The coactivator CBP stimulates human T-cell lymphotrophic virus type I Tax transactivation in vitro.

Tax interacts with the cellular cyclic AMP-responsive element binding protein (CREB) and facilitates the binding of the coactivator CREB binding protein (CBP), forming a multimeric complex on the cyclic AMP-responsive element (CRE)-like sites in the human T-cell lymphotrophic virus type I (HTLV-I) promoter. The trimeric complex is believed to recruit additional regulatory proteins to the HTLV-I long terminal repeat, but there has been no direct evidence that CBP is required for Tax-mediated transactivation. We present evidence that Tax and CBP activate transcription from the HTLV-I 21 base pair repeats on naked DNA templates. Transcriptional activation of the HTLV-I sequences required both Tax and CBP and could be mediated by either the N-terminal activation domain of CBP or the full-length protein. Fluorescence polarization binding assays indicated that CBP does not markedly enhance the affinity of Tax for the trimeric complex. Transcription analyses suggest that CBP activates Tax-dependent transcription by promoting transcriptional initiation and reinitiation. The ability of CBP to activate the HTLV-I promoter does not involve the stabilization of Tax binding, but rather depends upon gene activation properties of the co-activator that function in the context of a naked DNA template.

quences in the HTLV-I LTR known as the Tax responsive elements or 21-bp repeats (4 -10). Tax increases the DNAbinding activity of the CREB transcription factor, associating with CREB to form a ternary complex on the DNA (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). It has been proposed that the cooperative binding of CREB and Tax to the promoter positions the latter protein to interact with basal transcription factors. In support of this model, Tax has been shown to interact with the basal transcription factors TFIIA and TFIID (22)(23)(24). The role of CREB in this model is to allow the binding of Tax to DNA rather than to provide a direct transcriptional activation function because nonphosphorylated (inactive) CREB can substitute for the phosphorylated (active) form of the transcription factor. In contrast, the activation of cellular promoters is believed to require phosphorylated CREB, and CREB mutants that cannot be phosphorylated function as dominant negatives.
Recent studies have suggested that the transcriptional coactivator CBP also contributes to Tax-mediated activation of the HTLV-I LTR (25)(26)(27)(28). CBP was first identified as a component of the CREB activation pathway (29,30) and was shown to interact specifically with the phosphorylated form of CREB. Subsequently, CBP was shown to interact with Tax, leading to the hypothesis that Tax, CREB, and CBP form a trimeric complex on the CRE-like sequences in the HTLV-I LTR (25)(26)(27)(28). Interestingly, in the presence of Tax, CBP binding and transactivation of the viral LTR does not require CREB to be phosphorylated (25,31). Presumably, CBP is recruited through Tax and not through the activation domain of CREB.
The presence of CBP in the HTLV-I transcription complex provides several additional possible mechanisms for gene activation. CBP has been reported to interact with the basal transcription factors TFIIB and TBP (32)(33)(34). Moreover, the Nterminal half of CBP, which contains the TBP binding site, has been shown to be important for CBP-mediated transcription (32). It has also been reported that CBP interacts with the RNA polymerase holoenzyme (holo pol II) (35) and several histone acetyltransferases, in addition to containing an intrinsic histone acetyltransferase domain (36,37). The fact that mutations in the acetyltransferase domain block transcriptional activation in vivo suggests that this enzymatic activity is important for CBP function on some promoters (38). Given the atypical mode of CBP recruitment utilized by the HTLV-I LTR, it is unknown which, if any, of these functions contribute to HTLV-I gene activation. It is possible, for example, that CBP could serve to stabilize Tax binding to the HTLV-I promoter and that the transcriptional activation properties of Tax, rather than CBP, are responsible for gene induction. Given the importance of CBP in CREB, NF-B, AP1, Stat-1/-2, c-Fos, MyoD, MAPK, and nuclear hormone receptor signaling, elucidating the mechanism of action of this coactivator has important implications for gene regulation (29, 30, 39 -59).
