This Article Abstract Full Text (PDF) All Versions of this Article: 17/7/1296    most recent Author Manuscript (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager NURSA Molecule Pages Link Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Dutertre, M. Articles by Smith, C. L. Search for Related Content PubMed PubMed Citation Articles by Dutertre, M. Articles by Smith, C. L.

Ligand-Independent Interactions of p160/Steroid Receptor Coactivators and CREB-Binding Protein (CBP) with Estrogen Receptor-: Regulation by Phosphorylation Sites in the A/B Region Depends on Other Receptor Domains

Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, Texas 77030

Address all correspondence and requests for reprints to: Carolyn L. Smith, Ph.D., Department of Molecular and Cellular Biology, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030. E-mail: carolyns{at}bcm.tmc.edu.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
Estrogen receptor (ER) and ERß are transcription factors that can be activated by both ligand binding and growth factor signaling. Estradiol increases ER activity in part by enhancing interactions between its carboxy-terminal, ligand-binding domain (LBD) and the p160/SRC (steroid receptor coactivator) and p300/CBP (cAMP response element binding protein (CREB) binding protein) families of coactivators. In the absence of ligand and the LBD, these cofactors can also interact with the amino-terminal (A/B) domain of ERs in vitro. SRC-1 also enhances the ligand-independent activity of the full-length receptor. Both ligand-independent and estradiol-induced ER activity are increased by phosphorylation at specific serine (Ser) residues in the A/B domain (Ser^104/106 and Ser¹¹⁸ in ER). In the case of ERß, phosphorylation enhances the ligand-independent recruitment and action of SRC-1. We show here that unliganded ER can activate endogenous gene expression in MCF-7 cells, and that this activation is mediated in part by a p160 coactivator. In transfected HeLa cells, we show that the full-length ER can interact physically and functionally with p160/SRCs and CBP in the absence of ligand and that mutation of Ser^104/106/118 affects these interactions. Accordingly, ER dephosphorylation decreases its ligand-independent interaction with SRC-1 and CBP in vitro. In HeLa cells, both Ser^104/106 and Ser¹¹⁸ impinge on SRC-1 action by two mechanisms: 1) a seemingly indirect effect on SRC-1 recruitment that requires other receptor domains in addition to the A/B, consistent with our finding that the ligand-independent interaction between the A/B and the LBD and its enhancement by SRC-1 depend in part on Ser^104/106/118; and 2) an effect on SRC-1 coactivation that can be observed in the absence of the LBD. Ser^104/106/118 can also affect coactivation by a subset of coactivators in the presence of hormone, albeit to a lesser extent than in its absence. Altogether, our observations suggest that the enhancement of ER activity by p160/SRCs and CBP can be regulated by phosphorylation and stress the importance of using full-length receptors to assess the role of the activation function-1 in cofactor recruitment.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
ESTROGENS PLAY AN important role in both female and male reproductive function, as well as in female cancers, and they have multiple effects on the nervous, skeletal, and cardiovascular systems (1, 2). Although some of these effects are initiated by receptors present at the cell surface, many estrogen effects in the female reproductive and skeletal systems are mediated by regulation of gene transcription by estrogen receptor (ER) and ERß in the cell nucleus (3, 4, 5). ER and ERß are related molecules that belong to the nuclear receptor superfamily of transcription factors, whose activity is regulated by ligands, such as steroid and thyroid hormones and vitamins A and D (3). Nuclear receptors exhibit similar, yet distinct, structural and functional features, with a centrally located DNA-binding domain (DBD or domain C) and a C-terminal ligand-binding domain (LBD or domain E; see Fig. 2A). Both ER and ERß bind the hormone 17ß-estradiol (E2) and DNA sequences called estrogen-response elements (EREs). Detailed characterization of ER indicates that it increases the transcription of target genes through two transcription activation functions, AF-1 and AF-2, that reside in the A/B and E domains, respectively (see Fig. 2A). When these AFs are analyzed separately, the AF-2 activity is dependent on ligand, whereas the AF-1 activity is constitutive (6). However, both the AF-1 and AF-2 may participate in the E2-induced activity of the full-length receptor, with their relative contribution depending on the cell and promoter context (7, 8).

View larger version (31K): [in this window] [in a new window]   Figure 2. Mutation of Amino-Terminal Serine Phosphorylation Sites to Alanine Decreases ER Activity in the Presence of E2 and in the Absence of Ligand A, Schematic of human ER. Amino acid positions, structural domains (A–F), functional domains (see text), and serine (Ser) phosphorylation sites discussed in this study are indicated. B, Ligand-independent, ER-mediated activation of reporter gene activity in transfected cells. HeLa cells were placed in either charcoal-stripped serum-containing medium (5% sFBS) or chemically defined medium (CD-CHO), transfected with 10 ng pCMV5 or pCMV5-ER along with 1000 ng ERE-E1b-CAT and 100 ng pCR3.1-ßgal, and treated with vehicle (0.1% ethanol), 1 nM E2 or 100 nM ICI 182,780 for 24 h. CAT activity in cell extracts was normalized to ß-gal activity and expressed as a percentage of activity in the presence of stripped serum and ER without E2. Duplicate samples were measured in all experiments, and data are presented as the average ± SEM of several experiments. C, ER is phosphorylated at Ser118 in the absence of E2 treatment. Extracts from HeLa cells cultured in 5% sFCS and transfected with 1 µg pCMV5, pCMV5-ER(wt), or pCMV5-ER(S104/106/118A) were analyzed by Western blot with antibodies against phosphoserine118 ER (16J4, top) or total ER (H222, bottom). D and E, The S118A and S104/106/118A mutations reduce ER activity, but not its expression. HeLa cells grown in 5% sFBS were transfected with 10 ng pCMV5-ER (wt, S118A or S104/106/118A), along with 1000 ng ERE-E1b-CAT and 100 ng pCR3.1-ßgal, and treated with either 1 nM E2 or vehicle for 24 h. Harvested cells were split into two halves, one for measurement of reporter gene activity (D), and one for Western blot analysis of ER expression using the H222 antibody (E). In D, CAT activity was normalized to ß-gal activity and expressed as a percentage of wt ER activity in the presence of E2. The inset shows the fold induction by E2. Duplicate samples were measured in all experiments, and data are presented as the average ± SEM of three experiments. E, ER expression levels in a representative experiment.

