Phosphatidylinositol 3-Kinase/AKT-mediated Activation of Estrogen Receptor a A NEW MODEL FOR ANTI-ESTROGEN RESISTANCE*

Estrogen receptors (ERs) mediate most of the biological effects of estrogen in mammary and uterine epithelial cells by binding to estrogen response elements in the promoter region of target genes or through protein-pro-tein interactions. Anti-estrogens such as tamoxifen in-hibit the growth of ER-positive breast cancers by reduc-ing the expression of estrogen-regulated genes. However, anti-estrogen-resistant growth of ER-positive tumors remains a significant clinical problem. Here we show that phosphatidylinositol (PI) 3-kinase and AKT activate ER a in the absence of estrogen. Although PI 3-kinase increased the activity of both estrogen-independent activation function 1 (AF-1) and estrogen-dependent activation function 2 (AF-2) of ER a , AKT increased the activity of only AF-1. PTEN and a catalytically inactive AKT decreased PI 3-kinase-induced AF-1 activity, suggesting that PI 3-kinase utilizes AKT-dependent and AKT-independent pathways in activat-ing ER a . The consensus AKT phosphorylation site Ser-167 of ER a is required for phosphorylation and activation by AKT. In addition, LY294002, a specific inhibitor

Estrogen-induced proliferation of mammary and uterine epithelial cells is primarily mediated by estrogen receptors (ERs), 1 which belong to steroid/thyroid hormone superfamily of transcription factors (1). ERs, through their estrogen-independent and estrogen-dependent activation domains (AF-1 and AF-2, respectively), regulate transcription by recruiting coactivator proteins and interacting with the general transcriptional machinery (2). Tamoxifen, which functions as a cell type-specific anti-estrogen, competitively binds to ER␣ and inhibits estrogen-stimulated growth of mammary epithelial cells. Depending on the concentration of tamoxifen, growth-arrested cells undergo apoptosis within 24 h or after 72 h of tamoxifen treatment (3). Although most ER␣-positive breast cancers initially respond to tamoxifen therapy, tamoxifen-resistant tumors eventually develop (4). It has been shown previously that growth factors such as epidermal growth factor (EGF), insulinlike growth factor (IGF-1), and heregulin confer estrogen-independent growth properties to ER␣-positive breast cancer cells (5)(6)(7). It is suggested that EGF-and IGF-1-induced mitogenactivated protein kinase (MAPK) phosphorylates Ser-118 of ER␣, increases the activity of AF-1, and confers hormoneindependent growth (6). However, a recent study indicates that, although prolonged activation of MAPK is growth inhibitory in breast cancer cells, parallel activation of the PI 3-kinase/AKT pathway by EGF and IGF-1 is sufficient to overcome such inhibition (8). Therefore, these growth factors may utilize the PI 3-kinase/AKT pathway to activate ER␣ and confer hormone-independent growth.
Growth factor-dependent survival of a wide variety of cultured cells types ranging from fibroblasts to neurons is dependent on PI 3-kinase pathway (9). Growth factor-induced activation of transmembrane receptors (IGF-1, EGF, platelet-derived growth factor, basic fibroblast growth factor, and heregulin) results in recruitment of PI 3-kinase to the plasma membrane (9). In the plasma membrane, PI 3-kinase promotes generation of 3Ј-phosphorylated phosphoinositides, which in turn bind to AKT. AKT bound to phosphoinositides is translocated from cytoplasm to the plasma membrane, where it is activated through phosphorylation. Apart from growth factors, estrogen also activates AKT by triggering the binding of ER␣ to the p85 regulatory subunit of PI 3-kinase (10). However, induction of AKT by estrogen is cell type-specific, as it is not observed in MCF-7 breast cancer cells (11). Activated AKT promotes cell survival by phosphorylating and modulating the activity of various transcription factors in the nucleus. The tumor suppressor PTEN gene, which dephosphorylates 3Ј-phosphoryl-* This work was supported by the Catherine Peachy Fund, Inc. and by American Cancer Society Grant RPG-00-122-01-TBE (to H. N.). 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.
