Aspartate 351 of Estrogen Receptor α Is Not Crucial for the Antagonist Activity of Antiestrogens*

The antagonist activity of antiestrogens is due to the presence of a long carbon side chain at positions 7α or 11β or equivalent on their steroid or steroid-like skeletons. These side chains establish hydrophobic interactions with amino acids of the estrogen receptor α (ERα) ligand binding domain. In addition, a hydrogen bond formed between amino acid Asp-351 and the tertiary amine present at the end of the side chain of partial antiestrogens is considered to be crucial for their antiestrogenicity. Here, we have investigated the role of Asp-351 in antiestrogen action in transiently transfected HeLa and MDA-MB-231 cells. Our results indicate that disruption of the negative charge at position 351 does not increase the agonist activity of partial antiestrogens and thus that the hydrogen bond with the antiestrogen side chain is not determinant in positioning the side chain in an antagonist position. The negative charge at position 351 was not required for transcriptional activity in the presence of hormone, but its presence was necessary for basal activity of the wild-type receptor and constitutive activities of mutants L536P and Y537A, suggesting a role of Asp-351 in stabilizing the active conformation of ERα. This stabilizing role of Asp-351 could be due to interaction of Asp-351 with the amide group of the peptide bond between Leu-539 and Leu-540 in helix 12 observed in the active conformation of the ERα ligand binding domain.

Estrogen regulates target gene expression by binding to specific nuclear receptors that function as ligand-dependent transcription factors. Estrogen receptors (ERs) 1 contain two transcription activation domains, AF1 at the N terminus and AF2 in the C-terminal ligand-binding domain (1)(2)(3). Several proteins interact with AF2 in the presence of estrogen, some of which have the properties of transcriptional coactivators (4 -6).
The crystal structures of several nuclear receptor ligandbinding domains (LBDs) have now been determined (14 -19) and have revealed a striking conservation despite modest sequence homology (20). The LBD folds into a structure described as a sandwich of ␣-helices with a central hydrophobic ligandbinding pocket. In the presence of ligand, helices 3, 5, and 12 form a hydrophobic groove (21)(22)(23)(24) important for interaction with the LXXLL motifs (9,(25)(26)(27) found in the p160 family members and also in other coactivators. However, the crystal structure of estrogen receptor ␣ (ER␣) revealed that helix 12 is repositioned in the presence of the antagonists tamoxifen (Tam) or raloxifene (Ral), thereby disrupting the surface of interaction with coactivators (16,23). The side chain of these antiestrogens plays an important role in displacing helix 12. This suggests that amino acids of the ligand-binding domain that interact with the antiestrogen side chain play an important role in the transcriptionally inactive conformation of this domain. It has been suggested that integrity of aspartate 351, which forms a hydrogen bond with the tertiary amine present at the end of the side chains of Tam and Ral, is the key to the antiestrogenic character of these analogs (28). Indeed, a mutation of Asp-351 to tyrosine was isolated from an MCF7 tumor variant that was not inhibited but rather stimulated by Tam (29,30). Both Tam and Ral also behaved as agonists for expression of the estrogen target gene transforming growth factor-␣ in MDA-MB-231 cells stably transfected with this mutant of ER␣, while the full antiestrogen ICI182,780 remained inactive (28).
Here we have introduced several mutations at position 351 and tested the functional consequences of these changes on ER␣ transactivation properties in the presence of estrogen and of antiestrogens. Our results demonstrate that Asp-351 can be mutagenized to Gly, Ala, or Val without diminishing the antagonist activity of antiestrogens in HeLa cells. However, we provide evidence for a stabilizing effect of Asp-351 on the active conformation of the wild-type ER LBD in the absence of hormones.

