Mutations in the Estrogen Receptor DNA-binding Domain Discriminate between the Classical Mechanism of Action and Cross-talk with Stat5b and Activating Protein 1 (AP-1)*

Estrogen receptors (ERs) efficiently potentiate the transcriptional activity of prolactin-activated Stat5b through a mechanism that involves the ER DNA-binding domain (DBD) and the hinge domain. We have identified residues within the DBD of ER that are critical for the functional interaction of ER with Stat5b. We show that disruption of the second zinc finger structure abrogated cross-talk between ER and Stat5b, while the structure of the first zinc finger was not important. Furthermore, we confirm that intact DNA binding activity was not required for potentiation of Stat5b activity and that the dimerization of ER did not seem to be involved. Ligand-bound ERs also modulated activating protein 1-dependent transcription, and our data demonstrate that both zinc finger structures of the ER DBD are important for an intact response. We show that introduction of various point mutations within the DBD altered the response of the receptor to 17β-estradiol and to the estrogen antagonists 4-hydroxytamoxifen and ICI 182,870 on the collagenase promoter. These findings provide new insights into the mechanisms by which ERs act in cross-talk with non-related transcription factors.

Estrogens are powerful mitogens that promote growth and proliferation in many target organs. Their effects are mediated by two related nuclear hormone receptors, estrogen receptor ␣ (ER␣) 1 and estrogen receptor ␤ (ER␤). These receptors belong to a large superfamily of nuclear hormone receptors that share a well conserved DNA-binding domain (DBD) and a structurally conserved ligand-binding domain. The N-terminal domains of these receptors, on the other hand, do not resemble each other (1,2). The classical mechanism of activation of ERs depends on ligand binding to the receptor following which the receptor dimerizes and binds to estrogen response elements (EREs) located within the promoters of estrogen-responsive genes (for review, see Ref. 3). Ligand binding also induces a conformational change in the ligand-binding domain of the receptor, which allows the recruitment of co-activator proteins (4,5).
The ER is also able to regulate gene expression in the absence of DNA binding by modulating the activities of other transcription factors. This mechanism is referred to as crosstalk and is common for several nuclear receptors (for review, see Ref. 6). We recently showed that ER␣ and ER␤ efficiently potentiate the transcriptional activity of Stat5b when Stat5b is bound to the ␤-casein promoter upon prolactin (Prl) stimulation (7). We demonstrated that ER␣ and ER␤ can interact with Stat5 through the DBD/hinge domain, and furthermore, we showed that the interaction of ER with classical co-activator proteins is dispensable for the potentiation of Stat5 activity. Ligand-bound ERs have also been demonstrated to up-and down-regulate the transcription of genes that contain AP-1 sites, binding sites for the Jun⅐Fos complex, in a cell-and ER subtype-specific manner (8 -11). In addition, ER enhances the transcription of genes containing SP1-binding sites (12) and, conversely, represses the transcription of NF-B (13,14), GATA-1 (15), and CCAAT/enhancer-binding protein (14) when these transcription factors are bound to their cognate DNAbinding sites. The mechanism by which such cross-talk controls the expression of genes is not completely understood, but it is believed that the DNA binding activity of ER is not involved. The discovery of this mechanism would explain how estrogens regulate genes in which no consensus ERE has been found.
In this study, we further examined the domain of ER required for a functional interaction with Stat5b. We show that specific residues within the second helical structure of the ER DBD are essential for the potentiation of Stat5b transcriptional activity, whereas intact DNA binding activity per se is not required. We also analyzed the influence of various ER DBD mutations on the regulation of AP-1-dependent transcription and propose that distinct parts of the ER DBD distinguish between the classical mechanism of action and cross-talk with Stat5b and AP-1.

