A Functional Interaction with CBP Contributes to Transcriptional Activation by the Wilms Tumor Suppressor WT1*

The Wilms tumor gene WT1 encodes a zinc finger transcription factor that is required for normal kidney development. WT1 was identified as a transcriptional re-pressor, based on its suppression of promoter reporters, but analysis of native transcripts using high density microarrays has uncovered transcriptional activation, rather than repression, of potential target genes. We report here that WT1 binds to the transcriptional coactivator CBP, leading to synergistic activation of a physiologically relevant promoter. The physical interaction between WT1 and CBP is evident in vitro and in vivo , and the two proteins are co-immunoprecipitated from embryonic rat kidney cells. The WT1-CBP association requires the first two zinc fingers of WT1 and the adenovirus 5 E1A-binding domain of CBP. Overexpression of this domain of CBP is sufficient to inhibit WT1-medi-ated transcriptional activation of a promoter reporter, as is co-transfection of E1A. Retrovirally driven expression of either the CBP fragment or of E1A in human hematopoietic cells suppresses the induction by WT1 of its endogenous target gene, p21 Cip1 . These observations support a model of WT1 as a transcriptional activator of genes required for cellular differentiation. The tumor suppressor gene WT1 provides a paradigm for the link between normal organ development and cancer. WT1 was originally identified by its inactivation in a subset of Wilms tumors and in the germline of children with genetic predispo-sition

The tumor suppressor gene WT1 provides a paradigm for the link between normal organ development and cancer. WT1 was originally identified by its inactivation in a subset of Wilms tumors and in the germline of children with genetic predisposition to this pediatric kidney cancer (1)(2)(3)(4). Expression of WT1 is restricted to specific cell types in the fetal kidney, gonads, mesothelium, and hematopoietic lineages (5)(6)(7)(8). In the developing kidney, WT1 is present at low levels in blastemal stem cells, and very high levels of expression are observed as these cells differentiate to form glomerular podocytes. The critical role of WT1 in normal renal development is demonstrated by the apoptosis of renal blastemal cells in WT1-null mice, leading to complete failure of kidney differentiation (9).
The DNA-binding domain of WT1 is encoded by four Cterminal Cys-His zinc fingers, which mediate recognition of both GC-and TC-rich sequences (10,11). The affinity of WT1 for these motifs is greatly reduced by the product of an alternatively spliced transcript, in which the three amino acids KTS are inserted between zinc fingers three and four (10,12). The WT1(ϪKTS) isoform modulates transcriptional activity of re-porter constructs, and its ectopic expression leads to either cell cycle arrest or apoptosis (13)(14)(15). In contrast, the WT1(ϩKTS) isoform is inactive in such assays, and its discrete subnuclear localization has been linked to a potential role in pre-mRNA splicing (16). The N terminus of WT1 encodes a proline/glutamine-rich domain similar to the transactivation domain of other transcription factors, such as Sp1 (17). Fusion of this transactivation domain to a GAL4 DNA-binding domain results in transcriptional repression of a reporter construct (18), and WT1 itself has been shown to repress GC-and TC-rich promoters in transient transfection assays (for reviews, see Refs. 19 and 20). The observation that the transcriptional activity of promoters from many growth-promoting genes, including early growth response 1 (18), insulin-like growth factor 2 (21), insulin-like growth factor receptor (22), platelet-derived growth factor-A (23,24), and epidermal growth factor receptor (13), is suppressed by WT1 has led to the model that WT1 functions as a tumor suppressor by transcriptional repression of genes required for cellular proliferation (19). However, variations in experimental conditions, including promoter context (11), presence or absence of p53 (25), and even choice of expression vector (26), appear to modulate the properties of WT1 in transient transfection assays, leading to transcriptional activation as well as repression. Furthermore, analysis of cells with inducible expression of WT1 demonstrated that few endogenous genes with putative WT1-repressible promoters are in fact regulated by WT1 in vivo (13).
