Physical and functional interactions among AP-2 transcription factors, p300/CREB-binding protein, and CITED2.

The transcriptional co-activators and histone acetyltransferases p300/CREB-binding protein (CBP) interact with CITED2, a transcription factor AP-2 (TFAP2) co-activator. p300/CBP, CITED2, and TFAP2A are essential for normal neural tube and cardiac development. Here we show that p300 and CBP co-activate TFAP2A in the presence of CITED2. TFAP2A transcriptional activity was modestly impaired in p300(+/-) and CBP(+/-) mouse embryonic fibroblasts; this was rescued by ectopic expression of p300/CBP. p300, TFAP2A, and endogenous CITED2 could be co-immunoprecipitated from transfected U2-OS cells indicating that they can interact physically in vivo. CITED2 interacted with the dimerization domain of TFAP2C, which is highly conserved in TFAP2A/B. In mammalian two-hybrid experiments, full-length p300 and TFAP2A interacted only when CITED2 was co-transfected. N-terminal residues of TFAP2A, containing the transactivation domain, are both necessary and sufficient for interaction with p300, and this interaction was independent of CITED2. Consistent with this, N-terminal residues of TFAP2A were required for p300- and CITED2-dependent co-activation. A histone acetyltransferase-deficient p300 mutant (D1399Y) did not co-activate TFAP2A and did not affect the expression or cellular localization of TFAP2A or CITED2. In mammalian two-hybrid experiments p300D1399Y failed to interact with TFAP2A, explaining, at least in part, its failure to function as a co-activator. Our results suggest a model wherein interactions among TFAP2A, CITED2, and p300/CBP are necessary for TFAP2A-mediated transcriptional activation and for normal neural tube and cardiac development.

cules between transcription activators and the general transcriptional machinery, CBP and p300 have intrinsic histone acetyltransferase (HAT) activity and very likely play a role in chromatin remodeling. Acetylation of transcription factors by p300 and CBP also modulates their activity (2).
Mice lacking p300 or CBP develop neural tube closure and cardiac and skeletal defects (3)(4)(5). Mutation of a single allele of CBP in humans causes Rubinstein-Taybi syndrome, characterized by mental retardation and cardiac, cranio-facial, and skeletal malformations (6). p300 and CBP have overlapping functions, as indicated by the fact that mice lacking one allele of p300 and one allele of CBP (double heterozygotes) show early embryonic lethality, with defects in neural tube closure. However, they also have clearly distinct functions, as well. For instance, retinoid receptor function requires p300 rather than CBP, whereas CREB function requires CBP rather than p300 (5,7). Moreover, the plant homeodomain finger of p300 but not CBP is essential for histone acetyltransferase activity (8).
The similar phenotypes observed in mice lacking p300/CBP and TFAP2A/B led us to hypothesize that p300 and CBP may function as TFAP2 co-activators. Here we describe, for the first time, physical and functional interactions between TFAP2 isoforms and p300/CBP. Our results suggest a model wherein interactions among TFAP2A, CITED2, and p300/CBP are nec-essary for TFAP2A-mediated transcriptional activation and for normal neural tube and cardiac development.

EXPERIMENTAL PROCEDURES
Standard molecular biology protocols were used for all procedures (28). Reagents were from Sigma unless otherwise indicated.
Antibodies-RK5C1 (Santa Cruz Biotechnology, Inc.) is a monoclonal antibody against GAL4-DBD. Anti-CITED2 polyclonal antibody has been described previously (9). PAB419 and 9E10 are monoclonal antibodies against SV40 Tag and the myc epitope, respectively, and were gifts from Jim DeCaprio (Dana-Farber Cancer Institute, Boston, MA). Mouse monoclonal 12CA5 antibody was used to detect hemagglutinintagged proteins. Anti-tubulin antibody was obtained from Sigma (T-5293). Anti-TFAP2A (sc-184X) and anti-proliferating cell nuclear antigen (sc-56) antibodies were obtained from Santa Cruz Biotechnology, Inc. Western blots were performed as described previously (9).
Immunostaining-Hep3B cells were plated at 2 ϫ 10 5 cells per sixwell plate and transfected the following day with 2 g of CMV-p300myc or CMV-p300(D1399Y)-myc and 0.5 g of CMV-CITED2 or control vector using FuGENE 6 (Roche Molecular Biochemicals). 0.5 g of CMV-TFAP2A or the control vector were also co-transfected. Immunostaining was performed 48 h after the transfection using anti-TFAP2A and anti-CITED2 polyclonal antibodies at 1:150 dilution, essentially as described (9). The anti-myc 9E10 monoclonal antibody was used as neat culture supernatant. Nuclei were counterstained with TOPRO (Molecular Probes). Cells were mounted in Vectamount (Vector) and were visualized using a Bio-Rad MRC 1024 confocal microscope. Data from TOPRO (blue), rhodamine (red), and fluorescein isothiocyanate (green) channels were accumulated sequentially.
