cAMP-independent activation of the adenovirus type 12 E2 promoter correlates with the recruitment of CREB-1/ATF-1, E1A(12S), and CBP to the E2-CRE.

Expression of the transcription unit early region 2 (E2) is of crucial importance for adenoviruses because this region encodes proteins essential for viral replication. Here, we demonstrate that the E1A(12S) protein of the oncogenic adenovirus serotype 12 activates the E2 promoter in dependence of the N terminus and the conserved region 1. Activation is mediated through a cAMP-response element that is bound by CREB-1 and ATF-1. Moreover, the Ad12 E2 promoter is inducible by protein kinase A and repressed by either a dominant-negative cAMP-response element-binding protein (CREB) mutant or the highly specific protein kinase A inhibitor protein underscoring the participation of CREB-1/ATF-1 in promoter activation. E1A(12S) binds to CREB-1 and ATF-1 in dependence of the N terminus and CR1 and is recruited to the E2 cAMP-response element through both cellular transcription factors. Most interestingly, point mutations revealed that E1A(12S) domains essential for binding to CREB-1/ATF-1 and for activation of the Ad12 E2 promoter are also essential for binding to the CREB-binding protein. Due to these data and results obtained in DNA-dependent protein-protein interaction assays, we propose a model in which the cAMP-independent activation of the Ad12 E2 promoter is mediated through a ternary complex consisting of CREB-1/ATF-1, E1A(12S), and CREB-binding protein, which assembles on the E2 cAMP-response element.

Proteins encoded by the early transcription unit 1A (E1A) 1 of Ads play an essential role for the viral life cycle because they are required for the activation of the expression of all other viral genes (1)(2)(3). In addition, these proteins modulate the expression of specific cellular genes in infected cells to facilitate viral reproduction (4). E1A proteins are also able to act as oncoproteins that cooperate with Ad early region 1B oncogene products to transform rodent cells in culture and, depending on the serotype, to induce tumors in immunocompetent animals (e.g. Ad12; see Ref. 5).
Ad12 E1A gives rise to five proteins from which the 266R protein (translated from a 13 S mRNA; henceforth referred to as E1A 13S ) and the 235R protein (translated from a 12 S mRNA; E1A 12S ) are the predominant species (6). Both proteins are translated in the same reading frame but differ in a short stretch of 31 aa called CR3 that is absent in E1A 12S . CR3 is one of three E1A regions (CR1, CR2, and CR3) that are highly conserved among Ad serotypes (7). Besides the non-conserved N terminus, the CRs contain most of the gene regulatory functions necessary for viral reproduction and transformation (6).
Due to the lack of a sequence-specific DNA binding activity (8), E1A proteins modulate gene expression through protein/ protein interactions with cellular factors. The bZip family of transcription factors that is involved in cellular proliferation, differentiation, transformation, and stress response (9 -11) is one important target for E1A. This family includes the Jun, Fos, and ATF proteins as well as cAMP-responsive transcription factors like CREB-1. To fulfill their gene regulatory functions these factors form either homo-(e.g. CREB-1/CREB-1) or selected heterodimers (e.g. CREB-1/ATF-1), which bind to specific response elements in the promoter of target genes (11,12). Several studies have shown that transcriptional regulation through this family of transcription factors is accompanied by binding to cellular cofactors like p300 and CBP (13).
The cAMP/PKA pathway is involved in a number of physiological processes including intermediary metabolism, gene expression, and cellular proliferation (12,14). In response to cell surface signals, intracellular cAMP levels are raised enabling the rapid dissociation of the regulatory and catalytic subunits of the PKA holoenzyme in the cytoplasm followed by the passive translocation of the active catalytic subunit into the nucleus. Here, PKA phosphorylates the cellular transcription factor CREB at Ser-133. Phospho-CREB on its turn recruits the cellular coactivator CBP to the CRE which leads to the transcriptional activation of cAMP-dependent target genes (12).
E1A positively or negatively modulates the PKA pathway through interaction with CBP. By targeting CBP and disrupting transcription factor-coactivator complexes, E1A is able to repress cAMP-dependent transcription (15,16). For instance, by binding to p300 or CBP the Ad2/Ad5 E1A 12S protein prevents the transactivation function of CREB in transient expression assays using either the rat somatostatin CRE in a heterologous promoter context (15), a truncated somatostatin promoter (16), or the intact interleukin-6 promoter (15). On the other hand, it has been shown that the Ad2/Ad5 E1A 12S protein activates the human PCNA promoter via CREB (17). Moreover, using the CREB/ATF recognition sequence of the fibronectin promoter, Lee et al. (18) have demonstrated that the Ad2/Ad5 E1A 12S protein activates or down-regulates CREB-mediated reporter gene expression in a cell type-specific manner.
Although many Ad promoters contain CREs, viral gene expression can occur independently of cAMP stimulation (19,20). It is therefore of interest to understand how viral genome expression takes places to find out (i) the composition of protein complexes binding to these CREs and (ii) the regulation of the activity of these complexes through E1A irrespective of cAMP.