In this report, we utilized in vitro transcription assays to determine the requirement for CBP in Tax-mediated activation of the HTLV-I LTR. We found that CBP activates both transcriptional initiation and reinitiation. This ability of CBP to activate transcription does not occur simply through the stabilization of Tax binding, but rather through activation properties of CBP itself. These studies also demonstrate that the ability of CBP to activate transcription does not depend entirely on its intrinsic or associated histone acetyltransferase functions.

EXPERIMENTAL PROCEDURES
In Vitro Transcription Assays-The G-free DNA templates used in the in vitro transcription assays were pLovTATA and pTRE-1 Id (23) and 4 TxRE (a generous gift of Dr. Mark Anderson, Medical College of Georgia). Escherichia coli-expressed Tax protein was purified by ammonium sulfate precipitation as described (60). For the in vitro transcription reactions, preincubation was at 30°C for 30 or 45 min, followed by the addition of 2 l of [␣-32 P]UTP (Amersham Pharmacia Biotech, 400 Ci/mmol), and incubation at 30°C for 60 min. In experiments containing Sarkosyl, 0.02-0.1% Sarkosyl was added with the [ 32 P]UTP after preincubation. Reactions contained HeLa whole cell extract (25 l; 12.5 g/l), 1.0 -1.5 g of supercoiled DNA, 100 ng of Tax protein, 0.75-2.25 g of CBP-(1-682), or 2.5 g of full-length CBP in a total volume of 50 -65 l. Transcription buffer (30.5 l/reaction) contained 3 l of 20% PEG (6000), 3 l of 50 mM MgCl 2 , 3 l of 1 mM dithiothreitol, 1 l of 0.2 M creatine phosphate (Boehringer Mannheim), 1.5 l of 50 mM ATP/CTP, 1 l of 20 mM 3Ј-O-methylguanosine 5Јtriphosphate (Amersham Pharmacia Biotech), 20 units of RNase T1 (100 units/l, Boehringer Mannheim), and 18 l of Buffer D containing a final concentration of 20 mM HEPES (pH 7.9), 100 mM KCl, 12.5 mM MgCl 2 , 0.1 mM EDTA, 17% glycerol, and 1 mM dithiothreitol. For pulsechase assays, Tax, CBP-(1-682), and whole cell extract were incubated in the presence of [ 32 P]UTP for 15 min. A 10-fold molar excess of cold UTP was added to the reaction, and the polymerase complexes were allowed to elongate for 15-60 min. 3Ј-O-methyl-GTP was omitted in these assays to allow pol II elongation. Sarkosyl (0.03%) was added to inhibit reinitiation complexes in duplicates of the 15-and 60-min chase samples.
Protein Electroporation Assays-Wild-type and Tax mutant M47 protein were electroporated with the HTLV-I LTR-CAT or HIV-LTR-CAT plasmid into Jurkat cells as described previously (61).
Expression and Purification of Full-length CBP from SF9 Cells-Two copies of the FLAG epitope were added to the C terminus of full-length CBP subcloned into pFAST Bac1 (Life Technologies, Inc.). The pFAST-Bac-CBP-2x FLAG vector was then transformed into DH10Bac competent cells. The recombinant bacmid DNA was prepared from the positive colonies and transfected into SF9 cells. The expression of fulllength CBP in SF9 cells was confirmed by immunoprecipitation using anti-CBP and anti-FLAG antibodies. The FLAG-tagged CBP was purified on a FLAG affinity column (Kodak) and eluted with FLAG peptide. The mobility of the FLAG-tagged CBP determined on a 6% polyacrylamide electrophoresis gel was identical to that of CBP isolated from HeLa cells.
