in vitro, and these mutants displayed reduced or no transcriptional activity, respectively, upon transfection into C33A cells. Overexpression of TFIIB specifically restored transactivation by N127Y to close to wild-type levels, while W99C remained inactive. To further demonstrate the functional interaction of TFIIB with the wild-type E2 TAD, this region was fused to a bacterial DNA binding domain (LexA:E2:1-216). Upon transfection with increasing amounts of LexA:E2:1-216, there was reduction of its transcriptional activity, a phenomenon thought to result from titration of limiting factors, or squelching. Squelching of LexA:E2:1-216, or the wild-type E2 activator, was partially relieved by overexpression of TFIIB. We conclude that a specific region of the E2 TAD functionally interacts with TFIIB. The complex of TFIIB with TBP at a promoter represents the minimal requirement for formation of the preinitiation complex but is not sufficient for activation of gene expression. Transcriptional activators, including viral activation domains such as VP16 (29), EBNA-2 of Epstein-Barr virus (30), and large T antigen of simian virus (27), bind TFIIB directly. This interaction is thought to catalyze the assembly to the preinitiation complex. The TFIIB interaction with activators may also stimulate mRNA elongation by the RNA polymerase IIassociated complex (7, 47, 49). In this report, we present in vitro and in vivo results that demonstrate that TFIIB binds and functionally cooperates with BPV-1 E2 to stimulate transcription. Their association is mediated by two distinct regions of E2. As reported previously by Rank and Lambert (34), the E2 C-terminal DBD binds TFIIB. In addition, a region within the E2 activation domain encompassing aa 74 to 134 specifically and directly binds TFIIB. Our results indicate that the interaction of TFIIB with the E2 TAD is necessary for transcriptional activation.
The papillomavirus E2 proteins regulate viral transcription and replication (1, 4, 8, 18, 21, 50). The bovine papillomavirus type 1 (BPV-1) E2 gene product has been most widely studied, and because there is moderately strong homology among all animal papillomavirus E2 proteins, it is likely that they share important properties and characteristics. The 410-amino-acid (aa) BPV E2 protein contains modular domains (19). The C-terminal 125 aa of BPV-1 E2 serve as the DNA binding and dimerization domain (18, 19, 26, 32, 33). This C-terminal region has been shown to interact with TATA-binding protein (TBP) and TFIIB (34). However, the E2 transactivation domain (TAD) can stimulate transcription in eukaryotic cells when cloned onto a heterologous DNA binding domain (DBD) (9, 25, 46). The N-terminal ;220-aa transcriptional activation domain was predicted by computer algorithms to form two putative acidic amphipathic helices followed by a hydrophobic beta sheet. Mutations in each of these motifs render E2 inactive for transcriptional activation (1, 8, 18). Our goal is to identify the network of connections formed between E2 and cellular proteins, and we have isolated a novel cellular protein that functionally interacts with BPV-1 E2 and mapped its binding to the hydrophobic region of the TAD (9). Interactions between enhancer-activating proteins such as E2 and cellular factors have been demonstrated for other activators. One of the best-studied models is the small (;80 aa) and potent strong TAD of the herpes virus VP16 gene product (29). The VP16 TAD has been shown to bind to TFIIB, TBP, dTAF40, ADA2, and the p62 subunit of TFIIH (5, 13, 20, 41, 48). The general transcription factor IIB (TFIIB) is an essential component of the RNA polymerase II transcription apparatus.
MATERIALS AND METHODS Plasmids. The glutathione S-transferase (GST) fusion plasmid pGEXN:TFIIB was constructed by subcloning the TFIIB cDNA from pFIIB-11d, kindly provided by C.-M. Chiang and R. G. Roeder (12). GST:TFIIB deletion mutants were kindly supplied by D. Reinberg (23). For mammalian cell expression, TFIIB was transferred into pcDNA3 and pCG vectors. Wild-type and mutant pCG E2 expression vectors have been previously described (8, 21). The pGEX2T:E2:1134 wild-type and mutant vectors were constructed by cleavage of pGEX2T:E2: 1-286 (9) by digestion with AccI and KpnI followed by Klenow polymerase treatment. The following GST-E2 fusions were generated by cloning PCR fragments into pGEX2T: pGEX2T:E2:1-98, pGEX2T:E2:54-134, pGEX2T:E2:1591, pGEX2T:E2:15-134, pGEX2T:E2:74-134, pGEX2T:E2:91-134, and pGEX2T: E2:74-113. The E2:113-286 insert was cleaved from pUC19: NarbE2 (8) with BamHI and EcoRI and inserted into pGEX2T cleaved with the same enzymes. pGEX2T: E2:113-134 was derived from pGEX2T:E2:113-286 by digestion with AccI and KpnI followed by Klenow polymerase treatment. pGEX2T:E2:215-410 was made by transferring the insert encoding these amino acids from YEplac112G:E2:215-410 (8) to pGEX2T. To make six-histidine-tagged E2:1-216, oligonucleotides encoding a sixhistidine motif were placed in frame 59 to E2:1-216 in pET8C. The pcDNA3:
* Corresponding author. Mailing address: Department of Dermatology, New England Medical Center Box 166, 750 Washington St., Boston, MA 02111. Phone: (617) 636-1493. Fax: (617) 636-6190. Email:
[email protected]. 1013
1014
YAO ET AL.
