Estrogen Receptor α and the Activating Protein-1 Complex Cooperate during Insulin-like Growth Factor-I-induced Transcriptional Activation of the pS2/TFF1 Gene*

Insulin like growth factor I (IGF-I) displays estrogenic activity in breast cancer cells. This activity is strictly dependent on the presence of estrogen receptor α (ERα). However the precise molecular mechanisms involved in this process are still unclear. IGF-I treatment induces phosphorylation of the AF1 domain of ERα and activation of estrogen regulated genes. These genes are characterized by important differences in promoter architecture and response element composition. We show that promoter structure is crucial for IGF-I-induced transcription activation. We demonstrate that on a complex promoter such as the pS2/TFF1 promoter, which contains binding sites for ERα and for the activating protein-1 (AP1) complex, transcriptional activation by IGF-I requires both ERα and the AP1 complex. IGF-I is unable to stimulate transcription of an estrogen-regulated gene under the control of a minimal promoter containing only a binding site for ERα. We propose a molecular mechanism with stepwise assembly of the AP1 complex and ERα during transcription activation of pS2/TFF1 by IGF-I. IGF-I stimulation induces rapid phosphorylation and an increase in protein levels of the AP1 complex. Binding of the phosphorylated AP1 complex to the pS2/TFF1 promoter allows recruitment of the chromatin remodeling factor Brg1 followed by binding of ERα via its interaction with c-Jun.

Control of breast cancer cell proliferation is a complex, multifactorial, and interactive process. Estradiol, a steroid hormone, which acts via its nuclear receptors (ER␣ 2 and -␤ for estrogen receptor ␣ and ␤) and growth factors such as insulinlike growth factor I (IGF-I) or IGF-II, epidermal growth factor, or transforming growth factor ␣, whose action is mediated by tyrosine kinase transmembrane receptors, controls gene expression and cellular proliferation of breast cancer cells (1). Although it was discovered more than 10 years ago that ER␣ could be activated in a ligand-independent fashion by growth factors like epidermal growth factor or IGF (2), the cross-talk between estrogens and growth factor pathways remains to be elucidated.
In the classical model of ER action, the receptor binds as a homodimer (3) or a heterodimer (4 -7) to estrogen response elements (EREs) within the promoters of estrogen-responsive genes. Once present on the promoter of estrogen responsive genes, ER recruits an array of transcriptional cofactors (coactivators and corepressors) that bind to the receptor and interact with other transcription factors, including components of the general transcription machinery. Some of these cofactors possess chromatin-remodeling activities or histone modifying activities or recruit additional proteins to the complex to mediate transcription (for review, see Ref. 8). ER␣ may also modulate transcription without receptor-DNA interaction by functional interference with other transcription factors such as activating protein-1 (AP1) (for review, see Refs. 9 -12).
The AP1 complex, which is implicated in diverse cellular processes, including differentiation, cell proliferation, and transformation (for review, see Ref. 13), predominantly consists of various combinations of Jun (c-Jun, JunB, JunD) and Fos (c-Fos, Fra-1, Fra-2, FosB) proteins. This complex can result from many different combinations of heterodimers and homodimers, and the combination determines the genes that are regulated by AP1 (14,15). The AP1 complex formed of heterodimers or homodimers of Jun and Fos proteins has DNA binding activity via the TPA-responsive elements (TREs).
Furthermore transcriptional activity of the AP1 complex is post-translationally regulated by several kinases. Extracellular signal-regulated kinases 1 and 2 phosphorylate c-Fos on threonines 325 and 331. These phosphorylations regulate the transforming activity of c-Fos (16). Moreover, phosphorylation of c-Jun by c-Jun NH 2 -terminal kinases on residues serine 63 (17) and 73 potentiates its ability to activate transcription either as a homodimer (18,19) or a heterodimer together with c-Fos.