Fluorescence Polarization Assays-Association of Tax with the CREB⅐TxRE complex was measured by fluorescence polarization as described previously (25,27). Binding reactions (1 ml) contained 25 mM Tris-HCL (pH 7.5), 50 mM NaCl, 5 mM MgCl 2 , 0.5 mM EDTA, 1 mM dithiothreitol, 5% glycerol, 5 g/ml poly(dI:dC), 50 g/ml bovine serum albumin. Fluorescence anisotropy measurements utilized a fluoresceintagged duplex oligonucleotide derived from the promoter proximal Taxresponsive sequence (F-TxRE) (25). Binding reactions containing 5 nM F-TxRE and 30 nM CREB341 were titrated with increasing concentrations of purified Tax in the absence or presence of either 2.1 M CBP-(451-682) or 2 M CBP-(1-682). The K d for the association of CBP with Tax under these conditions is approximately 90 nM (25,27), thus all Tax added to the binding reactions is effectively bound to CBP prior to association with the CREB⅐TxRE complex. Binding affinity parameters were determined by nonlinear regression (SigmaPlot) as described previously (25,27). Fig. 1A shows a schematic of the HTLV-I promoter and enhancer (HTLV-I). The three Tax-responsive 21-bp elements (TRE-1) are indicated by the hatched boxes. For the in vitro transcription assays, three templates, each containing a G-free cassette, were utilized: 4TxRE contains four copies of the Tax-responsive 21-bp repeat I (TRE-I) inserted upstream of the HTLV-I TATA box and promoter (Ϫ52 to Ϫ1), pTRE-1 Id contains two copies of the 21-bp repeat upstream from the chicken ovalbumin TATA box, and pLovTATA (23), which also contains the chicken ovalbumin promoter, lacks the TRE-1 sequences. Our initial transcription studies utilized a purified fragment of CBP, CBP-(1-682) (51), which can activate transcription when fused to a Gal 4 DNA binding domain (32). This fragment has protein interaction domains for TBP, CREB, and Tax (25, 30, 32) ( Fig. 2A) and is sufficient for transcriptional activation in vivo and formation of the Tax⅐CREB⅐CBP complex (25,32). Tax protein was purified as described previously (60). Addition of purified Tax and a 10-to 20-fold molar excess of purified CBP-(1-682), compared with endogenous CBP, to the in vitro transcription reaction significantly induced transcrip-  (Fig. 1B, lanes 1-4), demonstrating that transcriptional activation depends upon the TRE-1 sequences.
As additional controls to demonstrate the specificity of the in vitro transcription assays, antibody blocking experiments were performed. Addition of either CBP or Tax antibodies, but not control antibodies, inhibited the in vitro transcription (data not shown).
Tax Mutant M47 Fails to Stimulate Transcription in the Presence of CBP-(1-682)-To further demonstrate the specificity of the in vitro transcription assays, we tested the Tax mutant M47 (62) (Leu-319 and Leu-320 converted to Arg and Ser), which does not activate CREB-dependent promoters in vivo. The addition of wild-type Tax and CBP-(1-682) significantly increased transcription from the template containing two copies of the 21-bp repeats (Fig. 3A, lanes 1-4). In contrast, the M47 Tax protein failed to activate transcription in the presence of CBP-(1-682) (Fig. 3A, lanes 5-7).
It could be argued that the M47 protein becomes inactivated during purification. To demonstrate the biological activity of the Tax M47 mutant protein, we electroporated the wild-type and mutant Tax protein along with a CREB-dependent or NF-B-dependent reporter plasmid. This assay takes advantage of the fact that M47 Tax has a mutant phenotype for CREB-dependent transcription, but retains a wild-type activity for transcriptional activation through the NF-B element (62). Equal amounts of the wild-type and M47 mutant protein were electroporated with either pU3R-CAT (HTLV-I LTR; CREBresponsive) or LTR-CAT (HIV LTR; NF-B-responsive) (61). The wild-type Tax protein, which can activate both the CREB and NF-B pathways, transactivated both the HTLV-I and HIV LTR (Fig. 3B, lanes 2 and 5). In contrast, the M47 mutant protein only minimally transactivated the CREB-responsive HTLV-I LTR, but transactivated the NF-B-responsive HIV LTR normally (Fig. 3B, lanes 3 and 6). These experiments demonstrate that the M47 protein was biologically active.