Transfections and Luciferase Assays-Cells were plated in 24-well plates at 2.5 ϫ 10 4 cells per well and were transfected in duplicate the following day using FuGENE 6 (Roche Molecular Biochemicals) or Transfast reagent (Promega). CMV-lacZ (100 ng) or pRL-CMV were co-transfected in all experiments. Firefly luciferase and ␤-galactosidase (lacZ) activities were measured as described (28). Renilla luciferase activity was measured using the DUAL-luciferase reporter assay kit (Promega). The ratio of firefly luciferase to lacZ or to Renilla luciferase activity (relative luciferase activity) was calculated to correct for variations in transfection efficiency. All expression plasmids were driven by the CMV or the RSV promoter, and an appropriate amount of CMV or RSV control vector was added to each transfection mix so that the total plasmid amount and CMV or RSV promoter per transfection were constant for each experiment. Amounts of DNA used per transfection refer to the amounts added per well of a 24-well plate.
Far Western Assays-The TFAP2 variants used as probes in Far Western (FW) assays were translated in vitro using the TNT coupled reticulocyte lysate system (Promega). The proteins were labeled radioactively in 100-l reactions using L-[ 35 S]methionine/L-[ 35 S]cysteine Promix (Amersham Biosciences). Unincorporated amino acids were removed using Microcon-MWCO 3000 (Amicon) centrifugal filter devices, according to the manufacturer's instructions, and the buffer was changed simultaneously to FW buffer (25 mM Tris-HCl, pH 7.5, 100 mM KCl, 5 mM MgCl 2 , 0.1% Tween 20, 5% glycerol, 1 mM DTT, Complete protease inhibitor mix (Roche Molecular Biochemicals). The successful synthesis and purification of the probes were verified by SDS-PAGE and autoradiography; all the protein probes used were synthesized to a comparable efficiency. Assays were performed against a TriplEx phage clone encoding the C-terminal residues 234 -270 of the human CITED2. The Escherichia coli strain XL1-Blue was transduced with the TriplEx-CITED2 phage and plated at ϳ100 plaque-forming unit per 10-cm Petri dish, according to the TriplEx library manual (Clontech PT3003-1). The plates were incubated at 42°C for 3-4 h until the plaques were just visible and then overlaid with Hybond-C Extra (Amersham Biosciences) membranes, prewetted in 10 mM isopropyl-␤-D-thiogalactopyranoside, and further incubated at 37°C for another 4 h. The membranes were then removed from the Petri dishes and placed in blocking solution (FW buffer containing 5% dry milk) for 2 h, shaking gently at room temperature. After blocking, the membranes were placed in FW buffer, supplemented with an in vitro translated, 35 S-labeled protein probe, and incubated overnight at 4°C, shaking gently. The membranes were washed several times in FW buffer, and each wash was performed for 30 min at room temperature, shaking gently. The washed membranes were air-dried and exposed to BioMax MR-1 (Eastman Kodak Co.) film for several days.
Yeast Two-hybrid Assays-SFY526 yeast were transformed simultaneously with full-length CITED2 cloned into pGAD424 and TFAP2C constructs cloned into pGBT9 as described previously (33). Liquid ␤-galactosidase assays were performed according to the Clontech Matchmaker manual. To verify that TFAP2C deletion mutants containing the dimerization domain were still able to self-associate, additional clones were made in pGAD424 to allow this to be tested in the yeast twohybrid system (data not shown). For those mutants that also contained the DNA binding domain, electrophoretic mobility shift assays were also performed using unlabeled in vitro translated protein to confirm DNA binding and hence dimerization (data not shown).