As a strategy to identify physiologically regulated WT1 target genes, we recently used high density oligonucleotide arrays representing 6,800 genes and expressed sequence tags to compare the expression profile of cells before and shortly after inducible expression of WT1 (27). Ectopic expression of WT1(ϪKTS) did not result in reduced expression of any transcripts represented on the microarrays. However, a small number of genes were strongly induced following WT1(ϪKTS) expression, notably amphiregulin (AR), 1 encoding a secreted growth factor of the epidermal growth factor family capable of stimulating cellular differentiation in organ culture systems, and the cyclin-dependent kinase inhibitor p21 Cip1 . A physiologically relevant interaction between WT1 and AR was suggested by their precise co-localization within differentiating glomeruli of the developing kidney, and by the ability of recombinant AR to induce epithelial differentiation in cultured kidney rudiments (27). WT1(ϪKTS) binds to a high affinity site (WRE) in the AR promoter, adjacent to a CRE site. The synergistic transactivation mediated through the WRE and CRE sites raised the possibility that WT1 itself might interact with CBP/P300, a coactivator known to enhance CRE-dependent transcriptional activation by CRE-binding factors (28). We show here that WT1 binds to CBP stoichiometrically both in vitro and in vivo, and that this protein interaction contributes to WT1-dependent transcriptional activation. These observations support the role of WT1 as a transcriptional activator, which may participate in the induction of target genes directly, as well as through potential interactions with other transcriptional regulators.

MATERIALS AND METHODS
Cell Culture and Expression Constructs-NIH3T3 cells were grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum. RSTEM cells with inducible WT1 expression (27) were grown at 32°C in the same media, supplemented with 1 g/ml tetracycline. For WT1 induction, cultures at 70% confluence were extensively washed with phosphate-buffered saline and allowed to grow for an additional 30 h in medium without tetracycline before protein extraction. CMV-driven constructs encoding WT1(ϪKTS) and WT1(ϩKTS) have been previously described (13), as have WT1-deletion constructs (29) and constructs encoding CBP (30), P300 (31), and E1A (32). Constructs encoding different CBP domains were generated by inserting the appropriate polymerase chain reaction fragments into pCMV-myc vector (Invitrogen). The AR promoter reporter (luciferase) PGL2-B-⌬CRE contains the WT1-responsive site and lacks an adjacent CRE site (27).
Transient Transfections and Luciferase Reporter Assays-NIH 3T3 cells were transfected using the calcium phosphate DNA precipitation method. For protein extraction following transfection with WT1 deletion constructs, 5 g of each plasmid was used. For luciferase reporter assays, 0.5 g of CMV-WT1(ϪKTS) and/or CBP expression plasmids FIG. 1. Enhancement of WT1dependent activation by CBP and P300. Relative luciferase activity following transient transfection of the amphiregulin promoter-reporter PGL-2B-⌬CRE (1 g) into NIH 3T3 cells, together with CMV promoter-driven constructs encoding WT1(ϪKTS) (0.5 g), CBP (0.5 g), and P300 (0.5 g). Equal transfection efficiency was confirmed by co-transfection of a human growth hormone construct, and the total amount of CMV promoter sequence transfected was equalized by addition of empty vector. Experiments were performed in triplicate.

FIG. 2.