Here we present evidence that Ad12 E1A 12S activates the E2 promoter through a CRE (E2-CRE, located at nucleotides Ϫ99 to Ϫ92 relative to the transcriptional start site) dependent on the N terminus and CR1. The E2-CRE is bound by CREB-1 and ATF-1. Mutational analyses revealed that the transactivation of the Ad12 E2 promoter through E1A 12S correlates with the binding of the adenoviral protein to CREB-1/ATF-1 and with the ability of the viral protein to associate with the CREB coactivator CBP. Most interestingly, CREB-1 and ATF-1 recruit E1A 12S to the E2-CRE, and the amount of E1A 12S bound to the CREB-1⅐E2-CRE complex is increased 2-4-fold in the presence of CBP. On the basis of these results we propose a model in which transcriptional activation of the Ad12 E2 promoter is attributed to the assembly of a quaternary complex consisting of CREB-1/ATF-1, E1A 12S , the cellular coactivator CBP, and the E2-CRE.
Bacterial expression vectors encoding the various E1A 12S proteins as fusions were constructed by cloning their cDNAs amplified by PCR into the BamHI site of the prokaryotic GST fusion expression vector pGEX-2T (Amersham Pharmacia Biotech). His 6 -E1A 12S was cloned by introducing the cDNA for the Ad12 E1A 12S into the BamHI site of the vector pQE-8 (Qiagen). To express CREB-1 by in vitro transcription/ translation, its cDNA was cloned by PCR into the pcDNA3.1 vector (Invitrogen) using pRc/RSV-CREB-1 as template. A HindIII-NotI fragment obtained from pRc/RSV-CBP was cloned into the vector pCR3.1 (Invitrogen) to generate pCR3.1-CBP. This plasmid was used for in vitro transcription/translation of full-length CBP. The plasmid pRc/ CMV-A-CREB encodes a dominant-negative inhibitor of CREB (22). phc-JUN-(Ϫ1600/ϩ740)-CAT (spanning nt Ϫ1600 to ϩ740 of the human c-JUN promoter) was described elsewhere (23). PGEX-CBP was constructed by cloning a BamHI fragment from pRc/RSV-CBP into the bacterial expression vector pGEX-2T. CBP fragments expressing exclusively the KIX-domain (CBP KIX , aa 461-682) or the minimal KIX domain interacting with the KID domain of CREB-1 (CBP mKIX ; aa 586 -666; see Ref. 12) were generated by PCR using respective primers and cloned into pGEX-2T.
The sequence of all constructs was verified using the ALF sequencing system (Amersham Pharmacia Biotech).
Transient Expression Assays-For transient expression assays 1 g of the respective Ad12 E2 reporter construct was cotransfected with different amounts of the respective expression vectors into 2.5 ϫ 10 5 KB cells as indicated in the figures using the LipofectAMINE method according to the manufacturer's instructions (Life Technologies, Inc.). The transfection was stopped after 5 h. 24 h later cells were harvested and lysed by the freeze and thaw treatment. The protein concentration was analyzed by the Bradford method (28). Equal amount of protein was used to determine CAT activity (29). The percentage of acetylated chloramphenicol was quantified using the automatic TLC linear analyzer of Berthold, Bad Wildbad, Germany.
For competition experiments a 10-or 100-fold molar excess of the respective unlabeled competitor was added simultaneously with the 32 P-labeled Ad12 E2-CRE-oligonucleotide to the binding reaction. As competitors the following were used: a CRE somatostatin promoter oligonucleotide (5Ј-CGCCTCCTTGGCTGACGTCAGAGAGAGAGTTT-3Ј, nt Ϫ60 to Ϫ29 of the rat somatostatin promoter; see Ref. 34), a mutated CRE somatostatin promoter oligonucleotide (5Ј-CGCCTCCT-TGGCTGAATTCAGAGAGAGAGTTT-3Ј; CG substituted by AT in the core binding motif), and the c-JUN2-TRE (5Ј-AGCATTACCTCATC-CCG-3Ј; nt Ϫ194 to Ϫ178 of the human c-JUN promoter; see Ref. 35).
Induction of His 6 -E1A 12S expression and purification of the fusion protein via nickel-chelate affinity chromatography were performed as described by Qiagen. GST-pull-down assays using purified full-length GST-CBP or respective GST-CBP fragments (50 g) and purified His 6 -E1A 12S (2 g) were performed as described above. Interaction was analyzed by Western blotting with an anti-Ad12 E1A antiserum (see below).
Immunoprecipitations and Western Blot Analyses-9.2 ϫ 10 6 COS7 cells were transfected by electroporation with 15 g of the wild type or mutant E1A 12S expression vectors as described (37) except that 10 mM glucose and 0.1 mM dithiothreitol were included in the electroporation buffer. For immunoprecipitation, cells were lysed 72 h later in lysis buffer (50 mM Hepes/KOH, pH 7.9, 100 mM KCl, 1 mM EDTA, 5 mM MgCl 2 , 0.1% Nonidet P-40, 10 mM NaF, 0.5 g/l BSA, 0.5 mM Pefabloc, and 10 g/ml aprotinin) by rocking for 30 min at 4°C. Whole cell extracts (1 mg) were precleared with normal rabbit IgG (Santa Cruz Biotechnology) and 60 l of protein A-Sepharose for 1 h at 4°C. Precleared extracts were then incubated with an anti-CBP rabbit polyclonal antiserum (A-22; Santa Cruz Biotechnology), an anti-CREB antiserum (Ab240; see Ref. 38), or an anti-ATF1 mouse monoclonal antibody (FI-1, Santa Cruz Biotechnology) and 30 l of protein A-Sepharose at 4°C overnight. Immune complexes were collected, washed three times in 1 ml of lysis buffer without BSA, eluted in 1ϫ SDSsample buffer, and resolved on 13% SDS-polyacrylamide gels. After electrophoresis proteins were transferred to a Hybond C-extra nitrocellulose membrane (Amersham Pharmacia Biotech), and the membrane was probed with the anti-Ad12 E1A antiserum as described previously (39). Enhanced chemiluminescence was carried out using the Super Signal Ultra Chemiluminescence detection system (Pierce). For Western blot analyses 10% of the whole cell extract prepared from transfected cells was used.
