Identification and characterization of a novel transcriptional activation domain in the CREB-binding protein.

The CREB-binding protein (CBP) plays a central role in the regulation of gene expression by several different second messenger pathways including serum growth factors, cAMP and phorbol esters. CBP specifically binds to the phosphorylated forms of CREB and c-Jun and is thought to activate transcription through a C-terminal activation domain. In this report, we demonstrate that the C terminus of CBP is dispensable for its ability to stimulate phospho-CREB activity, and, further, that the deletion of this domain produces highly active, mutant forms of CBP. The novel N-terminal activation identified by this deletional analysis consists of the first 714 amino acids of CBP and is sufficient for high levels of transcriptional activity. This domain is also capable of stimulating the activity of a second cAMP-regulated factor, ATF-1. Surprisingly, ATF-1 activity is not significantly stimulated by full-length CBP suggesting that the C-terminal domain of CBP may also serve to regulate ATF-1/CBP activity. Additionally, the demonstration that one of our hyperactive CBP mutants is able to activate a nonphosphorylatable mutant of CREB (M1 CREB) provides the first evidence that CBP may play a role in regulating the basal transcriptional activity of CREB.

The CREB-binding protein (CBP) plays a central role in the regulation of gene expression by several different second messenger pathways including serum growth factors, cAMP and phorbol esters. CBP specifically binds to the phosphorylated forms of CREB and c-Jun and is thought to activate transcription through a Cterminal activation domain. In this report, we demonstrate that the C terminus of CBP is dispensable for its ability to stimulate phospho-CREB activity, and, further, that the deletion of this domain produces highly active, mutant forms of CBP. The novel N-terminal activation identified by this deletional analysis consists of the first 714 amino acids of CBP and is sufficient for high levels of transcriptional activity. This domain is also capable of stimulating the activity of a second cAMPregulated factor, ATF-1. Surprisingly, ATF-1 activity is not significantly stimulated by full-length CBP suggesting that the C-terminal domain of CBP may also serve to regulate ATF-1/CBP activity. Additionally, the demonstration that one of our hyperactive CBP mutants is able to activate a nonphosphorylatable mutant of CREB (M1 CREB) provides the first evidence that CBP may play a role in regulating the basal transcriptional activity of CREB.
Hormonally induced changes in the levels of intracellular cyclic AMP (cAMP) regulate the expression of many cellular and viral genes through a common promoter element termed the CRE 1 (1). The activity of the CRE is, in turn, governed by a large family of transcription factors known as CREB/ATF proteins. These proteins share several common structural features including a DNA binding and dimerization motif, known as the bZIP domain, and transcriptional activation domains located in the N terminus of the molecule. Although poorly conserved overall, these activation domains share consensus phosphorylation sites for a variety of protein kinases suggesting that their activity, like that of many bZIP factors, is highly regulated by phosphorylation (2,3).
The transcription factor CREB (cAMP response elementbinding protein) represents the prototypical member of this family and it, along with the related factors ATF-1 and CREM, mediates many aspects of cAMP-regulated gene expression (4).
The activation domains of all three of these proteins are highly homologous and can be divided into three basic elements, a kinase-inducible domain (KID), which contains a consensus phosphorylation site for the cAMP-dependent kinase (PKA), and two glutamine-rich domains (Q1 and Q2) flanking the KID. The phosphorylation of the CREB KID at a consensus PKA site (serine 133) is required for full transcriptional activity and mutation of this site effectively blocks stimulated activity (5). Although CREB is capable of interacting directly with several components of the basic transcriptional apparatus including TFIIB and TFIID, these interactions are not regulated by phosphorylation (6,7), and it has been unclear how phosphorylation of the CREB activation domain results in enhanced transcriptional activity. The recent identification of a phosphorylationdependent transcriptional co-activator of CREB has provided exciting new insights into this significant problem.