Activation of the IGF-I receptor by IGF-I binding leads to a conformational change of its receptor, the auto phosphorylation of a tyrosine residue in the intracellular part of the receptor, and the subsequent activation of downstream signaling molecules. Then the IGF-I receptor is able to bind signaling adaptor proteins like insulin receptor substrate 1 (IRS1) and Shc, linking the receptor to downstream signaling pathways like the phosphatidylinositol 3-kinase and the mitogen-activated protein pathways (for review, see Ref. 20). Growth factors such as IGF-I are able to elicit estrogenic responses in target tissues in the absence of estradiol, and they cooperate with the hormone in its presence (21). Several studies have demonstrated that estradiol, anti-estrogens, and growth factors induce a rapid hyperphosphorylation of ER␣ on serine 118, located at the N terminus within transcriptional activation function AF-1 (2,22,23). This phosphorylation is essential for activation of ER␣ bound or not to its ligand (24). However, this phosphorylation is not sufficient to elicit estrogenic responses in the absence of hormone, and there may be other ER␣ domains or partners that fulfill additional requirements for growth factor signaling through ER␣ (22). Supporting this, MCF7 breast cancer cells with a constitutively active MEK1 show increased interaction between ER␣ and its coactivators (SRC2 and SRC3) and hormone hypersensitivity (25). In addition, phosphatidylinositol 3-kinase, which is activated by IGF-I, has been shown to phosphorylate ER␣ on serine 167 within 1 h after treatment with growth factors. Active phosphatidylinositol 3-kinase increases the transcription-stimulating activity of both the AF-1 and AF-2 domains of ER␣, whereas active protein kinase B (PKB), a downstream target of phosphatidylinositol 3-kinase, only increases the activity of AF-1, suggesting that phosphatidylinositol 3-kinase utilizes PKB-dependent and -independent pathways in activating ER␣. In addition, LY294002, a specific inhibitor of the phosphatidylinositol 3-kinase/protein kinase B pathway, reduced phosphorylation of ER␣ in vivo (26).
The goal of the present study was to better understand the cross-talk between the IGF-I and estradiol pathways at the molecular level. We studied the pS2/TFF1 gene whose regulation serves as a paradigm for estrogen regulation of gene expression in breast cancer cells. The pS2/TFF1 gene has a composite promoter with an ERE (located between Ϫ405 and Ϫ393) and a TRE (AP1 response element, located between Ϫ338 and Ϫ332), and it has been reported that its activation can be achieved by estradiol or IGF-I (27). Our laboratory has shown that both estradiol and IGF-I treatments induce a rapid chromatin remodeling of the pS2/TFF1 promoter in a region encompassing the ERE and the TRE (28). The role of the AP1 element in the response of pS2/TFF1 to estradiol has been analyzed in detail by others in a cell line derived from the HepG2 hepatocarcinoma cell line, re-expressing ER␣. They demonstrate that the ERE is essential for the response of pS2/TFF1 to estrogens and that the TRE is involved in both estrogen and phorbol-ester responses (29).
Here we demonstrate that activation of ER␣ induced by IGF-I is not sufficient to allow activation of a minimal estrogenregulated promoter containing only an ERE sequence. We show that in response to IGF-I treatment, the AP1 complex is phosphorylated and rapidly recruited to the pS2/TFF1 promoter, whereas recruitment of ER␣ is delayed. On the pS2/TFF1 promoter, ER␣ and the AP1 complex cooperate during IGF-I-induced activation of the transcription, but have different functions. The AP1 complex is implicated in chromatin remodeling by recruiting chromatin-remodeling factors such as Brg1, whereas ER␣ is involved in later stages of transcription.

EXPERIMENTAL PROCEDURES
Chemicals-Estradiol was purchased from Sigma. ICI 184780 (Fulvestrant) was purchased from Zeneca Pharmaceuticals. IGF-I was purchased from Euromedex.
Cell Lines and Tissue Culture-MCF7 cells were grown in Dulbecco's modified Eagle's medium/F-12 containing 50 g/ml gentamycin and 10% heat-inactivated fetal calf serum at 37°C in a humidified atmosphere containing 10% CO 2 .
To study the effects of estrogens and IGF-I, cells were grown for 2 days in media without phenol red without gentamycin and without serum. Cells were treated or not with 10 nM estradiol (E2), 5 nM IGF-I, or 100 nM Fulvestrant for the indicated times.
Luciferase Assays-Cells were grown in 3.5-cm diameter dishes and treated as indicated in the figure legends. Cells were then lysed in 200 l or Promega lysis buffer and frozen. Luciferase activity was determined using the Luciferase assay reagent (Promega E397A) on a Centro LB 960 (Berthold). Luciferase activities were normalized to plated cell number or to total protein content measured by the Amido Schwarz technique (33). All experiments were performed at least in triplicate, and S.D. were calculated.
