Enhanced Estrogen Receptor (ER) (cid:1) , ERBB2, and MAPK Signal Transduction Pathways Operate during the Adaptation of MCF-7 Cells to Long Term Estrogen Deprivation*

The mechanisms involved in resistance to estrogen deprivation are of major importance for optimal patient therapy and the development of new drugs. Long term culture of MCF-7 cells in estrogen (E2)-depleted medium (long term estrogen deprivation; LTED) results in hypersensitivity to E2 coinciding with elevated levels of estrogen receptor (ER) (cid:1) phosphorylated on Ser 118 and MAPK, together with several of its downstream targets associated previously with ER (cid:1) phosphorylation. Our data suggest elevated MAPK activity results from enhanced ERBB2 expression in the LTED cells versus the wild-type (wt), and treatment with the tyrosine kinase inhibitor ZD1839 revealed increased sensitivity in both transcription and proliferation assays. Similarly the MEK inhibitor U0126 decreased transcription and proliferation in the LTED cells and reduced their sensitivity to the proliferative effects of E2, while having no effect on the wt. However, the complete suppression of MAPK activity in the LTED cells did not inhibit ER (cid:1) Ser 118 phosphorylation suggesting that ER activity remained ligand-dependant. The LTED cells also expressed elevated levels of insulin-like growth

The knowledge that steroids play a pivotal role in the development of breast cancer has been exploited clinically by the development of endocrine treatments (1). These have sought to perturb the steroid hormone environment of the tumor cells, predominately by withdrawal or antagonism of estrogen. Unfortunately, the beneficial actions of existing endocrine treatments are attenuated by the ability of tumors to circumvent the need for steroid hormones, while in most cases retaining the nuclear steroid receptors (2). The identification of the factors and pathways responsible for the development of these resist-ant conditions is therefore paramount for the design of new diagnostics and therapeutic regimes (reviewed by Ali and Coombes (3)).
In an attempt to elucidate these mechanisms, our laboratory and others (4 -6) have developed in vitro models to study the molecular changes associated with long term estrogen deprivation (LTED). 1 Our previously published studies demonstrated that MCF-7 cells deprived of E2 for over 80 weeks passed through three distinct phases: quiescent (LTED-Q) followed by a hypersensitive phase (LTED-H), where basal cell growth was stimulated by the addition of E2 at concentrations below 10 Ϫ13 M, and finally an apparent independent phase (LTED-I) in which exogenous E2 no longer affected their growth. Our studies also revealed that the LTED MCF-7 cells expressed elevated levels of ER␣ that was phosphorylated on Ser 118 in the absence of exogenous E2 suggesting this was a contributing factor to the acquired hypersensitivity of these cells (6).
It has been postulated that in the majority of endocrineresistant tumors, control over growth is assumed by locally acting autocrine or paracrine peptide growth factors. These in turn activate the cell signal transduction pathways by binding to receptors on the cell surface. Several studies have suggested a potential role for the p42/p44 MAPK signaling pathway in the initiation and pathogenesis of breast cancer. For instance p42/ p44 MAPK activity appears to be elevated in primary breast cancer compared with benign breast tissue (7). Activation of the p42/p44 MAPK cascades modulates the phosphorylation and hence activity of several nuclear transcription factors that in turn regulate a series of genes. The various MAPK family members play a complex role in the determination of cell growth, differentiation, and programmed cell death, and this is thought to involve a balance between competing MAPK pathways (8).
ER␣ is functionally regulated via phosphorylation by several protein kinases (9 -17). These phosphorylation events are believed to play a pivotal role in regulating many aspects of steroid hormone receptor function including DNA binding and transcriptional activation. Ser 118 within the AF-1 domain of the ER has generated much interest. Phosphorylation of Ser 118 * 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.
§ To whom correspondence should be addressed. Tel.: 44-0-207-808-2883; Fax: 44-0-207-376-3918; E-mail: martinl@icr.ac.uk. is mediated by cyclin-dependent kinase-7 in response to E2 and is also phosphorylated by p42/p44 MAPK in a ligand-independent manner (9,18). Interestingly Ser 167 is phosphorylated by p90 RSK , which itself is activated by p42/p44 MAPK; hence increased MAPK activity could result in endocrine resistance. Another pathway thought to play an important role in ER activity and endocrine resistance is the PI3 kinase signaling pathway, which, together with its downstream target AKT (19), promotes cellular proliferation and anti-apoptotic responses. Recent studies have demonstrated phosphorylation of ER by AKT on Ser 167 also results in ligand-independent activation (17,20). Although ER activity has traditionally been associated with transcriptional regulation, several studies have implicated ER in non-genomic effects and demonstrated the ability of estrogens mediated by ER to activate the MAPK and PI3 kinase signaling pathways (21)(22)(23)(24).