* ated phosphoinositides in vivo, inhibits AKT activation (12). Several clinical and laboratory observations prompted us to study whether ER␣ is the target of AKT and whether AKT protects breast cancer cells against tamoxifen-induced apoptosis. For example, PI 3-kinase and AKT amplification is observed in breast and ovarian cancer (13)(14)(15). Cowden disease patients, who have a germ line-inactivating PTEN mutation, display increased breast cancer risks (16). Similarly, PTEN mutation is observed in endometrial cancer (17). Moreover, IGF-1 and heregulin, whose overexpression correlates with tamoxifen resistance, activate AKT (5, 7). Finally, serine at amino acid position 167 of ER␣ is a consensus AKT phosphorylation site (RXRXX(S/T)) (9). In this report, we show that AKT phosphorylates Ser-167 of ER␣ and confers tamoxifen resistance.
Recombinant Plasmids, DNA Transfections, and CAT Assays-MCF-7 and COS-1 cells were passed 24 h prior to transfection in PR-free DMEM ϩ 5% charcoal/dextran-treated serum (CCS) and transfected via the calcium phosphate method. The ␤-galactosidase expression vector RSV/␤-galactosidase (2 g) was cotransfected as an internal control. In all transfections, the total amount of expression vector was kept constant by substituting with appropriate empty expression vector. Fresh media (PR-free MEM ϩ 5% CCS for MCF-7, PR-free DMEM ϩ 5% CCS for COS-1 cells) along with appropriate drugs were added 24 h after transfection. Cells were harvested 36 h after transfection and CAT activity in equal numbers of ␤-galactosidase units was determined as described previously (18). ERE3-TK-CAT, GAL-AB and GAL-EF were gifts from P. Chambon and described previously (19). The CA-AKT expression vector that lacks pleckstrin homology domain (amino acids 4 -129) of AKT and KD-AKT expression vector that contains S179M FIG. 1. AKT specifically increases estrogen-independent activity of ER␣. A, MCF-7 cells were transfected with 5 g of ERE3-TK-CAT and 1 g of pcDNA3, CA-AKT, or KD-AKT. Cotransfection of AKT led to increases in estrogen-independent activity and further increases in estrogen-stimulated (E2, 10 Ϫ8 M) activity of ER␣. *, p ϭ 0.0019; **, p ϭ Ͻ0.0001; and ***, p ϭ 0.07 for CA-AKT versus pcDNA3 in the absence and presence of estrogen and 4-HT, respectively. Western blot analysis of extracts used in a representative CAT assay for CA-AKT expression is shown. B, PI 3-kinase-dependent increases in ERE3-TK-CAT activity. Western blot of p85 subunit of PI 3-kinase that coprecipitates with p110 subunit is also shown. *, p ϭ 0.0019 for untreated PI 3-kinase versus untreated pcDNA3 cells. **, p ϭ 0.0008 for estrogentreated PI 3-kinase versus estrogen-treated pcDNA3 cells. ***, p ϭ 0.0238 for 4-HT-treated PI 3-kinase cells versus 4-HT-treated pcDNA3 cells. C, AKT had no effect on retinoic acid (10 Ϫ6 M) inducible-expression of ␤-RARE-TK-CAT. In contrast, PI 3-kinase increased ␤-RARE-TK-CAT activity in the absence of retinoic acid. substitution have been described previously (20) and were gifts from R. Roth. COS-1 cell transfections were performed with the original SV40 enhancer-promoter-driven CA-AKT and KD-AKT expression vectors. pcDNA3 derivatives of CA-AKT and KD-AKT were generated and used in MCF-7 cell transfections. Constitutively active PI 3-kinase vector, which contains murine p110␣ cDNA with avian Src myristoylation signal at the N terminus (21), was a gift from S. Boswell. PTEN expression vector was a gift from J. Dixon.
Phosphorylation of GST-AB by AKT in Vitro-Amino acids 1-184 of ER␣ were cloned in frame to pGEX-2T vector to obtain recombinant GST-AB. AKT-kinase assay with recombinant GST, GST-AB, or AB obtained after cleaving GST-AB with thrombin in the presence of [␥-32 P]ATP was performed as per AKT manufacturer's recommendations (Upstate Biotechnology, Inc.).
