Characterization of an E1A-CBP Interaction Defines a Novel Transcriptional Adapter Motif (TRAM) in CBP/p300

Abstract

The adenovirus E1A protein subverts cellular processes to induce mitotic activity in quiescent cells. Important targets of E1A include members of the transcriptional adapter family containing CBP/p300. Competition for CBP/p300 binding by various cellular transcription factors has been suggested as a means of integrating different signalling pathways and may also represent a potential mechanism by which E1A manipulates cell fate. Here we describe the characterization of the interaction between E1A and the C/H3 region of CBP. We define a novel conserved 12-residue transcriptional adapter motif (TRAM) within CBP/p300 that represents the binding site for both E1A and numerous cellular transcription factors. We also identify a sequence (FPESLIL) within adenovirus E1A that is required to bind the CBP TRAM. Furthermore, an E1A peptide containing the FPESLIL sequence is capable of preventing the interaction between CBP and TRAM-binding transcription factors, such as p53, E2F, and TFIIB, thus providing a molecular model for E1A action. As an in vivo demonstration of this model, we used a small region of CBP containing a functional TRAM that can bind to the p53 protein. The CBP TRAM binds p53 sequences targeted by the cellular regulator MDM2, and we demonstrate that an MDM2-p53 interaction can be disrupted by the CBP TRAM, leading to stabilization of cellular p53 levels and the activation of p53-dependent transcription. Transcriptional activation of p53 by the CBP TRAM is abolished by wild-type E1A but not by a CBP-binding-deficient E1A mutant.

The major role of the adenovirus E1A protein is to induce cell entry into the S phase of the cell cycle, thus providing an optimal environment for viral replication (48). The biological activity of E1A is dependent on interactions with a number of cellular proteins, most notably, the pocket-containing proteins characterized by the Rb tumor suppressor protein and members of the CBP/p300 family of transcriptional coactivators (22, 34).

CBP and the closely related protein p300 are thought to play a fundamental role in a variety of signal-modulated cellular events (20, 31, 38). First described as a coactivator of the cyclic AMP-responsive regulator CREB (8), CBP has since been shown to interact with a large and diverse set of transcription factors (21).

The activation of gene expression by CBP has been attributed to two very different properties. First, CBP interacts with proteins that contain acetyltransferase (AT) activity (55) and possesses intrinsic AT activity (3, 41) capable of modifying histones (3, 41) and nonhistone transcription factors (12, 19). This intrinsic AT activity of CBP was very recently shown to be directly involved in stimulating gene transcription (32). For histone acetylation, activation has been suggested to result from an increase in the accessibility of the target promoter to transcription factors (27, 37). A second mechanism by which CBP has been suggested to activate transcription is by bridging the gap between DNA-bound transcription factors and components of the general transcription machinery. CBP is a large protein (2,441 amino acids), and its ability to interact simultaneously with a number of factors has led to its description as a transcriptional adapter protein (20). An important insight into how CBP might activate transcription as an adapter molecule was recently provided when CBP was shown to activate CREB-dependent expression through an interaction with RNA helicase A, a component of an RNA polymerase II complex (36). Both of the above mechanisms of activation depend on the recruitment of CBP to a particular promoter by specific protein-protein interactions.

Evidence from studies of CBP-associated disease (11, 43) and knockout mice (50, 56) suggests that cellular levels of CBP may be rate limiting. Support for such an idea comes from observations that different signal transduction pathways with a mutual dependence on CBP antagonize one another, but not when intracellular CBP levels are artificially raised. These observations were first noted for the repression of AP-1 activity by nuclear receptors (23), and subsequent examples have been described for Stat2 and NFκB (18) and the AP-1 and JAK/STAT pathways (17). Together, these observations have led to the proposals that CBP acts as an integrator of different signalling pathways and that sequestration of CBP by particular transcription factors functions to select which set of genes is to be expressed (6, 23).

This integration model for the regulation of signal transduction pathways is also consistent with results obtained from studies with the adenovirus E1A protein. An increasing body of evidence suggests that one mechanism by which E1A functions to subvert cellular processes is to displace cellular transcription factors from CBP (2, 10, 13, 28, 29, 45, 49). Consistent with this idea is the initial observation that E1A binds a region of CBP (between amino acids 1621 and 1877 and often referred to as the C/H3 region) (9) that is recognized by a number of important transcriptional regulators, including p53 (1, 13, 29, 49), E2F (51), TFIIB (25), MyoD (45, 47, 57), RNA helicase A (36), and P/CAF (55).

