The Hormonal Response of Estrogen Receptor β Is Decreased by the Phosphatidylinositol 3-Kinase/Akt Pathway via a Phosphorylation-dependent Release of CREB-binding Protein*

The hormonal response of estrogen receptors (ER) α and ERβ is controlled by a number of cofactors, including the general transcriptional coactivator CREB-binding protein (CBP). Growing evidence suggests that specific kinase signaling events also modulate the formation and activity of the ER coactivation complex. Here we show that ERβ activity and target gene expression are decreased upon activation of ErbB2/ErbB3 receptors despite the presence of CBP. This inhibition of ERβ involved activation of the phosphatidylinositol 3-kinase/Akt pathway, abrogating the potential of CBP to facilitate ERβ response to estrogen. Such reduced activity was associated with an impaired ability of ERβ to recruit CBP upon activation of Akt. Mutation of serine 255, an Akt consensus site contained in the hinge region of ERβ, prevented the release of CBP and rendered ERβ transcriptionally more responsive to CBP coactivation, suggesting that Ser-255 may serve as a regulatory site to restrain ERβ activity in Akt-activated cells. In contrast, we found that CBP intrinsic activity was increased by Akt through threonine 1872, a consensus site for Akt in the cysteine- and histidine-rich 3 domain of CBP, indicating that such enhanced transcriptional potential of CBP did not serve to activate ERβ. Interestingly, nuclear receptors sharing a conserved Akt consensus site with ERβ also exhibit a reduced ability to be coactivated by CBP, whereas others missing that site were able to benefit from the activation of CBP by Akt. These results therefore outline a regulatory mechanism by which the phosphatidylinositol 3-kinase/Akt pathway may discriminate nuclear receptor response through coactivator transcriptional competence.

Estrogen mediates many aspects in growth, development, and reproduction, through its interaction with estrogen receptors ER 2 ␣ and ER␤. Although encoded by unique genes, the two ERs share the functional domains characteristic of the nuclear hormone receptor family (1). These consist of an N-terminal region (also termed AB region), which confers ligandindependent activation of ERs through its activation function (AF)-1, a highly conserved DNA-binding domain (C) that allows specific binding to genomic response elements, a flexible hinge region (D) that includes signals for nuclear localization and the binding of heat shock proteins, and finally a C-terminal region (EF) that contains the ligand binding domain, and the AF-2 function that mediates hormone-dependent activation.
Increasing evidence suggests that, beside hormonal activation, ER function can be modulated by phosphorylation-dependent mechanisms, involving a wide variety of protein kinases that mostly target the AF-1 domain (2,3). In particular, direct phosphorylation of ER␣ AF-1 by MAPK/ERK in response to EGF was shown to induce ER␣ transactivation in the absence of ligand (4,5). Similarly, phosphorylation of Ser-167 by pp90 RSK1 was described to promote ER␣ AF-1 activity (6). Activation of phosphatidylinositol 3-kinase (PI3K) and Akt/protein kinase B also contributed to phosphorylate ER␣ and mediate its ligand-independent activation, an effect shown to oppose the tamoxifen-induced apoptosis in breast cancer cells (7). Although phosphorylation of ER␤ has not been examined in detail, ER␤ has been proposed as a potential target for intracellular kinases that modulate its transactivation properties. It was found that the ability of EGF and the oncogene Ras to activate ER␤ resulted from the MAPK-directed phosphorylation of Ser-106 and Ser-124 within the AF-1 domain leading to favored recruitment of coactivators SRC-1 and CBP (8,9). Furthermore, the ligand-dependent activation of ER␤ by the protooncogene Brx was shown to involve phosphorylation of ER␤ in a p38-dependent manner, although the exact site(s) were not described (10). More recently, we reported that activation of ErbB2 and ErbB3, which belong to the EGFR/ErbB receptor tyrosine kinase family, by growth factor heregulin resulted in a decrease in the estrogen-dependent cell growth and activity of ER␣ and ER␤ in breast cancer cells (11). However, unlike ER␣, this transcriptional repression of liganded ER␤ by heregulin was dependent upon ER␤ AF-1 function, thereby supporting a repressive role for kinase-mediated pathways in regulating ER␤ AF-1 and AF-2 functions. Taken together, the regulation of estrogen receptor activity by phosphorylation is intricate and could dictate receptor function, whether it involves activation or repression.
Recent evidence has emerged suggesting that nuclear receptor coactivators may also serve as points of convergence between the ER and growth factor signaling pathways. Phosphorylation of SRC coactivators has been described to modulate their intrinsic activities in mediating nuclear receptor transcription (12). Coregulatory proteins are often present in limiting concentrations in the nucleus so that modifications of their level of expression as well as their activity can lead to alterations in nuclear receptor signaling. The transcriptional coactivators CREB-binding protein (CBP) and p300 are evolutionary highly conserved proteins, and genetic evidence supports their availability to be critical. In humans, loss of one functional copy of cbp leads to Rubenstein-Taybi syndrome, a haploinsufficiency disorder resulting in mental retardation (13). Through their extremely versatile ability in bridging numerous transcription factors, including most nuclear receptors, with the basal transcription machinery, recruitment of CBP/p300 is important to maintain appropriate transcriptional events (14). One of the likely mechanisms responsible for CBP/ p300 recruitment involves phosphorylation. It was reported that phosphorylation of CBP promotes its interaction with several transcription factors, including CREB, Smad3, NFB p65 subunit, and p53 (15,16). We have recently shown that MAPKdependent phosphorylation of ER␤ also facilitates the recruitment of CBP to potentiate the ligand-independent activation of ER␤ in response to growth factors (9). Given such diversity in the signaling pathways integrated by CBP, it is believed that phosphorylation-mediated events may compete at various levels for the limited availability of CBP.