In vivo association between WT1 and CBP. Immunoprecipitation Western analysis of cellular lysates extracted from RSTEM cells, 30 h following withdrawal of tetracycline and induction of WT1 expression. Equal amount of lysates from cells expressing either WT1(ϪKTS) or WT1(ϩKTS) were immunoprecipitated with antibodies against WT1, CBP, or FasL (nonspecific), followed by immunoblotting using the complementary antibody. Total lysate (1%) was analyzed directly to show the migration position of the endogenous protein; for the WT1 immunoblot, a longer exposure of the total lysate lane from the same gel is shown. The stoichiometry of the protein interaction is estimated by comparing the amount of CBP co-precipitated using ␣-WT1 antibody with that directly immunoprecipitated using ␣-CBP antibody (upper panel), and the amount of WT1 co-immunoprecipitated using ␣-CBP antibody with that directly immunoprecipitated using ␣-WT1 antibody (lower panel).
were transfected along with 1 g of the PGL2-B-⌬CRE luciferase reporter construct. Transfections with CBP domains were done using 2 g of PGL2-B-⌬CRE, 4 g of CMV-WT1(ϪKTS), and increasing amounts (2-8 g) of each of the four CBP expression constructs (CMV-CBP1-4). Cells were collected 40 h following transfection and equal transfection efficiency was confirmed by co-transfecting a human growth hormone expression plasmid and quantitation of human growth hormone levels in the medium (Nichols Institute). Equal amounts of CMV promoter sequences were transfected in all cases by addition of vector plasmid, and experiments were performed in triplicate. In transfection experiments with E1A, cell viability was ascertained by vital dye staining, to ensure that differences in transcriptional activity were not the result of selective cell killing.
Immunoprecipitation and Western Blotting-For immunoprecipitation experiments, RSTEM or NIH 3T3 cells were lysed on ice for 20 min in hypotonic buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Nonidet P-40, 5 mM EDTA) supplemented with protease inhibitors. Cellular extracts were incubated with either anti-WT1 antibody (C19, Santa Cruz) or anti-CBP antibodies (SC-369, SC-583, Santa Cruz; 06 -294, Upstate Technology) at 4°C for 1.5 h. Immunoprecipitates were resolved by 7.5% SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting (ECL) using monoclonal antibodies against WT1 (13) or CBP. For GST pull-down assays, DNA fragments encoding the four relevant domains of CBP were generated by polymerase chain reaction, containing a primer-derived T7 polymerase-binding site and Kozak consensus sequence (GCCGCCATGGCT) at their 5Ј end. Proteins were in vitro translated in the presence of [ 35 S]methionine using TNT-coupled transcription/translation kit (Promega), and incubated with GST-WT1 affinity matrix (27) in phosphate-buffered saline plus 1% Nonidet P-40 for 1 h at room temperature. After extensive washing, the precipitates were resolved by SDS-polyacrylamide gel electrophoresis and analyzed by autoradiography.
Retroviral Infection-For retroviral expression, DNA sequences encoding E1A, E1A⌬CR2, CBP3, and WT1 were cloned into murine stem cell virus vectors as previously described (33). Virus encoding these constructs or vector control was prepared from supernatants of 293T cells transfected with the appropriate plasmids as well as plasmids expressing viral envelope proteins. For the first infection, 1 ml of viral supernatant was added onto 1 ml of U937 cells (10 5 cells/ml) grown in 24-well plate. Cells were centrifuged for 1 h and returned to the incubator. Viral supernatant was replaced with fresh media after 12 h and cells were allowed to recover for another 12 h. The procedure was repeated once, and cells were then selected with G418 for 4 -7 days to remove non-infected populations. For the second infection, G418-resistant cells were treated with viral supernatants according to the same procedure. 48 h after the second infection, cell lysates were prepared and the expression level of WT1 and p21 Cip1 was examined by Western blotting.