AKT-overexpressing Clones and Growth Analysis-MCF-7 cells were transfected with either empty pcDNA3 or pcDNA3 vector encoding CA-AKT. Transfected cell populations were selected using G418 (0.6 mg/ml) and CA-AKT expression in individual G418-resistant colonies was verified by Western analysis. MCF-7 control (pcDNA-1) and CA-AKT (CA-AKT-4) cells were plated in MEM with PR and 10% FCS at a density of 2 ϫ 10 5 cells/60-mm plate. After 24 h, fresh medium containing either 10 Ϫ8 M estradiol or 10 Ϫ6 M 4-HT or 10 Ϫ7 M ICI-182,780 was added and replaced every 48 h with appropriate drugs. At 72, 96, and 120 h, cells were collected by trypsinization, stained with trypan blue, and counted twice using a hemocytometer. All experiments were performed in triplicate.
Cell Cycle Analysis and Quantitation of Apoptosis-MCF-7 control and CA-AKT cells were plated at a density of 1 ϫ 10 5 cells/60-mm dish in MEM with PR ϩ 10% FCS. After 24 h, fresh medium was added, along with 10 Ϫ6 M 4-HT. Cell cycle analysis utilizing propidium iodide staining was performed at specific intervals. Apoptosis of untreated and 4-HT-treated (10 Ϫ6 M for 72 h) control and CA-AKT cells was measured by annexin V labeling and fluorescence microscopy, as per manufacturer's recommendations (annexin V-EGFP, BioVision).
Orthophosphate Labeling and Immunoprecipitation of ER␣-ER␣ cDNA was first cloned into pCMV4 expression vector, which is similar to pcDNA3, except that it contains a translational enhancer from alfalfa mosaic virus 4. 293 cells were transfected with either empty expression vector or pCMV4-ER␣. After maintaining cells for 48 h in PR-free DMEM with 5% CCS, cells were serum-starved in PR-free, phosphatefree DMEM for 24 h. Cells were incubated with orthophosphate and either Me 2 SO or 20 M LY294002. IGF-1 (100 ng/ml, R&D Systems) was added after 1 h, and cells were harvested after an additional 1 h. ER␣ was immunoprecipitated and subjected to autoradiography, followed by Western blotting with ER␣ antibody.
Statistical Analysis-Data were analyzed with StatView (version 4.1). Analysis of variance was employed to determine p values between mean measurements. A p value Ͻ 0.05 was deemed significant. Error bars on all histograms in this text represent standard deviations between measurements from duplicate experiments.

RESULTS
AKT Confers Ligand-independent Activity to ER␣-To investigate the ability of AKT to directly regulate ER␣ activity, ER␣-positive MCF-7 breast cancer cells were transiently transfected with a CAT reporter gene under the control of a thymidine kinase promoter and three ERE elements (ERE3-TK-CAT). Estrogen increased ERE3-TK-CAT activity 6-fold, whereas 4-hydroxytamoxifen (4-HT) had no effect (Fig. 1A). Cotransfection of a constitutively active AKT expression vector (CA-AKT) resulted in 4.5-and 15-fold increases in ERE3-TK-CAT activity in the absence and presence of estrogen, respectively (Fig. 1A). Cotransfection of a catalytically inactive AKT vector (KD-AKT) affected neither estrogen-dependent nor estrogen-independent ERE3-TK-CAT activity. These results suggest that AKT confers estrogen-independent activity and further increases estrogen-stimulated activity of ER␣. Note that AKT-mediated estrogen-independent ERE3-TK-CAT activity was reduced to 2.5-fold by 4-HT.