Western Blots-Western blots were performed as previously described (34). The quantity of extract loaded on the gels was normalized to total protein content, assayed by the Amido Schwarz technique. All antibodies were used at a 1/200 dilution, except for anti-phospho-c-Fos antibody, in which the dilution was 1/1000.
Co-immunoprecipitations-For co-immunoprecipitations of cross-linked protein complexes, 1 ϫ 10 6 cells were submitted to immunoprecipitation as in the ChIP procedure, except that after immuno-recovery proteins were eluted by the addition of 30 l of H 2 O and 10 l of 4ϫ electrophoresis loading buffer (200 mM Tris, pH 6.8, 8% SDS, 20% glycerol, 8 mM EDTA, 6% ␤-mercaptoethanol, 0.016% bromphenol blue). The totality of the immunoprecipitated material was electrophoresed on SDS gels and analyzed by Western blotting.
For co-immunoprecipitations of native protein complexes, 1 ϫ 10 6 MCF7 cells, grown for 3 days in steroid-free medium, were washed twice in PBS, resuspended in 500 l of lysis buffer (50 mM Tris, pH 8, 0.4% Nonidet P-40, 300 mM NaCl, 10 mM MgCl 2 , anti-protease from Roche Applied Science (1 mini-tablet/10 ml)), phosphatase inhibitors from Sigma (P2850), and incubated for 15 min at 4°C. After centrifugation for 20 min at 16,000 ϫ g, the lysate was adjusted to 1 ml with 500 l of dilution buffer (50 mM Tris, pH 8, 0.4% Nonidet P-40, 2.5 M CaCl 2 , 1 l of DNase I). A total of 40 l (v/v) of protein A/protein G beads were added, and the lysate was incubated for 45 min at 4°C on a rotating wheel. The supernatant was collected by centrifugation, 30 l was withdrawn for input assays, and the remaining material was incubated with 1 g of anti-c-Jun for 2 h at 4°C on a rotating wheel. A total of 20 l (v/v) protein A/protein G beads were added again, and the incubation was continued for 2 more hours. Beads were washed 3 times with wash buffer (50 mM Tris, pH 8, 0.4% Nonidet P-40, 150 mM NaCl, 5 mM MgCl 2 ), and proteins were eluted by the addition of 20 l of H 2 O and 10 l of 4ϫ electrophoresis loading buffer (200 mM Tris, pH 6.8, 8% SDS, 20% glycerol, 8 mM EDTA, 6% ␤-mercaptoethanol, 0.016% bromphenol blue). The totality of the immunoprecipitated material was electrophoresed on SDS gels and analyzed by Western blotting.
Cell Electroporation and RNA Interference-Five million cells were electroporated by siRNA to a final concentration of 20 M. Cells were electroporated on a gene pulser Xcell (Bio-Rad) at 250 V and 960 microfarads in 0.4-mm cuvettes. After electroporation, cells were plated and grown for 2 days in media without phenol red, without gentamycin, and without serum. Cells were treated or not with 5 nM IGF-I for 1 day.

IGF-I-induced Activation of the pS2/TFF1 Gene Requires
Both ER␣ and the AP1 Complex-IGF-I is an activator of estrogen-regulated genes in breast cancer cells (21). It promotes chromatin remodeling at the promoter of the estrogen-regulated gene pS2/TFF1 (28). However, its mechanism of action is not elucidated, and it is not known whether the architecture of the promoter is important for this activation. This gene has a complex promoter that contains one binding site for the ERE and one binding site for the AP1 complex (TRE) (Fig. 1A). We first determined the importance of these two transcription factors in IGF-I-induced activation of the pS2/TFF1 gene. As previously described (21) in MCF7 cells, IGF-I alone induced transcriptional activation of the estrogen-regulated gene pS2/TFF1 (Fig. 1B). To assess the importance of ER␣ in this activation, we treated cells both with IGF-I and the pure antiestrogen Fulvestrant (ICI 182 780). When bound to Fulvestrant, ER␣ is rapidly degraded in the nucleus (35,36) and consequently never binds DNA (37). Here we found that Fulvestrant blocked the IGF-Iinduced transcription of pS2/TFF1 (Fig. 1B) confirming that ER␣ is necessary for IGF-I-induced transcription of this promoter (27). To go further in the role of ER␣ in IGF-I mediated activation of the pS2/TFF1 gene, we knocked down ER␣ using siRNAs directed against ER␣ mRNA. MCF7 cells were then stimulated by IGF-I or not for 24 h, and RNA was purified. The specificity of siRNA depletion was analyzed by Western blot (Fig. 1C) and showed that siRNA directed against ER␣ mRNA caused a specific and dramatic decrease in ER␣ protein levels compared with scramble siRNA. The administration of ER␣ siRNA to MCF7 cells abolished the IGF-I-mediated activation of the pS2/TFF1 gene compared with cells electroporated by scramble siRNA (Fig. 1D). This result reinforces the results obtained with Fulvestrant and demonstrates that ER␣ is required for IGF-I-mediated activation of the pS2/TFF1 gene.