J. VIROL.
LexA202B and pcDNA3:LexA:E2:1-126 constructs and pDBL8 reporter (9), the negative control GST:SD21-HC8 subunit gene of human proteasome (11), and luciferase reporter plasmid pJSLuc (40) will be described separately. The pcDNA3: LexA:E2:1-133 and 162-410 constructs were made by transferring the LexA:E2 fusion from the analogous yeast expression vector (8) as HindIII/SpeI fragments to HindIII/XbaI-cleaved pcDNA3 plasmid. pcDNA3:LexA:VP16 (VP16 aa 410 to 490) was constructed from the VP16 activation domain in pSD10 (16) as a BamHI/XbaI fragment into pcDNA:LexA202B. Protein expression and binding assays. GST, GST:VP16, GST:TFIIB, and E2 fusion proteins were expressed in pLysS from pGEX plasmids (Pharmacia). Fifty milliliters of cells was grown to an optical density of 0.6 to 0.9 at 600 nm, induced with 0.5 mM isopropyl-b-D-thiogalactopyranoside (IPTG) for 2.5 h, pelleted, washed, and frozen. Cell pellets were resuspended in 2 ml of NETN (100 mM NaCl, 0.1 mM EDTA, 20 mM Tris-HCl [pH 8.0], and 0.1% Nonidet P-40) with 1 mM phenylmethylsulfonyl fluoride (PMSF). Cells were disrupted by sonication on ice, and insoluble material was removed by centrifugation at 15,000 3 g. E2 C-terminal (aa 215 to 410) GST fusion proteins were soluble at these conditions and were collected with glutathione-agarose beads added directly to the supernatant. Most N-terminal (aa 1 to 216) E2 fusion proteins were not soluble under these conditions. To purify these proteins, pellets were solubilized in 1 ml of 8 M urea. After sonication, Triton X-100 was added to 1%, and this fraction was diluted with 30 ml of NETN. After removal of debris by centrifugation at 15,000 3 g, the solubilized GST:E2 fusion proteins were collected by addition of 0.25 ml of glutathione-agarose beads. After being washed with NETN, beads were resuspended in NETN with the protease inhibitors PMSF, leupeptin, and pepstatin. FLAG-tagged TFIIB was prepared from Escherichia coli and purified according to the method of Chiang and Roeder (12). In vitro-synthesized proteins were produced and radiolabeled with Sp6 or T7-based rabbit reticulocyte lysate-coupled transcription-translation kits (Promega). Concentrations of unlabeled proteins were measured with the bicinchoninic acid protein assay kit (Pierce). For binding reactions, 2.5 mg of purified GST or GST fusion proteins was incubated with in vitro-translated 35S-labeled protein in 100 ml of IP buffer (Tris-HCl buffer at pH 7.9, 100 mM NaCl, 1% Nonidet P-40, 1 mM PMSF, and 1 mM dithiothreitol) (42) at 4°C for 1 h. The beads were washed three times in IP buffer containing 100 mM KCl and then in IP buffer containing 500 mM KCl three times. The bound proteins were analyzed by sodium dodecyl sulfate (SDS)polyacrylamide gel electrophoresis (PAGE) and quantified by a PhosphorImager (Bio-Rad). GST or GST fusion proteins were purified from bacteria and used in proteinprotein interaction assays under the conditions described above. Bound proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The filter was blocked with 5% milk in TBS-T buffer (10 mM Tris, 150 mM NaCl, 0.05% Tween 20); incubated either with an anti-TFIIB (43), an anti-FLAG (IBI), or an anti-E2 (33) antibody; and subsequently reacted with a secondary antibody conjugated with peroxidase. Immunoblots were developed by enhanced chemiluminescence (ECL) as described by the manufacturer (Amersham). Cell transfection and transcriptional activation assay. Human C33A cells were plated at 106 cells per 60-mm dish 1 day prior to calcium phosphatemediated transfection (10). Vector DNA was added to each transfection to normalize the total amount of expression vector. Each transfection included 1 mg of the luciferase reporter plasmid pJSLuc containing three E2 binding sites or pDBL8 with eight LexA binding sites. Cell extracts were harvested 48 h later with reporter lysate buffer (Promega). The luciferase activity in a 5-ml sample of lysate was determined by adding 25 ml of luciferase assay reagent as measured in a luminometer (Lumat LB9501; Berthod). The basal luciferase activity obtained with the reporter plasmid and expression vector without insert (pcDNA3 or pCG) was set to 1.0. Transfections were conducted in triplicate, and the indicated values are averaged from at least three independent experiments.
RESULTS Two distinct regions of E2 bind TFIIB in vitro. Initially, we sought to determine whether human TFIIB can specifically bind BPV-1 E2 in vitro. GST:TFIIB was expressed in E. coli, purified with glutathione-agarose beads, and incubated with in vitro-translated [35S]methionine-labeled full-length E2. Plasmids encoding the E2 TAD (aa 1 to 216), the C-terminal repressor of E2 (aa 162 to 410), or luciferase were also translated and reacted with GST:TFIIB (Fig. 1A). Approximately 10% of the input 35S-labeled wild-type E2 (lane 2) bound to GST:TFIIB. No binding was observed with the luciferase-negative control (lane 1). GST:TFIIB complexed with both E2:1216 (lane 3) and E2:162-410 (lane 4) (6 and 3% of input, respectively). It has previously been reported that the C-terminal DBD spanning aa 310 to 410 bound TFIIB in vitro (34), and while this was likely to explain the interaction with E2: 162-410, association with the E2 TAD has not been previously
FIG. 1. E2 interacts with immobilized TFIIB in vitro. (A) GST:TFIIB immobilized on glutathione beads was incubated with in vitro-translated 35S-labeled proteins. Bound proteins were visualized and quantitated by a PhosphorImager. GST:TFIIB retained ;10% of full-length E2 (lane 2), 6% of E2:1-216 (lane 3), and 3% of E2:162-410 (lane 4). No binding was observed between luciferase and GST:TFIIB (lane 1). (B) Binding of six-histidine-tagged E2:1-216 to GST:TFIIB deletion mutants. Wild-type GST:TFIIB and deletion mutants were incubated with 0.5 ml of crude cell extract of six-histidine-tagged E2:1-216 expressed in E. coli. Bound proteins were analyzed by SDS-PAGE and visualized by Western blotting with anti-E2 monoclonal antibody B201. E2:1-216 bound to wild-type GST:TFIIB (lane 2), D202-269 (lane 7), and D238-316 (lane 8) but not to GST alone (lane 1), the negative control GST:SD21-HC8 subunit gene of human proteasome (lane 9), and four other GST:TFIIB deletion mutants: D4-85, D45-123, D118-174, and D178-210 (lanes 3 to 6, respectively).