A substantial (4-fold) increase in estrogen-independent activity was observed when ERE3-TK-CAT was transfected along with PI 3-kinase, the upstream activator of AKT (Fig. 1B). PI 3-kinase-induced ERE3-TK-CAT activity was reduced to 2.5fold by 4-HT. CA-AKT had no effect on basal and retinoic acid-inducible expression of ␤-RARE-TK-CAT (a TK promoter with a retinoic acid response element from the ␤-retinoic acid receptor) (18). In contrast, PI 3-kinase increased ␤-RARE-TK-CAT activity in the absence of retinoic acid (Fig. 1C). These results suggest that, although AKT overexpression leads to a FIG. 3. PI 3-kinase activates AF-1 and AF-2 via AKT-dependent and AKT-independent mechanisms, respectively. COS-1 cells were transfected with the indicated plasmids and pG5/CAT reporter. A, note that PI 3-kinase increased both GAL-AB and GAL-EF activity. In addition, KD-AKT reduced PI 3-kinase-induced GAL-AB activity, indicating that PI 3-kinase utilizes AKT to activate AF-1. B, PTEN reduced PI 3-kinase-induced GAL-AB activity, suggesting that activation of AF-1 by PI 3-kinase requires the presence of AKT. Expression levels of endogenous and transfected genes are shown. specific increase in ER␣ activity, PI 3-kinase overexpression may lead to a global change in nuclear receptor activity, possibly through activation of coactivators (2,22).
AKT Increases AF-1 Activity-To determine the activation domains of ER␣ required for estrogen-independent activation by AKT, we utilized expression vectors that code for a fusion protein containing the DNA binding domain (DBD) of yeast transcription factor GAL-4 and either the AF-1 domain (GAL-AB) or AF-2 domain of ER␣ (GAL-EF) (19). A CAT reporter gene under the control of a TATA box and five binding sites for GAL-4 DBD (pG5/CAT) was cotransfected with either GAL(DBD) alone (pM), GAL-AB, or GAL-EF with or without CA-AKT. COS-1 cells were used because they provide an ER␣free background. CA-AKT increased GAL-AB activity by ϳ3.5fold, which was not influenced by estrogen (Fig. 2). GAL-EF activity was observed only in estrogen-treated cells and was not affected by AKT. These results demonstrate that AKT specifically increases AF-1 activity and are consistent with previously described EGF-and IGF-1-regulated enhancement of AF-1 activity (6).
PI 3-Kinase Increases Both AF-1 and AF-2 Activity-PI 3-kinase, apart from AKT, activates other signaling pathways including Jun N-terminal kinase, protein kinase C and protein kinase C, which may contribute to a PI 3-kinase-mediated increase in ER␣ activity (23)(24)(25). To test this hypothesis, we transfected COS-1 cells with pG5/CAT, GAL-AB, or GAL-EF and a constitutively active PI 3-kinase (Fig. 3A). PI 3-kinase increased GAL-AB activity by ϳ8-fold. Cotransfection with KD-AKT resulted in reduction of PI 3-kinase-induced GAL-AB activity. PTEN, a negative regulator of PI 3-kinase-mediated activation of AKT (12), also showed a concentration-dependent reduction of PI 3-kinase-induced GAL-AB activity (Fig. 3B). For unknown reasons, we observed reduced levels of PI 3-kinase p110⅐p85 complex in cells transfected with 5 g of PTEN (Fig. 3B, bottom panel). Therefore, it is not clear whether reduction in GAL-AB activity by PTEN involves only its lipid Orthophosphate-labeled ER␣ from 293 cells transfected with either vector alone or pCMV4-ER␣ was identified by immunoprecipitation and autoradiography (lanes 1-4). Western blot analysis of the same blot for ER␣ protein is also shown. Relative phosphorylation in each reaction was calculated by densitometric scanning and normalizing for ER␣ level. Orthophosphate-labeled proteins that coprecipitate with ER␣ are also shown (lanes 5-8). NS, nonspecific. C, serine 167 in the AB domain is essential for AKT-mediated activation. COS-1 cells were transfected with either GAL-AB or mutants as in Fig. 2. AB-S167A mutant in which serine 167 is mutated to alanine was not activated by CA-AKT. Wild type and mutant GAL-AB proteins are expressed at a similar level in transfected cells (data not shown). phosphatase activity or other biological activity. Nonetheless, PTEN overexpression resulted in reduction in AKT activity in transfected cells (data not shown). Interestingly, GAL-EF, which was not induced by AKT (Fig. 2), was activated by PI 3-kinase both in the absence and presence of estrogen (Fig. 3A). Moreover, KD-AKT failed to reduce GAL-EF activation by PI 3-kinase, suggesting that activation of AF-1 (but not AF-2) by PI 3-kinase is, at least in part, AKT-dependent (Fig. 3A).