The exact mechanism responsible for the development of the LTED phenotype and the presence of phosphorylated ER␣ is poorly understood. We postulate that the mechanism involves a dynamic interplay between several signal transduction pathways participating in ER␣ phosphorylation and cell growth. To address this, we characterized and compared the p42/p44 MAPK and PI3 kinase signal transduction pathways, in LTED-I and wild-type MCF-7 cell lines, to determine whether changes in activity of these phospho-proteins contributed to the development of the apparent estrogen independent phenotype noted in the LTED-I cell line (6).
Tissue Culture-Wild-type (wt) MCF-7 human breast cancer cells were maintained in phenol red-free RPMI 1640 supplemented with 10% fetal bovine serum (FBS), 10 g/ml insulin, 2 mM glutamine, 100 units of penicillin-streptomycin, and 10 Ϫ9 M estradiol. Long term estrogendeprived (LTED-I) MCF-7 cells were maintained in medium depleted of steroids and referred to as DCC medium. This comprised RPMI 1640 medium containing 10% (v/v) dextran charcoal-stripped (DCC) FBS, 2 mM glutamine, 100 units of penicillin-streptomycin, and 10 g/ml insulin. Cells were passaged weekly, and medium was replenished every two to three days.
Cell Growth Assays-Wt MCF-7 cells where depleted of steroids for 3 days and then seeded into 12-well plates at a density of ϳ1 ϫ 10 4 cells per well in DCC medium. LTED-I cells were treated similarly. The cells were left for a further 2 days, to acclimatize. Cell monolayers were subsequently treated with steroids or inhibitors for 6 days with daily changes. The cell number was determined using a Z1 Coulter counter (Beckman Coulter). In studies determining the effect of insulin on cell growth, both wt and LTED-I cells were cultured in DCC medium minus insulin for 3 days prior to seeding 12-well plates. The cells were allowed to acclimatize in DCC medium minus insulin for a further 2 days prior to stimulation with DCC medium containing increasing doses of E2 plus or minus 10 g/ml insulin. In studies determining the effect of ZD1839 both wt MCF-7 and LTED cells were seeded at a lower density of 5 ϫ 10 3 cells per well. This was to ensure that cells remained subconfluent as ERBB2 expression is cell density-dependent (25).
Cell Stimulation Assays-For stimulation assays wt MCF-7 cells (which had previously been stripped of steroids for 3 days by culturing in DCC medium) and LTED-I cells were seeded at a density of 1 ϫ 10 6 cells per 10-cm dish. The following day monolayers were washed once with phosphate-buffered saline and then transferred to serum-free me-dium for 24 h. Cells were then stimulated with the appropriate agent as indicated in the figure legends.
Preparation of Whole Cell Extracts for Immunoblots-Cell monolayers were washed with ice-cold phosphate-buffered saline and then lysed in extraction buffer (1% (v/v) Triton X-100, 10 mM Tris-HCl, pH 7.4, 5 mM EDTA, 50 mM NaCl, 50 mM sodium fluoride, 2 mM Na 3 VO 4 , and one tablet of Complete TM inhibitor mix (Roche Applied Science) per 10 ml of buffer) and homogenized by passage through a 26-gauge needle six times. The lysate was incubated on ice for 10 min and then clarified by centrifugation (14,000 rpm for 10 min at 4°C). The protein concentration was then quantified using bovine serum albumin reagent (Bio-Rad). Equal amounts of protein (50 g unless stated otherwise) were resolved by SDS-PAGE and then subjected to immunoblot analysis. Antigen-antibody interactions were detected with Super-signal reagent (Pierce). Chemiluminescence was quantified using Fluro-S and analyzed using Quantity One software (Bio-Rad).
Transcriptional Analysis-LTED-I and wt MCF-7 cells (previously stripped of steroids for 3 days) were seeded in 24-well plates at a density of ϳ4 ϫ 10 4 and ϳ8 ϫ 10 4 cells per well respectively, in DCC medium. The following day the cells were transfected by Lipofectin (Invitrogen) with 0.25 g of EREIItkluc (luciferase reporter plasmid) and 0.25 g of pCH110 (␤-galactosidase for normalizing luciferase data) for 4 h. The cells were subsequently fed with DCC medium and left to recover overnight, before treatment with the appropriate concentration of E2, modulators, or vehicle. After treatment for 24 h the luciferase and ␤-galactosidase activity were measured using a luminometer.

Removal of Insulin from the Growth Medium Restored
Hypersensitivity to E2 in LTED-I Cells-Our previous study (6) of the development of the LTED phenotype showed that the cells pass from a hypersensitive phase to an apparent independent phase where the addition of E2 had no effect on cell growth. We hypothesized that rather than being ligand-independent, the LTED-I cells were in fact super-sensitized to the residual E2 in the DCC medium, to the degree that no further growth stimulation was possible with added E2.
