Estrogen Receptor- (cid:1) Binds p53 Tumor Suppressor Protein Directly and Represses Its Function

Estrogen receptor- (cid:1) (ER (cid:1) ) promotes proliferation of breast cancer cells,whereastumorsuppressorproteinp53impedesproliferationofcellswithgenomicdamage.Whetherthereisadirectlinkbetweenthesetwoantagonisticpathwayshasremainedunclear.HerewereportthatER (cid:1) binds directly to p53 and represses its function. The activation function-2 (AF-2) domain of ER (cid:1) and the C-terminal regulatory domain of p53 are necessaryfortheinteraction.Knockingdownp53andER (cid:1) bysmallinter-fering RNA elicits opposite effects on p53-target gene expression and cell cycle progression. Remarkably, ionizing radiation that causes genomic damage disrupts the interaction between ER (cid:1) and p53. Ionizing radiation together with ER (cid:1) knock down results in an additive effect on transcription of endogenous p53-target gene p21 ( CDKN1 ) in human breast cancer cells. Our findings reveal a novel mechanism for regulating p53 and suggest that suppressing p53 function is an important component in the pro- proliferative role of ER (cid:1) . As a tumor suppressor, p53 plays a

As a tumor suppressor, p53 plays a central role in cellular processes such as cell cycle arrest, apoptosis, senescence, and differentiation (1,2). Although these functions of p53 are essential to prevent cells from becoming cancerous, left uncontrolled, they can lead to consequences deleterious to normal cells. Mutations in the p53 gene or aberrations in the mechanisms to balance p53 function pave the way to tumorigenesis (3). p53 elicits its biological functions mainly by functioning as a transcriptional regulator of various cellular genes with p53-response elements. On the other hand, estrogen receptor-␣ (ER␣) 5 regulates growth and development of various tissues and promotes proliferation of breast cancer cells (4 -8). ER␣ is a transcriptional regulator that is recruited to the promoter regions of target genes directly through binding to estrogen response elements (EREs) or indirectly through other DNA-binding factors, such as AP1 and Sp1 (7,9). The opposing functions of p53 and ER␣, while stringently controlled in normal cells, are likely disrupted in cancer cells. Various observations have alluded to the potential for a cross-talk between p53 and ER␣ signaling pathways. For example, in murine models, early exposure to 17␤-estradiol (E 2 ) and progesterone to mimic pregnancy induced nuclear p53 enabling resistance to carcinogenesis by blocking proliferation of apparently ER␣-positive cells (10). In breast cancer cells, increased expression of ER␣ led to elevated levels of p53 and MDM2, an inhibitor of p53 function (11), whereas overexpression of MDM2 enhanced the function of ER␣ (12). However, whether there is a direct link between the p53 and ER␣ pathways has remained unclear. To address this important issue, we investigated whether ER␣ directly interacts with p53 and affects its function.

EXPERIMENTAL PROCEDURES
Cell Culture and Irradiation-MCF7 cells and Saos2 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) (Invitrogen) or 10% dextran-charcoal-treated FBS at 37°C under 5% CO 2 . MCF7 cells were irradiated at a dose of 12 grays in a ␥-cell 40, S/N 20 irradiator. Control cells were mock-irradiated.
Transient Transfection and Inhibition of Endogenous ER␣ and p53 Expression by siRNA-Saos2 cells maintained in Dulbecco's modified Eagle's medium containing 10% dextran-charcoal-treated FBS were transfected with Ϫ1265 PCNA-luc, p53, and wild type and mutant ER␣ expression plasmids or empty vector using FuGENE 6 (Roche Diagnostics). After 24 h, cells were processed for measuring transcriptional activity by luciferase assay (Promega). For knock down experiments, ER␣, p53, and nonspecific (NS) siRNA (Dharmacon) were transfected into MCF7 cells with Lipofectamine 2000 (Invitrogen). To ascertain the specificity of siRNAs used, the mRNA and the protein levels as well as effects on functional targets of the targeted protein were monitored.