RESULTS AND DISCUSSION
Enhancement of WT1-mediated Transactivation by CBP-We made use of AR, as a potentially physiological WT1target gene, to test the effect of CBP expression on WT1mediated transcriptional activation. The WT1-responsive element (WRE) in the AR promoter, 5Ј-CCGTGGGTGG-3Ј, is located at position Ϫ292 to Ϫ283, adjacent to a CRE element at position Ϫ274 to Ϫ267 (27). To restrict our analysis to WREdependent transactivation, we deleted the CRE site from a minimal reporter (pGL2-B-⌬CRE), which was co-transfected into NIH 3T3 cells together with cytomegalovirus (CMV)driven constructs encoding WT1(ϪKTS) and CBP. 2-Fold transcriptional activation of the reporter was observed following transfection of small amounts (0.5 g) of WT1(ϪKTS) plasmid alone. Coexpression of CBP (0.5 g) resulted in 10-fold activation of the reporter, while CBP alone had no effect (Fig. 1). Transfection of CMV-driven P300, a transcriptional coactivator that is closely related to CBP, also enhanced WT1-dependent transactivation, but to a lesser extent.
In Vivo Physical Association between WT1 and CBP-In the fetal kidney, physiological expression of WT1 is restricted to FIG. 3. Interaction between CBP and zinc fingers 1-2 of WT1. A, immunoprecipitation Western analysis of NIH 3T3 cells, transfected with plasmids encoding HA-epitope tagged WT1 deletion constructs along with full-length CBP. Equal amounts of cellular lysates were immunoprecipitated using anti-CBP antibody, followed by immunoblotting analysis using antibody to the HA epitope. B, comparable expression of the WT1 deletion constructs is shown by direct Western analysis of the cellular lysates. A background band is present at 50 kDa. C, a schematic representation of the deletions within WT1 is shown, demonstrating that the ⌬282-364 construct lacks zinc fingers 1-2.
blastemal stem cells and differentiating glomerular precursors (8). To identify WT1-interacting proteins in an appropriate cellular context, we used RSTEM cells, derived from day 12.5 embryonic rat kidney. Expression of endogenous WT1 in these cells is low but clearly detectable by immunoblotting, and the presence of a stably transfected tetracycline-regulated WT1 construct allows a 5-fold increase in WT1 protein levels without inducing any discernible alterations in cellular properties (27). Extraction of RSTEM cellular lysates using nonionic detergents, followed by immunoprecipitation with antibody against the C terminus of WT1 (C19), and immunoblotting using anti-CBP antibody demonstrated co-immunoprecipitation of these two proteins (Fig. 2). Similarly, immunoprecipitation of lysates with anti-CBP antibody, followed by Western blotting with anti-WT1 antibody showed co-precipitation of WT1 with endogenous CBP. Comparison of the amounts of co-precipitated WT1 or CBP with the total cellular proteins directly immunoprecipitated from cellular lysates (Fig. 2) indicated that ϳ10% of cellular CBP is co-precipitated with WT1, while a similar fraction of cellular WT1 is associated with CBP. The two major isoforms of WT1, resulting from the KTS alternative splice between zinc fingers 3 and 4, demonstrated comparable binding to CBP. In many cell types, CBP is thought to be limiting in its cellular concentration, resulting in possible competition for binding with different transcription factors. The levels of WT1 expression achieved in RSTEM cells are comparable to those seen in glomerular precursor cells, in which WT1 is presumed to play an important role in mediating cellular differentiation. The stoichiometry of the interaction between WT1 and CBP observed in RSTEM cells may therefore be physiologically relevant in the developing kidney.
Domains Involved in the Interaction between WT1 and CBP-We used five deletion constructs spanning the entire length of the WT1 coding region to map the site required for its interaction with CBP. The HA epitope-tagged deletion constructs were transfected into NIH 3T3 cells along with a plasmid expressing CBP, and whole cell extracts were subjected to immunoprecipitation Western analysis (Fig. 3). Equal expression of the transfected constructs was determined, and anti-CBP immunoprecipitates were analyzed for the presence of WT1 using antibody against the HA epitope. A WT1 product lacking the first two zinc fingers failed to co-immunoprecipiate with CBP, indicating that this portion of the WT1 DNA-binding domain is required for this protein interaction.