Evidence suggests that insulin and insulin-like growth factors are potent breast cancer cell mitogens able to act synergistically with E2 (26). Because the wt MCF-7 and LTED cells were maintained in medium containing insulin, we examined whether removal of insulin would affect the basal and E2stimulated growth rate of the two cell lines. Removal of insulin resulted in an ϳ50% drop in the basal growth rate of the LTED-I cells while having no effect on the wt MCF-7 (compare Fig. 1, A and B). In the absence of insulin, E2 stimulated the growth of the wt cells, and this effect was enhanced by addition of insulin (Fig. 1D). In the absence of insulin, the hypersensitivity of the LTED-I cells to the effects of E2 was restored with doses as low as 10 Ϫ12 providing an ϳ50% increase in cell growth (Fig. 1C). However, in the presence of insulin, addition of 10 Ϫ12 M E2 had no effect on growth compared with the untreated control whereas doses in excess of 10 Ϫ10 M E2 resulted in a decrease in cell growth (greater than 50%). In the absence of insulin, these elevated doses of E2 had no marked effect on cell growth compared with the control.
The removal of insulin appears to reveal a persistent hypersensitivity in the LTED-I cells, which is imperceptible in the presence of insulin, as the cells have reached their maximum growth rate. Hence cross-talk between the insulin-like growth factor and steroid signaling pathways may be involved in the adaptation of the cells to E2 deprivation.
The Pure Anti-estrogen ICI 182780 Inhibits LTED-I Cell Growth-The pure anti-estrogen ICI 182780 prevents activation of AF-1 and AF-2 and reduces the half-life of the ER. Thus if the growth of the LTED-I cells was ER-independent they would be resistant to the anti-proliferative effect of the drug.
ICI 182780 alone had no major effect on the wt MCF-7 cell growth, whereas concentrations in excess of 10 Ϫ9 M suppressed the cell growth stimulated by 10 Ϫ9 M E2 ( Fig. 2A). In contrast treatment of the LTED-I cells with ICI 182780 alone markedly inhibited cell growth. These data were consistent with the hypothesis that the LTED-I cells were super-sensitized to residual E2 in the DCC medium. We next considered the possibility that ICI 182780 may indirectly effect growth by inhibiting growth factor-mediated transcription. To test this, LTED-I cells were cultured in the presence of 10 Ϫ8 M ICI 182780 (demonstrated to inhibit cell growth) and increasing concentrations of E2. Doses in excess of 10 Ϫ10 M E2 were able to reverse the inhibitory effect of 10 Ϫ8 M ICI (Fig. 2B). From this we concluded that cell growth was mediated wholly or in part via the ER-ERE pathway.
LTED-I Cells Utilize the Classical ERE Pathways in Part for Proliferation and Transcription-Our previous studies (6) indicated that the LTED-I cells contained an elevated level of ER␣, which was phosphorylated on Ser 118 . We envisaged this could influence the basal ER-ERE-driven transcription sensitizing the cells to the effects of residual E2 in the medium. To test this hypothesis wt and LTED-I cells were co-transfected with a luciferase reporter construct regulated by two EREs and a plasmid that constitutively expressed ␤-galactosidase to normalize transfection efficiency. Basal transcription was elevated 7.5-fold in the LTED-I compared with the wt MCF7 cells (Fig.  3A). We next investigated the effect of varying doses of E2 and the pure anti-estrogen ICI 182780 on ER␣-driven transcription in the LTED-I versus the wt MCF-7 cells. The LTED-I and wt cells both showed dose-related increases in transcription in response to E2 (Fig. 3B). The LTED-I cells, however, showed an enhanced sensitivity to E2, and it is important to note that this stimulation is layered upon the increased basal activity described above. Thus for 10 Ϫ13 M E2 the transcription of the LTED-I cells is ϳ7.5-fold higher than the wt MCF-7 cells in absolute terms. Treatment with 10 Ϫ11 M E2 resulted in a 20fold increase in luciferase activity in the LTED-I cells compared with a 10-fold increase in the wt MCF-7. ICI 182780 antago-nized E2-stimulated transcription in both wt MCF-7 and LTED-I cells and reduced the basal transcription in the LTED-I cells by over 50% at doses as low as 10 Ϫ10 M whereas the wt cells remained largely unaffected with only a 20% decrease in basal transcription at 10 Ϫ8 M (Fig. 3C). This, combined with our previous cell growth data, suggests elevated levels of ER and the resultant ER-driven transcription are important components of the LTED-I cell phenotype.