A wide variety of non-DNA binding molecules, called coactivators, have been identified that are able to enhance ligand-induced activity of steroid receptors, including ER, through direct or indirect binding to these receptors (21). In addition, in the case of the p160/SRC (steroid receptor coactivator) and p300/CBP [cAMP response element binding protein (CREB)-binding protein] families of coactivators, neutralization experiments using specific antibodies or antisense oligonucletoides suggest that these coregulators are critical for ligand-induced, nuclear receptor-mediated transcription activation (22, 23, 24, 25). There are three p160/SRC family members, SRC-1/nuclear receptor coactivator-1 (NCoA-1) (26), transcription intermediary factor-2 (TIF2)/glucocorticoid receptor-interacting protein-1 (GRIP1)/SRC-2/NCoA-2 (27, 28), and receptor-associated coactivator-3 (RAC3)/activator for thyroid hormone and retinoid receptors (ACTR)/p300/CBP-cointegrator-associated protein (pCIP)/amplified in breast cancer-1 (AIB1)/thyroid hormone receptor-activator molecule-1 (TRAM-1)/SRC-3/NCoA-3 (23, 29, 30, 31). SRC-1 and CBP can synergize with each other in enhancing E2-induced ER activity (32), which is consistent with their ability to interact with each other physically (33). The p160s and p300/CBP can interact directly with the AF-2 region through LXXLL motifs in coactivators (where L stands for leucine and X for any amino acid) (21), and studies examining the isolated E region indicate that E2 binding induces or stabilizes a conformation in this domain that increases its affinity for these motifs (34). However, clearly there are other mechanisms that control the interactions of these cofactors with ER, because SRC-1 can enhance the ligand-independent activity of the full-length receptor, either basal or stimulated by elevated cAMP levels (35, 36). In agreement with these data, both the p160 and p300/CBP families of coactivators can interact with the isolated A/B domain and enhance its AF-1 activity (24, 37, 38, 39, 40). Furthermore, they enhance the functional and physical interactions between the A/B and LBD regions induced by E2 (39, 40, 41).

Besides p300/CBP and p160s, so far only a few other cofactors have been shown to interact physically and/or functionally with the ER A/B domain. These include the unrelated p68/p72 (42, 43) and steroid receptor RNA activator (SRA) (44) coactivators, which can cooperate with each other and with p160s in enhancing E2-induced ER activity, owing to the ability of p68/p72 to physically interact with the ER A/B domain, p160s, and steroid receptor RNA activator (SRA) (43). The human homolog of the yeast DNA repair and transcription regulator MMS19 is also an AF-1-specific coactivator of ER that interacts in a ligand-independent manner with the receptor and also binds to RAC3 (45).

One hypothesis is that Ser^104/106/118 may regulate ER activity by influencing its interactions with AF-1 coactivators. Indeed, Endoh et al. (42) showed that phosphorylation of Ser¹¹⁸ increases the physical and functional interactions of p68 with the isolated ER ABC region. These data, together with p68/p72-p160 and p160-CBP interactions, raise the possibility that the recruitment and/or action of p160s and CBP may also be regulated by Ser¹¹⁸ phosphorylation. This hypothesis is also supported by the observation that Ala mutation of the Erk2 phosphorylation sites (Ser¹⁰⁶ and Ser¹²⁴) in murine ERß affects both its physical and functional ligand-independent interactions with SRC-1 in response to Ras activation (46).

In this study, we show that a p160 coactivator contributes to the ligand-independent ER activation of a target gene in a cellular model in which ER, coactivator, and target gene are endogenous. Using transfected cells, we further show that the full-length ER can interact physically and functionally with all three p160/SRCs and CBP in the absence of ligand in vivo and that mutation of Ser^104/106/118 to Ala residues in ER affects these interactions. In addition, mutation of these residues affects ER coactivation by a subset of coactivators in the presence of E2, albeit to a lesser extent than in the absence of hormone. Further analysis reveals that mutations of both Ser^104/106 and Ser¹¹⁸ decrease ligand-independent SRC-1 coactivation of ER activity by two mechanisms. First, there is a seemingly indirect effect on SRC-1 recruitment that, surprisingly, requires other receptor domains in addition to A/B, which is consistent with our finding that SRC-1 enhancement of the ligand-independent interaction between the A/B and DEF regions is regulated by the Ser^104/106/118 phosphorylation sites. Secondly, we observe an effect on SRC-1 coactivation of the A/B domain that does not depend on the remainder of the molecule.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
Coactivator Contribution to ER Activation of Gene Expression in the Absence of Ligand
Expression of the pS2 gene in the MCF-7 breast cancer cell line is regulated by estrogens through activation of the endogenously expressed ER (47). To determine whether the receptor also regulates gene expression in the absence of exogenous ligand, we examined whether treatment with a pure ER antagonist, ICI 182,780, affects pS2 mRNA levels in these cells. Cells previously grown in charcoal-stripped serum-containing medium for 24 h were treated with vehicle, E2, or ICI 182,780 for another 20 h; total RNA was extracted; and pS2 mRNA levels were measured by quantitative real-time RT-PCR. As expected, E2 treatment produced an approximately 2.8-fold increase in pS2 expression in comparison to vehicle-treated cells (Fig. 1A). In contrast, ICI 182,780 treatment decreased pS2 levels by approximately 4-fold, suggesting that the ligand-independent pS2 expression is mediated by ER. Note that the serum used in the experiments described in this report was charcoal-stripped to deplete steroids from it, and cells were rinsed three times in serum-free medium before transfer to stripped serum-containing medium. To ensure that similar results were obtained in media lacking any potential estrogen contamination, we used a chemically defined medium, CD-CHO, which contains no serum, proteins, or estrogens. Cells were washed three times with DMEM and placed in CD-CHO medium 24 h before hormonal treatments and were maintained in this medium for the duration of the experiment. Again, E2 treatment produced an approximately 2.3-fold increase in pS2 expression in comparison to vehicle-treated cells, whereas ICI 182,780 treatment decreased pS2 levels by approximately 3-fold (Fig. 1B), confirming the ligand-independent activation of pS2 expression by ER.