Quantitative Real-time (qRT) PCR-Total RNA from MCF7 cells was isolated using the "Absolutely RNA Miniprep Kit" (Stratagene). For analyzing transcription of p21, 1 g of total RNA was reverse-transcribed in 20 l of reaction using the "SuperScript III First-Strand Synthesis System" (Invitrogen). One l of the resulting cDNA was used in a total volume of 25 l of PCR reaction. Real-time PCR was carried out in an Applied Biosystems Prism 7900 Sequence Detection System using TaqMan PCR master mixture, probes, and primers (Applied Biosystems). The relative mRNA levels in MCF7 cells transfected with specific versus NS siRNA were calculated using the ⌬⌬C t method with the endogenous ␤-actin mRNA as control.
Immunoprecipitation-MCF-7 cells irradiated or mock-irradiated were washed twice with phosphate-buffered saline, lysed in NENT buffer (100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 20 mM Tris-HCl, pH 8.0) containing 1ϫ Halt protease inhibitor mixture (Pierce). The extracts were cleared by centrifugation for 10 min at 14,000 rpm at 4°C and 2.5 mg of each lysate was precleared with agarose-conjugated mouse IgG (2 h, 4°C) and subsequently incubated with agarose-conjugated mouse IgG or mouse monoclonal p53 antibody (DO-1, Santa Cruz Biotechnology) overnight at 4°C. Agarose beads with immunoprecipitated proteins were washed four times with NENT buffer, boiled in SDS sample buffer, and resolved by SDS-PAGE followed by Western blotting using ECL method (Amersham Biosciences).
Chromatin Immunoprecipitation (ChIP)-ChIP assays were performed on MCF7 and Saos2 cells using the ChIP kit (Upstate Biotechnology) as per the manufacturer's instructions with minor modifications. In the case of Saos2 cells, 24 h prior to cross-linking, cells were transfected with p53 and ER␣ expression plasmids either individually or in combination. DNA-p53-ER␣ complexes were immunoprecipitated using antibodies against p53 (DO-1) or ER␣ (HC-20) (Santa Cruz Biotechnology). DNA after decross-linking, and purification was subjected to PCR with AccuPrime TaqDNA polymerase * This work was supported by National Institutes of Health Grant CA-79911, the Charlotte Geyer Foundation, and Wendy Will Case Cancer Fund, Inc. (to G. M. D.), and by the NCI/National Institutes of Health Comprehensive Cancer Center Grant to the Roswell Park Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental data and Figs. S1-S5. 1 These authors contributed equally to this work. 2  (Invitrogen). PCR products resolved on 2% agarose gels were visualized after staining with ethidium bromide. GST Pull-down Assay-BL21(DE3) pLySs bacteria expressing GST or wild type and mutant p53 proteins fused to GST were lysed in NENT buffer containing protease inhibitor mixture (Roche Diagnostics). Cells were sonicated, and the homogenate was centrifuged at 10,000 ϫ g for 15 min at 4°C. Proteins were bound to glutathione-Sepharose 4B beads (Amersham Biosciences). In vitro translated, [ 35 S]Methionine-labeled (TNT Kit, Promega) wild type or mutant ER␣ was incubated with immobilized GST or GST fused to wild type or mutant p53 proteins in lysis buffer containing 0.5% nonfat dry milk at 4°C for 1 h. Beads were washed four times and analyzed by SDS-PAGE followed by fluorography.
Cell Cycle Analysis-MCF7 cells were harvested 24 or 48 h after siRNA transfection. Cell cycle distribution was assayed by staining total cellular DNA with propidium iodide, and results were analyzed by ModFit software. DNA synthesis was determined by the BrdUrd incorporation assay using BrdUrd Flow Kit (Pharmingen). Cells were labeled with BrdUrd for 45 min at 37°C. After fixation, cells were treated with DNase for 1 h at 37°C to expose incorporated BrdUrd, followed by incubation with fluorescein isothiocyanate-conjugated anti-BrdUrd antibody. Total DNA was stained with 7-aminoactinomycin D. Flow cytometry was performed on FACScan (Pharmingen). Data were analyzed using Winlist (Verify Software House) software.