Here we describe a molecular mechanism by which ErbB2/ ErbB3 and PI3K/Akt signaling impairs the activity of ER␤ by reducing its ability to recruit and use CBP as a coactivator. The repression by Akt was also found for other nuclear receptors, for which a conserved Akt site may also participate in a manner similar to ER␤. In contrast, nuclear receptors that do not share such homology yielded increased responsiveness to CBP and benefit from the enhanced intrinsic activity of CBP by Akt.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-Expression of pCMX plasmids coding for ER␣, ER␤, CBP, ErbB2, its constitutive variant V659E and ErbB3 receptors, and luciferase reporter constructs vitA 2 -ERE-tkLuc and UAStkLuc have been described previously (8,9,11). ER␤ fragments corresponding to the AB (amino acids 1-167) and DEF (amino acids 234 -549) regions were obtained by PCR amplification and fused in-frame with the Gal4 DNA binding domain. The ER␤ Ser-255 to alanine and the CBP Thr-1872 to alanine mutants were generated by PCR mutagenesis using Pfu polymerase (Stratagene). All constructs were verified by automated sequencing. The expression plasmid coding for the constitutively active PI3K p110␣ catalytic subunit was a kind gift from J. Downward, and plasmids expressing Akt and K179M kinase dead Akt were generously provided by T. Chan and P. Tsichlis.
Cell Culture, DNA Transfection, and Luciferase Assay-Human embryonic kidney 293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal bovine serum (FBS). The cells were maintained at 37°C in a humidified atmosphere with 5% CO 2 . For transient transfection, cells were seeded in phenol red-free DMEM supplemented with 5% charcoal dextran-treated FBS, and plasmid constructs were introduced into cells using the calcium phosphate precipitation method as described (11). Typically, 50 -60% confluent cells were transfected with 2 g of DNA per well, which include 500 ng of reporter plasmid, 100 ng of receptor expression vector, 250 ng of CMX-␤gal, 100 ng each of PI3K and Akt expression vector, and 30 ng of CBP plasmid when indicated. After 5-8 h, the medium was changed, and cells were stimulated with 10 nM estradiol (E 2 ; Sigma) and/or 50 ng/ml heregulin-␤ (R&D Systems) for 16 -20 h or left untreated. For luciferase assay, cells were lysed in potassium phosphate buffer containing 1% Triton X-100, and light emission was measured using a luminometer (Wallac) after the addition of luciferin. Luciferase assays were performed in duplicate from at least three independent experiments, and values were expressed as relative light units normalized to the ␤-galactosidase activity of each sample.
Western Analysis and Immunoprecipitation Assay-Western analysis for the determination of phosphorylated and total Akt was performed as described with minor modifications (11). Briefly, transfected 293T cells were treated with 50 ng/ml heregulin-␤ for 20 min, washed in ice-cold PBS, and lysed in PBS containing 0.5% sodium deoxycholate, 0.1% SDS, 1% Triton X-100, 1 mM sodium orthovanadate, 1 mM sodium fluoride, 1 mM phenylmethanesulfonyl fluoride, and protease inhibitors (Roche Applied Science). Cell lysates were then subjected to SDS-PAGE and proteins transferred to nitrocellulose for immunoblotting. Membranes were incubated at 4°C with blocking reagent (Roche Applied Science) in TBS, probed with either a rabbit polyclonal antibody against phosphorylated Akt (Santa Cruz Biotechnology) or a mouse anti-Akt monoclonal antibody (Cell Signaling Technology), and signals revealed by ECL using appropriate horseradish peroxidase-conjugated secondary antibodies. The same procedure was used to determine the levels of ER␤, except that cells were transfected with HAtagged ER␤ (WT or S255A) and analyzed by Western using an anti-HA antibody (12CA5). For immunoprecipitation assay, transfected cells were washed in ice-cold PBS and lysed as described above. Cell lysates were precleared before incubation with an anti-CBP antibody (Santa Cruz Biotechnology) and protein A-Sepharose beads at 4°C. Immunoprecipitates were then washed in lysis buffer, resolved by SDS-PAGE, and analyzed by Western blotting using an anti-HA antibody. Membranes were also probed with an anti-CBP antibody for standardization of CBP levels in each well.
Generation of Hs-ER-stable Clones and RT-PCR-ER-negative Hs578t breast cancer cells were maintained in DMEM containing 10% FBS and transfected with expression vectors for ER␣ and ER␤ as described previously (11), and resistant clones were isolated in the presence of G418 (0.6 mg/ml; Invitrogen) to generate Hs-ER␣ and Hs-ER␤ cell lines, respectively. Stable clones were functionally validated for their respective expression of ER␣ or ER␤ by Western analysis and for their estrogenic response by luciferase assay, compared with mock-transfected Hs-578t cells. Total RNA was isolated from cells using TRIzol reagent (Invitrogen), and RT-PCR analysis was performed as described (17). The relative signal intensity was analyzed (Alpha Innotech, San Leandro, CA) from three separate experiments.
In Vitro Phosphorylation Assay-Bacterially expressed and purified GST fusions of wild type and S255A mutated ER␤ were prepared as described (18). For in vitro phosphorylation assay, GST-ER␤ fusions immobilized on glutathione-Sepharose 4B beads were resuspended in kinase buffer containing [␥-32 P]ATP (Amersham Biosciences) and active Akt1 (Cell Signaling) and incubated at 30°C for 30 min according to the manufacturer's instructions. Beads were then washed twice in kinase buffer and twice in PBS, and 32 P incorporation was determined following SDS-PAGE and autoradiography. Gels were stained with Coomassie Blue to monitor for equal loading.
Fluorescence Microscopy-Cells were seeded on coverslips in a 6-well plate overnight prior to transfection in phenol red-free DMEM supplemented with 5% charcoal dextran-treated FBS. Transient transfections were carried as above using the expression plasmids YFP-CBP and CFP-ER␤. 20 h after transfection, cells were washed twice with cold PBS and fixed in 4% formaldehyde. The coverslips were mounted on microscope slides and examined in fluorescence with excitation/emission filters of 435/470 nm (for CFP) and 480/535 nm (for YFP) using a Nikon TE-2000 inverted microscope.