The CREB-binding protein (CBP) is a large nuclear phosphoprotein capable of selectively interacting with the phosphorylated form of CREB by recognizing specific residues found in the CREB KID (8 -10). This interaction is required for PKA-dependent transcriptional activation as disruption of the complex by microinjection of either specific peptide fragments of the phospho-CREB binding domain or anti-CBP blocking antibodies is able to block stimulated expression of a CRE-dependent reporter gene in fibroblast cells (10,11). CBP is highly related to the E1a-associated protein p300, and both CBP and p300 are capable of acting as CREB co-activators and binding to E1a (12,13). The binding of E1a to CBP/p300 is believed to be responsible for mediating many aspects of E1a-dependent transformation and has recently been shown to result in a strong inhibition of transcriptional activity of CBP/p300. Fusion of either the CBP or p300 cDNA to a heterologous DNA binding domain demonstrates that CBP is a potent transcriptional activator on its own (8,12). This transcriptional activity is thought to reside in the C-terminal portion of the molecule and seems to be related to the ability of CBP to interact with the basic transcription factor TFIIB. Indeed, fusion of the C-terminal portion of CBP, including the TFIIB binding site, to a heterologous DNA binding domain produces a transcriptional activator which is more potent than the fulllength CBP fusion protein (9). These results have suggested a simple model of CBP activity in which it binds selectively to the phosphorylated form of CREB proteins and activates transcription by recruiting TFIIB (14). CBP activity is not regulated by phosphorylation-dependent binding alone, however, and mutant forms of CREB have been identified which bind to CBP in a phosphorylation-dependent fashion but are not activated by CBP (15).
CBP/p300 proteins have also been implicated as potential co-activators for several other important transcriptional activators including c-Jun, c-Fos, and YY-1 (11,16,17). Although the binding of CBP/p300 to both c-Jun and CREB appears to be phosphorylation-dependent, the binding of CBP/p300 to c-Fos and YY-1 is not, and it remains unclear if the transcriptional activity of CBP is regulated in these cases. Because CBP/p300 is capable of interacting with a wide variety of transcription factors, it is likely that the regulation of its transcriptional activity will play a central role in the overall regulation of gene expression by second messengers. To better understand how CBP activity is regulated, we have generated a series of mutant CBP molecules and assayed them for activity in a co-transfection assay with PKA-dependent transcription factors. In this report, we demonstrate that the previously identified C-terminal activation domain of CBP is dispensable for co-activator function in F9 cells and that the N terminus of CBP is sufficient for transcriptional activation.

MATERIALS AND METHODS
Plasmids-The expression vectors coding for the full-length mouse CBP protein (Rc/RSV mCBP) has been previously described (8,9) and was a kind gift of Dr. J. Chrivia, St. Louis University. The plasmid CBP ⌬E is identical to mCBP with the exception that the sequences between the unique ScaI and SmaI sites have been removed deleting amino acids 1569 -1891 while retaining the open reading frame. The CBP Xba⌬ expression vector (aa 1-1109) was constructed by completely digesting RSV mCBP with XbaI to remove the 3Ј end of the cDNA and then religating the backbone. The Sph⌬ vector (aa 1-714) was constructed by subcloning a HindIII/SphI fragment from RSV mCBP into pGEM 3Z and then removing a HindIII/XbaI fragment from this plasmid and recloning it into Rc/RSV. The two internal CBP deletion mutants (⌬N and S⌬⌬P) are identical to Sph⌬ with the exception that the sequence between the PvuII sites at position 1068 and 1494 (aa 357-498) have been removed in S⌬⌬P and the sequences between the DraI site and the EcoRI site at positions 160 and 1354 (aa 54 -452) have been removed from ⌬N. The EcoRI site at position 1494 was blunted with Klenow fragment in the ⌬N construct in order to retain the correct reading frame. The ⌬CBD vector (aa 1-452) was constructed by cloning the 5Ј-most EcoRI fragment of mCBP into pBluescript KS (Stratagene) and subsequently removing the HindIII/XbaI fragment from this plasmid into Rc/RSV. The cloning junctions of all mutant plasmids were confirmed by dideoxy sequencing using an unmodified T7 polymerase (Pharmacia Biotech Inc.). The RSV-M1 CREB expression vector was a kind gift of Dr. Richard Goodman, OHSU, Portland, OR. All other plasmids used in this work have been described previously (18).