demonstrated. In comparison to the N terminus of E2, GST: TFIIB weakly bound an in vitro-translated carboxy-terminal 125 aa which include the DBD (data not shown). GST:TFIIB also bound the full-length E2 proteins from human papillomavirus types 11, 16, and 18, consistent with this representing a conserved function (data not shown). To confirm the direct association of this E2 domain with human TFIIB, a six-histidine-tagged E2:1-216 was expressed in E. coli and binding experiments were performed with crude cell extract. GST:TFIIB bound this form of the E2 TAD after several washes with 0.5 M NaCl (Fig. 1B). This result implies that the interaction between the N-terminal E2 TAD and TFIIB is not mediated by eukaryotic factors present in reticulocyte lysate. To localize the domain(s) of TFIIB important for interaction with the N terminus of E2, several deletion mutants were tested (23). As shown in Fig. 1B, C-terminal deletions of TFIIB such as D202-269 and D238-316 retained the ability to bind E2:1-216. Binding was greatly reduced with the N-terminal deletion TFIIB mutants D4-85, D45-123, D118-174, and D178-201. A series of in-frame E2 deletions were synthesized as GST fusion proteins to define the limits of the TFIIB binding domain(s) within the TAD (Fig. 2A). These proteins were captured on glutathione beads and reacted with epitope-tagged (FLAG) TFIIB purified from E. coli. GST itself did not bind FLAG-TFIIB (Fig. 2B). GST:E2:1-91 (lane 3), 54-134 (lane 7), and 74-134 (lane 9) bound TFIIB, but GST:E2:1-54 (lane 2), 15-91 (lane 4), 15-134 (lane 5), 54-91 (lane 6), 74-91 (lane 8), and 91-134 (lane 10) did not. These results indicate that the central region of the E2 TAD, including aa 74 to 134, is necessary for TFIIB binding to the E2 TAD. The aa 54 to 134 showed more efficient association, but E2:15-134 did not interact with TFIIB. Similarly, E2:1-91 effectively bound TFIIB but 15-91 did not. The simplest interpretation of these results is that E2 proteins lacking the first 15 aa may be significantly misfolded and hence unable to bind. The functional importance of the first 15 aa has been demonstrated in our laboratory (8) and by others (32). TFIIB binding by GST:E2:1-91 and GST:E2:74-134 suggests that aa 74 to 91, shared by these two
VOL. 72, 1998
FIG. 2. The N-terminal activation domain of E2 interacts with TFIIB. (A) Coomassie blue-stained polyacrylamide gel of representative GST:E2 fusion proteins used in the experiments: GST (lane 1), GST:E2:74-134 (lane 2), GST: E2:54-134 (lane 3), GST:E2:1-91 (lane 4), and GST:E2:1-134 (lane 5). (B) Affinity chromatography with purified FLAG-TFIIB and GST:E2 directly demonstrates binding. Truncated forms of BPV-1 E2 were expressed and purified from E. coli (pLysS) as GST fusion proteins. FLAG-TFIIB was purified from E. coli (pLysS) with FLAG-tagged M2 beads eluted with FLAG peptides. Bound FLAG-TFIIB was detected with FLAG antibody and ECL. GST:E2:1-91 (lane 3), GST:E2:54-134 (lane 7), and GST:E2:74-134 (lane 9) bound FLAG-TFIIB, but GST (lane 1), GST:E2:1-54 (lane 2), GST:E2:15-91 (lane 4), GST:E2:15-134 (lane 5), GST:E2:54-91 (lane 6), GST:E2:74-91 (lane 8), and GST:E2:91-134 (lane 10) showed no interaction with TFIIB. (C) Specific binding of TFIIB to the E2 TAD. 35S-labeled TFIIB or luciferase was synthesized in reticulocyte lysates and incubated with GST (lanes 1 and 2), GST:E2:54-134 (lanes 3 and 4), GST: E2:113-286 (lanes 5 and 6), GST:E2:215-410 (lanes 7 and 8), or GST:E2:74-134 (lanes 9 and 10). GST:E2:54-134 retained 26% (lane 3) and GST:E2:74-134 retained 9% (lane 9) of input TFIIB. (D) Summary of the E2 regions tested for binding to FLAG-TFIIB.