DNA-dependent Protein-Protein Interaction Analysis-For DNA-dependent protein-protein interaction analyses, a 5Ј-biotinylated oligonucleotide containing the Ad12 E2-CRE recognition motif (corresponding to nt Ϫ113 to Ϫ80 of the Ad12 E2 promoter) was synthesized with a C 24 -atom spacer to reduce steric hindrance (Metabion, Martinsried, Germany): 5Ј-TACTCATCTTGTTGTGACGTCACTTTTCCCGCC-3Ј. Purified, double-stranded oligonucleotide (1 g) was incubated with in vitro translated ATF-1 or CREB-1 and 25 l of a slurry of neutravidin-agarose (Pierce) in PCϩ150 buffer (20 mM Hepes/KOH pH 7.9, 150 mM KCl, 12.5 mM MgCl 2 , 0.2 mM EDTA, 0.05% Nonidet P-40, 0.05% Tween 20, 0.5 mM Pefabloc (Roche Molecular Biochemicals), 2 g/l BSA) for 1 h at RT. Thereafter DNA-protein complexes were collected and washed three times with 1 ml of PCϩ150 buffer. DNA-protein complexes were then incubated in PCϩ150 buffer with 35 S-labeled E1A-12S and/or CBP, respectively. Bound complexes were pelleted and washed three times with 1 ml of PCϩ150 buffer, eluted in 1ϫ SDS-sample buffer, resolved on 10% SDS-polyacrylamide gels, and visualized by fluorography.  (40). A CRE (E2-CRE; nt Ϫ99 to Ϫ92 relative to the transcriptional start site) as well as one consensus E2F recognition sequence (nt Ϫ89 to Ϫ82) were identified (Fig. 1). Moreover, a TATA box is located at position nt Ϫ29 to Ϫ24 (Fig. 1). To analyze to what extent these binding sites contribute to basal as well as E1A 12S -mediated activation of the Ad12 E2 promoter and whether promoter elements not identified by computer analyses participate too, we performed transient expression assays. 5Ј-Progressive Ad12 E2 promoter truncations and point mutants were constructed, and the activity of the resulting E2 Ad12 -CAT reporter constructs was determined after cotransfection with an E1A 12S expression vector (pRc/RSV-235R) in KB cells. E2 Ad12 -(420 bp)-CAT containing the full-length Ad12 E2 promoter displayed a low basal CAT activity if transfected without E1A 12S into KB cells (Fig. 1). Cotransfection of pRc/RSV-235R led to a 5.3-fold induction of CAT gene expression (Fig. 1). CAT gene expression strictly depends on E2 promoter sequences because the basic reporter construct pBL-CAT3 showed neither a basal nor an E1A 12S -induced CAT gene expression (Fig. 1).

A CRE Confers
Promoter truncations up to nt Ϫ140 (E2 Ad12 -(389 bp)-CAT, E2 Ad12 -(324 bp)-CAT, E2 Ad12 -(215 bp)-CAT, and E2 Ad12 -(140 bp)-CAT) did not interfere with basal or with E1A 12S -induced reporter activity (Fig. 1). However, deletion or point mutations of the E2-CRE (E2 Ad12 ⌬CRE-(140 bp)-CAT and E2 Ad12 pmCRE-(140 bp)-CAT) resulted in a transcriptional inactive reporter construct (Fig. 1). In contrast to the mutations in the CRE, deletion or point mutants of the E2F-binding site (E2 Ad12 ⌬E2F-(140 bp)-CAT and E2 Ad12 pmE2F-(140 bp)-CAT) did not prevent basal nor E1A 12S -activated CAT gene expression although E1A 12S -mediated activation of E2 Ad12 pmE2F-(140 bp)-CAT was reduced compared with the E2 wild type reporter construct (Fig. 1). These data indicate that the E2F-binding site is of secondary importance for E2 promoter activation in transient Functional Interaction of CREB-1/ATF-1 with E1A 12S expression assays in KB cells. As expected, point mutations in the TATA box led to a loss of basal as well as E1A 12S -induced reporter gene expression (Fig. 1). Thus, we conclude that (i) nt Ϫ420 to Ϫ140 are dispensable for the activation of the Ad12 E2 promoter through E1A 12S and (ii) the E2-CRE at position nt Ϫ99 to Ϫ92 and the TATA box at position Ϫ29 to Ϫ24 are essential elements for promoter activation. However, the participation of the E2F site in promoter activation is not yet clear and is currently under investigation.