The Combination of Tax and CBP-(1-682) Facilitates Transcription Initiation and Reinitiation-To distinguish the effects of CBP on transcriptional initiation and reinitiation, we performed in vitro transcription assays in the presence of low concentrations of the detergent Sarkosyl, which inhibits reinitiation and limits transcription to a single round (63,64). Polymerase II preinitiation complexes were formed on the 4 TxRE template in the presence or absence of Tax and CBP-(1-682), and initiation was allowed to proceed before Sarkosyl was added to prevent transcription reinitiation. The addition of 0.1% Sarkosyl to the reactions inhibited transcription (Fig. 4A,  lanes 7 and 8), indicating that a relatively small number of active templates are generated by the addition of Tax and CBP (1-682) in the initial round of transcription. PhosphorImager TM quantitation of gel indicates that each initiation template supports an average of 25-50 rounds of transcription.
To visualize transcripts arising from reinitiation, 3Ј-O-methyl-GTP was added to the in vitro transcription assays. As described by Szentirmay and Sawadogo (65), in the presence of 3Ј-O-methyl-GTP, the first round of pol II elongation complexes remain at the end of a G-free region, blocking the elongation of pol II complexes that result from reinitiation (Fig. 4B). As a result, successive polymerases stack up from the end of the FIG. 4. CBP-(1-682) and Tax stimulate transcriptional initiation and reinitiation. A, purified Tax (100 ng) and CBP-(1-682) (750 ng) were added to the transcription reactions containing 1.5 g of pTRE-1 Id template. Following preincubation for 30 min at 30°C, 0.1% Sarkosyl and [ 32 P]UTP were added to the transcription reaction. Samples were then incubated for an additional 60 min, and 32 P-RNA was purified and analyzed on a denaturing acrylamide urea gel. CBP-(1-682) and Tax were added to reactions as indicated. Lane M, molecular weight marker. B, schematic for colliding polymerase reinitiation assay (65). In the presence of 3Ј-O-methyl-GTP, the first round of pol II elongation complexes remain at the end of a G-free region, blocking the elongation of pol II complexes that result from reinitiation. This results in successive polymerases stacking up from the end of the cassette, thus producing shorter transcripts representing rounds of reinitiation. C, Tax and CBP-(1-682) stimulate synthesis of full-length (360 base) as well as discrete reinitiation transcripts. In vitro transcription reactions were set up in the presence of 1.5 g of pTRE-1 Id and 3Ј-O-methyl-GTP. CBP-(1-682) (750 ng) and/or Tax (100 ng) were added to the reactions as indicated. 32 3-6, respectively). Sarkosyl (0.03%) was added to inhibit reinitiation complexes in duplicates of the 15-and 60-min chase samples (lanes 7 and 8). 3Ј-O-methyl-GTP was omitted in these assays to allow pol II elongation. 32 P-RNA was purified and separated on a denaturing acrylamide urea gel. The position of the 360-base transcript is indicated by the arrow. cassette, producing shorter transcripts that represent rounds of reinitiation. Fig. 4C shows that addition of both Tax and CBP-(1-682) stimulate the appearance of the full-length 360base pTRE-1 Id transcript as well as discrete shorter transcript species of approximately 265, 230, 195, 165, 150, and 130 bases in the presence of 3Ј-O-methyl-GTP (Fig. 4C, lane 4). Hybridization analysis of the full-length and shorter RNAs confirmed that they originated from the G-free cassette (data not shown). These data suggest that Tax and CBP-(1-682) stimulate both transcription initiation and reinitiation.
To confirm that the shorter RNAs represented reinitiated complexes, pTRE-1 Id template DNA, whole cell extract, CBP-(1-682), and Tax were preincubated for 30 min, followed by the addition of Sarkosyl and [ 32 P]UTP. Consistent with the results of Szentirmay and Sawadogo (65), the Tax-CBP induced reinitiation complexes were preferentially sensitive to Sarkosyl (Fig. 4D, lanes 2-4). Addition of 0.02-0.04% Sarkosyl inhibited the reinitiated pol II complexes but failed to inhibit synthesis of the full-length RNAs. The fact that synthesis of the full-length RNA was largely inhibited by 0.1% Sarkosyl (Fig. 4A) suggests that some of the complexes responsible for these transcripts may also be because of reinitiation.