To identify the domain of CBP required for its interaction with WT1, we incubated a bacterially synthesized protein containing the four WT1 zinc fingers with in vitro translated domains of CBP. This in vitro approach was used to confirm the direct interaction between these two proteins. CBP encodes a protein of 300 kDa, but four domains mediate its known interactions with transcription factors (34 -39). Polymerase chain reaction-generated fragments encoding amino acids 1-436 (CBP1: binding site for nuclear hormone receptors), amino acids 437-786 (CBP2: binding site for CRE-binding factors), amino acids 1625-1991 (CBP3: binding site for E1A), and amino acids 1992-2442 (CBP4: binding site for p53) were translated in vitro, and incubated with GST-WT1 affinity matrix. Only CBP3 bound to WT1 (Fig. 4), indicating that the E1A-binding domain of CBP is also responsible for its interaction with WT1. Consistent with the stoichiometric protein interaction observed in vivo, the in vitro association between WT1 and CBP was quantitative, with ϳ1/3 of input CBP protein associated with WT1. Of note, the WT1(ϩKTS) isoform also showed high affinity binding to CBP (Fig. 4). The possibility that this alternative splice product of WT1, which has been previously linked to a role in pre-mRNA processing (16), also FIG. 4. In vitro interaction between WT1 and the E1A-binding domain of CBP. A, co-immunoprecipitation of radiolabeled in vitro translated domains of CBP, following incubation with GST-WT1 affinity matrix. Comparable expression of the four known protein-association domains of CBP was confirmed by direct analysis of 4% of input lysate. Incubation of these reticulocyte lysates with bacterially synthesized GST-WT1(ϪKTS) or GST-WT1(ϩKTS) was followed by GST pull-down and autoradiography. B, a schematic representation of the protein-interaction domains of CBP is shown, including the domains involved in its association with nuclear hormone receptors (NR), CRE-binding factors (CREB), and Jun, E1A, and p53. The E1A-binding domain of CBP (amino acids (aa) 1625-1991) also mediates its interaction with WT1. plays a role in transcriptional regulation warrants further investigation.
Disruption of the WT1-CBP Interaction Abrogates WT1 Target Gene Expression-Having defined the domains required for the interaction of WT1 with CBP, we sought to determine the effect of disrupting this protein association on WT1-mediated transcriptional activation. We first tested whether CBP3, the domain that mediates the interaction of CBP with WT1 but lacks functional histone acetyltransferase activity, could function as a dominant negative construct in reporter assays. Cotransfection of NIH 3T3 cells with the WRE-containing reporter (pGL2-B-⌬CRE) along with CMV-WT1(ϪKTS) alone (4 g) resulted in 8-fold transcriptional activation (Fig. 5, upper  panel). Coexpression of increasing amounts of CMV-CBP3 (2-8 g), fused to an SV40-derived nuclear localization signal, resulted in a dose-dependent reduction in activation. In contrast, co-transfection of CMV-driven constructs encoding other domains of CBP that fail to interact with WT1 had no effect (Fig.  5, upper panel), despite equal expression levels (Fig. 5, lower  panel). Transcriptional activation by WT1 may therefore be specifically inhibited by disruption of its interaction with endogenous CBP.
