p300 Mediates Functional Synergism between AF-1 and AF-2 of Estrogen Receptor α and β by Interacting Directly with the N-terminal A/B Domains*

Estrogen receptor (ER) α and β mediate estrogen actions in target cells through transcriptional control of target gene expression. For 17β-estradiol-induced transactivation, the N-terminal A/B domain (AF-1) and the C-terminal E/F domain (AF-2) of ERs are required. Ligand binding is considered to induce functional synergism between AF-1 and AF-2, but the molecular mechanism remains unknown. To clarify this synergism, we studied the role of reported AF-2 coactivators, p300/CREB binding protein, steroid receptor coactivator-1/transcriptional intermediary factor-2 (SRC-1/TIF2) family proteins and thyroid hormone receptor-associated protein-220/(vitamin D3 receptor-interacting protein- 205-(TRAP220/DRIP205) on the AF-1 activity in terms of synergism with the AF-2 function. We found that neither any of the SRC-1/TIF2 family coactivators nor TRAP220/DRIP205 is potent, whereas p300 potentiates the AF-1 function of both human ERα and human ERβ. Direct interactions of p300 with the A/B domains of ERα and ERβ were observed in an in vitro glutathioneS-transferase pull-down assay in accordance with the interactions in yeast and mammalian two-hybrid assays. Furthermore, mutations in the p300 binding sites (56–72 amino acids in ERα and 62–72 amino acids in ERβ) in the A/B domains caused a reduction in ligand-induced transactivation functions of both ERα and ERβ. Thus, these findings indicate that ligand-induced functional synergism between AF-1 and AF-2 is mediated through p300 by its direct binding to the A/B regions of ERα and ERβ.

scriptional control of target gene expression (1)(2)(3). ERs are ligand inducible transcription factors, which belong to the superfamily of steroid/thyroid hormones, vitamin A and D nuclear receptors (4 -6). Based on structural and functional similarities, the nuclear receptors are dissected into five to six functional domains (designated domain A to E/F). Two domains are well conserved among nuclear receptors, the cysteine-rich C domain serving to direct DNA binding (DBD) and the Cterminal E/F domain forming a pocket for ligand binding. For the ligand-induced transactivation function of nuclear receptors, the E/F domain and a variable N-terminal A/B domain are prerequisite (7)(8)(9)(10)(11). The transactivation functions of both the A/B (AF-1) and the E/F (AF-2) domain depend on promoter context and cell type, and the properties of these domains in transactivation are distinct. Furthermore, it is reported that phosphorylation of the Ser residues in the A/B domains of human ER␤ (2,12) as well as human ER␣ (13)(14)(15)(16)(17) by mitogenactivated protein kinase enhances the AF-1 activity. The deletion studies of ER␣ showed that the activity of AF-2 is ligandinduced, whereas AF-1 itself exhibits constitutive activity (7,8). Therefore, in ER␣ whole molecule, the AF-1 activity is considered to be suppressed by the ligand-unbound E/F domain but be restored upon ligand binding. Thus, ligand binding is considered to induce functional and complicated synergism between AF-1 and AF-2 activities (18,19); however, the molecular mechanism of this synergism is largely unknown.
Evidence indicates that, during the ligand-induced transactivation, a coactivator complex is recruited, forming a higher complex with the nuclear receptor to achieve activation and repression of transcription. Several coactivators directly interacting with the E/F domain in a ligand-dependent manner have been identified, including the SRC-1/TIF2 (20 -22) and p300/ CBP (23)(24)(25)(26) families, TIF1 (27,28), ARA70 (29), PGC1 (30), and many others (31,32,33). More recently, a coactivator complex TRAP/DRIP containing none of the reported coactivators has been identified, and one of the components, TRAP220/ DRIP205 is shown to directly interact with the E/F domain of some nuclear receptors including ER␣ in a ligand-dependent way (34,35). Thus, such AF-2 coactivators have been well studied, whereas little is known about the AF-1 coactivator (s).