ErbB2/ErbB3
Receptor Dimer Activation Impairs the Hormonal Response and Coactivation of ER␤ by CBP-Activation of the epidermal growth factor receptor EGFR/ErbB1, a member of the ErbB receptor tyrosine kinase family, is well recognized to promote ER␣ and ER␤ transcriptional activation (4,8). However, we have recently reported that activation of the ErbB2/ ErbB3 heterodimer combination led to a decreased transcriptional activity of ER␤ (11). Given the ability of CBP to associate and promote the activation of ER␤ by a growth factor such as EGF (9), we addressed how CBP could modulate ER␤ activity in response to ErbB2/ErbB3 activation. ER-negative human embryonic kidney 293T cells were transfected with an EREtk-Luc reporter and an ER␤ plasmid. Cotransfection with CBP conferred a 2-fold increase in ER␤ basal activity and a 7-fold increase in the presence of hormone (Fig. 1A). As reported previously (11), the activation of the ErbB2/ErbB3 heterodimer by growth factor heregulin-␤, which binds ErbB3, resulted in a reduced activation of ER␤ by estrogen. Such impaired response was also mimicked using a constitutive variant of human ErbB2 (V659E), which corresponds to the natural mutation found in the rat Neu oncogene (19). However, although CBP strongly transactivates ER␤ in control cells, it is unable to prevent the inhibition of the hormonal response of ER␤ when the ErbB2/ ErbB3 heterodimer is not only expressed but is also stimulated by heregulin-␤ (Fig. 1A). Despite the presence of CBP, the transcriptional activity of ER␤ was decreased in both a hormoneindependent and -dependent manner. This inhibition was even more pronounced in cells expressing the constitutive ErbB2 V659E mutant.
Signaling of the EGFR/ErbB family members involves the activation of a variety of kinase pathways. More specifically, activation of the ErbB2/ErbB3 heterodimer has been shown to efficiently couple with the PI3K/Akt pathway, mainly through the intrinsic ability of numerous Src homology 2-binding motifs within ErbB3 that recognize the p85 regulatory subunit of PI3K (20,21). To evaluate the impact of ErbB2/ErbB3 activation on the Akt pathway, the activity of endogenous Akt was determined by Western analysis using a phospho-specific antibody against Ser-473. Although treatment of mock-transfected 293T cells with heregulin-␤ did not lead to activation of Akt, indicating that endogenous expression of ErbB3 is negligible, if not absent, an increase in phosphorylated Akt was observed in cells expressing ErbB2/ErbB3 and treated with heregulin-␤ (Fig. 1B). Similarly, cells expressing the ErbB2 V659E variant in the presence of ErbB3 also showed increased levels of phosphorylated Akt.

FIGURE 1. ErbB2/ErbB3 signaling impairs the hormonal response and coactivation of ER␤ by CBP. A, 293-T cells were transfected with an ER␤ plasmid and an EREtkLuc reporter, and expression vectors encoding ErbB2
and ErbB3, or a constitutively active ErbB2 (V659E) mutant in the presence or absence of CBP. Cells were then treated with 10 nM E 2 and/or 50 ng/ml heregulin (Hrg-␤) for 20 h and harvested for transcriptional activity. Luciferase values were normalized to ␤-galactosidase activity and expressed as fold activation compared with untreated cells set at 1.0. B, 293-T cells were transfected with ErbB plasmids and treated with 50 ng/ml heregulin for 20 min as indicated. Total cell extracts were analyzed by electrophoresis, and endogenous Akt phosphorylation was monitored by Western blotting using a specific anti-phospho Akt. Cell lysates were also analyzed for Akt content using an anti-Akt antibody.

Activation of the PI3K/Akt Pathway Mimics the Inhibition of ER␤ Response to Hormone in the Presence of CBP through the C-terminal Region of ER␤-
The possibility that ErbB2/ErbB3 activation by heregulin-␤ decreases ER␤ activity and its coactivation by CBP by enhancing the activity of Akt was further tested by transient expression of a membrane-bound and constitutively active p110␣ subunit (CAAX) of PI3K. Expression of the p110␣ mutant was sufficient to activate endogenous Akt in 293T cells, which was further enhanced when cells were cotransfected with a plasmid for WT Akt, as determined by Western blot analysis (see Ref. 22; data not shown). Under these conditions, we found that the estrogen-dependent activation of ER␤ in the presence of CBP, which reached almost 12-fold compared with untreated cells, was strongly impaired dropping to a 3-fold response in Akt-activated cells ( Fig. 2A). As observed previously in ErbB2 V659E-expressing cells in response to Akt activation (Fig. 1A), the addition of CBP reduces further the response of ER␤ to estrogen when compared with cells without exogenous CBP. These results suggest that ectopic expression of CBP could not relieve the inhibition of ER␤ by the PI3K/Akt pathway, therefore mimicking the results in ErbB2/ErbB3-expressing cells. The repression of ER␤ by activated Akt in the presence of CBP was partially relieved in cells expressing a dominant negative form of Akt (K179M), suggesting that the effects of activated PI3K on ER␤ mainly transit through Akt ( Fig. 2A). We next performed Western analysis to ascertain whether the modulation of ER␤ activity was not a direct effect of its protein concentration under the conditions used. As shown in Fig. 2B, activation of the Akt pathway led to an accumulation of ER␤ in untreated cells. A similar increase was also observed in the presence of estrogen, although the levels of ER␤ were slightly lower compared with untreated cells, probably reflecting an increase in ER turnover in response to hormone as reported previously (23). These results suggest that the inhibition in ER␤ activity to Akt activation is not related to a decrease in ER␤ protein levels.