E2 TAD BINDS TO TFIIB
1015
proteins, form the core TFIIB interaction region. To further delineate the TFIIB-binding peptide, GST:E2 fusions including aa 1 to 54, 54 to 91, 74 to 91, 74 to 113 (data not shown), 91 to 134, or 113 to 134 (data not shown) were tested and found to be unable to bind purified TFIIB (Fig. 2B). These data imply that E2 aa 74 to 91 are not sufficient for TFIIB association and that surrounding residues influence their activity. The additional amino acids outside of the core 74-to-91 region may contribute directly to TFIIB binding or may mediate proper folding or stabilization of GST:E2 proteins. In some experiments, we used reticulocyte lysate-translated TFIIB or luciferase as a negative control to confirm the specificity of TFIIB association with this region of E2 (Fig. 2C). The TFIIB protein bound to both GST:E2:54-134 and 74-134 (26 and 9% of input, respectively; lanes 3 and 9, respectively), but luciferase did not (lanes 4 and 10). GST:E2:113-286 (lanes 5 and 6) and GST:E2:215-410 (lanes 7 and 8) did not bind either TFIIB or luciferase. We were unable to reproducibly detect association of GST:E2:215-410 with in vitro-translated TFIIB (lane 7) or with purified FLAG-TFIIB. The results of all TFIIB binding experiments are summarized in Fig. 2D. E2 TAD mutants with reduced binding to TFIIB are functionally defective in vivo. We have previously characterized a large series of both transactivation-competent and -defective mutations throughout the E2 TAD (8). GST:E2 proteins carrying these mutations were tested for their ability to associate with TFIIB in vitro (Fig. 3A to D). Recombinant GST:E2 proteins were synthesized in E. coli as GST:E2:1-134 or GST: E2:1-286 fusions (Fig. 3A), purified, and immobilized on glutathione-Sepharose beads. In vitro-translated or purified FLAGTFIIB was used in the association reactions, and results are shown in Fig. 3B to D. Notably, GST:E2:1-134:W99C (Fig. 3B), having one of four transactivation-defective mutations within the first 134 aa tested, did not bind TFIIB. The other three mutants tested, W92R, F87S, and Q15H, interacted with TFIIB as efficiently as did wild-type GST:E2. Five other transactivation-defective E2 mutants, Q66R, S93P, E105G, P106S, and W130R (Fig. 3C and D), showed approximately wild-type levels of binding to TFIIB. One mutant that has ;50% of wild-type transcriptional activation, N127Y, displayed a comparable reduction in ability to bind in vitro-translated (Fig. 3C) or purified (Fig. 3D) FLAGTFIIB. We questioned whether overexpression of TFIIB would reverse the transcription defect of N127Y (Fig. 3E). Cotransfection of E2 N127Y with pCG:TFIIB resulted in recovery of E2-dependent transactivation to nearly wild-type levels. In contrast, the TFIIB binding-defective E2 mutant W99C remained inactive with overexpression of TFIIB. The inability of W99C to activate transcription even when cotransfected with TFIIB is not likely due to complete misfolding or low expression levels because it is functional in E1-dependent viral replication (21). As a control, we used S181F, a partially defective transactivation mutant that is outside the TFIIB association domain and is fully active for TFIIB binding (data not shown). Transactivation by E2 S181F was unchanged upon cotransfection with 2 mg of pCG:TFIIB. pCG:TFIIB alone showed no effect in the absence of any E2 (data not shown). Western blot analysis of cell lysates from these cotransfection experiments showed no differences in E2 protein levels in the presence or absence of TFIIB. All mutants were observed to be nuclear by immunofluorescence (data not shown). In vivo interaction of TFIIB with the N-terminal E2 TAD. Our studies demonstrated that aa 74 to 134 within the E2 TAD directly bind TFIIB, and the genetic studies described above suggested that this interaction is necessary for E2 activity. To further study the functional significance of interaction between
1016
YAO ET AL.
J. VIROL.
FIG. 4. TFIIB functionally interacts with the E2 TAD in vivo. (A) Transactivation by LexA E2 TAD fusions. C33A cells were transfected with pcDNA3: LexA:E2:1-133 (bar 3), E2:1-216 (bar 4), or E2:162-410 (bar 5) or transfected with control plasmid pcDNA3:LexA (bar 2) without E2 with 1 mg of reporter plasmid. Luciferase activity was normalized to pcDNA3 (bar 1). (B) Overexpression of TFIIB relieves squelching by the E2 TAD. Increasing amounts of pcDNA3:LexA:E2:1-216 (0.1, 1, and 2 mg) were transfected into C33A cells with 1 mg of the pDBL8 reporter plasmid, with or without 2 mg of pcDNA3:TFIIB. Vector DNA was added as necessary to standardize the total amount of expression plasmid in each reaction.
FIG. 3. E2 TAD mutants with reduced binding to TFIIB in vitro are also defective for TFIIB interaction in vivo. Recombinant GST:E2 fusion proteins were synthesized in E. coli, purified, and immobilized on glutathione-Sepharose beads. FLAG-TFIIB was used in panels B and D, and in vitro-translated TFIIB was used in panel C. ECL Western blotting with anti-FLAG antibody was used to detect captured FLAG-TFIIB as described in Materials and Methods. (A) Coomassie brilliant blue-stained GST fusion proteins in the GST:E2:1-286 form used in panels C and D: Q66R (lane 1), S93P (lane 2), E105G (lane 3), P106S (lane 4), N127Y (lane 5), W130R (lane 6), GST (lane 7), and wild type (lane 8). (B) FLAG-TFIIB binding to E2 mutants (in GST:E2:1-134 form) and other fragments: GST (lane 1), GST:E2:74-216 (lane 2), GST:E2:1-134 (lane 3), GST: E2:1-98 (lane 4), GST:E2:1-134W92R (lane 5), GST:E2:1-134F87S (lane 6), GST:E2:1-134Q15H (lane 7), GST:E2:1-134W99C (lane 8), GST:E2:215-286 (lane 9), GST:E2:215-410 (lane 10), and 10% input FLAG-TFIIB (lane 11). (C) In vitro-translated TFIIB binding to GST (lane 1), the negative control GST: SD21-HC8 subunit gene of human proteasome (lane 2), and the following point mutants in the GST:E2:1-286 form: Q66R (lane 3), S93P (lane 4), E105G (lane 5), P106S (lane 6), N127Y (lane 7), W130R (lane 8), and wild type (lane 9). (D) FLAG-TFIIB binding to GST:E2 mutants (GST [lane 1], GST:SD21 [lane 2], GST:E2:1-286 wild type [lane 3], and GST:E2:1-134 wild type [lane 4]) and GST:E2:1-286 form mutants (Q66R [lane 5], S93P [lane 6], E105G [lane 7], P106S [lane 8], N127Y [lane 9], W130R [lane 10], and 10% input FLAG-TFIIB [lane 11]). (E) TFIIB interacts with E2 in vivo. C33A cells were transfected with the luciferase reporter pJSLuc (three E2 binding sites) and increasing amounts of pCG:E2 wild type (wt), pCG:E2W99C, pCG:E2N127Y, and pCG:E2S181F, with or without 2 mg of pCG:TFIIB. The amount of DNA in each transfection was adjusted with the empty expression vector pCG to a total of 5 mg. The luciferase activities of reporter in the absence of E2 were normalized to 1.