The E1A 12S Protein Activates the Ad12 E2 Promoter Dependent on the N Terminus and CR1-Recently, we have identified the very N terminus (aa 1-29) and CR1 of E1A 12S as independent transactivation domains (41). To determine whether both domains are involved in the activation of the Ad12 E2 promoter, we cotransfected E1A 12S deletion mutants lacking the N terminus (aa 1-29; ⌬N/235R), CR1 (aa 39 -79; ⌬CR1/235R), or aa 1-79 (⌬1-79/235R) with E2 Ad12 -(140 bp)-CAT in KB cells. CAT assays revealed that none of the E1A 12S deletion mutants was able to induce reporter gene expression (Fig. 2B). Furthermore, using point mutants we identified three aa in the N terminus whose mutations led either to a greatly reduced (D16A/235R; Fig. 2B) or to a loss (I18P/235R, L19S/235R; Fig.  2B) of the transactivation potential of the N terminus. Contrary to this result, mutation of the arginine residue at position 2 (R2G/235R) or of the aspartic acid residue at position 24 (D24A/235R) had no influence on the transactivation potential. Both point mutants induced reporter gene expression with E1A 12S wild type efficiency (Fig. 2B). Western blot analyses confirmed the comparable expression level of all E1A 12S mutants in transiently transfected cells (Fig. 8B) excluding that differences of the transactivation potentials were due to insufficient expression of the E1A mutant proteins.
CREB-1 and ATF-1 Bind to the Ad12 E2-CRE-As the E2-CRE is one of the decisive elements for the regulation of the Ad12 E2 promoter, we were interested to identify cellular transcription factors binding to this site. To address this question we performed gel retardation assays. Incubation of nuclear extract prepared from KB cells with a radioactively labeled Ad12 E2-CRE oligonucleotide gave rise to three complexes among which complexes 2 and 3 were the dominant ones (Fig.   3, B and C, lane 2; Fig. 3C is an overexposure of Fig. 3B to clearly demonstrate complex 1). To scrutinize the specificity of these complexes we performed competition experiments with a 10-or 100-fold molar excess of unlabeled competitor as depicted in Fig. 3A and at the top of Fig. 3, B and C. Complexes 1-3 were efficiently competed through a 10-and 100-fold molar excess of the unlabeled Ad12 E2-CRE oligonucleotide (Fig. 3, B and C, lanes 3 and 4) or through a 10-or 100-fold molar excess of an unlabeled oligonucleotide containing the somatostatin CRE, which represents the classical CREB/ATF-binding site (Fig. 3, B and C, lanes 5 and 6; see Ref. 34). The generation of complex 1 was also inhibited in the presence of a 10-and 100-fold molar excess of the unlabeled c-JUN2-TRE, which had been shown to bind predominantly c-Jun/ATF-2 heterodimers (35), whereas complexes 2 and 3 were only modestly affected (Fig. 3C, lanes  9 and 10). In contrast, no competition was observed using a point-mutated CRE oligonucleotide (Fig. 3, B and C, lanes 7 and 8) with a CG to AT mutation in the core binding motif (Fig. 3A).
To examine which transcription factors of the bZIP family are part of the complexes 1-3, we performed antibody perturbation assays. The presence of an anti-ATF-1 mouse monoclonal antibody resulted in a complete loss of complex 3, whereas complex 2 was partially affected (Fig. 3D, lane 5). In addition, at least two supershifted complexes were generated (Fig. 3D, lane 5, supershifted complexes are indicated by arrowheads). A third complex, which is marked by a closed circle in Fig. 3D, lane 5, comigrates with complex 1. Therefore, we were unable to differentiate between a supershifted complex or a stabilization of complex 1 through the anti-ATF-1 antibody. Preincubation with an anti-CREB-1 rabbit polyclonal antiserum which is cross-reactive with ATF-1 prevented the formation of complex 2 and reduced the amount of complex 3 (Fig. 3D,  lane 6). The lack of a supershifted complex might be due to the fact that the anti-CREB-1 rabbit polyclonal antiserum recognizes the DNA-binding domains of CREB-1 and ATF-1 thereby preventing their binding to the E2-CRE. A comparable result was obtained using a mouse monoclonal anti-CREB-1 antibody that specifically recognizes the DNA binding and dimerization domain of CREB although the effect on complex 3 was not as strong as in case of the anti-CREB-1 rabbit polyclonal antiserum (Fig. 3D, lane 7). From these results we conclude that complexes 2 and 3 contain CREB-1 and ATF-1. However, complex 1 contains most probably c-Jun as well as ATF-2 as indicated by the usage of an anti-c-Jun (Fig. 3D, lanes 3 and 9; lane 9 is an overexposure of lane 3 to show the supershift of complex 1) or anti-ATF-2 antibody (Fig. 3D, lanes 4 and 10; lane 10 is an overexposure of lane 4 to show the supershift of complex 1).