CBP is known to transactivate estrogen receptors through both its AF-1 and AF-2 activities (9,24). In an attempt to identify the functional domain within ER␤ responsible for its impaired ability to be coactivated by CBP in response to Akt, we used Gal4 fusions of truncated forms of ER␤ for which each respective AF-containing domain has been removed. Fig. 2C shows that in Akt-activated cells, the activation of a Gal4-AB␤ (corresponding to ER␤ amino acids 1-167) on a UAStkLuc reporter was further enhanced by CBP, reaching a near 5-fold increase compared with control cells. The N-terminal domain of ER␤ is known to contain several serine residues that are conserved within the recognition motifs for Ser/Thr kinases of the MAPK family, and phosphorylation of specific residues was shown to allow for coactivators such as CBP to be recruited and to potentiate ER␤ AF-1 activity (8,9). However, none of the potential phosphorylation sites within ER␤ AB region belongs to a consensus Akt site, suggesting that the enhanced activity of ER␤ AF-1 by CBP in response to Akt might possibly result from other kinase pathways activated by Akt or direct effects on CBP itself. We next tested the role of the C-terminal region of ER␤ in the same conditions. Cells transfected with a Gal4-DEF␤ (amino acids 234 -549) showed a reduced hormone-dependent activity to Akt activation in the presence of CBP, mimicking the response observed with full-length ER␤ (Fig. 2C, right panel). These results indicate that the repressive effect of activated Akt on CBP-mediated transactivation of ER␤ is mediated through a region contained in the C-terminal portion of ER␤, which in the context of the full-length receptor seems to counteract the positive effect on the AF-1 activity.
Serine 255 in the Hinge Region Mediates ER␤ Inhibition to ErbB2/ErbB3 Signaling-Our examination of the C-terminal sequence of mouse ER␤ revealed a consensus sequence RQRSAS 255 in the hinge region of ER␤ that corresponds to the recognition motif RXRXX(S/T) for the kinase Akt (Fig. 3A). To determine whether Ser-255 is a direct target for Akt-mediated phosphorylation, we used site-directed mutagenesis to convert the serine at position 255 into an alanine and performed an in vitro kinase assay. Fig. 3B shows that disruption of Ser-255 strongly abolished the phosphorylation of ER␤ by Akt compared with wild type, indicating that Ser-255 can be efficiently phosphorylated by Akt. We then tested whether Ser-255 was involved in the inhibition of ER␤ activity to ErbB2/ErbB3 activation as observed in Fig. 1A. Using the S255A mutant in luciferase assay, we found that the inhibition observed for WT ER␤ by either ErbB2/ErbB3 dimer expression or its activation with heregulin-␤ was completely abrogated by disruption of Ser-255 FIGURE 2. The effect of Akt activation on ER␤ response to CBP coactivation is mediated by the C-terminal region of ER␤. A, 293T cells were transfected with the EREtkLuc reporter and expression plasmid for ER␤ in the presence or absence of CBP. Cells were also transfected with plasmids for Akt or its kinase dead K179M mutant form in the presence of the constitutively active p110␣ subunit of PI3K, as indicated. Cells were then treated with 10 nM E 2 for 20 h and harvested for transcriptional activity. Luciferase values were normalized to ␤-galactosidase activity and expressed as fold activation compared with untreated cells set at 1.0. B, Western analysis of ER␤ in response to Akt activation. 293T cells were transfected with ER␤ in the absence or presence of p110␣ PI3K and Akt plasmids to trigger the Akt pathway. Cells were then treated with 10 nM E 2 or left untreated for 20 h and harvested for Western analysis using an anti-ER␤ antibody. Loading was monitored with ␤-actin for each sample. C, cells were transfected with a UAStkLuc reporter and truncated forms of ER␤ corresponding to the N-terminal or AB region (left) or the C-terminal or DEF region (right) fused to the Gal-4 DNA binding domain. Cells were also transfected with p110␣ and Akt plasmids and treated with 10 nM E 2 for 20 h prior to luciferase assay. Luciferase values are expressed as in A.
(compare Figs. 1A and 3C). Noticeably, the hormonal response of S255A was enhanced upon ErbB2/ErbB3 activation and further potentiated by CBP. This enhanced response to hormone by the S255A mutant was also observed in response to Akt activation using the constitutively active p110␣ PI3K construct in transfection (Fig. 3C). Therefore, the results indicate the hinge region of ER␤ contains a specific site that not only can be targeted by Akt but also dictates responsiveness of ER␤ to CBP coactivation in response to Akt signaling pathway. To determine whether Ser-255 is involved in the regulation of ER␤ in terms of protein levels, we next performed Western analysis on cells expressing the ER␤ S255A mutant. As compared with wild type ER␤ (Fig. 2B), the disruption of Ser-255 completely abrogated the accumulation of ER␤ in response to Akt activation (Fig. 3D), indicating that Ser-255 is a critical site in the regulation of ER␤ levels by the PI3K/Akt pathway.