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assays-Hep3B cells were plated at 2 ϫ 10 6 cells per p100 plate and transfected the following day with 15 g of CMV-p300-myc or CMV-p300(D1399Y)-myc and 5 g of CMV-CITED2 using FuGENE 6 (Roche Molecular Biochemicals). 5 g of CMV-TFAP2A or the control vector were also co-transfected. 48 h post-transfection the cell pellet was lysed for 5 min on ice by adding 5-10 volumes of lysis buffer (20 mM Tris-HCl, pH 8.0, 20 mM NaCl, 1 mM DTT, 0.5% Nonidet P-40, 320 mM sucrose, and Complete protease inhibitor mixture (Roche Molecular Biochemicals)). The nuclei were then pelleted by centrifugation and lysed for 30 min on ice by adding 5-10 volumes of nuclei lysis buffer (20 mM HEPES-KOH, pH 8.0, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 1 mM DTT, 25% glycerol, and Complete protease inhibitor mixture (Roche Molecular Biochemicals)). Before the binding assay, the nuclear extracts contained in the supernatant were diluted with two volumes of 20 mM HEPES-KOH, pH 8.0 containing the Complete protease inhibitor mixture (Roche Molecular Biochemicals). The synthetic oligonucleotide 5Ј-GATCGAACTGACCGCCCGCGGCCCGT-3Ј corresponding to the consensus TFAP2 binding site of the human metallothionein IIa promoter was end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase and annealed to a slight excess of the unlabeled complementary strand. Unincorporated nucleotides were removed using a Microspin TM G-25 column (Amersham Biosciences). Unlabeled oligonucleotide was annealed in a similar manner and used for competition experiments. The DNA binding reaction was performed in a volume of 25 l containing 1.5 g of nuclear extract, 1 g of poly(dI-dC)⅐poly(dI-dC), 25 mM HEPES-KOH, pH 7.9, 150 mM NaCl, 1 mM EDTA, 5 mM DTT, and 10% glycerol. This mixture was incubated for 10 min on ice before adding 10 -15 fmol of radiolabeled probe. The binding reaction was then carried out for 30 min at room temperature before loading the samples onto a 5% polyacrylamide gel (37.5:1, acrylamide:bisacrylamide) pre-run for at least 2 h in 25 mM Tris, pH 8.3, 25 mM boric acid, 0.5 mM EDTA. Electrophoresis was performed for 2-3 h at 200 V at 4°C. The gel was dried out and autoradiographed. For competition experiments, a 50-fold molar excess of unlabeled oligonucleotide was mixed with the probe before adding nuclear extract.

RESULTS
p300 and CBP Co-activate TFAP2A Only in the Presence of CITED2-To determine whether p300 and CBP act as TFAP2A co-activators, we transiently co-transfected Hep3B cells (a human hepatocellular carcinoma cell line that has low levels of endogenous TFAP2 and CITED2) with a luciferase reporter cloned downstream of TFAP2 binding elements (p3xAP2-Bluc) and vectors expressing TFAP2A, CITED2, CBP, or p300 (Fig.  1A). Under these conditions, the reporter was mildly activated by the transfection of TFAP2A expression vector alone. Consistent with our previous results (19), co-transfection of CMV-CITED2 resulted in further activation of the reporter, which was dependent on the presence of co-transfected CMV-TFAP2A. A further increase of the reporter activity (in a doseresponsive manner) was observed when CBP or p300 expressing plasmids were co-transfected with both CMV-TFAP2A and CMV-CITED2 (Fig. 1A). CMV-CBP or CMV-p300 showed no effect on reporter plasmid activity in the absence of cotransfected CMV-TFAP2A or CMV-CITED2. These results indicate that p300/CBP can co-activate TFAP2A but only in the presence of CITED2. Similar results were observed in HepG2 cells, another human hepatocellular carcinoma cell line (Fig. 1B), indicating that the CITED2-dependent co-activation of TFAP2A by CBP and p300 is not restricted to one particular cell line.
p300 and CBP Are Required for TFAP2A Transactivation-To determine whether endogenous p300 and CBP are necessary for TFAP2 transactivation, we examined the transcriptional activity of TFAP2 isoforms in MEFs. We isolated MEFs from embryos lacking either one allele of p300 (p300 ϩ/Ϫ ) or one allele of CBP (CBP ϩ/Ϫ ). We co-transfected the TFAP2 reporter gene and plasmids expressing TFAP2 isoforms into these MEFs and into wild-type control MEFs isolated from littermate embryos ( Fig. 1, C and D). As MEFs express CITED2 (9), it was not necessary to also transfect CITED2 in these experiments. In p300 ϩ/Ϫ and CBP ϩ/Ϫ MEFs we observed reduced transcriptional activities of TFAP2A and TFAP2C isoforms ( Fig. 1, C and D). TFAP2B transcriptional activity was also reduced in CBP ϩ/Ϫ MEFs (Fig. 1D), whereas only a weak effect was observed in p300 ϩ/Ϫ MEFs (Fig. 1C). Co-transfection of p300 and CBP expressing plasmids in p300 ϩ/Ϫ and CBP ϩ/Ϫ MEFs, respectively, successfully rescued the defective TFAP2 co-activation (Fig. 1, C and D). These results indicate that normal levels of endogenous p300 and/or CBP are necessary for full transcriptional activation by TFAP2.