We wished to establish a molecular basis for the mechanism by which E1A controls the CBP-dependent integration of signal transduction pathways. We therefore set about defining precisely the sequences responsible for the E1A-CBP C/H3 interaction. From this analysis, we have now defined an interaction motif within the CBP C/H3 region that is responsible for binding E1A as well as a number of diverse transcription factors. This transcriptional adapter motif (TRAM) is conserved in all members of the CBP/p300 family of proteins. Moreover, we present data showing that the sequence FPESLIL in E1A is required for E1A binding to the CBP TRAM and for the ability of an E1A peptide to competitively inhibit the interaction of cellular transcription factors with the CBP TRAM both in vitro and in vivo.

MATERIALS AND METHODS

Plasmids and fusion proteins.

Glutathione S-transferase (GST)–CBP constructs were created by cloning PCR-amplified fragments or double-stranded oligonucleotides into pGEX2TKP (a modified version of the Pharmacia vector pGEX2TK that contains a new polylinker). GST-E1A constructs contained either wild-type sequences from adenovirus type 12 (Ad12) E1A residues 14 to 72 or sequences containing alanine substitutions.

The GST-MDM2 (residues 1 to 125) construct was a gift from B. Li. GST fusion proteins were expressed in Escherichia coli, extracted with lysis buffer (50 mM Tris-HCl [pH 8.0], 0.5 mM EDTA, 5 mM dithiothreitol, 15% glycerol, 1 mg of lysozyme per ml, 1 mM phenylmethylsulfonyl fluoride), sonicated, centrifuged, and stored at −70°C.

Detection of protein-protein interactions with microaffinity columns.

Bacterial lysate containing GST fusion proteins was incubated with glutathione-Sepharose beads (Pharmacia) for 30 min at 4°C in 1× NENT buffer (100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, Tris-HCl [pH 8.0]). After being centrifuged and washed with 1 ml of 1× NENT, the beads were loaded into a yellow Gilson pipette tip containing a glass bead (BDH; catalog no. 332134Y) to create a 25-μl GST microaffinity column. In vitro transcription and translation (IVT) of proteins incorporating ³⁵S-methionine were performed with TNT kits (Promega) in accordance with the manufacturer’s recommendations. Forty microliters of a 50-μl IVT reaction mixture was diluted with 360 μl of IPD buffer (50 mM KCl, 40 mM HEPES [pH 7.5], 5 mM 2-β-mercaptoethanol, 0.1% Tween 20, 0.5% milk) before being passed through the GST microaffinity column. After the column was washed twice with 200 μl of wash buffer (IPD buffer containing 150 mM KCl), proteins were eluted from the column by adding 25 μl of 2× sodium dodecyl sulfate (SDS) loading dye, heating to 95°C for 5 min, chasing with 25 μl of water, and centrifuging in a microcentrifuge.

Peptide competition assays.

To study the influence of specific peptides on protein-protein interactions, GST fusion proteins were bound to glutathione-Sepharose beads as described previously. For E1A peptide competition, the washed beads were incubated (rotated) in 200 μl of IPD buffer containing a peptide (final concentration, 10 μM to 100 mM) at 4°C for 1 h. Forty microliters of an in vitro translation reaction mixture was diluted with 60 μl of IPD buffer and added to the sample before further incubation for 30 min at 4°C. The glutathione-Sepharose beads were centrifuged and washed as described above before the bound proteins were eluted by heating at 95°C for 5 min in 50 μl of 1× SDS loading dye. For the CBP TRAM peptide competition studies, the peptide was preincubated with diluted in vitro translation reaction mixture for 15 min at 4°C before being incubated with GST fusion proteins.

Transfections and chloramphenicol AT assays.

U-2 OS cells were plated onto 10-cm-diameter culture dishes and transfected at 50 to 70% confluence with Lipofectin reagent (GIBCO-BRL) as described previously (39). Chloramphenicol AT assays have been described previously (39), and the data presented represent between 3 and 10 independent transfection experiments.

p53 stability experiments.