ER␤ Modulates the Intranuclear Behavior of CBP in an Aktdependent Manner through Serine 255-Studies using fluorescent-tagged proteins have demonstrated that expression of the estrogen receptor, in particular ER␣, affects the intranuclear organization of coactivators of the SRC/p160 family in response to hormone or anti-estrogens (25)(26)(27). Based on our results on the transcriptional response of ER␤ to CBP in Akt-activated cells, we investigated whether ER␤ could modulate the intranuclear behavior of CBP in response to Akt activation. Expression plasmids encoding a YFP-tagged full-length CBP and a CFPfused ER␤ were generated and functionally validated in luciferase coactivation assay (data not shown). We first determined the intranuclear distribution of CBP by transfecting cells with YFP-CBP in the absence of ER␤. In these conditions, CBP mainly localized into discrete clustered nuclear regions or foci, with a subpopulation showing a more diffuse pattern throughout the nucleus (Fig. 4A). This particular behavior of CBP has been observed in different cell types under basal or nonactivated conditions, and although not fully characterized, such a pattern was associated to poorly transcribing or transcriptional inactive compartments devoid of nascent mRNA transcription and active RNA polymerase II (28 -30). Given our results on the effects of CBP on ER␤ activity, we tested whether ER␤ could modulate the intranuclear distribution of CBP by cotransfecting cells with YFP-CBP and CFP-ER␤. Both proteins were shown to colocalize to the nucleus, but the ectopic expression of ER␤ strongly diminished the formation of CBP-related speckles, resulting in a more dispersed distribution of CBP throughout the nucleus (Fig. 4A). Interestingly, when the Akt pathway was activated in cells expressing both YFP-CBP and CFP-ER␤, CBP appeared to readopt the formation into speck-  Fluorescence signals were visualized using filters for YFP and CFP shown alone and merged. Cell nuclei were also stained with 4,6-diamidino-2-phenylindole (DAPI). B, CBP is released from ER␤ through Ser-255 in Akt-activated cells. Cells were cotransfected with HA-tagged WT or S255A ER␤ in the presence of CBP and then treated or not with 10 nM E 2 for 20 h. To activate the Akt pathway, cells were also transfected with PI3K p110␣ and Akt plasmids. Immunoprecipitation was carried out with an anti-CBP antibody, and ER␤ was analyzed by Western blot. CBP was also monitored in each sample by Western analysis. IB, immunoblot.
les, whereas the dispersion of ER␤ remained unaffected, indicating that Akt can induce a relocalization of CBP within the nucleus in the presence of ER␤ (Fig. 4A). Given the role of ER␤ Ser-255 to impair CBP-mediated coactivation of ER␤ in response to Akt, we next tested a CFP-ER␤ S255A construct on CBP intranuclear distribution. We observed that as opposed to WT ER␤, expression of the S255A mutant did not favor CBP to fully reform into speckles, but instead CBP remained in a more diffuse pattern (Fig. 4A). This distinct behavior of CBP in response to WT versus S255A ER␤ expression was also observed in the presence of hormone (data not shown), indicating that both the unliganded and liganded receptor affect CBP nuclear distribution to Akt activation in a similar manner. These results suggest that CBP relocalizes within the nucleus in response to Akt activation and that this behavior depends on the presence of ER␤ in a manner specific to Ser-255.
Cellular Activation of Akt Releases CBP from ER␤ through Serine 255-The observation that CBP could relocalize within the nucleus in a manner dependent of ER␤, and that Ser-255 seems to modulate that behavior in response to Akt activation prompted us to determine the effect of activation of the PI3K/ Akt pathway on the interaction of ER␤ with coactivator CBP. We found that under basal conditions CBP potently coimmunoprecipitated with ER␤ and that this interaction was further stabilized in the presence of estradiol (Fig. 4B), thus correlating with the enhanced activation of ER␤ by hormone and CBP (Figs. 1A and 2A). However, such interaction was strongly disrupted in Akt-activated cells independently on the presence of hormone (Fig. 4B). We then tested the S255A mutant using similar conditions and found that, as opposed to WT ER␤, CBP could efficiently immunoprecipitate the mutant ER␤ in the absence or in the presence of estradiol despite activation of Akt in cells (Fig. 4B). These results therefore provide a role for ER␤ Ser-255 to induce a release of CBP from ER␤ in response to Akt activation.
Akt Promotes the Intrinsic Transcriptional Activity of CBP through Thr-1872-CBP can be phosphorylated by several kinase signaling pathways, such as cyclin-dependent kinase or MAPK/ERK, leading to up-regulation of its histone acetyltransferase activity and therefore its intrinsic potential to activate transcription (31). To determine how CBP could affect transcription by ER␤ in response to Akt, we generated a Gal4 fusion of full-length CBP which, by interacting onto a UAStkLuc reporter, allows us to monitor directly CBP transcriptional activity in a luciferase assay. Cells transfected with Gal4-CBP showed a 4-fold activation in luciferase activity compared with cells expressing an empty Gal4 plasmid (Fig. 5A), indicating that CBP was able to promote transcription under these conditions. A further increase in CBP activity, reaching 8-fold activation compared with control, was observed upon expression of constitutive p110␣ PI3K and Akt in cells, suggesting that the intrinsic ability of CBP to promote transcription can be enhanced by Akt. By looking at the sequence of CBP, only one putative site (Thr-1872) is contained within the consensus motif for Akt. Interestingly, this site corresponds to Ser-1834 of p300, which was recently described as a target of Akt that promotes p300 activity (32). We therefore substituted Thr-1872 by an alanine residue and tested a Gal4-CBP T1872A mutant for its activity. We found that not only was the response to Akt activation completely abrogated by the mutation, but the basal activity was also severely impaired (Fig. 5A), indicating that Thr-1872 is a crucial regulatory site for CBP activity. The T1872A mutation did not significantly affect the steady-state levels of CBP expressed in cells, and Akt activation did not modulate wild type or mutated CBP levels as shown by Western analysis (Fig. 5B). Given the ability of CBP to respond to Akt through Thr-1872, we next tested whether this site was involved in the response of ER␤ and of ER␣ to Akt. Although the CBP T1872A mutant was less efficient in promoting ER␤ response to estrogen, we found that it behaves similarly as WT CBP in the inhibition of ER␤ by Akt, indicating that these effects were independent of CBP Thr-1872 (Fig. 5C). However, the activation of ER␤ S255A by Akt in the presence of WT CBP was lost when CBP T1872A mutant was expressed in cells. Similar results were obtained with ER␣ (Fig. 5C), suggesting that in contrast to ER␤, ER␣ seems to benefit from the enhanced activity of CBP to Akt in a manner dependent on Thr-1872.