As transient expression assays demonstrated that the Ad12 E2-E2F recognition sequence plays a secondary role in the activation of the Ad12 E2 promoter through the E1A 12S protein, the question arises whether this motif is functional and recognized by E2F/DP transcription factors at all. Antibody perturbation assays using a radioactively labeled oligonucleotide spanning the Ad12 E2-E2F site (nt Ϫ91 to Ϫ80 of the Ad12 E2 promoter) in combination with nuclear extract prepared from KB cells in the presence of various antibodies directed against distinct E2F family members or the DP-1 protein revealed that the Ad12 E2-E2F-binding site can be bound by at least an E2F-4/DP-1 heterodimer (data not shown). The reason why this complex does not show a strong effect in the activation of the Ad12 E2 promoter is currently unclear.
CREB-1/ATF-1 Activates the Ad12 E2 Promoter-The results of the gel retardation assays have shown that the Ad12 E2-CRE can be bound by CREB-1/ATF-1 or c-Jun/ATF-2. To show which transcription factor complexes mediate transactivation of the Ad12 E2 promoter in response to E1A 12S , we made use of the DNA-binding defective, dominant-negative inhibitor A-CREB (22). In A-CREB an acidic amphipathic polypeptide was fused to the CREB leucine zipper domain. After forming heterodimers, A-CREB prevents the binding of endogenous CREB-1 and most probably ATF-1 to their target sequences. As CREB-1 and ATF-1 do not heterodimerize with ATF-2 or c-Jun (42,43), A-CREB specifically inhibits the activation potential of the cAMP-dependent transcription factors CREB-1 and ATF-1.
Cotransfection of A-CREB with the E2 Ad12 -(140 bp)-CAT led to a decrease of basal CAT activity in transient expression assays in KB cells (Fig. 4). Interestingly, A-CREB also prevents the activation of CAT gene expression from E2 Ad12 -(140 bp)-CAT through E1A 12S (Fig. 4). To verify that A-CREB does not

Functional Interaction of CREB-1/ATF-1 with E1A 12S
interfere with the transcriptional activation dependent on c-Jun/ATF-2, we cotransfected the A-CREB expression vector with the human c-JUN reporter construct (phc-JUN-(Ϫ1600/ ϩ740)-CAT carrying nt Ϫ1600 to ϩ 740 of the human c-JUN promoter) which can also be activated through E1A 12S (23,35). In contrast to E2 Ad12 -(140 bp)-CAT, neither basal nor E1A 12Smediated activation of CAT gene expression is inhibited from the c-JUN promoter in the presence of A-CREB (Fig. 4).
To confirm these results the CREB-1/ATF-1-specific activator PKA was used. Phosphorylation of CREB-1/ATF-1 through PKA is a key step that enables both cellular transcription factors to interact with CBP which results in a strong induction of target gene expression (12). Transfection of RSV-CHO-PKAc␣ which expresses the catalytic subunit ␣ of PKA gave rise to a 4.2-fold activation of CAT gene expression from E2 Ad12 -(140 bp)-CAT which is comparable to the activation mediated by E1A 12S in this experiment (4-fold; Fig. 5). Expression of E1A 12S in the presence of the catalytic subunit ␣ did not lead to an additive nor to a synergistic activation of the reporter construct under the conditions used (5.3-fold activation; Fig. 5). As expected, coexpression of the PKA-specific inhibitor PKI prevented the activation of the Ad12-E2-(140 bp) promoter through PKA. Interestingly, cotransfection experiments using E1A 12S and PKI also revealed no reporter gene activation (Fig.  5). These findings suggest that both inducers might use overlapping activation mechanisms and strongly support our conclusion that E1A 12S activates the Ad12 E2 promoter through CREB-1/ATF-1 and not through c-Jun/ATF-2 transcription factors.
CREB-1 and ATF-1 Bind E1A 12S -One mechanism how E1A 12S might activate the Ad12 E2 promoter is through binding to CREB-1 and/or ATF-1 allowing the transactivation domains of the adenoviral protein to interact with cofactors or factors of the general transcription machinery. To check whether both cellular transcription factors interact physically with the adenoviral protein, we performed GST pull-down assays with immobilized GST-235R fusion protein and 35 S-labeled in vitro transcribed/translated CREB-1 or ATF-1. In vitro transcription/translation of ATF-1 gave rise to two polypep-tides (Fig. 6B, lane 1) which is most probably due to different phosphorylated forms (44). As shown in Fig. 6, A and B, lanes  3, CREB-1 and ATF-1 efficiently bind to the adenoviral protein.