Initiation and reinitiation complexes were also analyzed by a pulse-chase experiment. In vitro transcription reactions containing Tax and CBP-(1-682) were incubated in the presence of [ 32 P]UTP to label nascent RNAs formed from the pTRE-1 Id template (Fig. 4E). Subsequently, a 10-fold molar excess of cold UTP was added to the reaction, and the transcription complexes were allowed to elongate. 3Ј-O-methyl-GTP was omitted in these assays to allow pol II elongation. Analysis of the products demonstrated a gradual increase in the level of fulllength transcripts at 15, 30, 45, and 60 min of chase (Fig. 4E,  lanes 2-6). Importantly, a duplicate of the 60-min chase sample, in which 0.03% Sarkosyl was added to inhibit reinitiation complexes, failed to show an increase in the level of full-length RNA (Fig. 4E, compare lane 7 with lane 8). This result supports the hypothesis that Tax and CBP-(1-682) increase the level of transcriptional reinitiation.
Full-length CBP Stimulates Initiation and Reinitiation-It was important to demonstrate that full-length CBP behaved similarly to CBP-(1-682) in the in vitro transcription system. A full-length CBP molecule tagged at the C terminus with two FLAG epitopes was expressed in SF9 cells and purified on a FLAG affinity column. Western blot analysis revealed a single band of approximately 270 kDa (Fig. 5A, lane 2). The activity of the full-length CBP was tested in the in vitro transcription assay using the 4TxRE template. Consistent with the results presented above, Tax and CBP-(1-682) induced transcription from the Tax-responsive promoter (Fig. 5B, lane 2). Tax alone or CBP-(1-682) alone failed to activate transcription (data not shown). Remarkably, the addition of Tax and an equivalent amount of purified full-length CBP resulted in a dramatic increase in the level of transcription from initiation (360 base) as well as reinitiation complexes (130, 150, and 165 base) (Fig.  5B, lane 3). By PhosphorImager analysis, we calculate that the stimulation with Tax and CBP-(1-682) was approximately 8-fold, whereas the stimulation with Tax and an equivalent amount of full-length CBP was approximately 30-fold. Importantly, the addition of full-length CBP alone failed to activate transcription (data not shown). These results provide evidence that full-length CBP also stimulates transcriptional initiation and reinitiation.
CBP Does Not Stabilize the Association of the CREB-Tax Complex with DNA-Previous studies by Lenzmeier et al. (28) suggest that CBP association with Tax may stabilize the association with the TxRE, assayed by enhancement of the Tax-CREB footprint on the HTLV-I LTR. Thus, we asked whether CBP enhanced Tax-dependent transcriptional activation by enhancing formation of a multiprotein complex at the TxRE. For these experiments, we utilized a solution-binding assay measuring fluorescence polarization of an F-TxRE to monitor assembly of a multiprotein-DNA complex composed of CREB and Tax, with or without a CBP fragment containing the Taxinteraction domain CBP-(1-682) (25,27). In the absence of Tax, CREB binds poorly to the TxRE, with a K d Ͼ 100 nM (25). Thus, the TxRE is not significantly bound at a CREB concentration of 30 nM unless Tax is present. The increase in fluorescence polarization of the CREB⅐F-TxRE solution seen with the addition of Tax may thus be used to quantitate the amount of Tax incorporated into a complex containing CREB and DNA. In the presence of CBP-(1-682), the half-maximal effective concentration of Tax for promoting assembly of a Tax⅐CREB⅐TxRE complex is 19 nM, compared with 5 nM in the absence of CBP (Fig. 6). These values are not significantly different and indicate that CBP does not enhance transcriptional activation by increasing the affinity of Tax for the Tax⅐CREB⅐TxRE complex. DISCUSSION The Tax protein of HTLV-I utilizes a novel mechanism for activating gene expression, facilitating the formation of a multimeric CBP⅐Tax⅐CREB complex on the Tax-responsive DNA element. Tax facilitates the binding of the cellular transcription factor CREB to the atypical CRE-like sequences within the viral LTR. Recent data indicate Tax and CREB both contact DNA, Tax interacting with GC sequences in the minor groove, and CREB associating with the CRE-like sequences in the major groove (13,28). In this configuration, only the B/ZIP domain of CREB is required for formation of the ternary Tax⅐CREB⅐DNA complex. Consequently, nonphosphorylated CREB, as well as the negative modulator CREM␣, are competent to mediate Tax binding. This model differs significantly from an earlier one which suggested that Tax functioned primarily by promoting CREB dimerization (15). The exact role of CBP in the process of Tax-mediated gene activation has been puzzling. Despite the fact that CBP was originally identified through its ability to interact specifically with phosphorylated CREB, the CREB activation domain appears to be dispensable for Tax-mediated induction of the HTLV-I LTR. The synergistic interaction of Tax and CREB is believed to recruit the co-activator CBP in a manner that does not require CREB phosphorylation probably because, in this context, CBP interacts with Tax rather than CREB. The mechanism of gene activation could depend upon Tax, which has been shown to interact with TFIIA and TFIID (22)(23)(24), or CBP, which interacts with TBP, TFIIB, RNA polymerase holoenzyme, and histone acetyltransferases (32,(35)(36)(37)66), or both.