Cell Culture and Transfection Assays-F9 cells were grown on 0.7% gelatin-coated plates in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Life Technologies, Inc.), penicillin (100 units/ml), streptomycin (100 g/ml), and L-glutamate (2 mM). Cells were plated at 5.5 ϫ 10 5 cells per 10-cm plate 18 h prior to transfection. DNA was prepared by alkaline lysis, doublebanded in CsCl gradients, phenol-extracted, and ethanol-precipitated twice prior to use. Transfections contained 4 g of a RSV expression vector coding for the indicated transcription factor, 4 g of RSV-PKA, 5 g of ⌬-70 SS CAT, 20 g of a RSV expression vector coding for the indicated CBP mutant, and 2 g of a RSV-GH expression plasmid (gift of L. Roselli-Rehfuss, IRCM, Montreal, PQ) to control for transfection efficiency. In transfections lacking PKA or CBP, an equal mass of Rc/RSV (Invitrogen) was substituted. DNA mixtures were transfected by calcium phosphate precipitation (18 h) without a glycerol shock. Cells were washed, refed, and allowed to grow for 30 h before harvesting. CAT activity was determined by the method of Seed and Sheen (19), and GH immunoreactivity was determined with a commercially available solid-phase RIA (Immunocorp, Montreal, PQ). Relative transcriptional activities were expressed as the ratio of CAT activity to GH activity in arbitrary units. Transfections were repeated three to six times in order to ensure reproducibility and are reported as the mean Ϯ S.E.
Western Blotting-Neuro 2A cells were grown to 50 -70% confluence on 60-mm plates and transfected by calcium phosphate precipitation for 4 h followed by a glycerol shock. Precipitates contained 9 g of the indicated expression plasmid and 1 g of RSV-GH to measure transfection efficiency. Following removal of the precipitate, cells were refed and allowed to grow for an additional 44 h before harvesting in NET (40 mM Tris (7.5), 10 mM EDTA, 150 mM NaCl). After pelleting, cells were lysed in 45 l of WCE lysis buffer (50 mM Tris (8.0), 400 mM KCl, 1% Triton X-100, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 2 M pepstatin, and 0.6 M leupeptin) on ice for 15 min and then cleared of insoluble debris by centrifugation. The indicated volume of extract was denatured by adding an equal volume of 2X SDS Buffer (117 mM Tris (6.8), 3.4% SDS, 10% glycerol, 0.2 M dithiothreitol, 2% bromphenol blue) and heating to 90°C for 5 min. Proteins were then separated by SDS-polyacrylamide gel electrophoresis on either an 8% or 10% gel and transferred to polyvinylidene difluoride (Millipore) membrane by electroblotting overnight (30 V). The transfected CBP proteins were detected by using an antibody specific for the N terminus of CBP (CBP-NT; Upstate Biotechnologies) and a chemiluminescence kit (BM Chemiluminescence; Boehringer Mannheim) according to the manufacturer's instructions.

RESULTS AND DISCUSSION
The TFIIB Binding Site of CBP Is Dispensable for Activity-To determine if the ability of CBP to act as a CREB coactivator was dependent on the presence of the TFIIB binding site, we examined the transcriptional activity of a mutant form of CBP which lacks this region. To do this, sequences which comprise the TFIIB binding site (aa 1569 -1891) were removed by constructing an in-frame deletion of the region (Fig. 1A) and the ability of the resulting mutant (CBP ⌬E) to potentiate the activity of CREB was determined by measuring its ability to stimulate CREB-dependent transcriptional activity in F9 cells. coding for the catalytic subunit of protein kinase A (RSV-PKA). Transcriptional activities were normalized to the basal CREB activity observed in the absence of PKA or CBP.