U-2 OS cells were transfected with 0.1 μg of CMV-p53, 2 μg of either CMV-MDM2 or empty cytomegalovirus (CMV) expression vector, and 5 μg of either CMV-GST-CBP (residues 1808 to 1852), CMV-GST-CBP (residues 1808 to 1852) Mut, or empty GST expression vector. At 24 h after transfection, cells were shifted to medium containing 25 μg of cycloheximide per ml and were harvested at various times.

RESULTS

E1A binds a 12-amino-acid motif in the CBP C/H3 region.

Many transcription factors that interact with CBP bind the same 257-amino-acid, cysteine-rich region (C/H3 region) between residues 1621 and 1877 (Fig. 1A). The adenovirus E1A protein has also been shown to bind this region, and it has been proposed that E1A might repress the activity of these cellular factors by competing for binding to CBP. We therefore sought to establish the binding site for E1A within CBP in the hope of providing some insight into the molecular mechanism of E1A action.

FIG. 1.

E1A binds a 12-amino-acid motif in CBP (amino acids 1811 to 1822). (A) Schematic representation of CBP and some of the CBP-interacting proteins. (B) Use of GST-CBP fusion constructs to define sequences capable of binding 12S E1A. Microaffinity column experiments are described in Materials and Methods. Approximately 10% of the E1A translation reaction mixture was run in the input lane, the GST lane represents a control column, and lanes 1 to 9 represent the eluate obtained after passage of E1A over columns containing the nine GST-CBP fusion constructs. (C) Further deletion analysis of CBP amino acids 1808 to 1826. Thick lines indicate constructs that bind E1A; thin lines those that do not. A minimal construct of 12 amino acids (construct 10; CBP amino acids 1811 to 1822) still retains E1A-binding activity.

Using glutathione-Sepharose microaffinity columns and a series of GST-CBP fusion proteins, we initially identified a 19-amino-acid region of CBP (1808 to 1826) that was sufficient for the binding of E1A (Fig. 1B, lane 8). Deletion into this region abolished E1A binding (Fig. 1B, lane 9). Further fine-deletion analysis of CBP (amino acids 1808 to 1826) identified a 12-residue sequence (1811 to 1822) that was sufficient for E1A binding (Fig. 1C), although this sequence bound with a slightly lower affinity than the larger, 19-residue sequence (1808 to 1826). A comparison of analogous regions of CBP/p300 from a number of species revealed a high degree of conservation of this sequence. We have therefore termed the sequence between amino acids 1811 and 1822 of CBP a transcriptional adapter motif (TRAM).

Identification of amino acids within E1A required for the interaction with the CBP TRAM.

Having identified the E1A-binding region within the C/H3 domain of CBP, we were interested in identifying the residues in E1A responsible for binding the CBP TRAM. Previous dissection of E1A implicated residues 63 to 67 in binding the C/H3 domain of CBP. Figure 2 represents an alanine substitution analysis of this E1A region and shows that the sequence FPESLIL could be defined as essential for binding to CBP (amino acids 1621 to 1877). Mutagenesis of any two residues within this sequence (FE, PS, EL, SI, or LL) to alanine abolished binding to CBP. However, while single-residue substitutions E67A, S68A, and L69A resulted in a small reduction in CBP binding, no single substitution was able to prevent the E1A-CBP complex, indicating that the interaction between these proteins is dependent upon a combination of residues. The mutations that altered residues outside the FPESLIL sequence (F64P66 and E63F65) resulted in a very slight increase in binding over that seen with the wild-type E1A sequence. However, corresponding variations in the loading of the GST-E1A proteins suggest that this difference in binding affinity is probably not significant.

FIG. 2.

Identification of the amino acids in E1A involved in the interaction with CBP. (Top) Schematic representation of E1A. Contained within conserved region 1 (CR1) are residues 63 to 67, previously implicated in the binding of CBP. GST-E1A proteins containing wild-type or mutated (alanine-substituted) sequences were tested for their ability to interact with ³⁵S-labelled CBP (amino acids 1621 to 1877). Double substitution mutants within sequences F65 to L71 failed to bind CBP. (Bottom) Coomassie blue-stained SDS-polyacrylamide gel revealing quantities of GST-E1A fusion proteins recovered from the microaffinity columns. WT, wild type.