RESULTS AND DISCUSSION
ER␣ Binds to p53-We used coimmunoprecipitation assay to analyze association between endogenous ER␣ and p53 proteins in MCF7 (human breast cancer cells containing wild type ER␣ and wild type p53) cells. ER␣ was detected by immunobloting in the cell lysate immunoprecipitated by p53 antibody (Fig. 1A). As genomic damage is known to activate p53 (15) and affect its interaction with proteins such as MDM2 and Pin1, we investigated whether the ER␣-p53 interaction is sensitive to genomic damage. When MCF7 cells were exposed to ␥-radiation, the interaction between endogenous ER␣ and p53 was disrupted (Fig. 1A). A GST-p53 pull-down assay showed that p53 bound to in vitro-translated wild type ER␣, A/B domain deletion mutant ER␣ ⌬1-179, and DNA-binding domain mutant ER␣ C201H/C205H (Fig. 1B, lanes  3-5), whereas ER␣ ⌬283-595 was unable to bind to p53 (Fig. 1B, lane 6), indicating that the 283-395 region containing the ligand-binding and activation function-2 domains is necessary for binding to p53. A similar experiment with different GST-mutant p53 proteins mapped the interaction domain of p53 to the 75 amino acids (319 -393) at the COOH terminus (Fig. 1C). It is intriguing that the COOH terminus that is extensively modified in a stress-dependent manner and critical for regulation of p53 function (16 -18) is targeted by ER␣. Purified ER␣ bound specifically to bacterially expressed GST-p53 resolved on a denaturing gel in a far western assay (supplemental Fig. S3A), further demonstrating direct ER␣-p53 interaction. An electrophoretic gel mobility shift assay with extracts of Saos2 (human osteosarcoma cells, p53 Ϫ/Ϫ ; very low ER␣) cells transfected with p53 and ER␣ expression plasmids showed that ER␣-p53 interaction can occur when either one is bound to its cognate binding site on DNA (supplemental Fig. S1).
Interaction of ER␣ with p53 Bound to Target Gene Promoters in Vivo-As a first step toward investigating how the ER␣-p53 interaction affects the transcription of p53-target genes, we used the ChIP assay to analyze their interaction in vivo on a p53-target gene, p21 (CDKN1), in MCF7 cells. Interaction at two p53 sites (5Ј site, Ϫ2.2 kb and 3Ј site, Ϫ1.3 kb) (19) was assayed along with a control nonspecific region (Ϫ4.3 kb) in the promoter. As expected, p53 bound to both the 5Ј and 3Ј sites in the non-irradiated (Fig. 2A, lane 3) and in the ␥-irradiated cells (Fig. 2A, lane 4). In the non-irradiated cells, ER␣ was also bound to both the 5Ј and 3Ј sites (Fig. 2A, lane 7). However, when cells were subjected to genomic damage by ionizing radiation, consistent with the immunoprecipitation data (Fig. 1A), ER␣ binding to both sites was reduced considerably ( Fig. 2A, lane 8 and quantitative ChIP data in supplemental Fig. S3B). A similar pattern (data not shown) of p53 and ER␣ binding was observed on the p53-response element of the PCNA gene, another transcriptional target of p53 (20). Immunoblotting analysis showed that the ER␣ expression was unaffected, and the p53 level, as expected, was increased by irradiation (Fig. 2B). Therefore, the reduced binding of ER␣ to p53 on the p21 and PCNA promoters in response to ionizing radiation is not because of altered protein levels. In MCF7 cells where p53 was knocked down, ER␣ was unable to interact with the p21 promoter (Fig. 2C). Similarly, in p53 Ϫ/Ϫ Saos2 cells, ER␣ was able to bind to the p53-target promoter only when exogenous p53 was expressed ( Fig. 2D; compare lanes 2 and 4). These results demonstrate that ER␣ is accessing the pro-moter via the p53 bound to DNA. Furthermore, E 2 augments the ER␣-p53 interaction, although it is not necessary for interaction (Fig. 2E). Consistent with such an effect of estrogen on the interaction, transcription of p53-target gene p21 is also reduced in the presence of E 2 (Fig. 2F). This data along with the finding from a global gene expression profiling in E 2 -treated MCF7 cells (31) suggest that E 2 has a direct repressive effect on p21 transcription. Interestingly, early exposure of rodents to pregnancy levels of E 2 and progesterone leads to nuclear accumulation of p53 leading to p53-dependent response to exposure to chemical carcinogens and radiation (21). Besides, E 2 has been reported to increase p53 gene expression in MCF7 cells (22), and p53, on the other hand, transcriptionally up-regulates ER␣ (23). These observations along with our data showing mutual regulation of ER␣ and p53 levels (Fig. 3D) suggest that delicate mechanisms are at play in balancing the antagonistic functions of these proteins.