View larger version (23K): [in this window] [in a new window]   Figure 1. Coactivator Contribution to ER Activation of Gene Expression in the Absence of Ligand A and B, Ligand-independent, ER-mediated activation of endogenous gene expression in MCF-7 cells. MCF-7 cells were placed in either 10% charcoal-stripped serum-containing medium (10% sFBS; A) or serum-free medium (CD-CHO; B) for 48 h and treated with vehicle (0.1% ethanol), 1 nM E2 or 100 nM ICI 182,780 for another 20 h, and RNA was analyzed for pS2 mRNA levels, which were quantitated by real-time RT-PCR and normalized to 18S rRNA levels. C, Contribution of endogenous TIF2 to ligand-independent ER activation of pS2 gene expression. The same procedure as in A was repeated, except that cells were transfected with 200 pmol of either a TIF2 antisense (AS) or random sense (RS) oligodeoxynucleotide 24 h before hormonal treatments (top), and that both pS2 and TIF2 mRNA levels were quantitated. In bottom panels, for each hormonal condition, pS2 (left) and TIF2 (right) mRNA levels in the presence of AS are expressed as a percentage of cognate mRNA levels in the presence of RS. D, TIF2 antisense decreases TIF2 protein levels. Aliquots of cells from the experiments in panel C were analyzed for TIF2 protein levels by Western blot. Veh., Vehicle.

The S104/106/118A Mutation in ER Decreases Its Ligand-Independent Coactivation by p160s and CBP
The preceding experiments indicated that coactivators contributed to ligand-independent ER activity in cells with endogenous levels of receptor and coactivators. To examine the molecular mechanisms by which ER and coactivators may functionally interact in the absence of ligand, ER activity was reconstituted in an ER-negative cell line by cotransfecting HeLa cells with a wild-type (wt) ER expression vector and a synthetic reporter gene [ERE-E1b-chloramphenicol acetyl transferase (CAT)] containing an ERE upstream of the adenovirus E1b gene TATA box and the chloramphenicol acetyl transferase (CAT) coding sequence. Cotransfection of the ER expression vector increased reporter gene activity in the absence of estrogen treatment, and this activity was blocked by the pure antiestrogen, ICI 182,780, further indicating that it was receptor-dependent (Fig. 2B, left). Similar results were obtained in the serum-free CD-CHO medium (Fig. 2B, right). As expected, whether serum-containing or serum-free media were used, addition of E2 further increased reporter gene activity in a receptor-dependent manner. In addition, parallel experiments with an E1b-CAT reporter lacking an ERE indicated that both ligand-stimulated and ligand-independent activities were ERE-dependent (data not shown). Altogether, these data demonstrate that ER can activate transcription in the absence of ligand in an ERE-dependent manner.

We then examined the role of ER phosphorylation in ligand-independent transcriptional activity. Mutation of Ser¹¹⁸ to Ala (S118A) has been shown to decrease ER activity induced by either E2 (16, 17) or EGF (15, 19), and the additional mutation of Ser^104/106 to Ala (giving the S104/106/118A mutation) has been reported to further reduce E2-induced activity (17). To compare the effects of the S118A and S104/106/118A mutations on both ligand-dependent and ligand-independent ER activity, HeLa cells were transfected with wt or mutant ER expression vectors along with the ERE-E1b-CAT reporter gene. As shown in Fig. 2C, wt ER, but not the S104/106/118A mutant is recognized by a phosphoserine¹¹⁸ antibody, indicating that Ser¹¹⁸ is phosphorylated under our ligand-independent conditions. This is consistent with our previous demonstration of ³²P incorporation into wt ER under basal conditions (48). In agreement with previous reports, both the S118A and S104/106/118A mutations decreased E2-induced, ER-mediated activity of the reporter gene by approximately 30% and 40%, respectively (Fig. 2D). Similar effects were observed in the case of vehicle-treated cells, in which the S118A and S104/106/118A mutations decreased activity by approximately 35% and 50%, respectively. Consequently, the fold induction of activity by E2 was not affected by the mutations (Fig. 2D, inset). Also note that the S104/106/118A mutant was consistently less active than the S118A mutant both in the absence of ligand and in the presence of E2 [83 ± 6% and 84 ± 4% of S118A activity (average ± SEM), respectively]. Importantly, the decreases in activity by the mutations were not due to reduction in receptor expression levels (Fig. 2E).

Because E2-induced ER activity is mediated at least in part by the p160/SRC and p300/CBP families of coactivators (22, 23, 24) and SRC-1 is also able to enhance the ligand-independent activity of the receptor (35, 36), we investigated whether the effects of the S118A and S104/106/118A mutations on ER activity could be due to an effect of these mutations on the ability of coactivators to enhance ER activity. For this, the effects of cotransfected coactivators on ER activity were assessed in the context of wt and mutant receptors. Because the p160/SRC family members significantly differ in their sequences, all three of them (i.e. SRC-1, TIF2, and RAC3) were studied. Furthermore, the two major isoforms of SRC-1 (SRC-1a and SRC-1e) were examined. SRC-1a differs from SRC-1e by a unique 56-amino acid sequence in its very C terminus that decreases the activity of its adjacent transactivation domain, reflected by a decrease in ER coactivation (36). Also of potential interest, whereas all p160/SRCs possess three centrally located, nuclear receptor AF-2-interacting LXXLL motifs, only SRC-1a has a fourth LXXLL motif in its C-terminal region. In contrast, the two p300 family members (CBP and p300 itself) appear to be functionally equivalent when transiently transfected into cells (21), and therefore only CBP was included in this study.