Deletion of aa 1-79 from the E1A 12S protein prevented the interaction with CREB-1 and ATF-1 (Fig. 6, A and B, lane 6). Moreover, deletion of either aa 1-29 (GST-⌬N/235R; Fig. 6, A  and B, lanes 4) or CR1 (Fig. 6, A and B, lanes 5) strongly reduces the binding capacity of E1A 12S . The GST leader sequence did not bind CREB-1 nor ATF-1 (Fig. 6, A and B, lanes  2). Control experiments using aliquots of the GST pull-down experiments in conventional SDS-polyacrylamide gel electrophoresis followed by Coomassie staining confirmed that comparable amounts of fusion proteins were used for protein/protein interaction assays (compare Fig. 6, A with C and B with D). Therefore these results indicate that E1A 12S binds to CREB-1 and ATF-1 through aa 1-79 and that both the N terminus and To demonstrate that CREB-1 and E1A 12S interact in vivo, we performed immunoprecipitations. Precleared whole cell extracts of COS7 cells transiently transfected with expression vectors encoding for either the E1A 12S wild type or E1A 12S mutant proteins were incubated with an anti-CREB antiserum (Ab240; see Ref. 38), and immune complexes were subjected to Western blotting using an anti-Ad12 E1A antiserum. Two E1Aspecific signals were obtained in the immunoprecipitation experiments (Fig. 7A, lane 3) which are most probably due to different phosphorylated adenoviral polypeptides (45)(46)(47). No E1A-specific signal was obtained in mock-transfected cells (Fig.  7A, lane 2). The mutant ⌬CR1/235R interacted only weakly with CREB-1 in vivo (Fig. 7A, lane 5; indicated by an arrow), whereas no interaction was observed with a 235R N-terminal deletion mutant (⌬N/235R, Fig. 7A, lane 4) or a mutant lacking the N terminus and CR1 (⌬1-79/235R) (Fig. 7A, lane 6). Furthermore, the anti-ATF1 mouse monoclonal antibody (FI-1) coprecipitated E1A 12S (Fig. 7B, lane 3). In contrast, no E1Aspecific signal was obtained with the mutants ⌬N/235R (Fig.  7B, lane 4) and ⌬CR1/235R (Fig. 7B, lane 5). The comparable expression of the E1A 12S full-length protein and derived mutants in transfected COS7 cells was confirmed by Western blotting experiments of aliquots of the extracts used for immunoprecipitations (data not shown and Fig. 8B). These data indicate that E1A 12S binds to CREB-1 and ATF-1 in vivo although we cannot rule out that the interaction is indirect and mediated through a third protein like CBP.
E1A 12S -mediated Activation of the Ad12 E2 Promoter Correlates with Its Ability to Interact with the Cellular Cofactor CBP-CREB-1 fulfills its transactivating function by recruiting the cellular coactivator CBP to the promoter (12). The data that E1A proteins of the oncogenic serotype Ad12 interact with CBP in dependence of the N terminus and CR1 (39) and that both E1A domains are also essential for the E1A 12S -mediated activation of the Ad12 E2 promoter (Fig. 2B) imply therefore that an interaction between CBP and the viral protein might be involved in promoter activation. Consequently we wanted to find out whether the E1A 12S -mediated transcriptional activation of the Ad12 E2 promoter correlates with the interaction of the adenoviral protein with CBP in vivo.
Whole cell extracts of COS7 cells transfected with wild type or mutant E1A 12S expression vectors were prepared, and immunoprecipitations were performed with an anti-CBP rabbit polyclonal antiserum. Immune complexes were analyzed by Western blotting using the anti-Ad12 E1A antiserum (39). As expected, wild type E1A 12S interacted with CBP (Fig. 8A, lane  3), whereas deletion of the E1A 12S N terminus (aa 1-29) prevented the E1A 12S /CBP interaction (⌬N/235R; Fig. 8A, lane 4), and removal of CR1 nearly completely abolished complex formation (⌬CR1/235R; Fig. 8A, lane 5). An E1A 12S mutant protein lacking the N terminus as well as CR1 did not bind CBP (⌬1-79/235R; Fig. 8A, lane 6). In addition, the E1A 12S point mutant D16A/235R did not associate with CBP (Fig. 8A, lane  8), and the point mutants I18P/235R and L19S/235R had almost lost their ability to bind to CBP (Fig. 8A, lanes 9 and 10). In contrast, the point mutants R2G/235R and D24A/235R interacted with CBP like the E1A 12S wild type protein (Fig. 8A,  lanes 7 and 11). All E1A 12S mutant proteins are expressed in comparable amounts as checked by Western blotting of extracts prepared from transfected COS7 cells (Fig. 8B) excluding that the loss of in vivo interaction between CBP and E1A 12S mutants is due to a lack of expression of the E1A 12S mutants. In addition, the E1A 12S /CBP interaction pattern was confirmed in vitro in GST pull-down assays using extracts of HeLa cells and respective GST-235R fusion proteins (data not shown and Ref. 39).
We next investigated if E1A 12S interacts directly with CBP or through an additional factor like CREB-1 or ATF-1. We therefore performed GST pull-down assays using bacterially expressed and purified GST-CBP and His 6 -E1A 12S fusion proteins. Interaction was analyzed by Western blotting with an anti-Ad12 E1A antiserum. These experiments demonstrate that E1A 12S binds directly to full-length CBP (Fig. 8C, lane 3; the full length of the CBP fusion protein was confirmed by Western blotting using an anti-C-terminal CBP antiserum, data not shown). Moreover, mutational analyses revealed that the adenoviral protein binds to the KIX domain of CBP (Fig.  8C, lanes 4 and 5). No binding was detected with the GST leader sequence (Fig. 8C, lane 2). These data clearly show that E1A 12S associates directly with CBP and that the KIX domain within CBP comprises one interaction domain. However, these data do not exclude that E1A 12S binds directly to other CBP domains as well.