ER␣ Represses Transcriptional Activation by p53-Having observed an ER␣-p53 interaction in vivo on the p53 response element of endogenous p53 target gene promoters, we analyzed the functional consequences of this interaction. In a transient transfection assay in Saos2 cells, p53, as expected (13), activated transcription of the PCNA-luciferase reporter with wild type p53binding site (Fig. 3A) but not the reporter with a mutant p53-binding site (data not shown). Wild type ER␣ repressed transcriptional activation of the PCNA promoter by p53. However, two transcriptionally defective ER␣ mutants (ER L539A, a coactivator-binding mutant, and ER ⌬283-595, a deletion mutant FIGURE 1. ER␣ interacts with p53. A, extracts of mock-or ␥-irradiated MCF7 cells were subjected to immunoprecipitation with agarose-conjugated mouse IgG or p53 antibody followed by immunoblotting with ER␣ antibodies. "Input" was 2% of total proteins used for immunoprecipitation. B, GST pull-down assay was performed using GST-p53 protein and 35 S-labeled ER␣ wild type (wt) and mutant proteins translated in vitro. A representation of wild type ER␣ protein is shown at the top (domains: AF-1 and -2, activation function-1 and -2; DBD, DNA-binding domain; LBD, ligand-binding domain). C, GST pulldown assay to map the domains of p53 interacting with ER␣ was performed using fulllength and mutant GST-p53 recombinant proteins and 35 S-labeled ER␣ translated in vitro. The bottom panel shows Coomassie-stained GST-p53 proteins resolved on SDS-PAGE. A representation of wild type p53 protein is shown at the top (domains: TA, transactivation; Tm, tetrameriztion; Reg, regulatory). that lacks the entire C terminus containing the activation function-2 domain) were not effective repressors of transcriptional activation by p53. Similar levels of p53 and ER␣ proteins in cells transfected with p53 and ER␣ expression constructs either alone or in combination (supplemental Fig. S4) indicate that repression was due to interaction of ER␣ with p53 bound to the PCNA promoter and not as a consequence of altered p53 or ER␣ expression. Immunofluorescence followed by confocal microscopy showed that both the wild type ER␣ and the ER L539A mutant that is incapable of repressing transcriptional activation by p53 (Fig. 3A), but capable of binding to p53 in vitro (data not shown), colocalized with p53 in the nucleus (supplemental Fig. S2). Thus, unlike in the case of antagonism between GR and p53 where p53 was sequestered in the cytoplasm (24), the repression of p53 function by ER␣ is not due to unavailability of p53 in the nucleus but rather is a consequence of, as shown by the ChIP assay (Fig. 2), interaction of ER␣ with p53 bound to the endogenous target gene promoter.