EXPERIMENTAL PROCEDURES
Plasmids-The ␤-casein (Ϫ344 to Ϫ1) luciferase reporter was provided by Bernd Groner (Frankfurt, Germany) (16), the human Stat5b expression vector was provided by Julian Ng (Imperial Cancer Research Fund, London, UK), the long form of the prolactin receptor (Prl-R) was provided by Paul Kelly (Paris, France) (17), the collagenase (Ϫ73 to ϩ63) luciferase reporter was provided by Peter Kushner (University of California, San Francisco, CA) (8), and the TRE-tk-luc reporter was provided by Sam Okret (Karolinska Institutet, Huddinge, Sweden). The following plasmids have been described previously: ERE-tk-luc (18), human ER␤ expression vector pSG5-ER␤ (19), and mouse ER␣ expression vector pMT2-MOR (20). The L206A, R207A/K208A, Y210A, K175A/ R176E, E167A/G168A, C149A/C152A, A187T, and P186T mutants were constructed by site-directed mutagenesis (QuikChange ® , Stratagene) with oligonucleotide primers designed to introduce specific point mutations in pSG5-ER␤. The C201A/C204A, S200A, S200E, and ER␤ ⌬122-226 mutants were constructed by PCR techniques and subsequent * This work was supported by the Swedish Cancer Society, the Karolinska Institutet, the Swedish Medical Society, the M. Bergwall Foundation, and the Å. Wiberg Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
subcloning into pSG5-ER␤. The DNA sequences of the constructed ER␤ mutants were verified by sequencing.
Cell Culture and Transient Transfection Techniques-COS-7 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) supplemented with 10% fetal bovine serum (FBS) (Invitrogen). HC11 cells were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% FBS, 5 g/ml epidermal growth factor (human recombinant, Sigma), and insulin (Actrapid, Novonordisk, Bagsvaerd, Denmark). For transient transfection assays, COS-7 and HC11 cells were seeded in DMEM free of phenol red (Invitrogen) supplemented with 5% dextran-charcoal-stripped FBS (HyClone Laboratories, Inc.) in 24-well microtiter plates 24 h before transfection. Cells were transfected with 1 g of reporter plasmid and 250 ng of pCMV-␤Gal plasmid as an internal control and various expression plasmids, as indicated in the figure legends, together with an empty expression vector to give a total amount of 2 g of DNA/well using a modified calcium phosphate coprecipitation method (21). The transfection medium was changed after 24 h to DMEM free of phenol red. The hormones 17␤-estradiol (E 2 ) (10 Ϫ8 M) (Sigma), 4-hydroxytamoxifen (OHT) (10 Ϫ7 M) (Sigma), ICI 182,780 (10 Ϫ7 M) (Tocris Cookson, Inc.), and ovine Prl (5 g/ml) (Sigma) were added as described in the figure legends. After 24 h, the cells were harvested, and luciferase was measured as described previously (7). Expression of the various ER␤ proteins was confirmed by Western blotting using an ER␤ antibody (Upstate Biotechnology, Inc.).

Mutations within the Second Helical Structure of ER␤ DBD Disrupt Functional
Interaction with Stat5b-We have previously reported that ER-mediated potentiation of Stat5b transcriptional activity through cross-talk in the nucleus requires the DBD/hinge domain (7). To determine which amino acids within this domain are important for this potentiation, we have introduced point mutations into the DBD region of ER␤. The mutant receptors were analyzed for their functional activity on a classical ERE reporter, and they were analyzed in cross-talk with Stat5b on the Stat5-responsive ␤-casein reporter by transient transfection assays in COS-7 cells. The ERE reporter includes two consensus ER DNA-binding sites (18), and the ␤-casein reporter comprises a fragment of the ␤-casein promoter (Ϫ344 to Ϫ1) and includes two Stat5-binding sites (16). The structure of the ER␤ DBD and the location of the mutations that we introduced are shown in Fig. 1A. Data in Fig. 1C confirm the results we have reported previously (7). The wildtype ER␤ efficiently potentiated transcription from the ␤-casein reporter, the