A comparison of these interaction data with the transient expression assays in which the E1A 12S mutants were analyzed with respect to their ability to activate the E2 Ad12 -(140 bp)-CAT reporter construct reveals that the binding of E1A 12S to CBP correlates well with the potential of the adenoviral protein to transactivate the Ad12 E2 promoter (Fig. 2 and Fig. 8A; summarized in Fig. 8D). The E1A 12S mutants ⌬N/235R, ⌬CR1/ 235R, ⌬1-79/235R, D16A/235R, I18P/235R, and L19S/235R have either lost their ability to bind CBP or have a greatly reduced binding capacity. These mutants do not or only very weakly (D16A/235R) activate the Ad12 E2 promoter. In contrast, the mutants R2G/235R and D24A/235R bind to CBP and activate the Ad12 E2 promoter like the E1A 12S wild type protein. These data suggest that an interaction of E1A 12S with CBP might be necessary for promoter activation.
CREB-1 and ATF-1 Recruit E1A 12S to the E2-CRE-Our data indicate that activation of the Ad12 E2 promoter includes the assembly of a complex containing CREB-1/ATF-1 and E1A 12S on the E2-CRE. This possibility was tested by using a DNA-dependent protein/protein interaction assay in which a 5Ј-biotinylated oligonucleotide spanning the Ad12 E2-CRE recognition motif was incubated with in vitro translated, unlabeled CREB-1 or ATF-1 and in vitro translated, 35 S-labeled E1A 12S . Protein-DNA complexes were collected using neutravidin-agarose and analyzed on SDS-polyacrylamide gels followed by fluorography. These experiments showed that both CREB-1 and ATF-1 recruit E1A 12S to the E2-CRE in vitro (Fig.  9A, lanes 6 and 7). In contrast, E1A 12S gave rise to only a very faint background signal in the absence of the cellular transcription factors (Fig. 9A, lane 5) which is in agreement with data showing that adenoviral E1A proteins do not bind sequence specific to DNA (8). Control experiments revealed that this interaction assay was strictly dependent on the E2-CRE oligonucleotide because neither CREB-1 nor ATF-1 were pulled down in its absence (Fig. 9A, compare lanes 3 and 4 with lanes  1 and 2).
The assembly of CREB-1/ATF-1, E1A 12S , and the CRE to a ternary complex prompted us to investigate if the addition of CBP allows the formation of a quaternary complex. Addition of FIG. 8. Activation of the Ad12 E2 promoter through the E1A 12S protein correlates with the binding of the adenoviral protein to the cellular coactivator CBP. A, the E1A 12S protein binds to CBP in dependence of the N terminus and CR1. COS7 cells were transiently transfected with either the wild type or mutant E1A 12S expression vectors as indicated. 72 h later cells were lysed, and precleared extracts were incubated with an anti-CBP (A-22) rabbit polyclonal antiserum. Immune complexes were analyzed on 13% SDS-polyacrylamide gels followed by Western blotting using the anti-Ad12 E1A antiserum.  5). Bound proteins were eluted and analyzed on 10% SDS-polyacrylamide gels followed by Western blotting using an anti-Ad12 E1A antiserum. Lane 1 represents 10% input of His 6 -E1A 12S . D, the transactivation of the Ad12 E2 promoter through the E1A 12S protein correlates with the binding of the adenoviral protein to CBP. ϩϩϩ, strong binding of the E1A 12S protein or E1A 12S mutants to CBP and strong transactivation of the Ad12 E2 promoter through E1A 12S or E1A 12S mutants; ϩ, very weak binding of an E1A 12S mutant to CBP and very weak transactivation of the Ad12 E2 promoter through an E1A 12S mutant; Ϫ/ϩ, marginal binding of the E1A 12S mutants to CBP (indicated by arrows in A); Ϫ, no interaction and transactivation detected. The transactivation of the Ad12 E2 promoter through the E1A 12S wild type/mutant proteins is shown in Fig. 2B. in vitro translated, 35 S-labeled CBP did not prevent the recruitment of E1A 12S through CREB-1 to the E2-CRE (Fig. 9B, lane  6) although identical E1A 12S domains are involved in the binding to the cofactor (Fig. 8A, lanes 4 -6) and to CREB-1 (Fig. 6A). Instead, quantitative evaluation using the LKB UltroScan XL Laser Densitometer revealed that the amount of E1A 12S bound to the CREB-1⅐DNA complex is increased 2-4-fold in the presence of CBP (Fig. 9B, compare lane 6 with lane 5 and data not shown) indicating that CBP might stabilize the CREB-1⅐E1A 12S complex or recruit additional E1A 12S to the E2-CREbound complex. CBP itself is unable to bind to DNA (Fig. 9B, lane 2) but is recruited through CREB-1 to the E2-CRE (Fig.  9B, lane 4). Taken together, these data suggest that E1A 12S is recruited by CREB-1/ATF-1 to the E2-CRE and that CBP either stabilizes this complex or tethers additional E1A 12S to this complex. DISCUSSION Expression of region E2 is of crucial importance for propagation of Ads as this region codes for proteins essential for viral DNA replication (48). Like many other adenoviral promoters the E2 promoter contains a CRE that is essential for E2 expression. However, the molecular mechanism how E1A facilitates promoter activation through CREs is poorly understood. Here, we show that the E1A 12S protein of oncogenic Ad12 activates the Ad12 E2 promoter through a cAMP-response element. This element is bound by CREB-1/ATF-1 transcription factors that in turn recruit E1A 12S to the E2 promoter. Most interestingly, the transactivating potential of E1A 12S correlates with its ability to bind to the CREB cofactor CBP implicating a critical role for CBP in the E1A 12S -mediated E2 promoter activation. This assumption is