The present study was thus undertaken to clarify the ligandinduced functional synergism between AF-1 and AF-2 in ER␣ as well as ER␤, with attention paid to the actions of known AF-2 coactivators in the AF-1. Consequently, we found that the core domains for both the ER␣ and ER␤ AF-1 activities in the A/B domains are indistinguishable from those responsible for the functional synergism with each of the AF-2s. Though known coactivators (SRC-1, TIF2, and AIB1) enhanced the AF-2 of ER␣ as well as ER␤, neither of them nor TRAP220/ DRIP205 potentiated the AF-1s of ER␣ and ER␤. However, consistent with the transcriptional potentiation by p300, direct interaction with the ER␣ and ER␤ A/B domains was detected only in p300 in vitro. Furthermore, the ligand-induced transactivation functions of ER␣ and ER␤ were impaired by the mutations of the p300 binding sites in their A/B domains. Thus, the present study indicates that p300 mediates, at least in part, the functional synergism between AF-1 and AF-2 through its direct binding of the A/B regions of ER␣ and ER␤.
Cell Culture and CAT Assay-COS-1 cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) without phenol red and supplemented with 5% fetal bovine serum, which had been stripped with dextran-coated charcoal in 10-cm dishes. At 40 -60% confluency, cells were transfected by calcium phosphate with 2 g of ERE-G-CAT or 17M2G-CAT reporter plasmid, either 0.3 g of ER expression vector, 1 g of GAL4-fused ER expression vector, or 3 g of coactivator expression vector, along with 3 g of transfection indicator construct pCH110 (␤-galactosidase reporter) (Amersham Pharmacia Biotech), and Bluescribe M13ϩ (Stratagene) was used as carrier DNA to adjust the total amount of DNA to 20 g (37). After a 24-h incubation in medium containing the precipitated DNA, the cells were washed with fresh medium and continued to grow for an additional 24 h with or without 10 Ϫ8 M E2. Cell extracts were prepared by freeze-thawing and assayed for CAT activity after normalization for ␤-galactosidase activity (37). Results from CAT assays were analyzed by TLC, and the TLC plate was quantified using an image analyzer (BAS1500, Fuji Film, Tokyo, Japan) and shown as mean Ϯ standard deviations calculated from the three independent experiments.
Mammalian Two-hybrid Assay-COS-1 cells were maintained as described above for the CAT assay. The cells were transfected with 2 g of 17M2G-CAT reporter plasmid, either 1 g of GAL4-or VP16-fused ER expression vector or 3 g of coactivator expression vector, along with 3 g of pCH110 (Amersham Pharmacia Biotech), and Bluescribe M13ϩ (Stratagene) was used as carrier DNA to adjust the total amount of DNA to 20 g (32). After a 24-h incubation in medium containing the precipitated DNA, the cells were washed with fresh medium and continued to grow for an additional 24 h with or without 10 Ϫ8 M E2. CAT activity was measured and shown as described above.
GST Pull-Down Assay-The GST fusion protein or GST alone was expressed in Escherichia coli as described (32), and the expression of proteins of the predicted size was then monitored by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. For GST pull-down assays, bacterially expressed GST fusion proteins or GST bound to glutathione-Sepharose-4B beads (Amersham Pharmacia Biotech) and incubated with 35 S-labeled proteins were expressed by in vitro translation using the TNT-coupled transcription-translation system, with conditions as described by the manufacturer (Promega). After 2 h of incubation, free proteins were washed away from the beads with NET-N ϩ buffer (150 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl, pH 7.5, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol). Specifically bound proteins were eluted by boiling in SDS sample buffer and analyzed by 10% SDS-polyacrylamide gel electrophoresis. After electrophoresis, radiolabeled proteins were visualized using an image analyzer (BAS1500, Fuji Film, Tokyo, Japan).