transcription being increased in the presence of Prl by a factor of 5 over the level seen when Stat5b was present alone. The transcription was increased even further in the presence of Prl plus E 2 . Interestingly, mutations introduced within the amphipathic helix beyond the second zinc binding motif (L206A, R207A/K208A, and Y210A) completely eliminated the ability of ER␤ to potentiate the activity of Prl-activated Stat5b (Fig. 1C), whereas the transcriptional activity on the classical ERE reporter was retained (Fig. 1B). These results show that this region of the ER DBD has an important function in cross-talk with Stat5b. Conversely, mutation of residues within the "P-box" of the recognition helix (K174A/R175E and E167A/G168A), which is known to be involved in direct interaction with DNA (22), resulted in the loss of activation of the ERE reporter (Fig. 1B), whereas the ability of the receptor to potentiate transcription from the ␤-casein reporter remained intact (Fig. 1C). These results demonstrate that the DNA binding activity of ER is not required for functional interaction with Stat5b. Also shown in Fig. 1C, and as reported previously (7), a DNA binding-defective mutant in which two cysteines in the second zinc finger (C201A/C204A) had been changed to alanines was unable to act in cross-talk with Stat5b, suggesting that an intact structural conformation of the DBD is required. However, a corresponding disruption of the first zinc finger (C149A/C152A) did not interfere with the ability of ER␤ to potentiate Stat5b activity, defining the region required for functional interaction with Stat5b to the second zinc binding motif of the DBD and the following amphipathic helix. As expected, neither of the ER␤ zinc finger mutants were able to activate the ERE reporter (Fig. 1B). We also analyzed the importance of residues known to participate in the dimer interface (22). Whereas substitution of serine 200 for alanine (S200A) did not affect the activity of the receptor on the ERE- driven reporter, substitution for glutamic acid (S200E) resulted in minimal activation of the reporter (Fig. 1B). Mutation of the corresponding serine residue within ER␣ to glutamic acid (S236A) prevents DNA binding by inhibiting the dimerization of ER␣ in the absence of ligand (23). The fact that the S200E mutant was still able to potentiate the transcriptional activity of Stat5b in response to Prl alone (Fig. 1C) indicates that potentiation of Stat5b activity is a function of an ER monomer. Likewise, substitution of alanine 187, which lies within the "D-box" of the second zinc finger, for threonine (A187T) also resulted in diminished transcriptional activity on the ERE reporter (Fig. 1B), although the mutated receptor was still able to potentiate transcription from the ␤-casein reporter (Fig. 1C). We also analyzed whether the proline residue adjacent to alanine 187 is important for dimerization since the corresponding position within the glucocorticoid receptor DBD is crucial for dimerization (24). However, substitution of this proline for threonine (P186T) did not affect the activity of the receptor on the ERE-driven reporter (Fig. 1B), indicating that mutation of this residue does not inhibit dimerization of ER. This substitution also did not affect cross-talk with Stat5b (Fig. 1C). As expected, deletion of the entire DBD (⌬122-226) completely eliminated potentiation of Stat5b activity (Fig. 1C) as well as activation of the ERE reporter (Fig. 1B), confirming that an intact domain is required for functional interaction with Stat5b. Notably, similar results to those described above were obtained in transient transfection assays using the mouse mammary epithelial cell line, HC11 cells (data not shown), confirming that the observed phenotypes of the ER␤ DBD mutants are not restricted to a certain cell type. Western blot analyses of cell extracts confirmed equal expression of the wild-type and mutant receptors (Fig. 1D).