Long Term Estrogen Deprivation Results in Alterations in the Expression of MAPK Family Members-The above data and our observation that the level of ER␣ phosphorylated on Ser 118 was elevated 3-fold in the LTED-I compared with the wt MCF-7 cells (Fig. 4) suggested that enhanced expression of ER and its activation via specific kinases played a central role in the apparent E2-independent proliferation. Several studies have shown that p42/p44 MAPK may be involved in ligand-independent activation of ER␣ (9,18). Similarly elevated levels of activated p42/p44 MAPK have been detected in cells during LTED (5,27). We therefore examined whole cell extracts isolated from our MCF-7 cells at intervals during E2 deprivation (Fig. 4). As the expression of the phosphorylated proteins Raf, MEK1/2, MAPK, and p90 RSK appeared to fluctuate we assessed median -fold increases during the LTED phases. Additionally we confirmed the reproducibility of changes among independent blots from the same time points (data not shown). Analysis of the LTED cell extracts using an antibody specific for phosphorylated p42/p44 MAPK showed a marked elevation compared with the parental MCF-7 cells. This equated to a median ϳ2-fold increase in the LTED-Q phase and an ϳ8-fold increase in the LTED-H and LTED-I phases compared with the wt MCF-7 cells. The level of total p42/p44 MAPK remained similar in the wt and LTED cells. We next investigated the expression of the upstream members of the MAPK pathway. Phosphorylated Raf-1 was elevated ϳ2-fold in the LTED-I cells, and MEK1/2 showed an ϳ5-6-fold increase during the LTED-H and LTED-I phases compared with the wt cells. Additionally phosphorylated p90 RSK , a downstream partner in the p42/p44 MAPK pathway demonstrated previously to phosphorylate ER␣ Ser 167 (14), was elevated ϳ4 -5-fold during the hypersensitive and independent phases. We next analyzed the level of the transcription factor c-Myc. This phospho-protein is involved in cell proliferation and differentiation and is phosphorylated by p42/p44 MAPK. c-Myc is also a classically ER-regulated gene in cells that are ER-positive and proliferate in response to steroids (28,29). During the first week of E2 deprivation c-Myc expression was lost. By week 3 the level of expression was elevated ϳ1.5-fold despite the cells being quiescent. At the onset of the LTED-H phase and during the LTED-I phase the level of phosphorylated c-Myc was ϳ2-fold higher compared with the wt MCF-7.
Elevated p42/p44 MAPK Is Associated with Increased Cell Growth and Transcription during LTED-I-We postulated that the elevation in p42/p44 MAPK activity in the LTED-I cells influenced their growth and transcription. To test this LTED-I cells and wt MCF-7 cells plus or minus E2 were treated with increasing doses of the MEK inhibitor U0126. The LTED-I cells showed a greater sensitivity to the MEK inhibitor compared with the wt MCF-7 ( Fig. 5A) with concentrations as low as 5 M resulting in an ϳ50% decrease in relative growth rate compared with the vehicle control. To determine whether the elevated level of p42/p44 MAPK influenced ER␣-driven tran-scription, LTED-I cells and wt cells were co-transfected with the ERE-driven luciferase and ␤-galactosidase reporter constructs as described previously and treated with U0126 plus or minus increasing doses of E2 (Fig. 5B). Basal transcription in the wt cells showed a small but statistically insignificant increase by the inhibitor whereas in the LTED-I cells ER␣ transactivation was reduced by ϳ50%. Addition of high doses of E2 (10 Ϫ9 M) readily overcame the suppressive effect of U0126 suggesting p42/p44 MAPK may be used during ER signaling in the LTED-I cells and that they are sensitized to the restorative effects of E2 consistent with the result found after treatment with ICI 182780.
Elevated p42/p44 MAPK Is Involved in the Mechanism of E2 Hypersensitivity-We postulated that if p42/p44 MAPK were directly or indirectly involved in the development of E2 hypersensitivity in response to LTED, then blocking its activity using the MEK inhibitor U0126 should shift the dose response curve to the right similar to the wt cells. To allow the continued sensitivity of the LTED-I cells to be visible, this experiment was conducted in the absence of insulin. The cells were treated over a 6-day period with increasing doses of E2 plus or minus U0126 (20 M). The E2 dose curve (Fig. 6) in the presence of U0126 shifted to the right when compared with E2 alone, with the sensitivity altering by one log. This implies that p42/p44 MAPK is involved in sensitizing the LTED-I cells to the effects of E2.
Induction of p42/p44 MAPK Activity by Insulin Does Not Play a Role in the Phosphorylation of ER␣ Ser 118 -The question remains as to what mechanism elevates MAPK activity. Studies have shown that in MCF-7 cells, insulin up-regulates the ER content and binding capacity and that this is blocked by specific tyrosine kinase inhibitors (30), and more recent studies have suggested that IGF-1R is elevated during LTED (31). As insulin has been shown to activate a spectrum of downstream signaling pathways such as MAPK, PI3 kinase, and AKT, we considered the possibility that insulin might stimulate p42/p44 MAPK activity in the LTED-I cells, leading to phosphorylation of ER␣ Ser 118 . Initially we screened whole cell extracts from the LTED cell pellets for expression of IGF-1R (Fig. 7A). This indicated that there was an ϳ2-fold increase in IGF-1R protein content during the LTED-H and LTED-I stages compared with the parental control.
To investigate this further we next stimulated the LTED-I and wt cells with insulin over a 30-min time course (Fig. 7B). Insulin increased p42/p44 MAPK activity as expected, within 5 min. However, treatment with insulin had no effect on the phosphorylation status of ER␣ Ser 118 in the wt or LTED-I cells.