p300 and TFAP2A Interact Physically-We next determined whether p300 and TFAP2A can physically interact in mammalian cells. We initially attempted to detect TFAP2A in anti-p300 immunoprecipitates. TFAP2A migrates at ϳ50 kDa, very close to the immunoglobulin heavy chain from the immunoprecipitating antibody, and this interfered with its detection by Western blotting (data not shown). To overcome this, we transfected U2-OS cells with CMV-p300-myc and CMV-GAL4-TFAP2A. GAL4-TFAP2A (a fusion of the yeast GAL4 DNA binding domain with TFAP2A) migrates at 70 kDa, away from the immunoglobulin heavy chain (Fig. 2). We used U2-OS cells as they express endogenous CITED2 (9). We immunoprecipitated p300-myc from the transfected cell lysates with an antimyc antibody (9E10) and detected co-immunoprecipitated pro- teins by Western blotting using an anti-GAL4 antibody (RK5C1). Anti-myc immunoprecipitates from cells transfected with CMV-p300-myc and CMV-GAL4-TFAP2A contained GAL4-TFAP2A ( Fig. 2A, lane 7) and endogenous CITED2 proteins (Fig. 2B, lane  7). Control primary antibody immunoprecipitates from the same cells (Fig. 2, lane 9) did not contain GAL4-TFAP2A or CITED2. Anti-myc immunoprecipitates from cells transfected with either CMV-p300-myc or CMV-GAL4-TFAP2A alone also did not contain GAL4-TFAP2A ( Fig. 2A, lanes 8 and 10, respectively). Antimyc immunoprecipitates from cells co-transfected with CMV-p300-myc and CMV-GAL4 also did not contain detectable GAL4 protein (Fig. 2C, lane 12). These results indicate that p300-myc complexes contain both GAL4-TFAP2A and endogenous CITED2 and that the TFAP2A residues in the GAL4-TFAP2A fusion protein are required for the interaction.
CITED2 Interacts with the First Helix of the TFAP2 Dimerization Motif-To delineate residues of TFAP2 required for the interaction with CITED2, we used Far Western and yeast two-hybrid assays. We used 35 S-labeled TFAP2A, TFAP2B, and TFAP2C protein probes to determine their ability to bind Cterminal residues of human CITED2 peptide (residues 234 -270), immobilized on nitrocellulose membranes (Fig. 3A, panels  1-3). Consistent with previous results (19), all TFAP2 isoforms bound strongly to CITED2. We next examined the ability of different fragments of TFAP2C to interact with CITED2. The deletion of N-terminal residues 1-219 of TFAP2C (TFAP2C⌬N220) did not affect the interaction with CITED2 (Fig. 3A, panel 4). Deletion of the C-terminal residues 431-450 (TFAP2C⌬N220C430) also did not affect the interaction (Fig.  3A, panel 5). A further N-terminal deletion that deleted residues 1-291 (TFAP2C⌬N292C430) also did not affect the interaction, but deletion of residues 1-352 (TFAP2C⌬N353C430) abrogated the interaction (Fig. 3A, panel 7). This deletion removes the first helix of the helix-span-helix dimerization domain of TFAP2C (35).
We then explored these interactions using yeast two-hybrid assays. Here we co-expressed GAL4DBD-TFAP2C "baits" with GAL4AD-CITED2 "prey" expression plasmids in yeast cells containing an integrated lacZ reporter downstream of GAL4 binding sites. Consistent with previous observations (19) we found that GAL4DBD-TFAP2C full-length bait strongly interacts with a GAL4AD-CITED2 prey (Fig. 3B). This assay also indicated that the N-terminal residues 1-180 of TFAP2C (GAL4-TFAP2C⌬C181) and the C-terminal deletion of the second helix of the dimerization domain (GAL4-TFAP2C⌬C413) are dispensable for the interaction with CITED2 (Fig. 3B). Compared with the interaction domain defined by far Western, the only discrepancy was that deletion of the residues 1-291 (GAL4-TFAP2C⌬N292) failed to interact with AD-CITED2 in yeast two-hybrid assay. We hypothesize that the fusion of the GAL4-DBD domain to the residues 292-450 of TFAP2C can lead to the occlusion of the CITED2 interaction domain in yeast two-hybrid assay. Notably, TFAP2 proteins do not need to be dimers to bind CITED2, as non-dimerizing mutants (e.g. TFAP2C⌬N220C413) can still interact (Fig. 3C) (data not shown).