Ser-167 of ER␣ Is Essential for Phosphorylation and Activation by AKT-Phosphorylation of the AF-1 domain of ER␣ by AKT was investigated using purified AKT and a recombinant protein containing the AF-1 domain fused to glutathione S-transferase (GST-AB). GST and GST-AB proteins were incubated with [␥-32 P]ATP and recombinant AKT. GST-AB but not GST was phosphorylated by AKT, thereby confirming direct phosphorylation of AF-1 by AKT (data not shown). Mass spectrometry followed by mutation analysis revealed that AKT fortuitously phosphorylates Ser-173 or Ser-178 of AF-1 within the context of GST-AB fusion protein (data not shown). Therefore, purified wild type AB or mutants lacking GST were used in in vitro kinase assays. Mutation of Ser-167 (AB-S167A) but not Ser-118 (AB-S118A), Ser-173 (AB-S173A), Ser-178 (AB-A178A), and Ser-173/Ser-178 (AB-S173A,S178A) to alanine prevented phosphorylation by AKT (Fig. 4A). Ser-167 is a consensus AKT phosphorylation site (RXRXX(S/T)), which is also phosphorylated by p90 RSK1 (26), and is the major phosphorylation site of ER␣ in MCF-7 cells (27).
We next determined whether basal or IGF-1-induced phosphorylation of ER␣ is dependent on PI 3-kinase/AKT pathway. Human embryonic kidney cells 293 transfected with ER␣ expression vector were labeled with [ 32 P]orthophosphate and treated with Me 2 SO, IGF-1, or LY294002 ϩ IGF-1. LY294002 is a specific inhibitor of PI 3-kinase/AKT pathway. Although IGF-1 treatment led to only marginal increase in overall ER␣ phosphorylation, LY294002 inhibited ER␣ phosphorylation (Fig. 4B). The effect of LY294002 is specific to ER␣, as phosphorylation of an unknown protein that coprecipitates with ER␣ was similar in both untreated and LY294002-treated cells (Fig. 4B, lanes 5-8, indicated by **). These results further suggest the involvement of PI 3-kinase/AKT pathway in phosphorylation of ER␣.
We next compared AKT-mediated activation of wild type GAL-AB and GAL-AB mutants lacking Ser-102, Ser-104, and Ser-106 (AB-S102N,S104P,S106A), Ser-118 (AB-S118A), Ser-167 (AB-S167A), or Ser-173 and Ser-178 (AB-S173A,S178A). Only AB-S167A did not respond to AKT, suggesting that Ser-167 is essential for activation of AF-1 by AKT (Fig. 4C). Mutation of Ser-167 to alanine also reduced estrogen-independent and/or estrogen-dependent activity of full-length ER␣ in COS-1, MDA-MB-231, and HeLa cells (data not shown). AKTmediated activation of AB-S102N, S104P, S106A, and AB-S118A was also lower than the wild type, which could be due to reduced basal activity of these mutants. Alternatively, activation of AF-1 by AKT is partially dependent on phosphorylation of AB by cdk2/cyclinA or MAPK (6,28). In this regard, we have observed that mutation of both cdk2/cyclin A and AKT phosphorylation sites severely compromised ligand-independent activity of full-length ER␣ in MDA-MB-231 cells. 2 Endogenous Estrogen-regulated pS2 Gene Expression Is Upregulated by AKT-To study whether CA-AKT induces the expression of endogenous estrogen-regulated genes, we generated MCF-7 cells overexpressing CA-AKT (Fig. 5A). We measured the expression of pS2 and c-Myc in untreated, estrogen-, 4-HT-, and estrogenϩ4-HT-treated cells. The basal level of pS2 was higher in CA-AKT-4 cells compared with pcDNA3-2 and CA-AKT-2 cells (Fig. 5B). Estrogen-inducible expression of pS2 was higher in both CA-AKT-2 and CA-AKT-4 cells compared with pcDNA3 cells. Note that CA-AKT-4 cells express higher level of AKT than CA-AKT-2 cells (Fig. 5A). As with the results of transient transfection assays, constitutive and estrogen-inducible expression of pS2 was repressed by 4-HT. Interestingly, AKT overexpression did not result in changes in c-Myc expression.