Both Zinc Finger Structures within the ER␤ DBD Contribute to Transrepression of AP-1-The finding that the introduction of specific point mutations into the ER␤ DBD abrogated potentiation of Stat5b transcriptional activity prompted us to analyze whether this was a common phenomenon in cross-talk between ER and non-related transcription factors. We examined the ability of ER␣ and ER␤ to modulate AP-1-dependent transcription using the coll-73-luc reporter, which comprises a fragment of the collagenase promoter (Ϫ73 to ϩ63) and includes a single AP-1-binding site (8). In HC11 cells, the coll-73-luc reporter was repressed 2-fold in the presence of E 2 , while transcription was induced 2.5-fold in the presence of the full estrogen antagonist ICI 182,870 when cells were cotransfected with ER␣ or ER␤ expression plasmids ( Fig. 2A). Interestingly, whereas the responses to E 2 and ICI 182,870 were similar in the presence of ER␣ and ER␤, only ER␤ was able to activate transcription from the coll-73-luc reporter upon stimulation with the selective estrogen receptor modulator OHT ( Fig. 2A), showing that the two receptor subtypes differ in their ligand preferences. Previous studies have shown that antagonistbound ERs induce the transcriptional activity of AP-1 (8 -11). Whether AP-1-regulated promoters are actively transrepressed or induced in response to E 2 may depend on the composition of the AP-1 complex (25). In accordance with our results, cellspecific ER␤-mediated transrepression of the collagenase promoter in response to E 2 has been demonstrated (9) and has been recently described for ER␣ (11). Fig. 2B shows the activities of the various ER␤ DBD mutants on the AP-1-regulated coll-73-luc reporter upon transient transfection of HC11 cells. Interestingly, cotransfection of the L206A, R207A/K208A, and Y210A mutants, which were all found to eliminate cross-talk with Stat5b (Fig. 1C), resulted in a reversed activity by E 2 with a 3-fold activation of the reporter. On the other hand, treatment with estrogen antagonists OHT or ICI 182,870 did not activate transcription. The two P-box mutants (E167A/G168A and K174A/R175E), which had no effect on the ERE reporter (Fig. 1B), displayed distinct effects on the AP-1-regulated reporter. The E167A/G168A mutant behaved as the wild-type receptor, repressing AP-1 activity in response to E 2 as previously shown with the corresponding ER␣ mutant (11). However, the K174A/R175E mutant displayed the same reversed activity in response to E 2 as the L206A, R207A/K208A, and Y210A mutants displayed. The K174A/R175E also did not respond to treatment with estrogen antagonists, indicating that the region between the two zinc binding motifs has an essential function in the modulation of AP-1-dependent transcription. Furthermore, the integrities of both zinc finger structures of the DBD were essential, and disruption of either the first or the second zinc binding motif (C149A/C152A and C201A/C204A) altered the response to ER ligands on the AP-1-regulated reporter. The requirement for both zinc fingers contrasts with the situation concerning crosstalk with Stat5b, and it suggests that the functional interaction between ER and AP-1 is sensitive to conformational changes within both zinc fingers of the ER DBD. Notably, an ER␣ zinc finger mutant (C241A/C244A) also induced activation of the coll-73-luc reporter in response to E 2 (data not shown), demonstrating that the reversed activity obtained upon disruption of the DBD is not restricted to ER␤. Interestingly, the S200E and A187T mutants both reversed the response to E 2 on the coll-73-luc reporter (Fig. 2B). These results show that residues participating in the dimer interface also contribute to E 2 -induced transrepression of AP-1. However, activation of the reporter in response to estrogen antagonists OHT or ICI 182,870 was retained, suggesting that these receptor mutants act at the promoter through distinct pathways. The S200A and P186T mutants, which both displayed intact activities on the EREdriven reporter (Fig. 1B), repressed and activated the AP-1regulated reporter to the same extent as the wild-type receptor in the presence of E 2 and estrogen antagonists, respectively. Finally, deletion of the entire DBD (⌬122-226) also resulted in a reversed activity by E 2 , and this deletion eliminated activation of the reporter in the presence of estrogen antagonists (Fig.  2B). These results demonstrate that, whereas sequences within the DBD are essential for the ability of ER␤ to transrepress AP-1-dependent transcription in response to E 2 , the presence of the DBD is not essential for mediating the reversed, E 2 -induced activation of AP-1.
The Activity of ER␤ DBD Mutants at AP-1 Sites Is Cellspecific and Is Blocked by Estrogen Antagonists-To confirm that the ER-dependent regulation of transcription from the authentic coll-73-luc reporter is mediated through the AP-1binding site in the collagenase promoter, a consensus reporter construct including two AP-1 binding sites, TRE-tk-luc, was cotransfected with ER␤ wild type and with the ER␤ DBD deletion mutant (⌬122-226) into HC11 cells. Fig. 3A shows that results similar to those obtained with the coll-73-luc reporter were obtained using the TRE-tk-luc reporter (Fig. 2A). These results show that AP-1 is the target transcription factor of ER actions on the promoter. We also confirmed that the TRE-tk-luc reporter was induced in response to E 2 when relevant ER␤ DBD point mutants were transfected into cells (data not shown).