The p300-TFAP2A Interaction Requires CITED2-To determine whether CITED2 is involved in the recruitment of p300 to TFAP2A, we used a mammalian two-hybrid assay in Hep3B cells to detect interactions between the bait p300-E2 (a fusion of p300 with the bovine papilloma virus E2-DNA binding domain) and the prey VP16-TFAP2A (Fig. 4A). In these experiments we used a reporter plasmid containing E2 binding sites upstream of the luciferase gene. Transfection of CMV-VP16-TFAP2A alone led to a weak activation of the reporter and may reflect the presence of cryptic TFAP2 binding sites in the reporter plasmid. In keeping with previous observations, transfection of CMV-p300-E2 (with CMV-VP16 control) activated the luciferase reporter gene (30). A marginal degree of further activation was observed when CMV-p300-E2 was co-transfected with the prey plasmid CMV-VP16-TFAP2A, indicating the lack of a significant two-hybrid interaction between p300-E2 and VP16-TFAP2A. However, when CMV-CITED2 was co-transfected with CMV-p300-E2 and CMV-VP16-TFAP2A, a marked increase in the activity of the reporter was observed, indicating that in the presence of CITED2, VP16-TFAP2A was able to interact with p300-E2. Co-transfection of CMV-p300-E2, CMV-VP16, and CMV-CITED2 did not activate the reporter, indicating that the TFAP2A residues in the VP16-TFAP2A fusion are necessary for the interaction between VP16-TFAP2A and p300-E2. These results indicate that CIT-ED2 is necessary for the interaction between full-length TFAP2A and full-length p300.
The p300-TFAP2A Interaction Requires an Intact p300-CH1 Domain-To determine the role of the p300-CH1 domain (which is required for binding CITED2), we also performed these experiments with CMV-p300CH1⌬-E2, which lacks the CH1 domain (Fig. 4A). Transfection of CMV-p300CH1⌬-E2 (with CMV-VP16 control) activated the reporter to the same extent as noted for CMV-p300-E2, and co-transfection with CMV-VP16-TFAP2A led to a marginal increase in reporter activity. However, in contrast to the results obtained for CMV-p300-E2, no increase in reporter activity was noted with cotransfected CMV-CITED2. This indicates that the p300-CH1 domain is essential for the CITED2-induced interaction between TFAP2A and p300.
Interactions between p300 and TFAP2A Domains That Do Not Require CITED2-Certain transcription factors, such as NF-B p65 and p53, interact with multiple domains of p300/ CBP (36 -39). To determine whether this is true for TFAP2A, we used a mammalian two-hybrid experiment using different p300-E2 baits and VP16-TFAP2A prey expression constructs (Fig. 4B). In these experiments, co-transfection of CMV-p300-E2 and N-terminal residues of TFAP2A (VP16-TFAP2A (1-165)), containing the transactivation domain (29), led to a marked increase of reporter activity, when compared with CMV-p300-E2 co-transfected with CMV-VP16 (control prey empty vector; see Fig. 4B). The level of activation was not affected by co-transfection of CMV-CITED2. This indicates that an interaction can occur between p300 and the isolated N terminus of TFAP2A and that this interaction is independent of CITED2. Notably, the isolated C-terminal domain of TFAP2A (202-437) did not interact with p300 in this assay regardless of the presence or absence of CITED2 (Fig. 4B). Thus, residues 1-201 of TFAP2A (containing the N-terminal transactivation domain) are essential for interaction with p300. Co-transfection of p300CH1⌬-E2 and VP16-TFAP2A (1-165) activated the reporter construct, indicating that this interaction does not require the p300-CH1 domain (Fig. 4B).
To further explore CITED2-independent interactions between TFAP2A and p300, we used GAL4-p300 baits that expressed different fragments of p300 (Fig. 4C). In these experiments, no interaction was observed among GAL4 (control bait empty plasmid), GAL4-p300 (residues 1-242) or GAL4-p300 (residues 1737-2414), and VP16-TFAP2A. The N terminus of p300 (GAL4-p300 residues 1-596 and 1-743) interacted with the N terminus of TFAP2A (VP16-TFAP2A residues 1-165) but not with full-length VP16-TFAP2A or the C terminus of TFAP2A (VP16-TFAP2A residues 202-437). This indicates that the isolated N terminus of TFAP2A can interact with the N terminus of p300, independently of CITED2. The lack of interaction between full-length TFAP2A and the N terminus of p300 is not clear but may reflect a structural constraint. We also found that the C-terminal residues of p300 (GAL4-p300 residues 964 -1922) interacted strongly with full-length VP16-TFAP2A and with VP16-TFAP2A (residues 202-437). These results indicate that the C-terminal DNA binding and dimerization domains of TFAP2A can interact with C-terminal residues of p300.