Transcriptional coactivation by both CBP and P300 is also known to be disrupted by their physical association with the adenovirus 5 early gene product E1A (31,40), and the binding of both WT1 and E1A to the CBP3 domain suggests that these proteins may compete for this binding site. We therefore examined the effect of E1A expression on WRE-dependent transcription. Co-transfection of CMV-driven E1A demonstrated a dosedependent suppression of CBP-mediated coactivation of the WT1-responsive promoter (Fig. 6). To extend these studies to an endogenous WT1-target gene, we examined the effect of E1A expression on induction by WT1 of the cyclin-dependent kinase inhibitor p21 Cip1 . WT1 directly activates the p21 Cip1 promoter, and induction of the endogenous p21 Cip1 gene is linked to WT1-mediated cell cycle arrest (14,21,41). More recently, we have found that high titer retroviral infection of U937 human leukemia cells with a mouse stem cell virus promoter-driven WT1 construct, leads to p21 Cip1 induction and triggers growth FIG. 5. Disruption of WT1-dependent transactivation by dominant negative CBP mutant. Relative luciferase activity following transient transfection of the amphiregulin promoter-reporter PGL-2B-⌬CRE (2 g) into NIH 3T3 cells, together with CMV promoter-driven constructs encoding WT1(ϪKTS) (4 g), and increasing amounts (2-8 g) of plasmids expressing each of the four CBP domains described in the legend to Fig. 4. Nuclear localization of the CBP fragments was ensured by insertion of a SV40-derived nuclear localization signal and comparable expression was demonstrated by Western blotting (data not shown). The total amount of CMV promoter sequence transfected was equalized by addition of empty vector. Data presented are derived from two sets of duplicate experiments and presented as fold induction relative to that of vector alone, with standard deviation shown. The expression level of each CBP deletion construct (5 g) is shown by Western blotting (lower panel).

FIG. 6. Inhibition of WT1-mediated transactivation by E1A.
E1A suppresses CBP-mediated enhancement of transcriptional activation by WT1. NIH 3T3 cells were transfected with the PGL-2B-⌬CRE reporter (2 g), along with CMV-driven WT1(ϪKTS) and increasing amounts (0.5 to 2.5 g) of CMV-driven E1A. Equal amount of DNA was transfected in each experiment and luciferase activity was compared with that of vector-transfected controls in each experiment. No difference in total cell numbers and viability were observed between samples. Experiments were performed in triplicate. arrest and differentiation. 2 We therefore used this in vivo assay to examine the effect of E1A, as well as the dominant negative fragment CBP3, on expression on a physiological WT1 target.
U937 cells were first infected with virus expressing E1A, CBP3, or vector control along with the neomycin resistance gene. Uncloned pools of infected cells were selected by treatment with G418, and then subjected to a second round of infection with virus encoding WT1(ϪKTS) or vector, linked to the green fluorescent protein (GFP) gene within a bicistronic construct. The expression level of each protein after selection was confirmed by Western blotting (Fig. 7A, lower panel). Following the second retroviral infection, ϳ35% of both vector-and WT1-infected cell populations expressed the GFP marker. No differences in cell viability were observed within the time studied (48 h after the second infection). The expression of p21 Cip1 in infected cells was quantitated by Western blotting and corrected for that of WT1 itself. As expected, U937 cells expressing WT1(ϪKTS) demonstrated dramatic induction of p21 Cip1 (Fig.  7B). In contrast, E1A-infected cells reproducibly demonstrated a 2-fold reduction in p21 Cip1 induction by WT1. Ectopic expression of E1A is associated with cellular toxicity, which limits the levels of protein expressed following retroviral infection. A variant E1A construct, E1A⌬CR2, lacking the CR2 domain between amino acids 121 and 139, has a preserved CBP interaction and reduced cellular toxicity, allowing for higher levels of protein expression (Ref. 33 and Fig. 7A, lower panel). In multiple experiments, retroviral infection of U937 cells with constructs encoding E1A⌬CR2 reproducibly demonstrated a 4-fold reduction in endogenous p21 Cip1 induction by WT1 (Fig.   7B). Infection of U937 cells with a construct expressing the dominant negative fragment CBP3 also resulted in a 2.5-fold reduction in p21 Cip1 induction by WT1. These observations suggest that disruption of the WT1-CBP interaction by either E1A or dominant negative CBP fragment, suppresses the transcriptional activation of an endogenous target gene by WT1.