Functional and Ligand-induced Interaction of the A/B Domain with the E/F Domain Is Indirect in hER␣ and hER␤-
For the ligand-induced transactivation in ERs, two domains in the N-terminal A/B and the C-terminal E/F domain act synergistically to stimulate transcription. However, the molecular mechanism of this synergism is largely unknown. Therefore, we first tested in vivo the interaction between the A/B and E/F domains in hER␣ and hER␤ by a mammalian two-hybrid assay with the hER deletion mutants fused to either the GAL4 DBD or VP16 transactivation domain (32). As expected from previous reports (18,19), ligand-induced interaction between the A/B and E/F domains in hER␣ was observed (Fig. 1A). Though the hER␣ A/B domain did not form a homodimer, it interacted with the E/F domain in a ligand-dependent way. Similar interactions between the two domains were detected also in hER␤ (Fig. 1B). Interestingly, ligand-induced interactions between the two domains derived from different ER subtypes (␣-␤) were also detected (Fig. 1C). Thus, the properties of the ligandinduced interactions in the A/B-E/F and the E/F-E/F domains are indistinguishable between hER␣ and hER␤. We then tried to delineate the regions of ligand-induced interaction of the A/B domain for the E/F domain with a series of A/B domain deletion mutants fused to a yeast GAL4 DBD. As shown in Fig. 2A, the interaction regions were mapped to the central region (aa 56 -127) of the hER␣ A/B domain and to the N-terminal region (aa 1-62) of the hER␤ A/B domain. Interestingly, these regions almost overlapped with the core regions for the full activities of the AF-1s of hER␣ and hER␤, though the AF-1 activity of hER␤ is half that of hER␣ ( Fig. 2A), as expected from previous reports (2,12). We confirmed that the expression levels of the A/B domain deletion mutants are almost the same by a Western blot analysis with an antibody for GAL4 DBD, when the cell extracts were normalized to the transfection efficiency (data not shown).
To test whether the observed in vivo interactions of the A/B domain with the ligand-bound E/F domain of hER␣ and hER␤ are physically direct or indirect, we applied a GST pull-down assay with bacterially expressed chimeric hER proteins fused with GST. Monitoring the properties of ligand-induced direct interactions of the E/F domains with coactivators, we chose a nuclear receptor coactivator, SRC-1, and consequently SRC-1 was found to associate with both the hER␣ and hER␤ E/F domains in a ligand-dependent manner (Fig. 3A, lanes 5-8) as reported (12,20). However, under this condition, physical in-teraction between the A/B and E/F domains was detected in neither hER␣ nor hER␤ even in the presence of ligand (Fig.  3A). Thus, these findings indicate that the ligand-induced interaction of the A/B domain with the E/F domain in hER␣ and hER␤ is indirect and possibly mediated through a factor bridging the two domains.
p300 Acts as a Coactivator for the AF-1 Activities of hER␣ and hER␤-To test this idea, we studied whether the well characterized AF-2 coactivators directly interacting with the E/F domains of various nuclear receptors potentiate the AF-1 activities of hER␣ and hER␤. For this study, all three 160-kDa coactivators, SRC-1, TIF2, and AIB1, and p300 were used. All four coactivators potentiated the ligand-induced transactivations of the full-length hER␣ and hER␤ (data not shown), as we expected from previous reports (21,22,23). Such potentiations by these coactivators were also confirmed for the AF-2 activities of hER␣ and hER␤ (data not shown), as reported (22,23). None of the 160-kDa coactivators potentiated the hER␣ and hER␤ AF-1 activities in the ER A-C domains (HE15, ER␣ A-C domain and HE␤15, ER␤ A-C domain) (Fig. 3B), whereas p300 was clearly potent (Fig. 3B).