The Transactivation Domain of TFAP2A Is Required for CITED2 Co-activation-The above results indicated that the N terminus of TFAP2A, containing the transactivation domain, is both necessary and sufficient for interaction with full-length p300. This interaction is independent of CITED2. To determine whether this domain is necessary for synergistic co-activation by CITED2 and p300, we co-transfected HepG2 cells with p3xAP2-Bluc and RSV-TFAP2A (expressing full-length TFAP2A) or RSV-TFAP2A⌬N165 (expressing a truncated protein lacking the transactivation domain; see Fig. 5). Co-transfection of CMV-CITED2 resulted in an increase of the transcriptional activity of the full-length TFAP2A but failed to enhance the transcriptional activity of TFAP2A⌬N165. Cotransfection of CMV-p300, CMV-CITED2, and RSV-TFAP2A led to a further increase of the reporter activity, consistent with the results in Fig. 1. However, co-transfection of CMV-p300, CMV-CITED2, and RSV-TFAP2⌬N165 failed to activate the reporter. This result indicated that co-activation by p300 is not only dependent on CITED2 but also requires the activation domain of TFAP2A and implies that CITED2independent interactions between TFAP2A and p300 may also play an important role in TFAP2A-mediated transcriptional activation.

AP-2 Co-activation
CMV-TFAP2A led to an increase of the reporter plasmid activity only in the presence of CMV-CITED2. However, co-transfection of CMV-p300(D1399Y)-myc failed to stimulate, and in fact appeared to reduce, transcriptional co-activation of TFAP2A by CITED2. Western blotting of nuclear extracts from transfected Hep3B cells showed that p300-myc and p300(D1399Y)-myc are expressed at similar levels (Fig. 6B) and that transfection of p300(D1399Y)-myc did not affect the expression of either co-transfected TFAP2A or CITED2. Thus the observed failure of p300(D1399Y)-myc to function as a TFAP2A co-activator is not because of a simple reduction in expression of p300, TFAP2A, or CITED2.
p300D1399Y Does Not Affect TFAP2A DNA Binding Activity or CITED2 Nuclear Localization-In addition to functioning as a histone acetyltransferase, p300 and CBP can also acetylate transcription factors and enhance their ability to bind DNA (2). We asked whether the HAT activity of transfected p300 could affect the DNA binding activity of co-transfected TFAP2A. We co-transfected Hep3B cells with CMV-TFAP2A, CMV-CITED2, and with either CMV-p300-myc or CMV-p300(D1399Y)-myc. We then made nuclear extracts from these cells and examined their binding to an oligonucleotide containing a high affinity binding site for TFAP2, by electrophoretic mobility shift assay (Fig. 6C). In cells transfected with CMV-TFAP2A, we observed a DNA-protein complex (Fig. 6C, lane 2 and 3, TFAP2A). This was not observed in cells that were not transfected with TFAP2A (lanes 4, 5 and 6). This DNA-protein complex was efficiently competed out by a 50-fold molar excess of unlabeled oligonucleotide (Fig. 6C, lanes 7 and 8), indicating its specificity for TFAP2 binding sites. No obvious difference was observed in the TFAP2A-DNA complex in extracts from cells transfected with CMV-p300-myc or CMV-p300(D1399Y)-myc. These results indicate that the HAT activity of p300 does not modulate the ability of TFAP2A to bind DNA.
ization. Similarly, we determined the effect of p300 on TFAP2A localization (Fig. 6D, panels e-h). In Hep3B cells, transfected TFAP2A was detected in the nuclei (Fig. 6D,  panel g), and co-transfection of either CMV-p300-myc or CMV-p300(D1399Y)-myc did not affect TFAP2A localization (Fig. 6D, panels e and f). These results indicate that the HAT activity of p300 does not modulate the cellular localization of TFAP2A or CITED2.