The peptide competition studies depicted in Fig. 3 confirmed the importance of the FPESLIL residues for the interaction between E1A and the CBP TRAM. Peptides containing wild-type or mutant E1A sequences were analyzed for their ability to prevent the binding of full-length radiolabelled 12S E1A protein to a GST-CBP fusion protein. While E1A binding was detected in the absence of a competitor peptide, increasing amounts of the wild-type E1A peptide abolished the interaction. In contrast, a peptide containing mutations in E67L69 (peptide Mut 1) was abrogated in this capacity. Additional mutations (in F65L71) within the FPESLIL residues (peptide Mut 2) resulted in a corresponding reduction in the ability of this peptide to inhibit the E1A-CBP TRAM interaction. Together, these results demonstrate that the interaction of E1A with the CBP TRAM involves this small E1A region from residues F65 to L71. It should be noted that previously published studies have also suggested the importance of L20 and R2 in the binding of p300 by E1A (53, 54). In our E1A mutagenesis studies and E1A peptide competition experiments, the L20 residue, but not the R2 residue, was present and contributed to the stability of the E1A-CBP TRAM interaction (40). Thus, while residues F65 to L71 play a critical role in the interaction with the CBP TRAM, L20 and other residues may also play an important role by stabilizing the interaction with the CBP TRAM.

FIG. 3.

Peptide competition assays confirming the involvement of sequences F65 to L71 in the interaction with CBP. Peptides of 30 amino acids and containing the sequences shown were used in competition studies (described in Materials and Methods). Only the wild-type (WT) peptide prevented the E1A-CBP interaction.

Peptides containing the FPESLIL sequence of E1A can prevent the interaction of cellular factors with the CBP TRAM.

The CBP TRAM is within a region of CBP (1621 to 1877) that is a “hot spot” for the binding of transcription factors (Fig. 1A). We wished to know whether the TRAM was also the target of these interactions. Figure 4A shows that three transcription factors, namely, p53, E2F-1, and TFIIB, that bind this region of CBP were able interact with the CBP TRAM (GST-CBP, 1808 to 1826). Figure 4A also demonstrates that the interaction of these cellular factors with the CBP TRAM could be prevented in competition experiments by an E1A peptide containing the wild-type CBP-binding site but not by a mutant E1A peptide that was unable to bind CBP (Mut 2).

FIG. 4.

Multiple cellular factors bind to the CBP TRAM and can be inhibited by an E1A FPESLIL-containing peptide. (A) p53, E2F, and TFIIB all bind to CBP amino acids 1808 to 1826, containing the TRAM. Binding can be inhibited by competition with the wild-type (WT) E1A peptide but not the Mut 2 (Mut) E1A peptide. −, no competitor. (B) Alignment of adenovirus E1A, p53, and E2F sequences showing the conservation (boldface) of FXE/DXXXL residues implicated in the interaction with the CBP TRAM.

Since E1A, p53, E2F, and TFIIB all recognize the same small motif within CBP, we wondered whether or not these proteins might have common sequences that facilitate the recognition of the CBP TRAM. In fact, mutagenesis studies have already defined residues in two of these transcription factors (p53 and E2F) that are required for the interaction with CBP (13, 14, 16, 30, 52). These residues in p53 and E2F show marked similarity to those identified in this study for the adenovirus type 12 E1A protein (and those of adenovirus type 2 and type 5) (Fig. 4B). In each case, a conserved sequence (FXE/DXXXL), when mutated, resulted in an inability to bind CBP. These findings provide a possible model to explain how E1A might regulate the activity of certain cellular transcription factors: that is, by utilizing the FXE/DXXXL sequence and competing for the same CBP-binding site.

A small region of CBP (amino acids 1808 to 1852) containing a functional TRAM is sufficient to bind to p53, stabilize cellular p53 levels, and activate p53-dependent transcription.

Recently, a number of reports have demonstrated that the expression of full-length CBP activates p53-dependent transcription in vivo and that this activation is abolished by E1A (1, 13, 29, 49). Our results shown in Fig. 4A demonstrated that p53 binds the CBP TRAM and that this interaction is disrupted in vitro by an E1A peptide containing a wild-type FPESLIL sequence but not a mutated version. We wished to extend our in vitro studies and investigate this particular interaction in more depth in vivo, since p53 is a potentially important target for adenovirus E1A function.