All five coactivators examined enhanced ER activity both in the presence and absence of E2. This was observed both in the chemically defined CD-CHO medium (Fig. 3A) and in stripped serum-containing medium (Fig. 3B). Note that coactivators did not enhance reporter gene activity in the absence of ER, indicating that their effects were receptor-dependent (data not shown). Because ligand-independent ER transcriptional activity (Fig. 2) and coactivation (Fig. 3) were similar in DMEM containing 5% stripped fetal bovine serum (sFBS) or CD-CHO medium, subsequent experiments were performed in the stripped serum-containing medium only. The effects of the S118A and S104/106/118A mutations on coactivation were then assessed. In the absence of exogenous ligand, the S104/106/118A mutation significantly reduced coactivation of ER activity by all p160/SRC family members and CBP, although to varying extents (SRC-1e, 25%; SRC-1a and TIF2, 50%; RAC3 and CBP, 65%; Fig. 3C). In contrast, in the presence of E2, the same mutation had no effect on ER coactivation by SRC-1 and TIF2 and only modestly decreased coactivation by RAC3 and CBP (by 30% and 20%, respectively; Fig. 3D). Although RAC3 and CBP were the coactivators most affected by the S104/106/118A mutation both in the vehicle and E2-treated cells, the effect of this mutation was much more pronounced in the absence of ligand. When compared with the S104/106/118A mutation, mutation of S118A alone had similar, but less pronounced effects on the action of coactivators (Fig. 3, C and D). The finding that the S104/106/118A mutations have bigger effects in the absence of E2 treatment likely reflects the lack of strong interactions between the AF-2 domain of ER and coactivators in these conditions.

View larger version (40K): [in this window] [in a new window]   Figure 3. Mutation of Amino-Terminal Serine Phosphorylation Sites to Alanine Decreases ER Ligand-Independent Coactivation by p160s and CBP A and B, Coactivation of wt ER activity both in the presence of E2 and in the absence of ligand. C and D, Effects of ER S118A and S104/106/118A mutations on coactivation by p160s and CBP in the absence of ligand (C), or in the presence of E2 (D). HeLa cells were transfected with 10 ng pCMV5-ER (wt, S118A or S104/106/118A), along with pCR3.1 or the indicated coactivators, 1000 ng ERE-E1b-CAT and 100 ng pCR3.1-ßgal, and treated with either 1 nM E2 or vehicle (0.1% ethanol) for 24 h. In one set of experiments (A), cells were placed in CD-CHO medium and 1000 ng coactivator plasmids were cotransfected. In a second set of experiments (B–D), cells were placed in 5% sFBS, and 2500 ng of coactivator plasmids were used. In both cases, duplicate samples were examined in each experiment, and CAT activity in cell extracts was normalized to ß-gal values. For both sets of experiments, a representative experiment (A and B, left) and the average ± SEM of the fold coactivation by each coactivator in the whole set of experiments (A and B, right) are shown for wt ER. In the experiments in 5% sFBS, the effects of mutations in ER were assessed and are shown in C (vehicle) and D (E2), in which the fold-coactivation elicited by each coactivator for ER mutants was expressed as a percentage of the fold-coactivation of wt receptor. These data are presented as the average ± SEM of at least three experiments.

The S104/106/118A Mutation in ER Decreases Its Ligand-Independent, Physical Interactions with p160s and CBP

View larger version (33K): [in this window] [in a new window]   Figure 4. Amino-Terminal Phosphorylation of Full-Length ER Decreases Its Ligand-Independent, Physical Interactions with p160s and CBP A, CBP and p160s can interact physically with unliganded ER in a mammalian two-hybrid assay. B, Quantitation of the effects of the S104/106/118A mutation in ER on its interactions with p160s and CBP. C, Effects of the S118A and S104/106A mutations alone and in combination on ER interactions with SRC-1e. HeLa cells were transfected with 500 ng of expression vector for VP16 or VP16-ER (wt or mutants), along with 100 ng of GAL4 or GAL4-coactivator, and 1000 ng pG5-Luc. In all experiments, duplicate samples were examined, and luciferase activity in cell extracts was normalized to protein amounts. In A, which shows a representative experiment, the activity of VP16-ER wt in the absence of coactivator was set as 100. In B and C, for each ER mutant the activity was expressed as a percentage of that observed for wt ER with the same coactivator, and data are presented as the average ± SEM of at least three experiments. D, The S118A and S104/106A mutations do not reduce VP16-ER expression. Extracts from HeLa cells transfected with 500 ng of plasmid for VP16 or VP16-ER (wt or mutants), along with 1000 ng of pG5-Luc were analyzed by Western blot using the H222 antibody. E, ER dephosphorylation reduces its ligand-independent interactions with SRC-1 and CBP in co-IP assays. Recombinant human ER was preincubated in -PPase buffer either in the absence (lanes 3 and 8) or presence of -PPase (lanes 4 and 9) or E2 (lanes 2 and 7), before its mixing with HeLa cell extract and co-IP with antibodies against SRC-1 or CBP. An aliquot of ER preincubated without -PPase or E2 was mixed with HeLa cell extract and -PPase before co-IP (lanes 5 and 10). Note that all IP were carried out in the presence of phosphatase inhibitors. Aliquots of input (1%; top) and IP supernatants (1%; middle) and eluates (15%; bottom) were analyzed together by Western blot with antibodies against total ER (left) or phosphoserine118 ER (ER-118P; right), as indicated. All blots analyzed with the same antibody were exposed for the same amount of time to allow comparison of IP samples with input. A lower exposure is also shown for eluates from IP with the CBP antibody.