As can be seen in Fig. 1B, CREB is about 3-fold more active in activating transcription from the CRE-dependent reporter plasmid in the presence of PKA stimulation as compared to unstimulated activity. This level of PKA responsiveness is consistent with previous reports of CREB activity in F9 cells after normalization to basal activity (9,19). Co-transfection of 20 g of an expression vector coding for either the RSV-CBP or RSV-CBP ⌬E construct in the absence of PKA stimulation had no effect on CREB activity. This is consistent with the idea that the full-length CBP molecule is not normally a potent activator of basal CREB activity. In the presence of PKA stimulation, both the wild type and mutant CBP constructs were able to stimulate CREB activity about 9-fold over basal levels. We were initially surprised by this result as we had anticipated that the absence of the TFIIB binding site would inhibit CBP's ability to co-activate CREB. This result demonstrates, however, that the TFIIB binding site is dispensable for activity, and, further, that an additional, previously uncharacterized, activation region must be present in CBP.
The N Terminus of CBP Acts as a Potent CREB Co-activator-To localize these additional activator sequences, we generated a series of internal and C-terminal deletions of CBP and tested each of them for their ability to stimulate phospho-CREB activity. As CBP and p300 have both been shown to act as phospho-CREB co-activators, we initially targeted the regions of CBP that were conserved with p300 ( Fig. 2A). These conserved regions consist of two small (approximately 100 aa) N-terminal domains and the large (approximately 1300 aa) C-terminal region which had previously been shown to be transcriptionally active (9). To determine if the presence of this large C-terminal domain was required for CBP activity, we generated two truncation mutants which contained either the first 1109 (CBP Xba⌬) or 714 (CBP Sph⌬) amino acids of CBP and assayed them for co-activator function as described previously (Fig. 2B). Surprisingly, both C-terminal truncation mutants were able to significantly stimulate CREB activity to similar levels in the presence of PKA. Indeed, both constructs stimulated phospho-CREB activity 20 -30-fold over basal levels and 2-3 times better than the wild type CBP. In addition, CBP Xba⌬ stimulated basal CREB activity about 5-fold, although CBP Sph⌬ did not.
To further investigate the nature of this novel N-terminal activation domain, we made additional mutations in the smallest active CBP mutant which were again guided by comparison with the sequence of p300. This region contains two domains which are highly conserved with p300. The first domain contains a putative zinc finger with no known function while the second appears to act as a phospho-CREB binding site. To determine if the first domain was involved in co-activation, we generated two in-frame deletion mutants of CBP Sph⌬ which removed either most (CBP S⌬⌬P) or all (CBP ⌬N) of these conserved sequences. As a control, we generated a third truncation (CBP ⌬CBD) which consists of the first 452 amino acids of CBP ( Fig. 2A). We anticipated that this latter mutant would be transcriptionally inactive as it lacks a phospho-CREB binding domain. Transcriptional activity was again assessed by co-transfection and is shown in Fig. 2B. In the absence of PKA stimulation, neither CBP S⌬⌬P nor CBP ⌬N were able to potentiate basal CREB activity. In the presence of PKA, both proteins were able to stimulate CREB activity 5-7-fold over basal levels and, surprisingly, are only slightly less active than the wild type CBP. As expected, deletion of the second conserved domain (⌬CBD) produced a molecule which was unable to significantly stimulate CREB activity, consistent with the idea that the CREB binding domain is required for the stimulation of phospho-CREB activity.