Cellular p53 protein levels and p53 transcriptional activity are, under normal circumstances, regulated by the MDM2 protein (26), since the binding of MDM2 to p53 results in both the repression of p53 transactivation capacity (33, 42) and the degradation of the p53 protein (15, 24). A previous study has shown that a double point mutation within the p53 FSDLWKLL sequence abolishes the binding of p53 to the C/H3 region of CBP (13). Interestingly, dissection of MDM2 has identified at the MDM2 N terminus a p53-binding site that recognizes p53 residues overlapping the p53 FSDLWKLL sequence (7, 44). In order to investigate whether identical contacts were made with p53 by MDM2 and the CBP TRAM, we tested whether a previously described p53 mutant (L14Q/F19L) that abolishes the MDM2 interaction had an effect on the binding of the CBP TRAM. As shown in Fig. 5A, the p53 mutant L14Q/F19L (30), while abolishing the binding of the MDM2 N terminus (residues 1 to 125), did not affect the binding of the CBP TRAM contained within residues 1715 to 1852. This result demonstrates that the MDM2 N-terminal domain and the CBP TRAM recognize overlapping but distinct residues within the p53 activation domain.

FIG. 5.

The CBP TRAM and the N-terminal domain of MDM2 recognize distinct but overlapping residues on p53. (A) Differential effects of the p53 mutant L14Q/F19L (p53 mut 14/19) on the MDM2 N-terminal domain and the CBP TRAM. Binding of p53 to the N-terminal domain of MDM2 was inhibited by the L14Q/F19L mutation, while binding to the CBP TRAM remained unaffected. (B) A CBP TRAM peptide can inhibit the binding of the N-terminal MDM2 domain to p53. CBP peptides of 27 amino acids (1806 to 1832) and containing either wild-type (WT) or mutant (Mut) (R1811, K1812, and N1814) TRAM sequences were used in competition assays to prevent the interaction of in vitro-translated p53 and GST-MDM2 (1 to 125). The wild-type peptide completely inhibited the p53-MDM2 interaction over the concentration range used (10 to 100 μM), while the ability of the mutant TRAM peptide to inhibit the interaction was severely impaired.

Since distinct residues are involved in the interaction with p53, we tested whether or not the binding of the CBP TRAM and the binding of MDM2 were mutually exclusive events. Figure 5B shows the ability of a peptide containing the CBP TRAM (residues 1806 to 1832) to successfully inhibit the binding of the MDM2 N-terminal domain (residues 1 to 125) to p53 in vitro. This competition was dependent upon an intact TRAM, since a peptide containing mutations in this motif (R1811A, K1812A and N1814A) that inhibit the binding of E1A (40) was severely abrogated in this capacity.

As the CBP TRAM inhibited the MDM2-p53 interaction in vitro, we next tried to establish if the CBP TRAM was able to reverse the functions of full-length MDM2 in vivo. Recently, the binding of MDM2 to p53 has been shown to result in the degradation of p53 (15, 24). Figure 6 shows that indeed, as reported, the introduction of MDM2 into U-2 OS cells resulted in a rapid degradation of p53. However, the simultaneous expression of the CBP TRAM sequence (residues 1808 to 1852) disrupted this effect. In contrast, a mutant version of the CBP TRAM that was compromised in MDM2 displacement (Fig. 5B) was unable to reverse the MDM2-mediated degradation of p53. These results support the conclusion that the CBP TRAM contained within residues 1808 to 1852 is capable of displacing MDM2 from p53 in vivo and, as a consequence, stabilizes cellular p53 protein.

FIG. 6.

Introduction of the CBP TRAM, but not a mutant version, into U-2 OS cells stabilizes cellular p53 levels. Transfection of U-2 OS cells and p53 stability experiments are described in Materials and Methods. The wild-type [CBP (1808-52)] and mutant [CBP (1808-52) Mut] proteins were expressed at similar levels (data not shown). Only the CBP protein containing the wild-type TRAM was able to alleviate the MDM2-mediated degradation of cellular p53.