TFIIB and the TAD, E2:1-133, 1-216, and 162-410 were placed in frame C-terminal to the LexA DBD (202-aa form). These fusions eliminate a potential contribution of the E2 C-terminal DBD and its interactions with TBP and TFIIB. These constructs were transfected along with a LexA-dependent reporter into C33A cells. LexA:E2:1-216 stimulated luciferase expression from the LexA reporter almost 13-fold, but LexA:202B, E2:1-133, and E2:162-410 reduced transcription by 40% in comparison with vector alone (Fig. 4A). These experiments confirm that the N-terminal 216 aa of E2 are necessary and sufficient for transcriptional activation. Furthermore, the Nterminal 133 aa which include the minimal E2 domain required for association with TFIIB (aa 74 to 134) were unable to stimulate transcription. This is consistent with the fact that aa 133 to 216 are also necessary for E2 function. Another means to test the interaction between the E2 TAD and TFIIB is based on the observation that high-level expression of wild-type E2 results in decreased promoter-specific activation. This is thought to result from squelching, in which excess E2 proteins not bound to DNA sequester factors required for transcriptional activation. It has been demonstrated elsewhere that overproduction of TFIIB reduced this self-inhibitory effect (34), and we observed analogous findings (Fig. 3E). To exclude the potential contribution of the TFIIB interaction with the E2 DBD, increasing quantities of LexA: E2:1-216 were transfected into C33A cells (Fig. 4B). Two micrograms of LexA:E2:1-216 showed less enhancement of transactivation than did 0.1 mg (3.5- versus 13-fold, respectively), reproducing this phenomenon. Cotransfection of TFIIB resulted in increased transcriptional activity at high levels of transfected LexA:E2:1-216. Totals of 0.5 and 1 mg of TFIIB also reversed squelching but to a lesser extent (data not shown). DISCUSSION The papillomavirus E2 proteins are composed of three domains (19, 25, 33, 38, 45). The N-terminal ;215 aa of BPV-1 E2 represent its TAD. This is followed by a nonconserved segment called the hinge. The C-terminal 125 aa form a
VOL. 72, 1998
dimeric b-barrel with two high-affinity DNA-binding alpha helices that recognize the sequence ACCGNNNNCGGT (2, 26). To discern how E2 engages the transcription machinery, we are attempting to catalog the cellular proteins which interact with the E2 TAD and DBD. These investigations led us to examine whether the general transcription factor TFIIB binds E2. TFIIB has been proposed to be one of the key targets for transcriptional activators. It was reported that the acidic type of activator binds TFIIB, disrupting N- and C-terminal intramolecular contacts and exposing binding sites for general transcription factors and RNA polymerase II (35). Activation domains have been shown to interact with multiple factors and probably enhance several steps in the assembly of functional preinitiation complexes. Activators like herpes simplex virus VP16 strongly stimulate both initiation and elongation of the nascent transcript by RNA polymerase II (7, 49). VP16 interacts with at least five different factors: TFIIB, TBP, dTAF40, ADA2, and the p62 subunit of TFIIH (5, 13, 20, 41, 48). We demonstrate physical and functional interactions between the E2 N-terminal TAD and TFIIB. Affinity chromatography with GST fusions to the E2 TAD retained purified FLAG-TFIIB, both produced in bacteria. Thus, this N-terminal E2 association with TFIIB occurs through direct proteinprotein interaction and does not require any other eukaryotic factors. In vitro binding assays with wild-type and mutant E2 proteins defined aa 74 to 134 as the core TFIIB binding region. A slightly larger segment (aa 54 to 134) binds TFIIB more efficiently, indicating that adjacent amino acids may contribute to their association. Smaller partially overlapping fragments of E2 such as aa 74 to 91, 74 to 113, 113 to 134, and 54 to 91 cannot complex with TFIIB, although others including aa 1 to 91 and 1 to 98 bound TFIIB. There are several interpretations for these data, such as interference from the GST moiety in close proximity to the core domain or misfolding of the short E2 moiety. The interaction of TFIIB with the E2 TAD may involve multiple contacts between amino acids in the 54-to-134 region, and loss of a subset with the small truncations may lead to reduced affinity and undetectable binding in vitro. None of the constructs that initiated at methionine codon 15 present in wild-type E2 bound TFIIB, even those incorporating the core TFIIB binding region (e.g., aa 15 to 91 or 15 to 134). This implies that the first 14 aa of E2TA may be critical for maintaining the active conformation of the E2 TAD. The importance of these 14 aa has been established elsewhere (8, 32). Our results indicate that the E2 TAD interacts with the N-terminal 200 aa of TFIIB. A Gal4-VP16 TAD fusion bound to the C terminus of TFIIB while the N-terminal region of TFIIB was not required (13, 36). One mechanism suggested is that activators interrupt TFIIB intramolecular interaction and expose binding sites for general transcription factors to enter the preinitiation complex through TFIIB (35), although the importance of VP16-TFIIB interaction has been questioned by others (22, 37). One group has reported that the vitamin D receptor interacted with the N-terminal domain of TFIIB, while apparently contradicting results localized vitamin D receptor binding to the C-terminal region of TFIIB (6, 31). In vitro binding and transient transfection assays demonstrated that the putative zinc finger structure in the N terminus of TFIIB was necessary for interaction with the GAL4-ftzQ (GAL4 DBD fused to the last 86 aa of the Drosophila homeodomain protein Fushi tarazu [14]). It seems that both the N and C termini of TFIIB may interact with an activator. Our findings confirm previous observations that the C-terminal DBD of E2 binds TFIIB (34). In the context of the GST:TFIIB fusion protein, we detected binding to truncated forms of the E2 DBD peptide. In our experiments, GST:E2:
E2 TAD BINDS TO TFIIB
1017
215-410 did not bind in vitro-translated TFIIB or purified FLAG-TFIIB. This may reflect the fact that the conditions we used were not optimum for these reactions. The biological relevance of the E2 DBD binding to TBP and TFIIB in vitro (34) has not been established. Another study found a functional interaction between TBP and a larger E2 region that included both the hinge and DBD, although physical interaction could not be established (39). The E2 TAD activates transcription when fused to GAL4 or LexA DBD in yeast and mammalian cells as shown before (3, 8, 46) or here (Fig. 4B). These observations suggest that the C-terminal interaction with TFIIB is dispensable for transcriptional activation in vivo. However, both N- and C-terminal E2 interactions with TFIIB could be important for activation of some promoters. Evidence for this possibility is the observation that BPV-1 E2 transcriptional repressor (aa 162 to 410) can partially activate the human papillomavirus type 18 early promoter (17). We used several in vivo assays of E2 function to underscore the significance of the interaction between the E2 N-terminal TAD and TFIIB. The E2 TAD has been reported to stimulate mammalian promoters in the absence of its DBD, presumably by activating or recruiting proteins that initiate the transcription process (25). We found that overexpression of TFIIB stimulated this generalized activation phenomenon, implying its participation in this DNA-binding-independent model (data not shown). Some studies suggested that the entry of TFIIB may be rate limiting for transcriptional initiation and that activators recruit or stabilize the interaction of TFIIB with the initiation complex (13, 24, 36). Others suggested that overexpression of TFIIB did not significantly affect promoter activity (14, 15, 44). Our experience indicated that transfection of TFIIB alone with an E2-dependent promoter into C33A cells had no significant effect on its expression. Upon cotransfection with low levels of E2, TFIIB had also had no detectable effect, suggesting that their interaction is not rate limiting under these conditions. However, expression of increasing amounts of E2 in mammalian cells leads to a peak of maximal promoter activation followed by a gradual decrease. This is believed to result from competition of E2 not bound to DNA with promoter-bound E2 for a rate-limiting factor, or squelching. While TBP and TFIIB were shown to bind the C-terminal DBD of E2 and their overexpression inhibited squelching, this in vivo assay was performed with the full-length E2 protein (34). In our experiments, TFIIB reversed the inhibitory effects of high-level expression of both E2TA and LexA:E2:1-216. This strongly suggests that TFIIB is being titrated by the E2 TAD. However, even at elevated concentrations of TFIIB in the transfected cells, transactivation did not achieve levels greater than the peak observed with lower-input E2. This implies that another cellular protein(s) is limiting in addition to TFIIB. The experiments presented here imply that TFIIB plays an important role in transcriptional activation mediated by the papillomavirus E2 protein. The single-amino-acid mutant E2 W99C lacks the ability to bind TFIIB in vitro (Fig. 3B) and correspondingly fails to activate transcription in vivo. Cotransfection of TFIIB with E2 W99C was unable to restore transcriptional activation. While this tryptophan may be essential for the proper conformation of the TAD and, more specifically, the TFIIB binding domain, E2 W99C retains the ability to bind E1 and shows synergistic stimulation of DNA replication, although not to wild-type levels. This finding implies that its activation domain is not uniformly denatured and suggests that TFIIB binding by the E2 TAD is not required for E2 activation of E1-dependent DNA replication. Notably, the mutant E2 N127Y displayed reduced affinity for TFIIB compared to wild type and other mutants. Overexpression of TFIIB led to
1018
YAO ET AL.
J. VIROL.
a dramatic gain of transactivation by the N127Y mutant to near-wild-type levels. Another E2 mutant, S181F, similarly has a partial transactivation defect but is outside the core TFIIB binding region and is fully competent for TFIIB binding. In contrast to N127Y, no effect was observed when S181F was cotransfected with TFIIB into C33A cells. These studies demonstrate that the reduced affinity of E2 N127Y for TFIIB can be functionally complemented by elevated TFIIB protein levels. The S181F mutant is likely to be defective for another interaction necessary for transactivation, and this cannot be replaced by overexpression of TFIIB. While we cannot formally exclude the possibility that N127Y affects interaction with another cellular factor, these studies suggest that E2 mediates at least some of its effects through TFIIB in vivo by recruiting and /or stabilizing TFIIB in the transcription initiation complex. Transcriptional activation is believed to include a complex and highly organized series of events, and we assume that the relatively large E2 TAD serves as a platform for multiple protein-protein interactions. At least three cellular proteins, Sp1, TBP, and TFIIB, have been reported to bind E2 (28, 34, 39). Using the yeast two-hybrid system, we have identified and characterized a novel human factor, AMF-1, that interacts with the E2 TAD (9). The TFIIB binding core domain is located between aa 74 and 134. AMF-1 interacts with the TAD region defined by residues 133 to 216 of E2. LexA:E2:1-133 contains the TFIIB binding domain and yet cannot activate transcription, suggesting that E2 must interact with other cellular factors, such as AMF-1, to activate transcription. In summary, we conclude that the papillomavirus E2 TAD binds specifically and directly to TFIIB and that this interaction is necessary for transcriptional activation. ACKNOWLEDGMENTS We thank C.