Estrogen-responsive Genes Are Regulated Differently by Heregulin in ER-expressing Stable Clones-Based on our results on the apparent difference between the ER␣ and ER␤ response to CBP when Akt is activated and to delineate each ER contribution, we generated ER␣and ER␤-expressing stable clones using ER-negative Hs578t breast cancer cells. Hs578t cells are an appropriate model to study the effect of Akt because they exhibit high basal Akt activity through ErbB receptor signaling and mutated active Ras (33,34). In addition, Akt can be further activated by heregulin-␤ in each Hs-ER stable clone in a time- dependent fashion (Fig. 6A), indicating that these cells maintain the ability to respond to heregulin-␤ (35). Stable expression of ER␣ or ER␤ also confers enhanced estrogen-dependent activation of Akt compared with negative cells (Fig. 6A). Using RT-PCR analysis on cathepsin D1 (CatD1) and progesterone receptor (PR), two recognized estrogen-responsive genes, we found their expressions were enhanced by estradiol in both ER stable clones, compared with negative control cells (Fig. 6B). However, these increases were severely impaired by the addition of heregulin-␤ to Hs-ER␤ cells, therefore correlating with the results obtained in luciferase assays. In contrast, treatment of Hs-ER␣ cells with heregulin-␤ further potentiated the estrogen-stimulated expression of both genes (Fig. 6B). This suggests that the regulation of CatD1 and PR expression by ER␤ was more dependent on the effect of heregulin-␤ than the one through ER␣ (Fig. 6B). Under these conditions, the CBP steadystate levels were not significantly modified in Hs-ER␣ and Hs-ER␤ cells (Fig. 6C).
A Conserved Akt Site Can Predict the Transcriptional Response of Nuclear Receptors to CBP-Based on our results on the critical role of Ser-255 in regulating the response of mouse ER␤ to Akt and CBP coactivation, we checked whether the Akt motif containing Ser-255 was conserved within the nuclear receptor family. It should be noted that Ser-255 is located within the hinge region of ER␤, which is generally more conserved between ERs and orphan estrogen-related ERRs than with other nuclear receptors. As such, the sequence alignment in Fig. 7A showed that although ER␣ and all three isoforms of ERR contain the necessary arginine residue at position Ϫ3, and FIGURE 6. Estrogen-responsive genes are regulated differently by ER␣ and ER␤ in response to heregulin-␤. A, activation of Akt in stable Hs-ER␣ and Hs-ER␤ clones in response to heregulin-␤ and estrogen. ER␣ and ER␤expressing stable clones have been generated using ER-negative Hs578t cells (control) and were treated with 50 ng/ml heregulin-␤ for the indicated time or 10 nM E 2 for 60 min. Endogenous Akt phosphorylation was monitored by Western blotting using a specific anti-phospho Akt. Cell lysates were also analyzed for Akt content using an anti-Akt antibody. B, RT-PCR analysis of ER-responsive genes from Hs-ER clones and Hs control cells treated with 10 ng/ml heregulin-␤ and/or 50 nM E 2 for 20 h prior to RNA isolation. Representative images are shown from at least three separate experiments. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression was used to normalize samples. C, Western analysis of CBP in Hs-ER stable clones and Hs control cells treated as above. Samples were normalized for protein loading with ␤-actin.

FIGURE 7.
A conserved Akt site dictates the transcriptional response of nuclear receptors to CBP. A, sequence alignment for predicted Akt phosphorylation site of nuclear receptors. Shown are the predicted phosphorylated serines (arrow) with the obligatory arginine residues at position Ϫ3 and the less stringent arginine/lysine residues at position Ϫ5 of the receptor sequences aligned with mouse ER␤. The predicted Akt site is conserved in human and mouse ERR␤ and GR. B, response of nuclear receptors to Akt and CBP coactivation. 293T cells were transfected with expression plasmids for the indicated nuclear receptors with their respective luciferase reporter as follows: EREtkLuc for ER␣ and the estrogen-related receptor ERR isoforms; GREtkLuc for GR and PR; PPREtkLuc for PPAR␥; and CREtkLuc for the cAMPresponsive binding protein CREB. Cells were also transfected with CBP and/or p110 ␣/Akt plasmids and treated with ligands as follows: 10 nM E 2 (ER␣), 10 nM dexamethasone (GR), 10 nM progesterone (PR), or 1 M troglitazone (PPAR␥) for 20 h prior to determination of luciferase activity. Normalized luciferase values are expressed as fold activation compared with untreated cells set at 1.0 for each receptor. the less stringent arginine/lysine residues at position Ϫ5 of the canonical site for Akt (36) in their respective hinge regions, only ERR␤ possesses the expected phosphorylated serine (Fig. 7A). It is interesting to note that as opposed to the mouse and rat isoforms, human ER␤ does not contain a serine at the corresponding position but rather has a negatively charged aspartic acid residue. In addition, human, mouse, and rat forms of ER␣ do not share the conserved serine residue, having a glycine or leucine when aligned with ER␤ Ser-255 (mER␣ is shown in Fig.  7A). Although no perfect consensus site for Akt could be found in glucocorticoid receptor (GR) and PR, a putative Akt site conserved in mouse and human GR was found with the required arginine residue at position Ϫ3 and was aligned with ER␤. To address how other nuclear receptors responded to Akt-and CBP-mediated coactivation and to find a possible correlation with respect to their sequence homology with ER␤ Ser-255, we tested various nuclear receptors in the luciferase assay. Using an ERE-driven luciferase reporter known to bind and respond to ERRs as dimers (37), we found that coexpression of CBP increased the activity of the three ERR isoforms by 2-3-fold in 293 cells (Fig. 7B). Interestingly, when Akt activation was induced with p110␣ PI3K expression, CBP only failed to further transactivate ERR␤, whereas ERR␣ and ERR␥ reached 3-and 5-fold activation, respectively, compared with controls. Under these conditions, the response of ERR␤ to Akt activation and the inability of CBP to potentiate transactivation strongly correlate with what we observed with ER␤, and therefore point to a shared role for the putative Ser-191 Akt site of ERR␤ that overlaps with Ser-255 of ER␤. This observation also applies to GR for which CBP-mediated coactivation was severely abrogated in response to Akt. Conversely, among the receptors tested that do not share homology with ER␤ Ser-255, we found that ER␣, PR, and peroxisome proliferator-activated receptor PPAR␥ were all further activated by CBP in Akt-activated cells in the presence of their respective ligand (Fig. 7B). CBP, which has originally been described to directly interact with the cAMPresponsive binding protein CREB, also potentiated CREB activation to Akt. Hence, the impaired ability of CBP to transactivate ER␤ in response to Akt can be transposed to other receptors that share an apparent homology with the ER␤ Ser-255-containing motif. Accordingly, at least for those receptors tested that do not fit into that category, they seem to benefit from the enhanced intrinsic activity of CBP in response to activation of the PI3K/Akt pathway.