We next wished to determine whether or not the stabilization of cellular p53 levels resulted in an increase in p53 transactivation potential, as has previously been reported (4). Figure 7 shows that expression of the CBP TRAM (residues 1808 to 1852) could stimulate the activity of a p53-responsive promoter (PG13CAT) in a dosage-dependent manner, whereas a TRAM mutant was compromised in p53 stimulation. Thus, the expression of a small domain containing the CBP TRAM is sufficient to functionally compete for the binding of MDM2 to p53 and supports a model in which MDM2 and the CBP TRAM occupy overlapping and mutually exclusive sites on p53 (Fig. 8).

FIG. 7.

A CBP fragment (residues 1808 to 1852) containing a functional TRAM can stimulate p53-dependent transcription. Transient transfection of U-2 OS cells was carried out with 2 μg of either a vector containing a p53-responsive promoter (PG13CAT) or a control vector (MG15CAT). Indicated is the cotransfection of 1, 2, or 4 μg of a CMV-GST-CBP (residues 1808 to 1852) vector containing either a wild-type TRAM or a mutant (Mut) version. Introduction of the wild-type TRAM resulted in a dose-dependent increase in p53-dependent transcription. Cotransfection of wild-type (WT) 12S E1A but not a CBP-binding-deficient mutant (del 63-67) of 12S E1A abolished the activation of p53, while no activation was obtained with the mutant TRAM construct. Error bars indicate standard error. CAT, chloramphenicol AT.

FIG. 8.

Model for the activation of p53-dependent transcription by the CBP TRAM. (A) p53 under normal physiological conditions is regulated by the MDM2 protein, which binds p53 and facilitates p53 degradation. (B) The expression of full-length CBP in vivo activates p53, possibly through multiple pathways. These may include the stabilization of the binding of p53 to its cognate DNA recognition sites through the acetylation of p53, the bridging of DNA-bound p53 and components of basal transcription factors, and the displacement of MDM2 from p53, preventing p53 degradation. (C) A small CBP fragment containing a functional TRAM can lead to the stabilization of p53 protein and the activation of p53-dependent transcription. This diagram suggests a role for the competitive inhibition of MDM2 binding to p53 during the activation of p53 by full-length CBP.

E1A containing the wild-type FPESLIL sequence can counteract p53 transcriptional activation by the CBP TRAM.

Also demonstrated in Fig. 7 is the ability of wild-type 12S E1A protein to abolish the CBP TRAM-mediated activation of p53. This result is consistent with studies showing that the activation of p53 by full-length CBP is also abolished by E1A (13, 29, 49). Moreover, cotransfection of the del 63-67 mutant of E1A (54) demonstrated that abrogation of this activation was dependent on a functional FPESLIL sequence. Together, these results provide evidence that both the activation of p53 transcriptional activity by CBP and the ability of E1A to inhibit this activation are based on FXE/DXXXL-TRAM interactions. Consequently, our in vivo evidence confirms the importance of the FXE/DXXXL-TRAM interactions demonstrated in our in vitro binding assays.

DISCUSSION

The identification of a small TRAM within the C/H3 region of CBP explains to a large extent why this region represents a hot spot for transcription factor binding. In addition to interacting with E1A, p53, E2F, and TFIIB, the CBP TRAM also binds to the cellular transcription factors MyoD, YY1, c-fos, c-jun, and P/CAF (40). In principle, E1A could inhibit the binding of these and other factors to the CBP TRAM, whether or not they interact with residues similar to the FXE/DXXXL sequence. At least for MyoD (47) and YY1 (40), sequences similar to those in E1A, p53, and E2F have been shown to be necessary for the binding of CBP. In MyoD, the sequence has been referred to as the FYD motif, which is present in the N-terminal region of a number of myogenic transcription factors (36).

The modification and/or concentration of any given protein interacting with the CBP TRAM is likely to be important because the increased binding of one transcription factor may alter cellular responses by competing for a limiting TRAM-containing regulator, as has been suggested by recent studies (17, 18, 23). This strategy would appear to have been adopted by adenoviruses, which express high levels of the FXE/DXXXL-containing E1A protein; this protein competitively inhibits the binding of CBP to cellular proteins, resulting in both the inhibition of cellular differentiation and the activation of cell proliferation. Our characterization of the E1A-CBP interaction now provides a molecular basis for these observations.