Mutant Forms of CBP Are Stably Expressed in Cells-To confirm that each of the mutated forms of CBP was stably expressed in cells, we performed Western blots on extracts of cells transfected with the expression vectors coding for the different forms of CBP using an antibody specific for the N terminus of CBP. As can be seen in Fig. 3, cells transfected with the empty expression vector alone (Rc/RSV) produced no specific immunoreactivity suggesting that the endogenous level of CBP expressed in these cells is too low to be detected under these conditions. Cells which had been transfected with each of the different expression vectors produced specific robust immunoreactive signals consistent with the predicted molecular weight of the mutated protein. This result demonstrates that all forms of CBP used in this study are stably expressed in cells at similar levels with the exception of CBP S⌬⌬P and CBP ⌬CBD which are expressed at approximately 10 -20-fold higher levels. Because the level of wild type CBP expression vector used in these experiments has previously been shown to provide saturating amounts of CBP under identical conditions (9), it is likely that all of the mutant forms of CBP, which are expressed at the same or greater levels, are also present in  Fig. 1A. B, relative activity of each of the C-terminal mutants in A to stimulate CREB activity in the presence or the absence of PKA stimulation. Transfections were performed as described and contained the same amount of expression vector either without an insert (Rc/RSV) or with the indicated mutant CBP co-activator. Transcriptional activities were expressed as in Fig. 1. excess. We confirmed this observation by demonstrating that co-transfection of up to 35 g of CBP expression vector did not result in any additional increase in transcriptional activity (data not shown). These results suggest that the differences in transcriptional activity observed between the different forms of CBP are a direct result of intrinsic differences in their abilities to activate transcription.
Taken together, our results demonstrate that the previously described C-terminal activation domain of CBP is dispensable for its ability to stimulate CREB activity and that the Nterminal 714 amino acids alone are sufficient for full activity. The in-frame deletion analysis suggests that this N-terminal activation domain consists of at least two sub-domains; the first is located in the first conserved domain of CBP (aa 357-498), while a second is likely to be located between amino acids 498 and 714. Because this latter region contains the second conserved domain of CBP, it is tempting to speculate that additional activator sequences might be located in this region (aa 586 -679). However, Parker et al. (10) have recently shown that the second conserved domain acts as a dominant negative inhibitor of CREB activity in NIH 3T3 cells. Therefore, we conclude that these additional activator sequences reside in a nonconserved portion of CBP (aa 680 -714). The demonstration that mutant forms of CBP lacking the C-terminal domain are significantly better co-activators of CREB suggests that the C terminus of CBP may play a role in regulating CBP activity. Although the nature of this regulation is unclear from these experiments, recent studies have confirmed the idea that the transcriptional activity of CBP is regulated independently of its binding to phosphorylated CREB (15,20).
CBP Xba⌬ Stimulates the Activity of a Nonphosphorylatable CREB Mutant-The ability of the CBP Xba⌬ mutation to stimulate CREB activity in the absence of PKA was unexpected as CBP is not believed to interact with CREB unless CREB is phosphorylated. In order to determine whether this activity was the result of CBP Xba⌬ binding to trace amounts of phosphorylated CREB, we tested the ability of CBP Xba⌬ to stimulate the activity of a nonphosphorylatable mutant of CREB termed M1 CREB (5). As a control, we also determined the ability of the wild type CBP and CBP Sph⌬ to activate M1 CREB (Fig. 4). As previously reported, M1 CREB was not able to stimulate a CRE-dependent reporter gene in response to PKA, and co-transfection of 20 g of either the RSV-CBP or RSV-CBP Sph⌬ expression vector did not enhance its tran-scriptional activity. In contrast, co-transfection of an RSV-CBP Xba⌬ expression vector stimulated M1 CREB activity about 4-fold in both the presence and the absence of PKA stimulation. Control transfections (data not shown) demonstrate that this effect is specific for the CREB transcription factor as CBP Xba⌬ is unable to stimulate the transcriptional activity of either a Gal4-VP16 transcriptional activator or the estrogen receptor under the same conditions. This result suggests that while CBP is clearly an important component of phospho-CREB activity, it may also play a role in regulating the basal activity of CREB. The hypothesis that CBP binds exclusively to the phosphorylated form of CREB is based on an analysis of the N-terminal 700 amino acids of CBP (8) and does not rule out the possibility of a second CREB interaction site elsewhere on the molecule. Indeed, studies of the role of CBP in the regulation of both c-Fos and YY-1 have concluded that both these proteins interact in a non-phosphorylation-dependent manner with unique binding sites located in the C terminus of CBP (16,17). Alternatively, it is possible that the ability of CBP Xba⌬ to activate M1 CREB is the result of an indirect interaction through a second adaptor protein. Because both CBP Sph⌬ and CBP Xba⌬ are able to activate phospho-CREB activity to the same extent and only CBP Xba⌬ stimulates basal CREB activity, it seems likely that this M1 CREB interaction site may be located between amino acids 714 and 1109. Preliminary experiments suggest that this region does not directly interact with M1 CREB in a yeast two-hybrid system (data not shown) supporting the idea of an indirect interaction between CBP and M1 CREB.