DISCUSSION
Increasing evidence suggests that besides ligand activation, nuclear receptors are responsive to kinase signaling mechanisms, and for estrogen-responsive tissues in particular, this may represent a mean to regulate the different ER-mediated transcriptional pathways (2,3). More recently the idea that signaling pathways can also mediate transcriptional repression of estrogen receptors has led us to further investigate how these pathways are tightly controlled (11,38). Here we show that activation of ErbB2/ErbB3 receptors and the PI3K/Akt pathway can impair the transcriptional response of ER␤ to estrogen and its coactivation by CBP. The mechanism underlying ER␤ inhibition involves Ser-255, which upon its phosphorylation by Akt prevents CBP from interacting with ER␤, therefore abrogating ER␤ activity.
Dimerization of ErbB3 with its preferred partner ErbB2 is considered the most potent combination of ErbB receptors in terms of cellular growth and transformation (39). Deregulated signaling by ErbB2/ErbB3 has been associated with detrimental mitogenic potential in a number of reproductive cancers and the correlation of ErbB2 with ER␣ status has served as a predictive factor in endocrine-based therapy (40,41). However, the response of ER␤ to ErbB2/ErbB3 activation is not clearly defined, and the exact role of ER␤ in tumorigenesis remains uncertain. We found that the transcriptional response of ER␤ to estrogen was diminished upon activation of ErbB2/ErbB3 with ErbB3 ligand heregulin or the constitutive ErbB2 variant V659E derived from the Neu oncogene. These results have been transposed to ER␤-expressing stable breast cancer cells, therefore altering endogenous ER-responsive genes, as observed with the down-regulation of CatD1 and PR. Both conditions were associated with increased cellular activation of Akt. Intriguingly, although the coactivator CBP potently contributes to enhance basal and estrogen-dependent response of ER␤, it became inefficient to render optimal activation of ER␤ in Aktactivated cells. These effects seem to be specific to CBP because we showed that the coactivator SRC-1 was able to relieve ER␤ inhibition to heregulin signaling (11). Based on our observations that SRC-1 and CBP can trigger ER␤ response to growth factors in an AF-1-dependent manner (8,9), it was predicted that both coactivators would behave similarly. In an attempt to delineate the role of the AF-1 domain, we found by using an N-terminal form of ER␤ that CBP promoted ER␤ activation to Akt, suggesting a positive effect of the Akt pathway that obviously did not correlate with the response of the full-length receptor. Although the N-terminal region of ER␤ contains phosphorylation sites described to be directly phosphorylated by MAPK conferring AF-1 activity of the receptor in response to EGF or Ras (8,18), it does not have a consensus site for Akt, and therefore up-regulation in AF-1 activity by Akt might relate to possible indirect effects, including activation of CBP itself, as predicted in Fig. 5A. Removal of the AF-1 region demonstrated a similar inhibitory pattern as observed with the full-length ER␤, and further identified Ser-255 as a functional site responsible for the inhibition of ER␤ to ErbB2/ErbB3 and Akt signaling. Together, these findings clearly demonstrate that many signaling events converge to ER␤ to regulate cofactor assembly and transcriptional activity either in a positive or negative manner.
Our observation that ER␤ cellular levels were augmented by the PI3K/Akt pathway in the presence or absence of estrogen raised the possibility that ER␤ turnover is regulated by Akt. Interestingly and consistent with this idea is the apparent opposite regulation of the S255A mutant in the same conditions, suggesting that Ser-255 is a determinant involved in ER recycling in response to Akt signaling. Studies using ER␣ have integrated the response to estrogen with the cellular degradation of the receptor, thus supporting a means by which target cells can sustain or limit a hormonal response through a continuous receptor turnover. ER␣ has been shown to be degraded through the proteasome pathway in a ligand-dependent manner (23,42), and blocking proteasome activity impaired the ability of ER␣ to mediate a transcriptional response to hormone (43)(44)(45), suggesting that ER turnover is necessary for receptor activity. Similarly, activation of the PI3K/Akt through platelet-derived growth factor stimulation of smooth muscle cells was shown to target CREB for degradation in a phosphorylation-dependent manner (46).