To evaluate the importance of c-Jun and c-Fos for the pS2/ TFF1 activation mediated by IGF-I, we knocked down the two proteins using siRNAs directed against c-Jun or c-Fos mRNA. They were administrated to MCF7 cells separately or in combination. Cells were then stimulated or not by IGF-I, and RNA was purified. The effect of siRNA depletion of c-Jun and c-Fos was analyzed by Western blot (Fig. 2A). SiRNA directed against c-Fos was able to specifically lower c-Fos protein levels, whereas siRNA directed against c-Jun lowered both c-Jun and c-Fos protein levels. Lowering the protein level of a transcription factor involved in the formation of heterodimers often induces a decrease in partner protein levels. Moreover in the case of the AP1 complex, c-Jun is able to homodimerize, preventing its degradation when c-Fos protein levels are lowered. c-Jun and c-Fos siRNA have a clear effect on pS2/TFF1 activation mediated by IGF-I (Fig. 2B), dramatically impairing its activation compared with cells electroporated with the scramble siRNA. Treatment of cells with both siRNAs also abolished IGF-I-mediated activation of the pS2/TFF1 gene. These results demonstrate that both ER␣ c-Jun and c-Fos are essential for IGF-I action and that in the cell the absence of one these proteins cannot be compensated by the presence of other members of their respective families.
Activation of Estrogen-regulated Genes by IGF-I Requires a Composite Promoter-We have shown that both ER␣ and the AP1 complex are required for IGF-I-mediated activation of the pS2/TFF1 gene. To better understand the role of each of the two promoter elements (ERE and TRE) in this transcriptional activation, we used the MELN cell line that has been derived from MCF-7 cells and engineered to express the reporter gene luciferase under the control of a synthetic promoter containing only an ERE upstream of a minimal ␤-globin promoter (Fig. 3A). These cells were untreated or treated with estradiol and IGF-I alone or in combination, and induction of luciferase, GAPDH mRNA, and pS2/TFF1 mRNA were measured. Fig. 3B shows a robust induction of luciferase activity by estradiol, whereas IGF-I had no effect. The increase in GAPDH gene expression (a gene known to be induced by IGF-I (38)) after treatment of the MELN cells by IGF-I (Fig. 3C) demonstrates that the IGF-I pathway is not impaired in this cell line. This was further confirmed by the induction of the endogenous pS2/TFF1 gene by estradiol as well as by IGF-I in the MELN cell line (Fig. 3D). These results indicate that IGF-I treatment is not sufficient to promote the activation of a promoter containing a single ERE.
To investigate the importance of the TRE element in IGF-Imediated activation of pS2/TFF1, we used a cell line engineered   APRIL 20, 2007 • VOLUME 282 • NUMBER 16 to express luciferase under the control of a promoter containing TREs, which are recognized by the AP1 complex (Fig. 4A). In this experiment we used TPA, which is a well known inducer of the AP1 complex (39, 40) as a positive control. Upon IGF-I treatment of the cells, luciferase activity was induced 3-fold when compared with untreated cells (Fig. 4B). We conclude that IGF-I can activate a minimal promoter containing TRE elements. Concomitant treatment of the cells with either IGF-I or TPA and Fulvestrant or estradiol did not change the level of activation of luciferase compared with cells treated with IGF-I or TPA alone (Fig. 4B, compare lanes ϩTPA with ϩTPA ϩICI and lanes ϩIGF-I with ϩE2 ϩIGF-I and with ϩICI ϩIGF-I). This result demonstrates that the effect of Fulvestrant on the activation of the pS2/TFF1 gene by IGF-I (see Fig. 1B) is not due to a defect in the signal transduction pathway mediated by IGF-I in the presence of Fulvestrant. This also demonstrates that AP1-mediated activation of TRE 3 -Tk-Luc promoter is independent of ER␣ action. Taken together, these results indicate that ER␣ is able to cooperate with the AP1 complex to promote transcription on composite promoters such as pS2/ TFF1 and that promoter architecture is crucial for the cooperation between the AP1 complex and ER␣.