One cellular regulator targeted by E1A in this way is p53. Previous studies have demonstrated that p53-dependent transcription is activated by CBP/p300 and that this activation is abrogated by E1A. Stimulation of p53 activity by CBP/p300 has been attributed to the properties of these transcriptional adapters, namely, the ability to provide AT activity (12) and/or to bridge the gap between p53 and components of the basal transcription machinery. In this study, we show for the first time that binding per se is an important part of the mechanism by which CBP/p300 activates p53-dependent transcription. Consequently, E1A binding to the CBP TRAM and the concomitant prevention of a p53-CBP interaction may be sufficient to explain the abrogation of CBP-mediated p53 transcriptional activity.

The mechanism by which p53-dependent transcription is activated by a relatively small region of CBP (residues 1808 to 1852) does not appear to involve either the acetylation of p53 (since CBP sequences responsible for AT activity are not present) or bridging to components of the basal transcription machinery. Rather, activation results from the competitive inhibition of MDM2 binding to p53 and the resulting stabilization of cellular p53 protein levels (as demonstrated in Fig. 6). Similar observations were made in vivo when the MDM2-p53 interaction was blocked by an alternative mechanism: that is, through the expression of a small molecule (thioredoxin) that contains the MDM2-binding domain of p53 in its active-site loop (4). From these observations, we could predict that any protein or protein fragment that inhibits the binding of full-length MDM2 to p53 should be able to activate p53-dependent transcription. We tested this hypothesis by expressing the N-terminal p53-binding domain of MDM2 that lacks the sequences required to target p53 for degradation (24). As predicted, the expression of the MDM2 N-terminal region from residues 1 to 125 in U-2 OS cells resulted in an almost identical level of activation of p53-dependent transcription as the expression of the CBP TRAM fragment (residues 1808 to 1852) (40).

Interestingly, E1A is not the only viral oncoprotein to bind the CBP TRAM. We have recently shown that human papillomavirus E6 proteins also target CBP/p300 and can interact directly with CBP amino acids 1808 to 1826 (58). This interaction results in the down-regulation of p53 transcriptional activity to a level comparable to E1A-mediated repression and is limited to E6 proteins from human papillomaviruses associated with cervical cancer (58). Moreover, given that the simian virus 40 large T antigen also binds the C/H3 region of CBP (9a) and down-regulates p53-dependent transcription (32a), it is likely that all three of these small DNA tumor virus oncoproteins target the CBP TRAM, strongly suggesting an important role for this motif in cell cycle regulation.

Other inhibitors of the cell cycle that bind the CBP/p300 TRAM and have been shown to be targeted by adenovirus E1A include MyoD and P/CAF. The interaction of MyoD with CBP/p300 has been shown to be essential for cell cycle arrest and muscle-specific gene expression (45). E1A, by binding to CBP/p300, inhibits these processes (5, 35). Our results suggest this activity may involve an E1A FPESLIL-TRAM interaction. Similarly, our unpublished finding that P/CAF binds CBP amino acids 1808 to 1826 suggests another TRAM interaction targeted by E1A. Consistent with this idea is the previously described ability of E1A to displace P/CAF from the C/H3 region of CBP (55). Like MyoD and p53, P/CAF possesses cell cycle inhibition and cellular differentiation properties (46, 55). Thus, by targeting the CBP/p300 TRAM, E1A may affect multiple cellular factors whose role it is to inhibit cell cycle progression.

Future studies that make use of TRAM mutants in the context of full-length CBP/p300 should prove useful in the analysis of CBP/p300-mediated integration of different signalling pathways. Such an approach may be facilitated by the interesting finding that not all CBP TRAM-interacting transcription factors are affected to the same extent by the same point mutations within the CBP TRAM sequence (40). Thus, by use of variants of the TRAM sequence, it might be possible to selectively block the binding of certain transcription factors to CBP/p300.

In summary, we initiated this study in order to gain greater insight into the molecular mechanism of E1A action and how this protein functions to subvert cellular pathways. Through detailed mapping of the E1A-CBP C/H3 interaction, we have identified a TRAM that is conserved in all CBP/p300 proteins. This motif, targeted by E1A, is used by various cellular factors and may prove to play an important role in the integration of multiple signal transduction pathways.