As ectopic expression of ER␣ can repress the transcriptional transactivation by p53 in transfected Saos2 cells, we examined the impact of knocking down endogenous ER␣ on p53 activity in MCF7 cells. Analysis of mRNA by qRT-PCR showed that transcription of the endogenous p21 gene started to increase as early as 8 h after transfecting ER␣ siRNA (Fig. 3B) suggesting that increase in FIGURE 2. In vivo interaction of ER␣ with p53 bound to target gene promoter is disrupted by ionizing radiation. A, ChIP assay was performed on non-irradiated or ␥-irradiated MCF7 cells with primers (horizontal arrows) abutting p53-binding sites and a nonspecific site on the p21 promoter. A representation of p53-binding sites and NS site tested is shown at the top. The middle panel shows agarose gel analysis of PCR products from ChIP assay. B, immunoblotting of p53 and ER␣ expression in mock-and ␥-irradiated MCF7 cells. C, interaction of ER␣ with p53 bound to human p21 promoter in MCF7 cells transfected with either NS or p53 siRNAs was analyzed by ChIP. D, interaction of ER␣ with p53 bound to human PCNA promoter in Saos2 cells transfected with p53 and ER␣ expression plasmids individually or together was analyzed by ChIP. E, effect of E 2 on ER␣-p53 interaction in MCF7 cells was analyzed by ChIP assay. F, effect of E 2 on p53-target gene (p21) expression in MCF7 cells was analyzed by qRT-PCR. Ab (or ab), antibody. p21 transcription is not a consequence of cell cycle arrest. Furthermore, transcription of p21-luciferase reporter was considerably enhanced when ER␣ was knocked down by siRNA, whereas transcription was severely reduced when either p53 alone or p53 and ER␣ in combination were knocked down (supplemental Fig. S3C). As expected, ␥-irradiation resulted in increased p21 transcript (Fig. 3C) and protein (Fig. 3D) levels in MCF7 cells. The combination of ER␣ knock down and irradiation produced an additive effect on induction of p21 transcription that led to increased levels of the protein (Fig. 3, C and D). Furthermore, knocking down p53 prevented the radiation-induced increase in transcription, whereas knocking down both p53 and ER␣ had very little additional effect on p21 transcription indicating that ER␣ is targeting p53-mediated activation of p21 transcription. Interestingly, transcription of another p53 target, the FAS/APO-1/CD95 (1) gene, was increased in response to ionizing radiation; however, unlike p21, ER␣ knockdown did not result in increased FAS gene transcription (data not shown). Thus, the ER␣ effect on transcriptional activation by p53 appears to be gene-selective, reminiscent of observations that promoter context-specific assembly of transcription complexes could influence function of transcriptional regulators such as p53 (25,26).
ER␣ Counteracts p53 during Cell Cycle Progression-As MCF7 cells are dependent on active ER␣ for proliferation, we tested the possibility that ER␣ may counter p53 during cell cycle progression. Flow cytometric analysis showed that the percentage of cells in the S phase decreased in ER␣ knock down cells, whereas it increased in the p53 knock down cells (Fig. 4A). This observation is consistent with increased p21 (Fig. 3D) transcription in ER␣ knock down cells. Furthermore, BrdUrd incorporation assay revealed decreased S-phase entry in ER␣ knock down cells, but an opposite effect was observed when p53 was down-regulated (Fig. 4B). Together, these data indicate that antagonizing p53 function is an important component in the proproliferative role of ER␣. It is likely that ER␣ also negatively regulates p53mediated apoptosis. Future experiments should address this issue.
In conclusion, we have uncovered a novel mechanism by which ER␣, generally up-regulated in breast cancer (6,8), suppresses p53 function. Increasing evidence indicates that p53 dysfunction is an important event in breast cancer (27,28). It is possible that interaction between p53 and ER␣ resulting in their reciprocal regulation plays an important role in regulating normal breast epithelial cell proliferation, and aberration of this control may lead to breast cancer onset and progression. Future studies should reveal the mechanisms and consequences of the interaction under different physiological and pathological contexts. Furthermore, whether ER␤ (29) plays any role in this regulatory mechanism remains to be elucidated. Besides unraveling a novel cellular pathway, these results have important clinical implications. ER␣ binding to p53 resulting in functional inactivation could be one of the reasons for the inability of wild type p53 to restrain the tumor growth and metastasis in ER␣-positive breast cancer. Our observations open the opportunity for clinical evaluation of the importance of ER␣-p53 interaction in tumors containing wild type ER␣ and p53. ER␣ and p53 status of tumors may be useful in developing a rational basis for combination therapy with ER␣ modulators and radiation. Furthermore, identification of agents that can specifically disrupt protein-protein interaction (30) between p53 and ER␣ might have cancer preventive and therapeutic potential.