Overexpression of EGFR or ERBB2 May Provide a Route to Increased p42/p44 MAPK Activity-Recent studies have suggested estrogens mediated by ER can rapidly stimulate p42/ p44 MAPK activation (21,22,32). We postulated that this may provide a potential mechanism for elevated p42/p44 MAPK in the LTED-I cells, based on their increased sensitivity for E2 and elevated ER content. Treatment of the LTED-I and wt MCF-7 cells with E2 resulted in rapid phosphorylation of Ser 118 but did not activate p42/p44 MAPK (data not shown). These results were similar to those reported by Joel et al. (33) and Lobenhofer and Marks (34).
Reports have suggested that cross-talk between growth factor and steroid receptors may play an important role in endocrine resistance (3). We postulated that the increase in p42/p44 MAPK activity in the LTED-I cells may be a result of increased EGFR receptor activity. Whole cell extracts from MCF-7 cells harvested at intervals during E2 withdrawal were immunoprobed with antibodies specific for phosphorylated and total EGFR. Levels of phosphorylated receptor during the LTED-Q phase were ϳ50% lower compared with those in the wt cells. However, the level of phosphorylated receptor returned to wt levels by week 19 and remained so during the LTED-H and LTED-I phases (Fig. 8A). The level of total EGFR remained largely unchanged compared with the wt control.
We next investigated the level of phosphorylated and total ERBB2 in the LTED whole cell extracts versus the wt MCF-7 (Fig. 8B). During the LTED-Q phase phosphorylated ERBB2 was undetectable compared with the parental control. However, during LTED-H and LTED-I phases the level of total and phosphorylated receptor was markedly elevated compared with the wt MCF-7 cells. Fluorescent in situ hybridization analysis (using the Vysis PathVysion TM HER-2 kit) indicated that ERBB2 was not amplified in the LTED-I cells (data not shown). To investigate this further we used the tyrosine kinase inhibitor ZD1839 (Iressa). This inhibitor has been shown to be selective for EGFR (IC 50 27-33 nM) but to also inhibit ERBB2driven signaling and to suppress the growth of ERBB2 overexpressing tumor cells (IC 50 2 M) (35). LTED-I and wt MCF-7 cells were cultured in the presence of increasing doses of ZD1839, and the breast tumor cell line SKBR3, which overexpresses ERBB2 and is particularly sensitive to ZD1839 (35), was used as a positive control. Low micromolar doses of ZD1839 had no effect on proliferation of the LTED-I or wt MCF-7 cells (Fig. 8C). This implied that up-regulation of EGFR activity was not involved in the development of the LTED-I phenotype in this experimental setting. However, reduced cell growth was notable at higher doses; the IC 50 of the wt cells was ϳ10 M compared with an IC 50 of ϳ5 M for the LTED-I cells and 0.5 M for the SKBR3 cells.
We next investigated the effect of ZD1839 on the basal ER-dependent transcription in the LTED-I and wt MCF-7 cells. At a dose of 20 M the basal transcription was reduced by 40% in the LTED-I cells but remained unaffected in the wt MCF-7 cells (Fig. 8D). E2 remained stimulatory in the presence of ZD1839 but was restricted by the inhibitor particularly at 10 Ϫ11 M E2. Although increases in growth factor receptor activity have been postulated previously, to our knowledge this is the first report showing elevated levels of phosphorylated ERBB2 associated with LTED phenotype, and it provides a potential mechanism for the elevated p42/p44 MAPK.

Increased Activity of p42/p44 MAPK in LTED-I Cells Is Not Responsible for ER␣ Ser 118 Phosphorylation-
The above results suggested that the elevated level of p42/p44 MAPK in the LTED-I cells (possibly mediated via ERBB2) might be linked to the phosphorylation of ER␣ Ser 118 . To further test the hypothesis that elevated levels of p42/p44 MAPK were responsible for phosphorylation of Ser 118 in the LTED-I cells, we compared the effects of various signal transduction pathway inhibitors on Ser 118 basal phosphorylation versus E2-induced phosphorylation in the wt MCF-7 and LTED-I cell lines. Fig. 9A shows that U0126, a specific inhibitor of MEK1/2, blocked p42/p44 MAPK activity but had no effect on the basal phosphorylation status of Ser 118 in the LTED-I or wt cells. More importantly even though U0126 abolished p42/p44 MAPK activity, it did not effect E2induced Ser 118 phosphorylation in the LTED-I or wt cells. To determine that the effect noted was not a time-related issue and that Ser 118 remained phosphorylated for more than 30 min, negating the effect of blocking MAPK, wt and LTED-I cells were treated with U0126 over an 8-h time course (Fig. 9B). No significant reduction in Ser 118 phosphorylation was noted in either cell line, whereas p42/p44 MAPK activity was abolished.