Recent studies derived from fluorescent-based approaches have revealed the dynamic nature of ER␣ within the nucleus and its behavior with transcriptional coactivators in response to hormonal stimuli (26,44,47,48). Under basal conditions and in the absence of ER␤, CBP adopted a speckled pattern with a subpopulation being more diffuse within the nucleus. The reason for such behavior is unclear, but the ability of CBP to form speckles has been observed in different cell types under nonactivated conditions, and has been associated with poorly transcribing or transcriptionally inactive compartments devoid of nascent mRNA transcription and active RNA polymerase II (28 -30). The speckled clustering of CBP has also been shown to not segregate with regions of histone hyperacetylation, suggesting a decreased activity of CBP (49). However, such compartmentalized pattern of CBP was not always related to transcriptional inactivity, as the promyelocytic leukemia protein was identified as a nuclear receptor coactivator that segregates CBP into nuclear bodies (50). Interestingly, the expression of ER␤ resulted in a marked decrease in speckle formation and a more diffuse pattern of CBP throughout the nucleus that overlapped with the distribution of ER␤. This colocalization of ER␤ and CBP occurred in the absence or presence of estrogen, therefore correlating with the enhanced activation of ER␤ by CBP in luciferase assays. However, the activation of the PI3K/Akt pathway has the distinct effect of driving CBP to readopt a speckled pattern, whereas ER␤ remained diffused, coinciding with a reduced ER␤ activity. Although these studies did not allow assessing directly the interaction between CBP and ER␤, it is interesting to note that although the S255A mutant was tested, the formation of CBP-related foci was greatly reduced in Aktactivated cells. The expression of ER␤ therefore allows for a redistribution of CBP in the nucleus, which implicates Ser-255 as a determinant in the response to Akt. Consistent with these observations, activation of Akt led to a release of CBP from ER␤ even in the presence of estrogen as determined in the coimmunoprecipitation assay, whereas disruption of Ser-255 was found to stabilize such interaction. These observations emphasize the role of Ser-255 in mediating CBP release from ER␤ in a phosphorylation-dependent process. Although phosphorylation provides an important mechanism by which steroid hormone receptors can be activated (3), increasing evidence suggests that phosphorylation also mediates nuclear receptor inhibition or repression involving various mechanisms and different kinase pathways. Phosphorylation of serine 236 by cAMP-dependent protein kinase was reported to impair ER␣ dimerization and hence transcriptional activity (51). Phosphorylation of AF-1 Ser-112 by MAPK reduced the ligand binding affinity and activity of PPAR␥ (52). In the case of the androgen receptor, Ser-210 and Ser-790 were identified as phosphorylation sites for Akt, which inhibited the association of AR with coactivator ARA70 (53). Our results therefore provide a mechanism by which ErbB2/ErbB3 and Akt signaling impairs ER␤ activity through a phosphorylation-dependent release of coactivator CBP.
CBP/p300 are general signal integrators common to many transcription factors, and evidence suggests that part of the mechanism that regulates their function involves direct phosphorylation (14). Interestingly, phosphorylation of Ser-1834 by Akt was shown to promote p300 histone acetyltransferase activity and its transcriptional potential (32). By mutating the corresponding site within CBP, we observed that Thr-1872 is essential to promote CBP enhanced transcriptional capacity in response to Akt activation. However, ER␤ was not able to benefit from this improved activity as opposed to the S255A mutant, suggesting that phosphorylation of ER␤ at Ser-255 may prevail in the response of ER␤ to Akt. Indeed, phosphorylation of Ser-255 impaired CBP recruitment to ER␤ and did not allow for proper CBP-mediated coactivation, therefore preventing any potential of CBP to activate ER␤. A similar mechanism was described in the inhibition of C/EBP␤-targeted gene expression by insulin, except that the phosphorylation of Ser-1834 in the C/H 3 domain of p300 by Akt prevented p300 to interact with C/EBP␤ (54). CBP Thr-1872 is also contained in the C/H 3 domain, which is described to mediate the recruitment of many transcription factors to CBP/p300 (14). However, our results suggest that Thr-1872 of CBP does not regulate the transcriptional response of ER␤ to Akt. A recent report has described the interaction of ER␣ with the CBP C/H 3 domain in the presence of an anti-estrogen, as opposed to the previously recognized N-terminal interaction domain of CBP for agonist-bound nuclear receptors, but whether phosphorylation of CBP was involved has not been determined (55).
Genetic studies have established that the cellular availability of CBP is critical for normal physiologic functions, and as a coactivator that integrates the effects of several transcription factors, this may represent a mean by which CBP can discriminate between various regulatory pathways (16,56). As such, while testing other members of the nuclear receptor family, we found that unlike ER␤, the activation of ER␣ by Akt was potentiated in the presence of CBP and further contributed to enhance the expression of known ER target genes such as those encoding CatD1 and PR, in stably ER␣-expressing breast cancer cells. ER␣ does not contain the corresponding Ser-255 found in ER␤, but an Akt site within ER␣ AF-1 domain, which is absent in ER␤, has been described to functionally activate ER␣ (7, 57). Such isoform-selective coactivation of ERs by CBP may represent a mechanism by which CBP can discriminate between ER␣and ER␤-regulated pathways in response to Akt signaling. This mechanism can become important in pathologic conditions such as early breast cancer, in which activation of Akt is extremely frequent as a consequence of ErbB2 amplification (58). Clinically, Akt activation strongly correlates with ER␣ in breast tumors, whereas the prognostic value of ER␤ is not established (40,59). It therefore seems interesting to propose that the negatively charged aspartic residue that corresponds to mouse Ser-255 could predict a reduced response of human ER␤ to CBP coactivation. Clearly, further studies are needed to unravel these distinctions.
The ER isoform-specific effect of CBP by the PI3K/Akt pathway has also been observed between ERR members. As opposed to ERR␣ and ERR␥, ERR␤ contains a consensus for Akt found within the same region as ER␤ and was found negatively regulated by Akt in the presence of CBP. Although structurally closely related to the ERs, the ERRs do not exhibit estrogen binding and are still considered orphan receptors without a known endogenous ligand. However, our results predict that ERRs can be selectively regulated by kinase signaling pathways such as PI3K/Akt. With the emerging role of ERR isoforms in modulating ER functions and target gene expression (37,60), it will be of interest to investigate whether such regulation might influence these aspects.
The present findings demonstrate a molecular mechanism by which the PI3K/Akt pathway may dictate the activity of ER␤ and other nuclear receptors, through their selective ability to use CBP as a coactivator. With the impact of ErbB2 signaling and/or Akt activation pathways to also affect CBP intrinsic coactivation properties, elucidation of the various regulatory signals that dictate nuclear receptor-coactivator functions might provide insights into their integrative function.