p300D1399Y Does Not Interact with TFAP2A-We next investigated the interaction of the p300-HAT-deficient mutant with TFAP2A and CITED2. We constructed a HAT-deficient version of CMV-p300-E2 (CMV-p300(D1399Y)-E2) and used it FIG. 6. The HAT-deficient p300 mutant (p300D1399Y) does not co-activate TFAP2A. A, transcriptional activation of p3xAP2-Bluc reporter (40 ng) in Hep3B cells by co-transfected CMV-TFAP2A or the control CMV-vector (60 ng each), CMV-CITED2 and the control CMV-vector (60 ng each), and increasing doses (0, 160, and 320 ng) of CMV-p300-myc (expressing wild type p300 fused to the myc epitope tag) or CMV-p300(D1399Y)-myc (expressing a HAT-deficient p300). CMV-lacZ (100 ng) was also co-transfected in each reaction. Results are presented as for Fig. 1A. B, Western blots were performed using nuclear extracts prepared from Hep3B cells transfected with plasmids expressing the proteins indicated. The relative expression levels of p300-myc (top panel, lanes 1 and 3), p300(D1399Y)-myc (top panel, lanes 2 and 4), TFAP2A (lanes 1 and  2), and CITED2 proteins (lanes 1-4), after transfection of the relevant expression plasmids into Hep3B cells, is compared with untransfected cells (lane 5). The anti-proliferating cell nuclear antigen Western blot was performed on the stripped membrane to demonstrate equal loading. C, electrophoretic mobility shift assays were performed with nuclear extracts prepared from Hep3B cells transfected with plasmids expressing the proteins indicated at the top (lanes [2][3][4][5][6][7][8][9][10] or no protein extract (lane 1). Nuclear extracts were incubated with a radiolabeled oligonucleotide containing the TFAP2 binding site of the human metallothionein IIa promoter, in the absence (lane [2][3][4][5][6] or the presence of 50-fold molar excess of unlabeled competitor oligonucleotide (lanes 7-10). The position of the TFAP2A-DNA complex is indicated. D, Hep3B cells were co-transfected either with the plasmids indicated in each panel. CMV-p300(DY)-myc is an abbreviation for CMV-p300(D1399Y)-myc, the HAT-deficient mutant. 48 h after transfection cells were co-immunostained with a monoclonal anti-myc antibody to detect myc-tagged p300 proteins (red) and with a polyclonal anti-CITED2 antibody (green; panels a-d) or with a polyclonal anti-TFAP2A antibody (green; panels e-h). Nuclei were counterstained using TOPRO (blue), and cells were examined by confocal microscopy. The merged image is shown in the panels on the extreme right.
in mammalian two-hybrid assays in Hep3B cells using a reporter plasmid containing E2 binding sites upstream of the luciferase gene. CMV-p300-E2 and CMV-p300(D1399Y)-E2 were co-transfected in parallel with plasmids expressing VP16-TFAP2A in the presence or the absence of CMV-CITED2 (Fig.  7A). Transfection of CMV-p300-E2 resulted in modest reporter gene activation (compared with CMV-vector control) which was not affected by CITED2. Co-transfection of VP16-TFAP2A did not affect reporter gene activity in the absence of CITED2, but in the presence of CITED2 there was a further increase in reporter activity, indicating a two-hybrid interaction that is dependent on CITED2. Co-transfection of VP16-TFAP2A (1-165) resulted in activation of the reporter gene regardless of CITED2 co-transfection. These results are consistent with those shown in Fig. 4A. Transfection of CMV-p300(D1399Y)-E2 led to a smaller increase of reporter activity compared with CMV-p300-E2, despite being expressed to a similar level (Fig.  7C), indicating that HAT-deficient p300 has lower transcriptional activity. No increase in reporter activity was observed when CMV-p300(D1399Y)-E2 was co-transfected with CMV-VP16-TFAP2A or CMV-VP16-TFAP2A (1-165), regardless of CMV-CITED2 co-transfection. This result indicated that the p300-HAT-deficient mutant D1399Y does not interact with either full-length TFAP2A or with TFAP2A N-terminal residues 1-165.
The p300-HAT-deficient Point Mutant Interacts with CITED2-We next determined whether the p300-HAT-deficient mutant interacts with CITED2. Hep3B cells were cotransfected with CMV-p300-E2 or CMV-p300(D1399Y)-E2 and CMV-VP16-CITED2 (Fig. 7B). Co-transfection of CMV-VP16-CITED2 with CMV-p300-E2 led to a ϳ4.5-fold increase of reporter activity, compared with CMV-p300-E2 alone, indicating that CITED2 interacts with p300. Co-transfection of CMV-p300(D1399Y)-E2 with CMV-VP16-CITED2 also led to a similar -fold increase of the reporter activity compared with the basal activation because of CMV-p300(D1399Y)-E2, indicating that the p300-HAT-deficient mutant interacts normally with CITED2. DISCUSSION In this study, we have shown that p300 and CBP can coactivate TFAP2A and that this co-activation requires the pres-ence of CITED2. Our experiments show that p300 and CBP heterozygote MEFs display modest defects in transactivation by TFAP2, which indicate that normal levels of endogenous p300 and/or CBP are necessary for full transcriptional activation by TFAP2. The modest impairment in TFAP2 transactivation observed in heterozygote MEFs is consistent with the fact that the total complement of p300/CBP is only reduced by one-fourth. The successful rescue of these defects by ectopically expressed p300 and CBP indicates that there is no other secondary genetic defect in these MEFs. Taken together these experiments show that p300/CBP function as TFAP2 co-activators and that this co-activation critically requires the presence of another TFAP2 co-activator, CITED2.