Our anisotropy studies show that CBP-(1-682) probably does not function simply by stabilizing Tax binding to the TRE sequences. Rather, a more direct role of CBP in gene activation is likely. While the histone acetyltransferase function of CBP and associated proteins could participate in the activation of the HTLV-I LTR in vivo, our studies show that CBP can also strongly activate transcription in the absence of chromatin. Our studies demonstrate for the first time that the N-terminal activation domain of CBP, in the presence of Tax, stimulates transcription of a Tax-responsive transcriptional template. Deletion of the TBP binding domain (32) abolishes the activity of CBP- (1-682) in vitro, suggesting that transcriptional activation is due, in part, to interaction with the basal transcription factors.
Our results do not imply that CBP/p300 histone acetyltransferase (HAT) activity is not important for the transcription of some genes (36,37). For example, steroid receptor induction of gene transcription has been proposed to involve two independent functions of coactivators, chromatin remodeling and enhanced stabilization of the preinitiation complex (67). Moreover, Kraus and Kadonaga (68) have reported that the CBP homologue p300 mediates estrogen receptor-activated gene induction in vitro only in the context of a chromatin template. Acetylated histones are a characteristic feature of transcriptionally active chromatin. Because CBP acetylates histones (HAT activity), it is likely that the coactivator increases transcription of some genes through remodeling of chromatin structure. In addition, the acetyltransferase domain may contribute to transcriptional activation through other mechanisms. For example, Gu and Roeder (69) have demonstrated that CBP increases p53 DNA binding through acetylation of the p53 C-terminal regulatory domain. In addition, Imhof et al. (70) recently demonstrated that p300 acetylates TFIIE p56 and TFIIF (FAT activity). The effect of acetylation on the function of these basal transcription factors remains to be determined.
Our data further suggest that Tax and CBP-(1-682) facili- tate both transcriptional initiation and reinitiation. The CREB-Tax-CBP complex may facilitate the recruitment of basal transcription factors into the initiation complex and, subsequently, stabilize the promoter complex from dissociation, thereby facilitating entry of pol II into the reinitiation complex. Zawel et al. (71) have analyzed the recycling of general transcription factors during RNA polymerase II transcription. All of the basal factors are found in the mature initiation complex, but following nucleotide addition, the complex becomes disrupted. TFIID remains bound to the promoter, and following release, TFIIB reassociates with TFIID (DB complex), forming the RNA pol II docking site for reinitiation. In agreement with Giebler et al. (26), we have found that the CREB⅐Tax⅐CBP-(1-682) initiation complex is quite stable, remaining associated with the transcription template for 2 h (data not shown). It will be of interest to analyze the DB complex dissociation and reassociation kinetics in the presence of Tax and CBP. Finally, it will be of interest to analyze the contribution of transcriptional reinitiation in the CBP/p300-dependent NF-B, AP1, MAPK, PKA, and nuclear hormone receptor signaling pathways.