supported by our finding that CBP activates the Ad12 E2 promoter in the presence of cotransfected PKA (39). A prerequisite for the binding of CBP to CREB-1 is the PKA-mediated phosphorylation of the cellular transcription factor at Ser-133 (12). The function of E1A 12S might therefore be to tether CBP to the promoter in the absence of phospho-CREB bypassing phosphorylation through PKA. However, our data showing that the potent PKA-specific inhibitor PKI inhibits E1A 12S -mediated E2 activation argues against such a mechanism as it indicates the importance of phospho-CREB for promoter activation. Therefore, CBP might associate with the E2-CRE-bound phospho-CREB⅐E1A 12S complex. In this model, CBP stabilizes the DNA-bound complex containing CREB-1/ATF-1 and the adenoviral protein by contacting both protein components. The observation that the amount of E1A 12S bound to CREB-1/ATF-1 is increased in the presence of CBP in vitro supports this hypothesis. On the other hand, these data do not exclude the possibilities (i) that E1A 12S binds to a preformed phospho-CREB⅐CBP complex or (ii) that CBP recruits additional E1A 12S molecules to the promoter, which might be concluded from our finding that E1A 12S binds directly to the CBP KIX domain. In these mechanisms, the combined action of CREB-1/ATF-1, E1A 12S and CBP is then responsible for the strong activation of the E2 promoter. The function of CBP might be to remodel chromatin structure through its intrinsic HAT activity thereby increasing the accessibility of general transcription factors (49). At the same time E1A 12S contacts factors of the general transcription machinery like TBP and/or the RAP30 subunit of TF II F. Both factors are important for promoter activation (50), and both factors bind to the N-terminal transactivation domain of E1A 12S (51). A second function of E1A 12S in this complex might be the induction of the CBP HAT activity. Recently, it has been shown that E1A 12S of Ad2/Ad5 enhances the HAT activity of CBP (52). Moreover, an E1A⅐CBP complex precipitated from Ad12-transformed mouse cell lines still displays HAT activity. 2 On the other hand, the HAT activity of CBP seems not to be involved in the transcriptional activation of all CREB-CBP-dependent target genes (53). Maybe the HAT activity is not involved, but one of the other transactivation domains of CBP located in the N or C terminus (53,54), which are targeted by E1A 12S (55), might be involved in the activation of the Ad12 E2 promoter. Currently, we are investigating whether Ad12 E1A 12S induces CBP HAT activity on the Ad12 E2 promoter or whether the adenoviral protein makes use of one of the other transactivation domains of this cellular coactivator to induce E2 expression.
In addition to the CRE, the Ad12 E2 promoter contains a single E2F-binding site that is bound by the potent activator E2F4/DP-1, at least in vitro. Surprisingly, destruction of the E2F site by point mutation only slightly affects basal and E1A 12S -dependent Ad12 E2 promoter activity in transient expression assays in KB cells demonstrating that promoter activation is primarily mediated through the CRE. In this respect, 2 P. Fax and D. Brockmann, unpublished data. the Ad12 E2 promoter differs significantly from the respective promoter of non-oncogenic Ad2 (Ad2 E2e). The Ad2 E2e promoter represents the classical model system for studying the regulation of E2F-dependent gene expression (48). In contrast to the Ad12 E2 promoter, Ad2 E2e contains two inverted E2F and one ATF transcription factor binding sites. All three ciselements are critical for Ad2 E1A-mediated transactivation. In addition, transcriptional activation is enhanced by the open reading frame 6/7 protein of the viral region E4 that enables the cooperative and stable binding of two E2F transcription factors. The factors binding to the Ad2 E2e ATF site, however, are not yet clearly defined, and cAMP inducibility of the Ad2 E2e promoter seems to be dependent on the cell system analyzed (19,56,57). In this context it is noteworthy that the Ad12 E2 promoter is inducible by cotransfected PKA in other cell systems as well, like F9 or HeLa cells (data not shown), indicating that the cAMP dependence of the Ad12 E2 promoter is most probably independent of the cell system.
The data presented here extend our understanding of the general mechanism how E1A 12S activates or represses CREdependent gene expression through the same set of cellular factors. Promoter repression might be attributed to the disruption of the CREB-1⅐CBP complex by E1A 12S as postulated in case of the somatostatin promoter (16). In contrast, activation might occur through recruiting E1A 12S to the promoter and generating a quaternary complex containing CREB-1/ATF-1, E1A 12S , CBP, and the E2-CRE. Consequently, the additional interaction of E1A 12S with other factors, e.g. sequence-specific and/or general transcription factors, which stabilize the complex on the CRE and thus prevent the E1A 12S -mediated dissociation of phospho-CREB/CBP, is essential for promoter activation. Such a mechanism might be correct for the activation of the viral E2 promoter and the human PCNA promoter, although in latter case the generation of a quaternary complex has not been demonstrated (17). The identification of these auxiliary factors that allow complex formation on the E2-CRE will therefore further enlighten the transactivation mechanism in the presence of the adenoviral E1A 12S protein.