We next analyzed the ability of estrogen antagonists to block E 2 -induced activation of AP-1 in the presence of ER␤ DBD mutant receptors. Both the selective estrogen receptor modulator OHT and the full antagonist ICI 182,780 completely blocked E 2 -induced activation of the TRE-tk-luc reporter in the presence of ER␤ DBD deletion mutants and point mutants ( Fig. 3A and data not shown). The altered responses to various ligands on AP-1-regulated reporters observed upon introduction of specific point mutations within the ER␤ DBD suggest that agonist-and antagonist-bound ERs are able to modulate AP-1 activity through distinct mechanisms. These mechanisms are determined by the structure of the DBD. The natural splice variant ER␤␦3, with a deletion of exon 3 that causes disruption of the DBD, activates transcription from the collagenase promoter in the presence of E 2 but not in the presence of OHT (26). This is consistent with our results obtained using point mutations within the ER␤ DBD.
When transient transfection experiments were repeated in COS-7 cells, overexpression of ER␤ wild type resulted in a 2-fold repression of the coll-73-luc reporter in response to E 2 , while transcription was induced 2.5-fold in response to estrogen antagonists OHT and ICI 182,870, resembling the activity observed in HC11 cells (Fig. 3B). Surprisingly, however, none of the ER␤ DBD mutants, which reversed the activity by E 2 in HC11 cells, affected the reporter in COS-7 cells neither in the presence of E 2 nor in the presence of estrogen antagonists ( Fig.  3B and data not shown). These results show that introduction of specific point mutations within the DBD eliminates ER␤-dependent modulation of AP-1 activity in COS-7 cells. However, the same mutations enable the receptor to efficiently induce AP-1 activity in response to E 2 in breast epithelial cells HC11 (Fig. 2B) and MCF-7 (data not shown). This activity appears to be mediated through a distinct mechanism that can be blocked by estrogen antagonists. One explanation for the difference in the way that ER␤ affects AP-1-dependent transcription in different cell types may be provided by subtle changes in the structure of the DBD. The ER␤ DBD may be involved in the recruitment of additional cofactors/co-repressors that bind to the liganded ER, and these may be specific for a certain cell type. Any subtle changes in the structure of the DBD may have profound effects on the recognition of such cell type-specific factors. It should be noted, however, that the level of ER␤mediated potentiation of Stat5b activity in COS-7 cells (Fig.  1C) is similar to that of ER␤-mediated potentiation of Stat5b activity in HC11 cells (data not shown).
Our results do not allow detailed descriptions of molecular mechanisms, and any such descriptions will only be speculations. We have, however, shown that cross-talk between ER␤ and Stat5b and between ER␤ and AP-1 depends on overlapping but not identical residues within the ER DBD as summarized in Fig. 4. We have defined a region within the second zinc finger of the DBD that is essential for the functional interaction of ER␤ with Stat5b. Furthermore, we have shown that the ER-dependent modulation of AP-1 activity is sensitive to structural changes within both zinc fingers of the DBD and that FIG. 4. Residues within the ER␤ DBD discriminate between classical activation on an ERE and cross-talk with Stat5b and AP-1. Asterisks (*) indicate residues that are essential for transcriptional activity on an ERE reporter. Squares (Ⅺ) indicate residues involved in cross-talk with Stat5b, and circles (q) indicate residues involved in cross-talk with AP-1. The boxed proline residue can be modulated without affecting any of the three activities.
introduction of point mutations within this domain alters the responses of the receptor to ER ligands on the collagenase promoter. We have identified ER␤ mutants that discriminate between classical activation on an ERE and cross-talk with Stat5b and AP-1, and our results contribute to understanding the mechanisms by which ERs act in cross-talk with nonrelated transcription factors.