Wild Type CBP Is Not a Potent Co-activator of ATF-1 in F9 Cells-To determine if the N-terminal activation domain of CBP would function in potentiating the activity of a second cAMP-regulated transcription factor, we tested the ability of the wild type and mutant CBPs to co-activate ATF-1 in F9 cells. To do this, co-transfection assays utilizing an ATF-1 expression vector were performed as described for CREB. As shown in Fig.  5, ATF-1 was 2-fold more active at stimulating the CRE-dependent reporter gene in the presence of PKA as compared to the unstimulated activity. This level of PKA-dependent stimulation is less robust than that observed with CREB; however, ATF-1 is known to be less PKA-responsive under these conditions (21). With the exception of CBP Sph⌬ and CBP Xba⌬, none of CBP constructs we tested were able to significantly stimulate ATF-1 activity in the presence or absence of PKA. Additional co-transfection assays performed with either twice the amount of ATF-1 or 1.5 times the amount of CBP confirm this result (data not shown) and suggest that the poor transcriptional activity we observed with CBP and ATF-1 is not the result of limiting amounts of either ATF-1 or CBP. This suggests that wild type CBP is not an effective co-activator of ATF-1 in F9 cells. In contrast, co-transfection of CBP Xba⌬ and CBP Sph⌬ mutants significantly enhanced the PKA-stimulated transcriptional activity of ATF-1. As observed with CREB, CBP Xba⌬ was also able to significantly stimulate basal ATF-1 activity. These results demonstrate that the N terminus of CBP is capable of interacting with the phosphorylated form of ATF-1, presumably through the CREB binding domain, and that the N terminus of CBP is capable of potentiating the activity of a second cAMP-regulated factor. The demonstration that only the C-terminally truncated forms of CBP act as potent activators of ATF-1 provides an independent line of evidence that the C terminus of CBP serves to regulate transcriptional activity.
Taken together, our results support an alternative model for CBP activity in which CBP may be constitutively associated with CREB and phosphorylation of the CREB⅐CBP complex by PKA induces a change in conformation to a more active state. Presumably, this conformational change might take place in two steps. The first would involve a deregulation of CBP activity while the second would involve the correct binding of CBP to the phosphorylated form of CREB. In this scenario, deletion of the C terminus would allow the N-terminal activation domain to assume this more active conformation in the absence of any additional regulatory signals, although the correct positioning of the N-terminal activation domain would still be dependent on CREB phosphorylation. For the case of the wild type CBP, these additional signals might include additional phosphorylation events, recruitment of additional regulatory proteins, or the requirement for CBP to adopt a specific conformation in order to interact with the basal transcription apparatus. CBP contains a consensus phosphorylation site for PKA (8), and it is possible that phosphorylation of this site might serve to regulate the activity of the N terminus. Experiments to test this hypothesis are presently in progress. The observation that CBP does not act as a good cofactor for ATF-1 unless the C terminus has been deleted also supports a role of this domain in regulating CBP activity. In this case, a negatively acting inhibitor of ATF-1, such as that recently postulated to exist in F9 cells (21), might serve to prevent CBP from adopting an active conformation. Deletion of the C terminus would again allow CBP to be transcriptionally active when bound to phosphorylated ATF-1. Further characterization of the events which regulate CBP activity will be required in order to fully understand the role CBP plays in control of gene expression. FIG. 5. Relative ability of wild type and mutant CBP to stimulate the activity of RSV-ATF-1 in the presence and the absence of PKA. Transfections were performed and results were plotted as described previously with the exception that they contained 4 g of RSV-ATF-1 instead of RSV-CREB.