-M. Chiang, R. G. Roeder, D. Reinberg, J. Strasswimmer, J. Chen, and N. E. Thompson for plasmids and antibody. We are also grateful to F. Sverdrup, J. Strasswimmer, and other E. Androphy laboratory members for helpful discussions. This work was supported in part by a Dermatology Foundation Research Fellowship to J.-M.Y. and grant RO1 CA58376 to E.J.A. REFERENCES 1. Abroi, A., R. Kurg, and M. Ustav. 1996. Transcriptional and replicational activation functions in the bovine papillomavirus type 1 E2 protein are encoded by different structural determinants. J. Virol. 70:6169–6179. 2. Androphy, E. J., D. Lowy, and J. Schiller. 1987. Bovine papillomavirus E2 trans-activating gene product binds to specific sites in papillomavirus DNA. Nature (London) 325:70–73. 3. Benson, J. D., and P. M. Howley. 1995. Amino-terminal domains of the bovine papillomavirus type 1 E1 and E2 proteins participate in complex formation. J. Virol. 69:4364–4372. 4. Berg, M., and A. Stenlund. 1997. Functional interactions between papillomavirus E1 and E2 proteins. J. Virol. 71:3853–3863. 5. Berger, S. L., B. Pina, N. Silverman, G. A. Marcus, J. Agapite, J. L. Regier, and S. J. Triezenberg. 1992. Genetic isolation of ADA2: a potential transcriptional adapter required for function of certain acidic activation domains. Cell 70:251–265. 6. Blanco, J. C. G., I.-M. Wang, S. Y. Tsai, M.-J. Tsai, B. W. O’Malley, P. W. Jurutka, M. R. Haussler, and K. Ozato. 1995. Transcription factor TFIIB and the vitamin D receptor cooperatively activate ligand-dependent transcription. Proc. Natl. Acad. Sci. USA 92:1535–1539. 7. Blau, J., H. Xiao, S. McCracken, P. O’Hare, J. Greenblatt, and D. Bentley. 1996. Three functional classes of transcriptional activation domains. Mol. Cell. Biol. 16:2044–2055. 8. Breiding, D. E., M. J. Grossel, and E. J. Androphy. 1996. Genetic analysis of the bovine papillomavirus E2 transcriptional activation domain. Virology 221:34–43. 9. Breiding, D. E., F. Sverdrup, M. J. Grossel, N. Moscufo, W. Boonchai, and E. J. Androphy. 1997. Functional interaction of a novel cellular protein with the papillomavirus E2 transactivation domain. Mol. Cell. Biol. 17:7208–7219. 10. Chen, C., and H. Okayama. 1987. High-efficiency transformation of mammalian cells by plasmid DNA. Mol. Cell. Biol. 7:2745–2752.
11. Chen, J. Personal communication. 12. Chiang, C.-M., and R. G. Roeder. 1993. Expression and purification of general transcription factors by FLAG epitope-tagging and peptide elution. Pep. Res. 6:62–64. 13. Choy, B., and M. R. Green. 1993. Eukaryotic activators function during multiple steps of preinitiation complex assembly. Nature (London) 366:531– 536. 14. Colgan, J., H. Ashali, and J. L. Manley. 1995. A direct interaction between a glutamine-rich activator and the N terminus of TFIIB can mediate transcriptional activation in vivo. Mol. Cell. Biol. 15:2311–2320. 15. Colgan, J., S. Wampler, and J. L. Manley. 1993. Interaction between a transcriptional activator and transcription factor IIB in vivo. Nature (London) 362:549–553. 16. Dalton, S., and R. Treisman. 1992. Characterization of SAP-1, a protein recruited by serum response factor to the c-fos serum response element. Cell 68:597–612. 17. Demeret, C., M. Yaniv, and F. Thierry. 1994. The E2 transcriptional repressor can compensate for Sp1 activation of the human papillomavirus type 18 early promoter. J. Virol. 68:7075–7082. 18. Ferguson, M. K., and M. R. Botchan. 1996. Genetic analysis of the activation domain of bovine papillomavirus protein E2: its role in transcription and replication. J. Virol. 70:4193–4199. 19. Giri, I., and M. Yaniv. 1988. Structural and mutational analysis of E2 transactivating proteins of papillomaviruses reveals three distinct functional domains. EMBO J. 7:2823–2829. 20. Goodrich, J. A., T. Hoey, C. J. Thut, A. Admon, and R. Tjian. 1993. Drosophila dTAFII40 interacts with both a VP16 activation domain and the basal transcription factor TFIIB. Cell 75:519–530. 21. Grossel, M. J., F. Sverdrup, D. E. Breiding, and E. J. Androphy. 1996. Transcriptional activation function is not required for stimulation of DNA replication by bovine papillomavirus type 1 E2. J. Virol. 70:7264–7269. 22. Gupta, R., A. Emili, G. Pan, H. Xiao, M. Shales, J. Greenblatt, and C. J. Ingles. 1996. Characterization of the interaction between the acidic activation domain of the VP16 and the RNA polymerase II initiation factor TFIIB. Nucleic Acids Res. 24:2324–2330. 23. Ha, I., S. Roberts, E. Maldonado, X. Sun, L. Kim, M. Green, and D. Reinberg. 1993. Multiple functional domains of human transcription factor IIB: distinct interactions with two general transcription factors and RNA polymerase II. Genes Dev. 7:1021–1032. 24. Hadzic, E., V. Desai-Yajnik, E. Helmer, S. Guo, S. Wu, N. Koudinova, J. Casanova, B. M. Raaka, and H. H. Samuels. 1995. A 10-amino-acid sequence in the N-terminal A/B domain of thyroid hormone receptor a is essential for transcriptional activation and interaction with the general transcription factor TFIIB. Mol. Cell. Biol. 15:4507–4517. 