IGF-I-induced Activation of the pS2/TFF1 Gene
The AP1 Complex Binding Precedes the Association of ER␣ on the pS2/TFF1 Promoter after IGF-I Stimulation-These data raise the question of the exact role of ER␣ and the AP1 complex in pS2/TFF1 activation upon IGF-I stimulation. In the context of estrogen-stimulation, ER␣ is recruited precociously to the pS2/TFF1 promoter, and it plays a central role in triggering transcription of the pS2/TFF1 gene (37). We wondered if in the case of IGF-I stimulation the situation was similar. Who triggers transcription, the AP1 complex or ER␣? To answer this question we performed ChIP experiments in which we measured the recruitment of c-Jun, c-Fos, and ER␣ on the pS2/TFF1 promoter after various treatment times of the cells with IGF-I. Fig. 5 shows that after 1 h of treatment ER␣ was not recruited to the pS2/TFF1 promoter. On the contrary c-Jun and c-Fos were present on pS2/TFF1. There was a 5-fold increase in c-Jun binding and a 2-fold increase in c-Fos binding to this promoter   compared with the untreated control (Fig. 5). After 3 h of IGF-I stimulation, c-Jun and c-Fos were still present on the pS2/TFF1 promoter roughly to the same extent as after 1 h of treatment. Their binding to DNA was ϳ4-fold higher than that of the untreated controls. After 3 h of IGF-I stimulation, a strong recruitment of ER␣ to the pS2/TFF1 promoter was observed (16-fold more than in the untreated control, Fig. 5). Using the irrelevant anti-IgG antibodies, no enrichment of pS2/TFF1 DNA could be detected in between the different experimental conditions. These results taken together with the fact that IGF-I treatment is not sufficient to promote the activation of a promoter containing a single ERE suggests that the AP1 complex could be the factor recruiting ER␣ to the pS2/TFF1 promoter.

IGF-I Induces Phosphorylation of the AP1 Complex and Promotes Its Interaction with ER␣-Results
shown above indicate that both ER␣ and the AP1 complex are recruited to the pS2/ TFF1 promoter but with different timing, AP1 complex being recruited earlier than ER␣. Therefore, we investigated the mechanisms involved in the delayed interaction of ER␣ with the pS2/TFF1 promoter and if AP1 complex could be responsible for this recruitment. First we studied the effect of IGF-I stimulation on the physical interaction between ER␣ and the AP1 complex. It has been previously shown that c-Jun and ER␣ interact directly in intact cells, forming a protein complex in a ligand-dependent manner (41). However, to date no interaction between ER␣ and c-Fos has been described. To assess if ER␣ and the c-Jun-c-Fos complex belong to the same protein complex after exposure to growth factors, we treated MCF7 cells for various times by IGF-I and cross-linked the proteins in whole cells to preserve the integrity of the complexes. This was followed by co-immunoprecipitation using anti-c-Fos and antic-Jun antibodies. Western blots using an antibody directed against ER␣ showed that both c-Jun and c-Fos co-immunoprecipitated with ER␣ upon stimulation by IGF-I (Fig. 6A). There was no detectable interaction between c-Jun and ER␣ after 1 h, a weak interaction after 2 h, and a strong interaction after 3 h of IGF-I stimulation. Co-immunoprecipitation of c-Fos and ER␣ was weaker than that of c-Jun and ER␣, also maximal after 3 h of treatment with IGF-I. To get further insights on the interaction between ER␣ and the AP1 complex, we performed co-immunoprecipitations on native protein complexes under stringent conditions. After 3 h of IGF-I stimulation a clear interaction between ER␣ and c-Jun could be detected (Fig. 6B), confirming the results obtained on cross-linked complexes and suggesting a strong interaction between these two transcription factors. In contrast, we were unable to detect any interaction between ER␣ and c-Fos under IGF-I stimulation (data not shown), suggesting that these two proteins, even if they belong to the same complex, as suggested by the cross-linking experiments, probably did not interact directly or have a weak interaction.