Concluding Remarks-We have described a functional interaction between the tumor suppressor gene product WT1 and the transcriptional coactivator CBP. Recruitment of CBP to a target promoter is thought to enhance transcription by providing a platform that facilitates the assembly of additional transcription factors and components of the basic transcriptional machinery, and by mediating histone acetylation (for review, see Refs. 42 and 43). The observation that WT1 encodes a bona fide transcriptional activator that associates with CBP in vivo was unexpected, since WT1 has long been thought to act as a transcriptional repressor, whose function as a tumor suppressor is linked to the repression of proliferation-inducing genes (19). However, the identification of physiologically induced WT1 target genes (27,44,45), and the analysis of alterations and naturally occurring mutations that specifically disrupt transcriptional activation but not repression by WT1 (41,46), have suggested an alternative role for WT1, as an inducer of genes involved in cellular differentiation. According to this model, inactivation of WT1 during Wilms tumorigenesis may result in the failure of renal cells to differentiate, leading to the persistence of pluripotent stem cells susceptible to malignant transformation.
The interaction between WT1 and CBP also has implications for understanding functional properties of WT1 that have been reported to date. The consequences of WT1 expression in baby FIG. 7. Suppression of endogenous p21 Cip1 induction by WT1 in cells expressing E1A or CBP3. A, upper panel: flow chart describing the sequential retroviral infection strategy. U937 cells were first infected with retroviral constructs encoding vector, CBP3, E1A, or the E1A⌬CR2 construct with reduced cell toxicity, all linked to the neomycin resistance gene. Following G418 selection (4 -7 days), uncloned pools of resistant cells were infected with retroviral constructs encoding either vector or WT1(ϪKTS) linked to GFP. Equal infection efficiency (ϳ35%) and cell viability was confirmed for these constructs, and expression of p21 Cip1 and WT1 were monitored by Western blotting (ECL). Lower panel, expression level of each retrovirally encoded construct following G418 selection of uncloned cell populations is shown by Western blotting. B, inhibition by E1A and CBP3 of endogenous p21 Cip1 induction by WT1. The relative intensity of p21 Cip1 expression was normalized to that of WT1 itself (p21/WT1 ratio) in U937 cell populations stably infected with vector, E1A, E1A⌬CR2, or CBP3, 48 h following infection with vector or WT1(ϪKTS). The p21/WT1 ratio in E1A-expressing cells is expressed as a fraction of that in vector-infected cells. A representative experiment is shown. rat kidney cells, a well established primary cell transformation model in which cellular immortalization is achieved by expression of adenovirus E1A, should be interpreted with caution, given the likely disruption of WT1-mediated transcriptional activation by E1A (47,48). These observations also provide insight into the functional and physical association between WT1 and p53. These two proteins do not interact directly in vitro, but are co-immunoprecipitated from cellular lysates, and expression of wild-type p53 suppresses transcriptional activation by WT1, while expression of WT1 inhibits p53-mediated apoptosis (29,25). CBP is likely to mediate this indirect interaction, since it binds to both WT1 and p53 through adjacent domains, and WT1 zinc fingers 1-2 are required for its interaction with both CBP and p53. Overexpression of either WT1 or p53 may thus modulate the transactivational properties of the other (29,25).
A similar indirect association may underlie the synergistic effect of WT1 and steroidogenic factor 1, an orphan nuclear receptor implicated in gonadal differentiation. WT1 and steroidogenic factor 1 act synergistically to activate the promoter of Mullerian inhibitory substance, but WT1 does not bind directly to the Mullerian inhibitory substance promoter, nor do WT1 and steroidogenic factor 1 co-immunoprecipitate from cellular lysates (49). Given the known interaction between nuclear receptors and CBP, the effect of WT1 may therefore also result from an indirect association mediated by CBP. Likewise, the activation by WT1 of the CRE site in the amphiregulin promoter may result from its enhancement of CBP-dependent transactivation mediated by CRE-binding factors (27). Finally, these observations raise the possibility that the WT1(ϩKTS) isoform, which does not appear to bind to a known DNA sequence with high affinity, may in fact play an indirect role in transcriptional activation by virtue of its association with CBP.