p300 Mediates the Ligand-induced Interactions between the A/B Domain and the E/F Domain in hER␣ and hER␤-As p300 appeared to act as a coactivator also for the AF-1s of hER␣ and hER␤, we examined if p300 directly associates with the A/B domain in vitro and in vivo. The GST pull-down assay clearly showed interactions of p300 with both the hER␣ and hER␤ A/B domains in vitro, whereas none of the SRC-1/TIF2 family proteins were able to bind (Fig. 3A, lanes 3 and 4). TRAP220/DRIP205 showed no interaction (Fig. 3A), reflecting its inability to potentiate the AF-1s (data not shown). Consistent with these results, chimeric hER␣ and hER␤ A/B domains fused to the GAL4 DBD exhibited in vivo interactions with p300, but not with the SRC-1/TIF2 family proteins (Fig. 3C). The binding sites of the A/B domains for p300 in vivo were further mapped to 16 aa residues (ER␣ aa 56 -72) and 72 aa residues (ER␤ 1-72) (Fig. 4, A and B). Though the p300 binding site of hER␣ in vivo was essential for direct binding of p300 in vitro (Fig. 4C, upper panel), the p300 binding to the hER␤ A/B domain in vivo (Fig. 4C) required more region than tested in vitro (only 10 amino acids (62-72) are required for in vitro binding), suggesting that the N-terminal region (1-62 aa) contributes a stable interaction of the ER␤ A/B domain with p300 in vivo. Note that the expression levels of the chimeric ER mutants with GAL4 DBD were confirmed at almost the same levels when estimated by Western blot analysis (data not shown). Thus, as p300 is able to bind to the A/B domains in both hER␣ and hER␤, we next tested whether p300 enhances the ligand-induced interactions between the A/B and E/F domains in vivo, as observed in Fig. 1. The mammalian two-hybrid p300-mediated Functional Synergism in ER␣ and -␤ system revealed that the overexpression of p300 potentiates the ligand-induced interactions (Fig. 5B). The interactions between the A/B and E/F domains were impaired when the p300 binding to the E/F domains is abolished by deleting the AF-2 cores (534 -595 aa or hER␣ (VP16-ER␣⌬AD) and 432-477 aa of hER␤ (VP16-ER␤⌬AD)) (23) (see Fig. 5B). Notably, these ligand-induced interactions between two domains were further detected in vitro only in the presence of p300 (Fig. 5A, lanes 10  and 15).
Lack of p300 Binding Sites in the A/B Domains Impaired the Ligand-induced Transactivation Function of hER␣ and hER␤-To test an idea that p300 functionally bridges the A/B domain and the E/F domain in ER␣ and ER␤, we made deletion mutants lacking the p300 binding sites (⌬ER␣-1 and ⌬ER␤-1 as illustrated in Fig. 6A). As shown in Fig. 6B, the lack of binding sites caused a clear reduction in the ligand-induced transactivation in hER␣ and hER␤. Thus, it is most likely that the p300 action of AF-1 by its direct binding contributes to the ligand-induced transactivation function of ER␣ and ER␤ in whole receptor. Taken together, these findings indicate that p300 mediates the ligand-induced interactions between the A/B and E/F domains in hER␣ and hER␤ by its direct binding to the A/B domains.

DISCUSSION
The tissue-specific activity of hER␣ is considered to play a significant role in the actions of estrogen. Especially tamoxifen, an estrogen partial agonist, is considered to exert its tissuespecific actions through activating AF-1 and inhibiting AF-2 in hER␣ (9,40). Therefore, it is of great interest to identify the cofactors responsible for the activities of the ER(s) AF-1, in terms of synergistic function with AF-2. In the present study, we identified the AF-1 activity in the A/B region of hER␤, though it was weaker than that in the hER␣ A/B domain in COS-1 cells. As the poor activity of hER␤ AF-1 was seen in other cell types, such as HeLa cells and MCF-7 cells (data not shown), it is likely that in response to E2, the AF-1 activity contributes less than the AF-2 activity of ER␤ as in ER␣ in most target cells in accordance with previous reports (2, 12). However, it is possible that the ER␤ AF-1 activity dominates in some target cells and in cells in a pathophysiological state, because its activity is cell-type specific.