IGF-I treatment causes the activation of the mitogen-activated protein kinases and phosphatidylinositol 3-kinase pathways, leading to the phosphorylation of many transcription factors and mediators. We then investigated if there was a correlation between the phosphorylation state of AP1 complex and the time course of interaction between AP1 complex and ER␣. Fig. 7 shows the effects of IGF-I treatment on the two main proteins implicated in the proliferative action of the AP1 com-plex, c-Jun and c-Fos. First, the levels of the two proteins were increased (Fig. 7A). The c-Fos concentration increased rapidly, and within 30 min the protein level reached 50% that of its maximal level. The increase in c-Jun was slower, reaching 50% that of its maximal level after 80 min of IGF-I treatment. These results are in agreement with the fact that IGF-I treatment induces an increase of mRNA levels of both c-Jun and c-Fos (42). Second, c-Jun and c-Fos are phosphorylated upon IGF-I treatment of the cells (Fig. 7B). Phosphorylation of c-Fos was rapid, occurring within 15 min and strong on threonine 325 (Fig. 7B), a key residue in c-Fos activation known to be phosphorylated by extracellular signal-regulated kinases (16). This phosphorylation level was maintained over time. Phosphorylation of the c-Jun protein on serine 63 (Fig. 7B), which is located within its transactivation domain, was slower and reached its maximal level after 3 h of treatment with IGF-I.
The increase in unphosphorylated and phosphorylated c-Jun and c-Fos levels was quantified from the blots, and the ratio of phosphorylated c-Jun and c-Fos over total c-Jun and c-Fos proteins was calculated. Fig. 7C showed a constant increase, up to 12-fold of the phospho-c-Jun/c-Jun ratio. This indicates that IGF-I induces an increase in both c-Jun protein levels and c-Jun phosphorylation but that phosphorylation increased more than the amount of the protein. For the phospho-c-Fos/c-Fos ratio, a dramatic increase (more than 100-fold) occurred within 15 min after treatment with IGF-I and decreased rapidly. After 2 h of treatment the phospho-c-Fos/c-Fos ratio remained constant and was approximately three times the basal level. This reveals that IGF-I induces rapid phosphorylation of c-Fos more rapidly MCF7 cells were treated for the indicated times with IGF-I. A, nuclear extracts containing cross-linked proteins were prepared as for ChIP experiments and immunoprecipitated (IP) with anti-c-Jun or anti-c-Fos antibodies. The presence of ER␣ in the eluates was analyzed by Western blot using anti-ER␣ antibody. B, native protein complexes were prepared as indicated under "Experimental Procedures" and subjected to immunoprecipitation with anti-c-Jun antibody. The presence of ER␣ in the eluates was analyzed by Western blot using anti-ER␣ antibody. than its synthesis. c-Fos phosphorylation is sustained during its synthesis occurs (Fig. 7, panel C, inset).
These data indicate that the IGF-I-dependent interaction between the c-Jun complex and ER␣ correlates with the time course of phosphorylation of c-Jun, suggesting a direct role of c-Jun phosphorylation in the interaction between ER␣ and the AP1 complex. Moreover this interaction of ER␣ with c-Jun after 3 h of IGF-I treatment coincides with the phosphorylated state of both c-Jun and c-Fos and a strong recruitment of ER␣ to the pS2/TFF1 promoter. We propose, therefore, that the AP1 complex is responsible for the recruitment of ER␣ to this promoter by IGF-I.
Chromatin Remodeling of the pS2/TFF1 Gene Activated by IGF-I Is Mediated by Direct Recruitment of Brg1 by AP1-Our results support a role for the AP1 complex in the recruitment of ER␣ to the pS2/TFF1 promoter upon IGF-I stimulation. However, the precise mechanism is still unknown. We have previously demonstrated in MCF-7 cells that IGF-I treatment induces chromatin remodeling of pS2/TTF1 promoter after 2 h of treatment (28). To assess if the chromatin remodeling of this promoter is due to the recruitment of ER␣ or of AP1, we analyzed by ChIP the presence of Brg1, a chromatin remodeling factor recruited precociously to the pS2/TFF1 promoter (37). MCF7 cells were treated for 1 h either with estradiol, IGF-I, or Fulvestrant alone or with a combination of IGF-I and Fulvestrant, and we measured the quantities of Brg1 associated to the promoter (Fig. 8). Upon estradiol stimulation, Brg1 was rapidly recruited to the pS2/TFF1 promoter, with a 10-fold enrichment compared with untreated cells. Upon IGF-I stimulation, Brg1 was also recruited to the pS2/TFF1 promoter (4-fold more than in the untreated control), suggesting that Brg1 may be responsible for IGF-I-induced chromatin remodeling of the pS2/TFF1 promoter. Furthermore co-treatment of the cells with IGF-I and Fulvestrant resulted in a recruitment of Brg1 similar to that obtained by treatment with IGF-I alone. This indicates an ER␣-independent recruitment of Brg1 upon IGF-I stimulation. Using the irrelevant anti-IgG antibodies, no enrichment of pS2/TFF1 DNA could be detected in between the different experimental conditions (Fig. 8). These data support a direct role for AP1 via the recruitment of Brg1 in chromatin remodeling of the pS2/ TFF1 promoter upon IGF-I stimulation.