Phosphatase Treatment of ER Decreases Its Ligand-Independent Interactions with SRC-1 and CBP in Vitro
Because we could not exclude the possibility that the effects of the S104/106/118A mutation on ER-coactivator interactions might be mediated by an effect on receptor conformation rather than on its phosphorylation, we then assessed whether treatment of ER with phosphatase affected its interactions with coactivators in an in vitro coimmunoprecipitation (co-IP) assay. For this, recombinant human ER, purified from Sf9 cells and preincubated alone, in the presence of -protein phosphatase (-PPase, a broad serine/threonine/tyrosine phosphatase), or in the presence of E2 as a positive control, was further incubated with HeLa cell extracts and antibodies against coactivators in the presence of phosphatase inhibitors. As expected from previous studies, ER coimmunoprecipitated with both SRC-1 and CBP in the presence of E2 (Fig. 4E, lane 2). Antibodies against both coactivators pulled down the receptor in the absence of ligand as well (Fig. 4E, lane 3), although to a lesser extent than in the presence of E2. Preincubation of ER with -PPase decreased receptor co-IP with both coactivators (Fig. 4E, compare lanes 3 and 4). In contrast, when the phosphatase was added to the immunoprecipitation (IP) mix only (Fig. 4E, lane 5), receptor-coactivator interactions were not altered, indicating that the effects of phosphatase on interactions were mediated by ER. Analysis of sample aliquots with a phosphoserine¹¹⁸ antibody before IP showed that recombinant ER used in these experiments was phosphorylated and that pretreatment with -PPase dramatically reduced this phosphorylation (Fig. 4E, Input). In contrast, analysis of ER phosphorylation status in IP supernatants shows that the -PPase was unable to dephosphorylate ER in the co-IP mixture (Fig. 4E, compare lanes 8–10). These data indicate that it is dephosphorylation of the receptor that produces alterations in ER-SRC-1/CBP interactions. Altogether, and in agreement with our previous two-hybrid experiments using phosphorylation site-mutated receptor, these results indicate that ER can physically interact with both SRC-1 and CBP in the absence of ligand and that these interactions depend, at least in part, on ER phosphorylation.

The S104/106/118A Mutation Decreases ER AF-1 Coactivation by p160s and CBP
Previous studies have shown that the S118A mutation decreases the activity of the ER AF-1 isolated from the LBD (14, 16). Because in our experiments the S118A and S104/106/118A mutations differentially affected the ligand-independent coactivation of the full-length ER, we then assessed the effects of these mutations on the isolated, AF-1-bearing ER A/B domain. HeLa cells were cotransfected with the GAL4-responsive pG5-Luc reporter gene and GAL4 DBD-ER A/B domain fusion constructs (GAL4-A/B), containing either a wt or phosphorylation site(s) mutant A/B domain (Fig. 5). In agreement with previous reports, the activity of GAL4-A/B was decreased by the S118A mutation (by 30%). Furthermore, it was also reduced by a S104/106A mutation (by 50%), and the combined S104/106/118A mutation decreased the activity more than the individual S118A and S104/106A mutations (by 75%). Conversely, mutants with Ser¹¹⁸ and/or Ser^104/106 mutated to glutamic acid (E), which mimicks phosphorylated residues on the basis of charge and size, were more active than the wt. The S118E, S104/106E, and S104/106/118E mutants were respectively 2.5-, 10-, and 26-fold more active than their Ala mutant counterparts. Thus, both Ser¹¹⁸ and Ser^104/106 affect the activity of the A/B domain.

View larger version (13K): [in this window] [in a new window]   Figure 5. Effects of Phosphorylation Site Mutations on the Activity of the Isolated ER A/B Domain HeLa cells were transfected with 40 ng GAL4, GAL4-A/B (wt or mutants), along with 1000 ng pG5-Luc. Duplicate samples were examined in all experiments. Luciferase activity in cell extracts was normalized to protein amounts and expressed as a percentage of GAL4-A/B(wt). Data are presented as the average ± SEM of three experiments.

View larger version (22K): [in this window] [in a new window]   Figure 6. Mutation of Serine Phosphorylation Sites to Alanine in the ER A/B Domain Decreases Its Coactivation by p160s and CBP HeLa cells were transfected with 40 ng GAL4, GAL4-A/B (wt or mutants), along with 2000 ng of pCR3.1 or the indicated coactivators, and 1000 ng pG5-Luc. In all experiments, duplicate samples were examined and luciferase activity in cell extracts was normalized to protein amounts. The increase in activity elicited by each coactivator for mutant GAL4-A/B was expressed as a percentage of its effect on GAL4-A/B(wt), and data are presented as the average ± SEM of at least three experiments.

View larger version (16K): [in this window] [in a new window]   Figure 7. Mutation of Serine Phosphorylation Sites to Alanine in the ER A/B Domain Decreases Its Coactivation by SRC-1 in Combination with CBP HeLa cells were transfected with 40 ng GAL4, GAL4-A/B (wt or S104/106/118A), along with 1000 ng pG5-Luc and the indicated amounts (in micrograms) of pCR3.1, pCR3.1-SRC-1e and/or pCR3.1-CBP. In all experiments, duplicate samples were examined, and luciferase activity in cell extracts was normalized to protein amounts. In A, which shows a representative experiment, the activity of GAL4-A/B(wt) in the absence of coactivator was set as 100. In B, the increase in activity elicited by the combination of SRC-1e and CBP for mutant GAL4-A/B was expressed as a percentage of its effect on GAL4-A/B(wt), and data are presented as the average ± SEM of three experiments.

View larger version (28K): [in this window] [in a new window]   Figure 8. Mutation of Serine Phosphorylation Sites to Alanine in the ER A/B Domain Does Not Affect Its Physical Interaction with SRC-1 HeLa cells were transfected with 40 ng GAL4, GAL4-A/B (wt or mutants), along with 1000 ng pG5-Luc and 250 ng VP16, VP16-SRC-1a or VP16-CBP. In all experiments, duplicate samples were examined, and luciferase activity in cell extracts was normalized to protein amounts. In A, which shows a representative experiment (S118A and S104/106A mutants not shown), the activity of GAL4-A/B(wt) in the absence of coactivator was set as 100. In B, because the mutations affect the activity of GAL4-A/B, the fold interaction of each ER derivative with coactivators was calculated as the ratio of the activity in the presence of coactivator to the activity without coactivator, and for each coactivator the fold interaction with wt A/B was set as 100. Data are presented as the average ± SEM of three experiments.