FIG. 3. Comparison of ER-directed transcription in wt MCF-7 and LTED-I cells. A, basal ER-mediated transcription is elevated in the LTED-I cells compared with the wt MCF-7. Cells
were co-transfected with EREIItkLuc and pCH110. After transfection cells were treated with DCC-FBS medium for 24 h. Cells were harvested, and luciferase and ␤-galactosidase activities were measured. To correct for differences in transfection efficiency, the luciferase activities were normalized to ␤-galactosidase activities. The data are reported as the -fold increase in activity compared with the wt MCF-7 cells. Average activity from triplicate wells Ϯ S.E. are shown. The results were confirmed in three independent experiments. B, effect of E2 on ER-mediated transcription. Wt and LTED-I cells were transiently co-transfected with EREIItk-Luc and pCH110 in serum-free medium followed by a 24-h incubation with DCC medium containing increasing doses of E2. Normalized luciferase activity from triplicate wells was expressed relative to the vehicle-treated control (Cont.). Bars represent Ϯ S.E. of the means. Results were confirmed in two independent experiments. C, ICI 182780 inhibits basal ERmediated transcription in LTED-I cells compared with wt MCF-7. Wt and LTED-I cells were transiently co-transfected with EREIItkLuc and pCH110 in serum-free medium followed by a 24-h incubation with DCC medium containing increasing doses of ICI 182780 plus or minus E2. Normalized luciferase activity from triplicate wells was expressed relative to the vehicle-treated control. Bars represent Ϯ S.E. of the means. Results were confirmed in two independent experiments.
It has been suggested that the protein kinase C and PI3 kinase pathways are capable of activating p42/p44 MAPK in a MEK1/2-independent manner (36), and PI3 kinase is involved in phosphorylation of ER␣ (17,37). We therefore postulated that an alternative signaling pathway was responsible for the phosphorylation associated with Ser 118 . To test this hypothesis we treated the LTED-I and wt MCF-7 cells with the PI3 kinase inhibitor LY294002 (Fig. 9C). The inhibitor had no effect on basal or E2-induced Ser 118 phosphorylation. From this we concluded that neither MAPK nor PI3 kinase were responsible for phosphorylation of Ser 118 in this setting, confirming our view that the LTED-I cells were not ligand-independent but supersensitized to residual E2 in the DCC medium. These data also suggested that alternative kinases were responsible for the phosphorylation of Ser 118 .
PI3 Kinase Is Associated with ER␣-directed Transcription-Our data suggested it was unlikely p42/p44 MAPK activation alone was responsible for the LTED-I phenotype. Recent data indicate that AKT is capable of phosphorylating ER␣ Ser 167 and activating transcription (17). We wished to investigate whether signaling pathways other than Ser 118 and p42/p44 MAPK were up-regulated in the LTED-I cells or whether there might be a dynamic interplay between phosphorylation events on AF-1 that could account for the LTED-I phenotype. First we assessed the whole cell extracts from time points during E2 withdrawal for expression of total and phosphorylated AKT (Fig. 10A). The overall median level of protein was unchanged when compared with the wt MCF-7 cell line.
To test this hypothesis further wt and LTED-I cells were co-transfected with the ERE-driven luciferase and ␤-galactosidase reporter constructs as described previously. The cells were treated subsequently with the PI3 kinase inhibitor LY294002 plus or minus increasing concentrations of E2. Basal transcription in the wt cells was marginally increased by the inhibitor whereas E2-stimulated transcription was blocked. In the LTED-I cells both basal and E2-stimulated ER␣-dependent transactivation was reduced by ϳ70% (Fig. 10B). This suggests that the PI3 kinase pathway was involved either directly or indirectly with ER␣ transactivation and that the LTED-I cells were more sensitive to the restorative effects of E2. We next considered whether there was a molecular cross-talk between the PI3 kinase and p42/p44 MAPK pathways in the LTED-I and wt cells. In this instance we treated the transiently transfected cells with a combination of the MEK inhibitor U0126 and LY294002 plus or minus E2. The basal transcription was reduced in the wt and LTED-I cells by 50 and 70%, respectively (Fig. 10C). Addition of E2 was unable to remove the block on transcription in the wt cells, whereas E2 rescued ER␣-driven transcription in the LTED-I cells by 3-fold compared with the untreated control. DISCUSSION Understanding the mechanism involved in the development of resistance to estrogen deprivation is now of paramount importance, because recent findings demonstrate that aromatase inhibitors are more effective than tamoxifen for the treatment of patients with advanced steroid receptor-positive breast tumors (37,38). To address this we have developed a LTED MCF-7 cell line that showed increased sensitivity to E2 and elevated levels of ER␣ phosphorylated at Ser 118 (6). In this paper we have extended our studies in an attempt to ascertain the cellular pathways involved in LTED and phosphorylation of Ser 118 . Our findings, similar to those by Santen and co-workers (39), demonstrated an increase in ER expression and basal ER functionality (assessed using an ERE reporter construct) in LTED-I versus wt MCF-7 cells. This increased activity was inhibited by the pure anti-estrogen ICI 182780 suggesting that classical ER/ERE-directed processes were wholly or partially responsible for adaptation to LTED conditions.