Screening of Atlas Human Cancer cDNA expression array using RNA from pcDNA3 and CA-AKT cells revealed up-regulation of macrophage inhibitory cytokine (MIC-1) by AKT (29). This was confirmed by Northern blotting (Fig. 5B). MIC-1 expression was observed only in CA-AKT-4 cells, which was increased modestly by estrogen. Interestingly, 4-HT did not suppress MIC-1 expression. Taken together, these results suggest that AKT increases either constitutive or estrogen-inducible expression of specific estrogen-responsive genes.
AKT Confers Tamoxifen Resistance to MCF-7 Cells-Because AKT increases AF-1 activity and AF-1 is implicated in tamoxifen resistance, we studied the 4-HT sensitivity of pcDNA3-1 and CA-AKT-4 cells (Fig. 6A). Growth rates for untreated, estrogen-(10 Ϫ8 M), 4-HT-(10 Ϫ6 M), and estrogenϩ4-HT-treated cells were determined at specific intervals. As expected, estrogen stimulated the growth of both pcDNA3 and CA-AKT cells. 2 H. Nakshatri, manuscript in preparation. Also note that we reproducibly observed ϳ20% increase in the basal growth rate of untreated CA-AKT compared with untreated pcDNA3 cells. Significantly, 4-HT was more efficient in inhibiting the growth of pcDNA3 compared with CA-AKT cells, indicating that AKT reduces 4-HT sensitivity of MCF-7 cells. In contrast to 4-HT, CA-AKT cells were significantly growth-in-  (Fig. 6B), which inhibits ER␣ activity by preventing DNA binding and enhancing degradation (30). These results suggest that modulation of ER␣ activity by AKT is responsible for 4-HT resistance of CA-AKT cells, although we can not completely rule out the involvement of ER␣-independent anti-apoptotic activity of AKT in 4-HT resistance.
Reduced sensitivity of CA-AKT cells to 4-HT could be due to either loss of 4-HT-induced cell cycle arrest at G 0 /G 1 or inhibition of 4-HT-induced apoptosis. To resolve this issue, we measured the cell cycle distribution of untreated and 4-HT-treated pcDNA3 and CA-AKT cells. Both pcDNA3 and CA-AKT cells accumulated at G 0 /G 1 phase at a similar rate after 4-HT treatment (Fig. 6C). Apoptosis of untreated and 4-HT-treated (10 Ϫ6 M for 72 h) pcDNA3 and CA-AKT cells were measured by annexin V labeling and fluorescence microscopy (Fig. 6D). A significantly higher percentage of pcDNA3 cells (31%) were apoptotic when compared with CA-AKT cells (12%) after 4-HT treatment. AKT therefore inhibits 4-HT-induced apoptosis but not cell cycle arrest of MCF-7 cells.
AKT Increases Bcl-2 but Not Cyclin D1 Expression-Western blot analysis of untreated and estrogen-treated cell extracts was performed to identify estrogen-regulated cell cycle and anti-apoptotic genes that are overexpressed in CA-AKT cells (Fig. 7). Cyclin D1 expression proved estrogen-inducible in both cell types. AKT had no significant effect on cyclin D1 expres-sion. The known anti-apoptotic and estrogen-inducible Bcl-2 gene (31) was expressed at very low levels and did not appear to be up-regulated by estrogen in pcDNA3 cells. In CA-AKT-2 cells, which expresses lower levels of AKT than CA-AKT-4 cells, estrogen dramatically increased Bcl-2 expression. In CA-AKT-4 cells, Bcl-2 is constitutively overexpressed. Expression of anti-apoptotic Bcl-X L and pro-apoptotic BAK and BAX was similar in both pcDNA3 and CA-AKT cells. These results suggest that AKT specifically increases Bcl-2 expression in MCF-7 cells. Because Bcl-2 overexpression in CA-AKT-2 cells was observed only in the presence of estrogen, it is likely that increased Bcl-2 expression in CA-AKT cells requires ER␣ activity.