The S104/106/118A Mutation Decreases the Ligand-Independent Interaction between the N and C Termini of ER and Its Enhancement by SRC-1

View larger version (22K): [in this window] [in a new window]   Figure 9. The S104/106/118A Mutation in the ER A/B Domain Decreases the Ligand-Independent Interaction between the N and C Termini of ER and Its Enhancement by SRC-1 A and B, The S104/106/118A mutation decreases the ligand-independent, physical interaction between the N and C terminus of ER. C and D, The S104/106/118A mutation decreases the enhancement of the ligand-independent interaction between the N and C terminus of ER by SRC-1e. HeLa cells were transfected with 10 ng GAL4, GAL4-A/B(wt), or GAL4-AB(S104/106/118A) (S:A), along with 150 ng VP16 or VP16-DEF, and 1000 ng pG5-Luc. In addition, 200 ng pCR3.1 or pCR3.1-SRC-1e were cotransfected in C and D. Duplicate samples were examined in all experiments, and luciferase activity in cell extracts was normalized to protein amounts. In A, activity was expressed as a percentage of the activity observed for wt A/B in the absence of the DEF domain. In B, the fold interaction between A/B and DEF (ratio of the activity with DEF to the activity without DEF) was calculated for wt and mutant A/B and was set as 100 for wt. In C, activity was expressed as a percentage of the activity observed for wt A/B in the absence of the DEF domain and of SRC-1e. In D, the enhancement of interaction between A/B and DEF by SRC-1e was calculated for each A/B derivative and was set as 100 for wt A/B. Data are presented as the average ± SEM of three experiments, except for C, which shows a representative experiment.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

We first confirmed that the full-length ER exhibits transcriptional activity in the absence of E2 treatment. This was observed in both MCF-7 cells and transfected HeLa cells, either in the presence of stripped serum or in a chemically defined, estrogen-free medium. Further studies in HeLa cells indicated that this activity was ER- and ERE-dependent. Several reports indicated that SRC-1 can enhance the ligand-independent activity of ER (35, 36), and we further demonstrate here that all three p160s, including both isoforms of SRC-1, share this characteristic. However, other studies failed to find so (46), and this may be due to differences in species (murine vs. human ER), cell type, target promoter, receptor/coactivator ratio, or the use of SRC-1a, which has been found to be less efficient than SRC-1e and other p160s in coactivating ER in certain contexts (Refs. 36 and 38 and this study). In addition to p160s, we found that CBP also coactivated the ligand-independent activity of ER. Furthermore, using a mammalian two-hybrid system, we provide evidence that the three p160s and CBP (full length) can physically interact with the unliganded, full-length ER in human cells. This is consistent with our previous observation that yellow fluorescent protein-tagged SRC-1 exhibits ligand-independent association with a green fluorescent protein-tagged ER bound on a genome-integrated artificial array of EREs in living cells (50). We also demonstrate ligand-independent ER interactions with SRC-1 and CBP in vitro, using full-length receptor and HeLa cell coactivators. Furthermore, we show by antisense oligodeoxynucleotide technology that a p160 coactivator contributes to ligand-independent ER activation of pS2 gene expression in MCF-7 cells in which ER, coactivators, and the target gene are endogenously expressed. Altogether, these data suggest a role for the p160 and CBP coactivators in the ligand-independent activity of the full-length ER.

A potential explanation for the ligand-independent activity of the receptor is the ability of the A/B (AF-1) domain to be constitutively coactivated by the p160/SRC and p300/CBP families of coactivators (24, 37, 38). In agreement with previous reports, all p160/SRCs and CBP coactivated the isolated ER A/B domain in our study. We further show that SRC-1 and CBP can synergize to enhance the AF-1 activity, indicating that their ability to cooperate, already observed in the case of the liganded receptor (32), does not require the E2-bound LBD. In previous studies, the isolated ER A/B domain displayed physical interactions with the three p160s in vitro (24, 38). Using a mammalian two-hybrid assay, we show here that its interaction with SRC-1a can take place in vivo and that CBP can also interact with the isolated ER A/B domain. However, in the case of CBP, whether this interaction is direct remains to be determined, because CBP can directly interact with p160s (21).

As shown by earlier studies and confirmed here, the S118A mutation reduces the activity of the isolated ER A/B domain and the full-length receptor. However, the reason for this has not been defined. Because endogenous p160/SRCs and CBP are expressed in HeLa cells (51), our finding that the S118A mutation decreases the ability of p160s and CBP to enhance the ligand-independent activity of both the isolated A/B domain and the full-length receptor potentially explains, at least in part, how this mutation affects ER activity. In a previous study, the S104/106/118A mutation did not decrease GRIP1 (mouse SRC-2) coactivation of the isolated ER A/B domain, but it should be noted that under the conditions used in those experiments the mutation had no significant effect on the activity of the isolated A/B domain (38). Consistent with our findings, Tremblay et al. (46) reported that mutation of the murine ERß Erk2 phosphorylation sites, Ser¹⁰⁶ and Ser¹²⁴, to Ala residues affected the ligand-independent coactivation of the receptor by SRC-1. However, whereas in that study SRC-1 action on ERß could be completely abolished by mutation of both Erk2 sites (but not by either site alone), in our analysis of the full-length ER the Erk2 site mutation did not reduce coactivation by any p160/SRC by more than 40%. It is not clear whether the differences in these results actually reflect functional differences between receptor subtypes or are merely due to variations in other experimental parameters, such as the target promoter, MAPK activity, or cell type. Nevertheless, these data suggest that serine phosphorylation of both ERs, by Erk2 or other kinases, can promote their functional interactions with p160s in a ligand-independent manner. Furthermore, we provide the first evidence that the S118A mutation also reduces CBP action, to the same extent as the most affected p160 (RAC3), which is consistent with the notion that p160s and CBP can cooperate in enhancing ER activity.