Several nuclear receptor coactivators have been studied, including the SRC-1/TIF2 (20 -22) and p300/CBP (23-26) family proteins, TIF1 (27,28), ARA70 (29), PGC1 (30), Smads (32), and others (31). It has been demonstrated that ligand binding induces the interactions of the SRC-1/TIF2 and p300/CBP proteins with ER␣ and other nuclear receptors. Most recently, a novel coactivator complex has been identified, and one of the components, TRAP220/DRIP205, is shown to bind the E/F domains of nuclear receptors in a ligand-dependent way (34,35). Further study revealed that helix 12 at the C terminus of the E/F domains serves as a direct interphase for these coactivators (38,39). Such ligand-induced functional interactions with the coactivators well explain the ligand-induced transactivation function of AF-2 in various nuclear receptors including ER␣ and ER␤ (40,41). However, little was known about the coactivators for AF-1, and in this respect it is of interest whether the AF-2 coactivators enhance the AF-1 activity or not. We found in the present study that all of the three SRC-1/TIF2 family proteins and p300 potentiate the AF-2 functions of ER␤ as well as ER␣ (data not shown), as expected from previous reports (21)(22)(23), reflecting the ligand-induced interactions of these coactivators with the E/F domains of ER␣ and ER␤ in vitro (Fig.  3A). Under the same conditions, we could detect neither any potentiation of the AF-1 activities of ER␣ and ER␤ by any SRC-1/TIF2 family proteins nor any direct interaction in vivo and in vitro, though previous reports showed that the SRC-1/ TIF2 family proteins potentiate the AF-1 activities of ER␣ and ER␤ (2,12,18). As these activities are dependent on cell types, we suspect that a cell type-specific factor specifies the actions of the SRC-1/TIF2 family proteins in the ER AF-1. In contrast, p300 enhanced the AF-1 activities of both ERs, even though in the Ser residue point mutants to be unphosphorylated by mitogen-activated protein kinase (data not shown). Supporting the p300 action, functional association in vivo and direct binding of p300 were detected in the A/B domains of ER␣ and ER␤. However, we observed a discrepancy between in vivo and in vitro in the hER␤ A/B regions required for p300 binding, suggesting the existence of an unknown factor acting with p300 on the ER␤ AF-1. Thus, taken together, these observations indicate that p300 is one of coactivators supporting the AF-1 activities of ER␣ as well as ER␤; however, unknown coactivator(s) also appears to be required for the AF-1 activities.
The present findings together indicate that p300 mediates, at least in part, the ligand-induced functional synergism between AF-1 and AF-2 through its direct binding to the A/B domains of ER␣ and ER␤, because direct interactions between the two domains in vitro were detected only in the presence of p300 (Fig. 5A). A previous study indicated that the SRC-1/TIF2 family proteins exhibit similar effects on the functional interaction between the two domains in vivo (18); however, in the present study, the SRC-1/TIF2 family proteins failed to induce such a ligand-induced interaction in vitro. Therefore, from the present study, it appears that such action of the SRC-1/TIF2 family proteins is mediated through p300 (possibly CBP) bound to the A/B domain. As a coactivator complex containing the SRC-1/TIF2 family proteins and p300/CBP is recruited for the ligand-induced transactivation functions of ER␣ and ER␤ during this ligand-dependent process, p300 (CBP) in this complex may bridge the A/B and E/F domains in association of other component(s). Likewise, a component(s) of the TRAP/DRIP coactivator complex may directly interact with the ER A/B domains when this complex is recruited by the ligand-bound E/F domain (34,35). It is moreover possible to speculate that a coactivator complex different from these two coactivator complexes interacts with the ER A/B domain to fulfill the ER AF-1 function.