DISCUSSION
In this report, we demonstrate that PI 3-kinase/AKT signaling pathway modulates ER␣ activity in vivo, which correlates with phosphorylation of Ser-167 by AKT in vitro. Recent studies indicate that estrogen promotes association of ER␣ with IGF-1 receptor and p85 subunit of PI 3-kinase in the plasma membrane, which leads to AKT activation (10,32). Our results suggest that AKT serves as a functional link between membrane-associated and nuclear ER␣. By phosphorylating Ser-167, AKT may modulate coactivator:AF-1 and/or corepressor: AF-1 interactions in the nucleus. Consistent with this possibility, previous studies have shown that phosphorylation of Ser-118 of ER␣ alters both coactivator:AF-1 and corepressor: AF-1 interactions (33,34).
AF-1 region of ER␣ contains phosphorylation sites for a number of kinases including MAPK and cyclin A/cdk2 (6,28,35). Some of these sites are conserved between ER␣ and ER␤ (36). However, AKT phosphorylation site is present only in ER␣ (36). ERs exist as homodimers as well as ER␣⅐ER␤ heterodimers, and these combinations have different affinity for EREs (37,38). Therefore, induction of ERE containing genes upon activation of AKT may be cell type-dependent. Furthermore, regulation of ER␣ activity by AKT may be controlled by p90 RSK1 as it can also phosphorylate Ser-167 (26). Finally, MAPK may indirectly control AKT-mediated activation of ER␣ as it can activate p90 RSK1 (39).
We observed up-regulation of pS2, MIC-1, and Bcl-2 but not c-Myc in CA-AKT-overexpressing cells (Fig. 5B). These results suggest that the ability of AKT to induce ERE-containing genes is promoter context-dependent. Because 4-HT repressed pS2 but not MIC-1 and Bcl-2 expression in CA-AKT cells, promoter context may also play a role in determining the ability of 4-HT to suppress AKT-mediated activation of estrogen-responsive genes ( Fig. 5B and data not shown). Both pS2 and Bcl-2 are well characterized estrogen-regulated genes, whereas MIC-1 promoter is yet to be characterized. While induction of pS2 by estrogen requires an ERE (40), activation of Bcl-2 by estrogen requires a SP-1 binding site, cAMP response element-binding protein binding site, and an ERE in the Bcl-2 promoter (31,41). Interestingly, cAMP response element-binding protein binding site is required for AKT-mediated increase in Bcl-2 (42). Therefore, it is likely that the ability of 4-HT to repress AKT-mediated activation of ERE-containing promoters is determined by promoter elements other than ERE. Alternatively, AKT may activate certain ERE-containing promoters independent of ER␣, which is not subject to repression by 4-HT. Additional experiments with dominant negative mutants of ER␣ (43,44) are essential to distinguish between these possibilities.
Our results indicate that activation of the PI 3-kinase/AKT pathway leads to increased Bcl-2 expression, which correlates with 4-HT resistance in breast cancer cells. Consistent with our results, Her2/Neu-mediated 4-HT resistance in MCF-7 correlates with Bcl-2 overexpression (45). However, the results of our cell culture studies are inconsistent with clinical observations that correlate high Bcl-2 expression with favorable response to tamoxifen therapy and prolonged disease-free intervals (46). It is possible that Bcl-2 works in concert with additional anti-apoptotic proteins to confer tamoxifen resistance, which is in accordance with the proposal that overall level of Bcl-2 family anti-apoptotic proteins dictates cellular response to therapy (47).
The PI 3-kinase/AKT pathway is the major survival pathway for a wide variety of cultured cell types (9). ER␣ may be the central element in this survival pathway, at least in certain cell types. Broader implications of our results are on endometrial cancers as well as on Cowden's disease patients, who show a 30 -50% incidence of breast cancer in affected females (16). In both neoplastic conditions, AKT activity is increased due to mutation of PTEN. The PI 3-kinase/AKT pathway, therefore, may be an ideal target for therapeutic intervention in these cancers.