Interestingly, mutation of the Ser⁷³⁶ MAPK phosphorylation site in GRIP1 to Ala compromises its ability to mediate EGF stimulation of an ER S118A mutant (52). In addition, MAPK phosphorylation of SRC-1 at Thr¹¹⁷⁹ and Ser¹¹⁸⁵ affects its coactivation of the progesterone receptor (53), and we observed that mutation of all seven potential MAPK sites (Ser⁵⁶⁹, Ser³⁹⁵, Ser¹⁰³³, Ser³⁷², Ser⁵¹⁷, Thr¹¹⁷⁹, and Ser¹¹⁸⁵; Ref. 53) to Ala residues also decreases SRC-1 coactivation of ER (54). Taken together, it has become apparent that phosphorylation of ER and its coactivators plays an important role in receptor-dependent gene expression. It remains to be determined whether MAPK phosphorylation of amplified in breast cancer-1 (AIB1)/RAC3 (55), p300 (56), and CBP (57) influences their ability to coactivate ER transcriptional activity.

In agreement with previous reports, we found that the S118A and S104/106/118A mutations also reduced the activity of ER in the presence of E2. However, whereas these mutations affected both ligand-independent and E2-induced ER-mediated transcriptional activity to the same extent, their negative impact on ER coactivation by p160s and CBP was much less in the presence of E2 than in the absence of ligand. Although SRC-1a, SRC-1e, and TIF2 can interact with the isolated A/B domain and enhance its activity, their coactivation of E2-bound ER is not affected by Ser^104/106/118. This may be explained by a relatively strong E2-induced interaction between these coactivators and the LBD of the receptor. However, our data indicate that mutation of Ser^104/106/118 to Ala residues can reduce the functional interactions of the full-length, E2-bound receptor with a p160 coactivator (RAC3) and with CBP (by 30% and 20%, respectively). These effects may contribute to the 40% decrease in E2-induced ER activity that results from the S104/106/118A mutation in experiments in which no coactivator is transfected. These observations indicate that although the LBD plays an important role in recruiting coactivators, the Ser^104/106/118 in the A/B domain, most likely through their phosphorylation, also contribute to coactivator action in the presence of E2. In addition, our findings add a new example of functional differences between the related p160/SRC proteins. These molecules also differ in their histone acetyltransferase activity, which is absent in TIF2 (21).

The S104/106/118A mutation has been reported to affect E2-stimulated, ER-mediated activity more strongly than the single S118A mutation (17). This was confirmed in our study and was extended to the ligand-independent activity of the receptor. The difference between these mutants was even more pronounced when looking at the ligand-independent coactivation of the full-length receptor by p160s and CBP. Clearly, both S104/106A and S118A mutations affected the activity of the A/B domain, with the S104/106A mutation seemingly having a prominent impact, and both mutations reduced AF-1 coactivation by SRC-1. Altogether, these data indicate that both the Ser^104/106 and Ser¹¹⁸ phosphorylation sites in the A/B domain control coactivator action.

Our study reveals the existence of several mechanisms by which the S104/106/118A mutation affects the ligand-independent coactivation of ER activity by p160s and CBP. The first mechanism is the reduction (by up to 50%) of the physical interactions of the full-length ER with these coactivators. Tremblay et al. (46) observed a similar effect of the S106/124A mutation in an AF-2-deficient ERß on its interactions with SRC-1 in response to Ras activation. Our data extend these concepts to the interactions of the wt ER with all three p160s and CBP. In addition, we show that both the S118A and S104/106A mutations decrease ER-SRC-1 physical interactions.

In contrast, the effects of these mutations (whether alone or combined) on ER-SRC-1 physical interactions could not be observed in the context of the isolated A/B domain (although it did interact with SRC-1), indicating that the Ser^104/106 and Ser¹¹⁸ phosphorylation sites are not primary interaction sites for SRC-1, at least in this context. In a recent study, in vitro interaction between SRC-1 and the isolated ER B region produced in bacteria was not affected by the S104/106A and S118A mutations (whereas the S118A mutation affected the recruitment of p68 in the same conditions) and was mediated at least in part by an AF-1 -helical core (residues 35–47) (Ref. 58). This is consistent with our data, and we further demonstrate here that the Ser^104/106 and Ser¹¹⁸ phosphorylation sites do not impinge on the physical interaction of the isolated A/B domain with SRC-1 in HeLa cells. In another study, residues 90–116 in the A/B domain were found to increase its interactions with p160s in vitro, whereas residues 117–145 seemed to decrease these interactions (38). Our data suggest that these modulations are not due to the phosphorylation sites located in these portions of the molecule.

Similar to what we observed with SRC-1, the S104/106/118A mutation affected the physical interaction of CBP with the full-length ER much more than with the isolated A/B domain (55 vs. 20%). Altogether, our observations suggest that the ligand-independent, physical interactions of the receptor with SRC-1 and CBP are controlled jointly by the A/B and CDEF regions. This is consistent with our finding that ternary interactions between the N and C termini of ER and SRC-1, which were previously observed in the presence of E2 (41), can also take place in the absence of ligand. The existence of such ternary interactions in the absence of ligand is also supported by the observations that the isolated A/B domain can constitutively recruit p160s (Refs. 24 and 38 and this study) and that the isolated LBD can physically interact with p160s in the absence of hormone in vivo, albeit to a much lesser extent than in the presence of E2 (Refs. 59 and 60 ; Jaber, B., and C. L. Smith, unpublished data). Furthermore, we show that these ligand-independent ternary interactions are affected by the S104/106/118A mutation, consistent with the effect of this mutation on SRC-1 recruitment by the full-length receptor.

Thus, although the Ser^104/106 and Ser¹¹⁸ phosphorylation sites appear not to be primary interaction sites for SRC-1 in the context of the isolated A/B domain, they affect SRC-1 recruitment in the context of the full-length receptor. Although we cannot exclude the possibility that the presence of the CDEF region enables the Ser^104/106 and Ser¹¹⁸ phosphorylation sites to directly interact with SRC-1, it seems more likely that the effects of these sites on SRC-1 interaction with the full-length receptor are indirect. However, the mechanisms underlying these effects remain to be determined. In particular, at this point it is not clear whether the Ser^104/106/118 phosphorylation sites specifically regulate the interactions of the AF-1 or AF-2 with SRC-1, or both. Importantly, Métivier et al. (61) recently showed that the A domain and the C-terminal helix 12 of ER compete with each other and with corepressors for binding to the same hydrophobic cleft within the receptor LBD, thereby regulating receptor activity. Helix 12 positioning is also involved in p160 and CBP recruitment by the receptor (34). Thus, these authors proposed that the receptor can adopt various (active and inactive) conformations in the absence of ligand, the stability of which can be regulated by cofactors (61). Our observations raise the possibility that Ser^104/106/118 phosphorylation modifies not only the overall intensity of the ER N-C-terminal interaction, but also the nature of this interaction, favoring an SRC-mediated B–E interaction over a direct A–E interaction.