Functional Cooperation between ER␣ and the AP1 Complex
Is Dictated by Promoter Structure-We have investigated the cross-talk between estradiol and IGF-I signaling pathways in breast cancer cell lines. For this work we used three different kinds of promoters, a complex promoter containing both ER␣ and AP1 response elements and two minimal promoters, one containing one binding site for ER␣ and the other containing three binding sites for the AP1 complex. We show that cross-talk between ER␣ and the AP1 complex is greatly influenced by the structure of the promoter. First, our data, obtained in a cell line derived from MCF-7 cells that express luciferase under the control of a synthetic promoter containing an ERE, demonstrate that IGF-I is unable to stimulate transcription of a gene driven by a promoter containing only an estrogen response element. This result contrasts with a previous report that showed that IGF-I was able to induce the expression of a luciferase reporter transgene driven by a promoter containing two EREs in the uterus  of transgenic mice (43). The difference could be due to an impairment of the IGF-I pathway in cell line MELN. Induction of GAPDH and pS2/TFF1 by IGF-I indicates that this is not the case. Thus, this discrepancy may come from the type of minimal promoter used. For our study we used a reporter gene under the control of the ␤-globin minimal promoter with an estrogen response element, whereas Klotz et al. (43) used the minimal thymidine kinase promoter. Another possibility could be that the promoter used to drive the transgene in mice contains a cryptic AP1 or SP1 response element. In addition, differences may be species-specific (human versus mouse) or tissue-specific (breast cancer cells versus normal uterus). The use of a cell line containing the luciferase gene under control of TPA response elements (that binds AP1) shows that IGF-I (and phorbol esters) induces luciferase activity and that this induction is not affected by anti-estrogens such as Fulvestrant. This finding confirms a previous report demonstrating that Fulvestrant does not inhibit IGF-I signaling (44) and that on this kind of promoter ER␣ does not play any role in IGF-I-induced activation of the AP1 complex. Studies on synthetic promoters report that the use of anti-estrogens stimulates AP1 transcriptional activity and that DNA binding activity of ER␣ is not required for this activation (45). Our studies on the endogenous promoter pS2/TFF1 demonstrate that the molecular mechanisms of cooperation between ER␣ and AP1 complex differ on endogenous estrogen-regulated promoters. First, a combined treatment of MCF7 cells by IGF-I, which activates the AP1 complex, and Fulvestrant, which blocks ER␣ action and causes its rapid degradation (35,36), inhibits pS2/TFF1 transcription. Lowering protein levels of ER␣ by the use of siRNA in cells treated with IGF-I leads to the same result, demonstrating that ER␣ is crucial for IGF-I mediated activation of this gene. Moreover, invalidation of the AP1 complex by siRNA also causes loss of IGF-I mediated activation of the pS2/TFF1 gene. These results clearly indicate that on an endogenous promoter like pS2/TFF1 both ER␣ and AP1 are required for ligand-independent activation of this gene. Furthermore, for full cooperation between these transcription factors after IGF-I stimulation, both AP1 and ER␣ response elements are required. This suggests that DNA binding activity of both transcription factors is necessary.
The AP1 Complex and ER␣ Play Different Roles during Transcription Activation of the pS2/TFF1 Gene-We have previously described that both estradiol and IGF-I are able to induce chromatin remodeling of pS2/TFF1 promoter (28). Here we show that both estradiol and IGF-I are able to induce recruitment of Brg1 to the pS2/TFF1 promoter. In the presence of estradiol, ER␣ may recruit Brg1. Upon IGF-I treatment, Brg1 could be recruited either by ER␣ or by AP1. However, the fact that Fulvestrant does not prevent IGF-I-induced recruitment of Brg1 and the fact that ER␣ is not recruited to pS2/TFF1 promoter after 1 h of IGF-I treatment support a direct role of AP1 in this process. This suggests that transcriptional activation of the pS2/TFF1 gene via ER␣ and/or via the AP1 complex obeys different mechanisms.