We have also shown, using co-immunoprecipitation and mammalian two-hybrid techniques, that TFAP2A, CITED2, and p300 can interact physically in vivo. This interaction requires the presence of CITED2 and an intact binding site on p300 for CITED2 (i.e. the CH1 domain). We have also established that CITED2 interacts with residues contained within the first dimerization helix of TFAP2C. Based on the high conservation of this region among TFAP2 isoforms, showing only conservative changes at three amino acid positions, and the fact that CITED2 also interacts and co-activates TFAP2A and TFAP2B (19), we predict that CITED2 also binds to the first dimerization helix of these TFAP2 isoforms.
Our data also show that TFAP2A can interact with certain  1). The p300-E2 are produced as hemagglutinin-tagged proteins and were detected using the mouse monoclonal 12CA5 antibody as described previously (9). Bottom, anti-tubulin Western blot to demonstrate equal loading.
FIG. 8. Cartoon summarizing interactions among CITED2, p300, and TFAP2A. This cartoon summarizes the results obtained in this and previous studies (9,19). The regions in CITED2 that are conserved in other CITED family members (16) are indicated as CR1-3. p300 domains independently of CITED2. These interactions are summarized in Fig. 8. However, these CITED2-independent interactions are not sufficient for an interaction between full-length TFAP2 and full-length p300 and instead may be secondary interactions that are also required for TFAP2Amediated transcriptional activation. Significantly, we found that the N terminus of TFAP2A, containing the transactivation domain, is both necessary and sufficient for interaction with full-length p300. Consistent with this we found that a TFAP2A peptide lacking N-terminal residues is not successfully co-activated by CITED2 and p300. Taken together, this suggests that the CITED2-independent interaction between the N termini of TFAP2A and that of p300 is likely to be functionally significant. An alternative explanation is that the N terminus of TFAP2A is required to recruit other co-activators such as PC4 (40) and that these other co-activators are necessary for CITED2-p300 mediated co-activation of TFAP2A.
We also found that a p300-HAT-deficient point mutant (D1399Y) failed to co-activate TFAP2A, suggesting that p300-HAT activity is necessary for TFAP2A co-activation. This mutation did not appear to affect the DNA binding activity, nuclear localization, or protein level of co-transfected TFAP2A. However, the p300D1399Y mutant did not interact with fulllength TFAP2A, despite being able to interact with CITED2. This explains, at least in part, its failure to function as a TFAP2A co-activator and suggests that p300-HAT activity may be required for interaction with TFAP2A. Surprisingly, the N-terminal residues of TFAP2A also failed to interact with the p300-HAT mutant. As the N-terminal residues of TFAP2A interact strongly with the isolated N terminus of p300 (also HAT-deficient), one possibility is that in the context of fulllength p300, the HAT-deficient mutant has a conformation that precludes binding to TFAP2A.
In summary, our results show that TFAP2A and p300/CBP can interact both physically and functionally and that these interactions require CITED2. They also show that p300/CBP and CITED2 are independently required for full transcriptional activation by TFAP2A. These results are supported by the observation that a HAT-deficient p300 point mutation fails to co-activate and interact with TFAP2A and that cells lacking an allele of either p300 or CBP display TFAP2 transactivation defects. Our results suggest a model wherein interactions among TFAP2A, CITED2, and p300/CBP are necessary for TFAP2A-mediated transcriptional activation. In experiments presented here and elsewhere, we showed that CITED2 binds directly to p300/CBP and also to TFAP2 isoforms independently of its interaction with p300 and CBP (9,19). A simple model is that CITED2 acts as a bridge, linking TFAP2A to p300/CBP. Alternatively, interactions with CITED2 may lead to modifications in p300/CBP and/or TFAP2A, allowing them to interact. Taken together, these results support the idea that physical and functional interactions among p300/CBP, CIT-ED2, and TFAP2A are essential for normal neural tube and cardiac development and that certain aspects of p300 and CBP haploinsufficiency (Rubinstein-Taybi syndrome) result from defective TFAP2 function.