As a second mechanism by which the S104/106/118A mutation affects p160 coactivator action, our data indicate that it markedly decreases (by 40%) the ability of coactivator (i.e. SRC-1) to enhance the activity of the A/B domain, without affecting quantitatively their physical interaction. The same conclusions can be drawn in the case of the S104/106A and S118A mutations taken separately. Apparently, in the context of the isolated A/B domain, Ser^104/106 and Ser¹¹⁸ are not important for the recruitment of SRC-1, but rather for its activity (either intrinsic or mediated by its associated proteins). Also, CBP activity was more affected than its recruitment by the S104/106/118A mutation (50% vs. 20%).

Collectively, our data also suggest that other factors may be involved in the regulation of ER activity by Ser^104/106/118. Indeed, in the presence of E2, the S118A mutation had no effect on ER coactivation by p160s and only had a very modest effect on CBP action, which is unlikely to be responsible for the 30% decrease in ER activity due to S118A mutation in experiments in which no coactivator was cotransfected. Thus, although E2-induced ER activity is mediated in part by p160s and CBP, our data suggest that its decrease upon S118A mutation is due to factors other than the p160 coactivators, possibly in combination with CBP. Similarly, in the context of the isolated A/B domain, the S104/106/118A mutation appears to have a bigger impact on the activity without cotransfected coactivator than on the coactivation by any p160 or CBP coactivator alone or by SRC-1 in combination with CBP, suggesting that other factors may be involved. These may include the p68 and related p72 coactivators. Indeed, p68 is expressed in HeLa cells, both its physical and functional interactions with the A/B domain are regulated by phospho-Ser¹¹⁸, and it is involved in E2-induced activity (42, 43). In our study, because coactivation of E2-bound ER by overexpressed p160s was not affected by the S118A mutation, it may not depend on the endogenous p68. However, because p68 can synergize with p160s in enhancing E2 activity in cotransfection experiments (43), we cannot exclude the possibility that p160s might mediate the regulation of E2 activity by the Ser¹¹⁸ phosphorylation site in conjunction with p68 under conditions in which p160s are not overexpressed relative to p68. Finally, besides coactivators, the S118A mutation may also affect the physical interactions of ER with corepressors, as has been shown for nuclear receptor corepressor (N-CoR) in the presence of antiestrogen (24).

In conclusion, our data suggest that the full-length ER can interact physically and functionally with p160/SRCs and CBP in the absence of ligand in vivo, and that these interactions can be affected by mutation of Ser^104/106/118 located in the A/B region of the receptor. Further analyses reveal that both Ser^104/106 and Ser¹¹⁸ impinge on SRC-1 action by two mechanisms: an effect on SRC-1 recruitment that seems to be indirect and requires other receptor domains in addition to the A/B, and an effect on SRC-1 coactivation that can be observed in the absence of the CDEF region. Altogether, this study suggests that the regulation of ligand-independent ER activity by Ser^104/106/118 phosphorylation is mediated in part by a modulation of p160/SRC and CBP recruitment and activity, whereas the impact of these phosphorylation sites on E2- induced activity seems to be primarily influenced by other factors. In addition, our observations suggest a complex interplay between receptor domains and cofactors, and stress the importance of using full-length receptors to assess the role of the AF-1 and its phosphorylation in cofactor recruitment.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
Reagents and Plasmids
E2 was obtained from Sigma (St. Louis, MO). ICI 182,780 was a gift of A. Wakeling (Zeneca Pharmaceuticals, Macclesfield, UK). The anti-ER antibody (H222) is a rat monoclonal antibody obtained from Abbott Laboratories (Abbott Park, IL). The antibodies against phosphoserine¹¹⁸-ER (16J4) and TIF2 are mouse monoclonal antibodies from Cell Signaling Technology (Beverly, MA) and BD Biosciences (San Jose, CA), respectively. The TIF2 antisense and random sense oligodeoxynucleotides (no. 29977 and no. 117226, respectively) were synthesized by ISIS Pharmaceuticals (Carlsbad, CA) and have been described in detail by Cavarretta et al. (25).

The estrogen-responsive reporter genes, ERE-E1b-CAT (62) and ERE-E1b-Luc (63), have been used in previous studies, and both contain nucleotides -331 to -87 of the vitellogenin A2 promoter linked upstream of the adenovirus E1b TATA box. The pCR3.1-ßgal plasmid contains the ß-galactosidase (ß-gal) cDNA under control of the cytomegalovirus (CMV) promoter (63). The pCR3.1 vector is from Invitrogen (Carlsbad, CA), and the pBIND (GAL4) and pACT (VP16) vectors are from Promega Corp. (Madison, WI).

The pCMV₅-ER wt and mutant (S118A and S104/106/118A) expression vectors were generous gifts of Benita Katzenellenbogen (17). To generate the pACT-ER wt construct, ER (full-length) was inserted into pACT downstream of and in frame with the VP16 activation domain. To generate the pBIND-A/B wt, S118A, S104/106A, and S104/106/118A constructs, the ER A/B domain was amplified by PCR from wt and mutant pCMV₅-ER templates and cloned at the BamHI site of pBIND in frame with the GAL4 DBD. The S118E and S104/106E mutations were introduced in the pBIND-A/B wt construct by PCR-based site-directed mutagenesis using oligonucleotide primers with the desired mutations. Serine-to-alanine mutations were introduced in pACT-ER by substituting its 430 nt BamHI/FseI fragment with the corresponding fragments of the pBIND-A/B