We also confirmed previous findings (5,27,40) that p42/p44 MAPK was elevated during LTED and demonstrated for the first time that this may result from increased ERBB2 activity within the LTED cells. Previous studies have shown that elevated levels of EGFR and ERBB2 are associated with tamoxifen-resistant MCF-7 cells and that these cells are more sensitive to the anti-proliferative effects of ZD1839 (1 M) compared with wt MCF-7 cells (41,42). We also saw increased sensitivity to ZD1839, in the LTED-I cells, but the IC 50 was reduced (5 M) in keeping with an ERBB2-dependent rather than EGFR-dependent effect (35). Fluorescent in situ hybridization analysis indicated that ERBB2 was not amplified in the LTED cells suggesting the up-regulation may be via a transcriptional mechanism.
Although our data show increased levels of activated p42/p44 MAPK in the LTED-I cells this appears not to be the sole mechanism controlling this phenotype, rather a regulating factor that allows optimal utilization of other pathways normally involved in ER-directed transcription and proliferation. This is supported by our transfection experiments with U0126, which demonstrate that p42/p44 MAPK may play a role in elevated basal ER activity but does not account for the total increase. We demonstrated that at a concentration of 20 M U0126 abolished MAPK activity on Western blots but only reduced ER␣ transcription by 50%. This suggests that other mechanisms, along with elevated MAPK, contribute to both the enhanced ER transcriptional activity and LTED-I phenotype.
Our data with U0126 suggested that Ser 118 was not phosphorylated by p44/p42 MAPK, and indeed further studies with a number of signal transduction inhibitors suggested Ser 118 is the target for multiple kinases, in confirmation of previous findings (33). This provided further supporting evidence that LTED-I cells are ligand-dependent rather than ligand-independent as phosphorylation of Ser 118 in response to E2 is MAPKindependent (33). It was notable, however, that p42/p44 MAPK was involved in sensitizing the LTED-I cells to the effects of E2. This was demonstrated in the shift in dose response to E2 when p42/p44 MAPK was blocked and similarly the decrease in proliferation in response to U0126 compared with the wt cells. These data are consistent with the theory of direct interaction between ER and growth factor signaling but also with the possibility that the pathways may operate independently with both MAPK and E2 providing mitogenic signals. However, our data showing inhibition with ICI 182780 implies that interactive signaling plays a significant role.
It is clear that blocking p42/p44 MAPK may also affect other factors. For instance recent evidence suggests that the transcriptional activity of AIB1, a ligand-dependent ER coactivator, is enhanced by p42/p44 MAPK phosphorylation leading to the recruitment of CBP and an associated increase in acetyl transferase activity (43). It is envisaged that the ability of growth factors to augment estrogen action may be mediated in part through p42/p44 MAPK activation of AIB1. This may also explain the selection of AIB1 amplification during progression in ER-positive breast cancers (44 -46). Such a model is consistent with our findings of elevated ERBB2 and MAPK activity.
Further evidence for multiple signaling pathways having a role in the LTED-I phenotype and for the theory that p42/p44 MAPK amplifies the sensitivity of the cells to E2 was provided by transcription assays. For instance LY294002 reduced basal transcription in the LTED-I cells by 70% while having a slight stimulatory effect on the wt cells (possibly because of compensatory alternative pathways). Furthermore a combination of U0126 and LY294002 reduced the basal transcription in the wt FIG. 8. ERBB2 but not EGFR is elevated during LTED. A, the level of phosphorylated and total EGFR in the LTED cells. Whole cell extracts (100 g) from representative weeks post-E2 withdrawal were separated through 7% SDS-PAGE gels and immunoprobed with antibodies specific for phosphorylated and total EGFR. B, the level of phosphorylated and total ERBB2 in the LTED cells. Whole cell extracts were immunoprobed with antibodies specific for phosphorylated and total ERBB2. C, the effect of ZD1839 on LTED-I, wt MCF7, and SKBR3 cell growth. Wt MCF-7, LTED-I, and SKBR3 cells were seeded at a density of 5 ϫ 10 3 cells per well and 48 h later treated with increasing doses of ZD1839, a specific inhibitor of EGFR. Wt cells were cultured in the presence of 10 Ϫ9 M E2 whereas LTED-I cells were grown in DCC-FBS medium. SKBR3 cells were cultured in RPMI 1640 containing phenol red and 10% FBS. The data represent triplicate readings, and the bars show mean Ϯ S.E. Results were confirmed in two independent experiments. D, the effect of EGFR inhibitor ZD1839 on basal and E2-mediated ER␣ transactivation. Wt and LTED cells were transiently cotransfected with EREIItkLuc and pCH110 in serum-free medium followed by a 24-h incubation in DCC medium containing increasing doses of E2 alone or in the presence of ZD1839. Normalized luciferase activity from triplicate wells was expressed relative to the vehicle-treated control (cont). Bars represent Ϯ S.E. of the mean. Results were confirmed in two independent experiments. and LTED-I cells by 50 and 70%, respectively. It is notable that even high doses of E2 were unable to remove the block on transcription in the wt cells but partially rescued ER␣-driven transcription in the LTED-I cells.