Numbers of publications assume that growth factors induce a change in ER␣ phosphorylation directly via the activation of mitogen-activated protein kinase and/or AKT pathways. However, whether or not the post-translational modifications of ER␣ induce its binding to the ERE in the absence of hormone remains unclear. Our results rule out the possibility that phosphorylation of ER␣ induced by IGF-I-is sufficient to directly induce ER␣ binding to the ERE in the absence of hormone. First, IGF-I does not activate transcription of a gene under the control of a promoter containing only an ERE, and secondly, we see a strong induction of the ER␣ binding to the pS2/TFF1 promoter after 1 h of estradiol treatment (data not shown) but not after 1 h of IGF-I treatment. AP1 is a target of the IGF-I signaling pathway. We observed, as previously described, an increase in the protein levels of the two subunits of AP1, c-Fos and c-Jun, after treatment with IGF-I c-Fos, increasing more rapidly than c-Jun. Activation of both AP1 subunits by phosphorylation occurred at different rates, c-Fos being very rapidly phosphorylated, whereas c-Jun phosphorylation was delayed. It has been described that estradiol induces an interaction between ER␣ and c-Jun, but not c-Fos (46). We observed that IGF-I treatment induced interaction of ER␣ with c-Jun, with a time course corresponding to c-Jun phosphorylation. Furthermore, the co immunoprecipitation studies on native protein complexes show a strong interaction between ER␣ and c-Jun upon IGF-I stimulation. IGF-I treatment also induced ER␣ interaction with c-Fos but with a time course that did not follow c-Fos phosphorylation. However, we were able to co-immunoprecipitate ER␣ and c-Fos in crosslinked protein complexes but not in native ones. This suggests that ER␣ and c-Fos belong to the same protein complex, but that their interaction is either indirect or weak. Furthermore, the extent of interaction between ER␣ and c-Jun corresponds to the level of phosphorylation of c-Jun observed over time, suggesting that this interaction may be mediated by c-Jun. Because we did not see an interaction between ER␣ and c-Jun after 1 h of treatment with IGF-I, whereas there was a clear interaction after a 3-h treatment, we analyzed ER␣ recruitment to the pS2/TFF1 promoter. Contrasting with what was seen after 1-h of treatment, after 3 h IGF-I induced recruitment of ER␣ to the pS2/TFF1 promoter, suggesting that this process is mediated by c-Jun. The delay in the recruitment of ER␣ to the pS2/TFF1 promoter (3 h) indicates that ER␣ does not play any role in the chromatin remodeling that precedes transcription initiation of this gene, but rather, in later stages of transcription. This hypothesis is reinforced by a recent report that investigated the distribution of a green fluorescent protein-ER␣ fusion transiently expressed in MCF-7 cells after treatment with estradiol or growth factors (47). Estradiol induced the formation of speckles, which are associated to a transcriptionally active form of ER␣ within 10 min. In contrast, after treatment with IGF-I, the maximum ER␣ redistribution was observed only after 90 min. Fig. 9 presents our current working hypothesis. IGF-I induces an increase in c-Fos and c-Jun protein levels as well as their phosphorylation and the binding of AP1 to the TRE of pS2/TFF1 promoter. When bound to DNA, the AP1 complex can recruit factors involved in chromatin remodeling, such as Brg1. Next, the increase in c-Jun protein levels and phosphorylation allows the recruiting of more ER␣ and to tether it, via its interaction with AP1, to the pS2/TFF1 promoter at the ERE sequence. Thereby, DNA-bound ER␣ can recruit coactivators and maintain a high level of transcription of the pS2/TFF1 gene.
In this study we did not investigate ER␣ phosphorylation status under IGF-I stimulation. Therefore, we cannot exclude the possibility that some IGF-I-induced ER␣ phosphorylation may contribute to the formation of the complex with c-Jun in the absence of hormone, favoring indirect recruitment of ER␣ to the promoter.
Our work provides new insights in the field of ligand-independent regulation of ER␣. It is well known that ER␣ and the AP1 complex cooperate during transcription activation. Here we demonstrate that after IGF-I stimulation both ER␣ and the AP1 complex are required for transcription activation of the pS2/TFF1 gene but play different roles in this molecular mechanism.