25. Haugen, T. H., L. P. Turek, F. M. Mercurio, T. P. Cripe, B. J. Olson, R. D. Anderson, D. Seidl, M. Karin, and J. Schiller. 1988. Sequence-specific and general transcriptional activation by the bovine papillomavirus-1 E2 transactivator require an N-terminal amphipathic helix-containing E2 domain. EMBO J. 7:4245–4253. 26. Hegde, R. S., S. R. Grossman, L. A. Laimins, and P. B. Sigler. 1992. Crystal structure at 1.7 Å of the bovine papillomavirus-1 E2 DNA-binding domain bound to its DNA target. Nature (London) 359:505–512. 27. Johnston, S. D., X.-M. Yu, and J. E. Mertz. 1996. The major transcriptional transactivation domain of simian virus 40 large T antigen associates nonconcurrently with multiple components of the transcriptional preinitiation complex. J. Virol. 70:1191–1202. 28. Li, R., J. D. Knight, S. P. Jackson, R. Tjian, and M. R. Botchan. 1991. Direct interaction between Sp1 and the BPV enhancer E2 protein mediates synergistic activation of transcription. Cell 65:493–505. 29. Lin, Y.-S., I. Ha, E. Maldonado, D. Reinberg, and M. R. Green. 1991. Binding of general transcription factor TFIIB to an acidic activating region. Nature (London) 353:569–571. 30. Manet, E., C. Allera, H. Gruffat, I. Mikaelian, A. Rigolet, and A. Sergeant. 1993. The acidic activation domain of the Epstein-Barr virus transcription factor R interacts in vitro with both TBP and TFIIB and is cell-specifically potentiated by a proline-rich region. Gene Expr. 3:49–59. 31. Masuyama, H., S. C. J. Jefcoat, and P. N. MacDonald. 1997. The N-terminal domain of transcription factor IIB is required for direct interaction with the vitamin D receptor and participates in vitamin D-mediated transcription. Mol. Endocrinol. 11:218–228. 32. McBride, A. A., J. C. Byrne, and P. M. Howley. 1989. E2 polypeptides encoded by bovine papillomavirus type 1 form dimers through the common carboxyl-terminal domain: transactivation is mediated by the conserved amino-terminal domain. Proc. Natl. Acad. Sci. USA 86:510–514. 33. Prakash, S. S., S. Grossman, R. Pepinsky, L. Laimins, and E. Androphy. 1992. Amino acids necessary for DNA contact and dimerization imply novel motifs in the papillomavirus E2 trans-activator. Genes Dev. 6:105–116. 34. Rank, N. M., and P. F. Lambert. 1995. Bovine papillomavirus type 1 E2 transcriptional regulators directly bind two cellular transcription factors, TFIID and TFIIB. J. Virol. 69:6323–6334. 35. Roberts, S. G. E., and M. R. Green. 1994. Activator-induced conformational
VOL. 72, 1998 change in general transcription factor TFIIB. Nature (London) 371:717–720. 36. Roberts, S. G. E., I. Ha, E. Maldonado, D. Reinberg, and M. R. Green. 1993. Interaction between an acidic activator and transcription factor TFIIB is required for transcriptional activation. Nature (London) 363:741–744. 37. Shen, F., S. J. Triezenberg, P. Hensley, D. Porter, and J. R. Knutson. 1996. Transcriptional activation domain of the herpesvirus protein VP16 becomes conformational constrained upon interaction with basal transcription factors. J. Biol. Chem. 271:4827–4837. 38. Skiadopoulos, M. H., and A. A. McBride. 1996. The bovine papillomavirus type 1 E2 transactivator and repressor proteins use different nuclear localization signals. J. Virol. 70:1117–1124. 39. Steger, G., J. Ham, O. Lefebvre, and M. Yaniv. 1995. The bovine papillomavirus 1 E2 protein contains two activation domains: one that interacts with TBP and another that functions after TBP binding. EMBO J. 14:329–340. 40. Strasswimmer, J. Personal communication. 41. Stringer, K. F., C. J. Ingles, and J. Greenblatt. 1990. An acidic transcriptional activation domain binds directly and selectively to the TATA-box factor. Nature (London) 345:783–786. 42. Sun, X., D. Ma, M. Sheldom, K. Yeung, and D. Reinberg. 1994. Reconstitution of human TFIIA activity from recombinant polypeptides: a role in TFIID-mediated transcription. Genes Dev. 8:2336–2348. 43. Thompson, N. E., L. A. Strasheim, K. M. Nolan, and R. R. Burgess. 1995. Accessibility of epitopes on human transcription factor IIB in the native
E2 TAD BINDS TO TFIIB
1019
protein and in a complex with DNA. J. Biol. Chem. 270:4735–4740. 44. Um, M., C. Li, and J. L. Manley. 1995. The transcriptional repressor evenskipped interacts directly with TATA-binding protein. Mol. Cell. Biol. 15: 5007–5016. 45. Winokur, P., and A. McBride. 1992. Separation of the transcriptional activation and replication functions of the bovine papillomavirus-1 E2-protein. EMBO J. 11:4111–4118. 46. Winokur, P. L., and A. A. McBride. 1996. The transactivation and DNA binding domains of the BPV-1 E2 protein have different roles in cooperative origin binding with the E1 protein. Virology 221:44–53. 47. Xiao, H., J. D. Friesen, and J. T. Lis. 1994. A highly conserved domain of RNA polymerase II shares a functional element with acidic activation domains of upstream transcription factors. Mol. Cell. Biol. 14:7507–7516. 48. Xiao, H., A. Pearson, B. Coulombe, R. Truant, S. Zhang, J. L. Regier, S. J. Triezenberg, D. Reinberg, O. Flores, C. J. Ingles, and J. Greenblatt. 1994. Binding of basal transcription factor TFIIH to the acidic activation domains of VP16 and p53. Mol. Cell. Biol. 14:7013–7024. 49. Yankulov, K., J. Blau, T. Purton, S. Roberts, and D. Bentley. 1994. Transcriptional elongation by RNA polymerase II is stimulated by transactivators. Cell 77:749–759. 50. Yasugi, T., J. D. Benson, H. Sakai, M. Vidal, and P. M. Howley. 1997. Mapping and characterization of the interaction domains of human papillomavirus type E1 and E2 proteins. J. Virol. 71:891–899.