These observations and the complex pattern of changes in protein expression and phosphorylation seen during the development of the LTED phenotype suggests that a network of kinases and molecular switches operate at different stages during E2 withdrawal. It has been suggested that a complex of p42/p44 MAPK and p90 RSK interact with the ER and that these kinases coordinately phosphorylate Ser 118 and Ser 167 (14). The presence or absence of co-ordinate phosphorylation of these residues may provide the ER with a mechanism of distinguishing signals from various second messenger pathways. To add a FIG. 9. The effect of U0126 and LY294002 on ER␣ Ser 118 phosphorylation. A, LTED-I and wt cells were serum-starved for 24 h prior to experimentation as described under "Experimental Procedures." Cell monolayers were pretreated with U0126 (20 M) or vehicle for 30 min, followed by treatment with 10 Ϫ8 M E2, U0126, or a combination of the two for a further 30 min. Cells were harvested, and the level of p44/p42 MAPK activity and Ser 118 phosphorylation was monitored by immunoblotting. The data shown are representative of three independent experiments. B, time course shows that blocking MAPK does not affect ER␣ Ser 118 phosphorylation. Serum-starved LTED-I and wt MCF-7 cells were treated with U0126 (20 M) or vehicle over 8 h. Protein extracts at the time intervals indicated were immunoprobed for activated p42/p44 MAPK and ER␣ Ser 118 phosphorylation. C, the effect of LY294002 on MAPK and ER␣ Ser 118 phosphorylation. Serum-starved LTED-I and wt MCF-7 cells were pre-treated with LY294002 (50 M) or vehicle for 30 min. Monolayers were then treated with 10 Ϫ8 M E2, LY294002, or a combination of the two for a further 30 min. p44/p42 MAPK activity and Ser 118 phosphorylation were monitored by immunoblotting. The data shown are representative of three independent experiments.
FIG. 10. AKT remains unchanged during LTED, but ER transcription is inhibited by LY294002. A, protein samples from representative weeks post-E2 withdrawal were separated through SDS-PAGE gels and immunoblotted with antibodies specific for phosphorylated or total AKT. Figures below the blots indicate median -fold changes during the three LTED phases compared with the wt MCF-7 (FBS) control. B, the effect of PI3 kinase inhibitor LY294002 on basal and E2-mediated ER␣ transactivation. Wt and LTED cells were transiently co-transfected with EREIItkLuc and pCH110 in serum-free medium followed by a 24-h incubation in DCC medium containing E2 alone or in the presence of LY294002. Normalized luciferase activity from triplicate wells was expressed relative to the vehicle-treated control. Bars represent Ϯ S.E. Results were confirmed in two independent experiments. C, the effect of combining LY294002 and U0126 on basal and E2-mediated ER␣ transactivation. Wt and LTED cells were transiently co-transfected with EREIItkLuc and pCH110 in serum-free medium followed by a 24-h incubation with DCC medium containing E2 alone or in the presence of LY294002 and U0126. Normalized luciferase activity from triplicate wells was expressed relative to the vehicletreated control. Bars represent Ϯ S.E. Results were confirmed in two independent experiments further layer of complexity recent findings have shown that AKT phosphorylates Ser 167 (17) and that regulation of ER␣ activity by AKT may be indirectly controlled by p42/p44 MAPK via its interaction with p90 RSK (47,48). It has also been reported that p90 RSK , when phosphorylated by p42/p44 MAPK in response to growth factors, is transported into the nucleus and forms a stable complex with coactivator CBP, which then regulates transcription (49). Evidence also suggests CBP complexes with ER and that ectopic expression of CBP enhances ER-driven transcriptional activity (50). These findings may account for the elevated basal transcription in the LTED-I cell line as these cells have elevated levels of ER, p42/p44 MAPK, and p90 RSK .
In summary we have demonstrated that the LTED-I cells are super-sensitive to the effects of E2, requiring ER for both proliferation and ER-directed transcription. This finding suggests that ICI 182780 might be more effective in the treatment of breast cancers that acquire resistance to estrogen deprivation rather than in de novo hormone responsive disease. We have shown elevated levels of members of the MAPK kinase signaling pathway together with elevated levels of phosphorylated ERBB2. However, although our data show that the MAPK and PI3 kinase pathways play an integral role in the development of the LTED-I phenotype and sensitization of these cells to residual E2, neither appear responsible for the phosphorylation ER␣ Ser 118 . We postulate an adaptive pathway similar to that shown in Fig. 11 in which the development of the LTED-I phenotype results from elevated levels of ER coupled with enhanced activation of the ER as a result of increased ERBB2 expression and p42/p44 MAPK activity. p42/p44 MAPK may be involved in ER activation (on sites other than Ser 118 ) and regulation of down-stream partners such as p90 RSK , c-Myc, and AIB1 leading to increased coactivator activity and consequently providing a hypersensitive reception to residual E2. In conclusion these data confirm the presence of cross-talk between the ER and growth factor signaling pathways during LTED and support the development of treatment strategies based on signal transduction inhibitors, which may be used to extend the duration of sensitivity to E2 deprivation or to reverse resistance at its time of emergence.