EXPERIMENTAL PROCEDURES
Materials-Cell culture media and fetal bovine serum were purchased from Life Technologies, Inc. Estradiol, 4-hydroxytamoxifen (OHT), and ICI182,780 were purchased from Sigma. RU39,411 and RU58,668 were kind gifts from P. Van  nucleotides used for mutagenesis is available upon request). Expression plasmids for mutant ERs were generated by subcloning of a HindIII/ BamHI fragment of 768 base pairs into the pSG5-HEGO expression vector. Two clones generated by independent polymerase chain reactions were isolated for each mutant and characterized by restriction digest and sequencing. Mutants ERG400V, ERL536P, and ERY537A have been described previously (31)(32)(33)(34)(35). Vector pSG5-TIF2.1 expressing high levels of a truncated TIF2 that contains both the LXXLL motifs and the Gln-rich region was described previously (9,36).
Cell Culture and Transient Transfection Experiments-For CAT assays, HeLa cells were transiently cotransfected with pSG5 estrogen receptor expression vectors (0.5 g), with the ERE3-TATA-CAT reporter vector (2 g; Ref. 37) and with the CMV-␤-Gal internal standard construct (2 g) using the calcium phosphate coprecipitation technique. Precipitates were washed 24 h after transfection, and cells were incubated in the presence of hormones (as indicated) for another 24 h before harvesting. Whole cell extracts were prepared in 0.25 M Tris-HCl, pH 7.5, by three cycles of freeze-thawing and were standardized for ␤-galactosidase activity. CAT assays were performed as described previously (37). Each transfection was carried out in duplicate or triplicate and repeated at least three times.
For Western blotting, gel shift, and hormone binding assays, the wild-type and mutant receptors were expressed in COS-7 cells by transient transfection of 15 g of pSG5 expression vectors containing wildtype or mutant ER␣ cDNAs using the calcium phosphate coprecipitation technique.
Western Blotting-Whole cell extracts from transfected COS-7 cells were analyzed by 8% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis, transfer onto nitrocellulose, and incubation with anti-ER␣ mouse monoclonal antibodies B10 and F3 (obtained from Prof. P. Chambon). Complexes were revealed by ECL (NEN Life Science Products) as recommended by the manufacturer.
Gel Shift Assays-Complex formation between wild-type or mutant ERs and a consensus ERE probe from the vitellogenin A2 gene was determined as described previously (38). Essentially, cell extracts from transfected COS-7 cells were preincubated with poly(dI-dC) (2 g) for 15 min on ice prior to addition of the probe (100,000 cpm) and incubation at 25°C for 30 min. Receptor-DNA complexes were resolved from unbound DNA in 6% nondenaturing polyacrylamide gels in 0.5ϫ Tris borate-EDTA buffer and visualized by autoradiography.
Ligand Binding Assays-Whole cell extracts from transfected COS-7 cells were prepared in high salt buffer (50 mM Tris-HCl, pH 8.0, 1.5 mM EDTA, 50 mM NaCl, 10% glycerol, 10 mM sodium molybdate, 2 mM ␤-mercaptoethanol, and protease inhibitors) by three cycles of freezethawing. Protein content was determined by Bradford assays (Bio-Rad). For each binding reaction, 80 g of protein extract were used; each assay was carried out in duplicate. To the protein extract was added tritiated estradiol (6,7-[ 3 H]estradiol, Amersham Pharmacia Biotech), with or without a 100-fold excess of cold estradiol, and high salt buffer to a final volume of 50 l. After incubation for 2 h at room temperature followed by 10 min on ice, 50 l of high salt buffer containing 2% charcoal (Aldrich, Activated Carbon Norit SA3-100 mesh) and 0.1% Dextran T-70 (Amersham Pharmacia Biotech) were added to the samples. After 15 min of incubation on ice and centrifugation, 90 l of supernatant were counted by scintillation counting.
Glutathione S-Transferase (GST) Pull-down Assays-pSG5 plasmids containing cDNAs for wild-type or mutant ERs were linearized with BamHI, transcribed, and translated in vitro in reticulocyte lysate (Promega) in the presence of [ 35 S]methionine (Mandel) according to the manufacturer's instructions. GST fusion proteins were expressed in Escherichia coli BL21 cells by isopropyl-1-thio-␤-D-galactopyranoside induction. GST fusion proteins were bound to glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) by incubation of the beads with 1 ml of crude bacterial extract in GST buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM MgCl 2 , 0.3 mM dithiothreitol, 5% glycerol, 0.1% Nonidet P-40, and protease inhibitors). Beads bound to GST fusion protein or GST alone were incubated with in vitro translated wild-type or mutant ERs in the presence of vehicle or 1 M ligand at 4°C, overnight. The beads were then washed three times with 800 l of GST buffer, boiled for 2 min in 2ϫ Laemmli buffer, and electrophoresed on a 10% SDS-polyacrylamide gel. Gels were fixed with 20% methanol, 10% acetic acid for 15 min, rehydrated for 15 min, washed for 15 min in a 1 M salicylate solution, and vacuum-dried. Autoradiography was performed overnight.

Mutations of ER␣ That Suppress the Negative Charge at Position 351 Have Wild-type Levels of Estradiol-induced Transcriptional Activity in HeLa cells-Several point mutations
were introduced in wild-type human ER␣ (HEG0) at position 351 in helix 3 of the ligand binding domain. The aspartate residue was replaced by the hydrophobic amino acids alanine or valine, by the polar amino acid tyrosine or by the longer negatively charged residue glutamic acid; in addition, the effect of complete removal of the amino acid lateral chain at position 351 was investigated by replacement with a glycine residue (Fig. 1A). HEG0 and all mutants were expressed at similar levels after transient transfection into COS-7 cells (Fig. 1B) or HeLa cells (data not shown). The effects of these mutations on ER␣ transcriptional activity was investigated by cotransfection of expression vectors for each mutant together with a CAT reporter vector containing the minimal ERE3-TATA promoter. Levels of transcriptional activity generated from the reporter construct in the absence of exogenous hormone differed significantly between wild-type and mutant receptors (Fig. 2, inset). Although basal levels of activity with mutant HED351E were similar to those observed with HEG0, all other mutants generated very low or undetectable levels of activity in the absence of added hormone. However, levels of transcription achieved with all tested HED351 mutants at saturating concentrations of estradiol (10 nM and above) were comparable to those observed with HEG0 (Fig. 2).
To investigate whether recruitment of coactivators is similar with wild-type or mutant receptors, we performed GST pulldown assays using a fusion protein between GST and the coactivator TIF2.1 (9,36). Background levels of binding were observed in the absence of hormone or in the presence of antiestrogens OHT or ICI182,780. In the presence of estradiol, recruitment by the fusion protein of the HED351G, HED351V, and HED351Y mutants was comparable to that of wild-type ER␣ (about 10% of input protein, see Fig. 3), consistent with the transactivation data.
Elimination of the Negative Charge at Position 351 Does Not Increase the Agonist Activity of Partial Antiestrogens-The full antiestrogen ICI182,780 (100 nM) completely repressed transcriptional activity of HEG0 and all the Asp-351 mutants on the ERE3-TATA promoter (Fig. 4A). Weak transcriptional activity was detected in the presence of the partial antiestrogen OHT (100 nM) with HEG0 and mutants HED351E and HED351Y (Fig. 4A, inset). Activity of mutant HED351Y in the presence of OHT was comparable to those of wild-type and HED351E, but was elevated relative to its undetectable levels of basal activity. Similar results were obtained with the partial antiestrogen RU39,411 (data not shown). On the other hand, transcriptional activity of mutants HED351G, HED351A, or HED351V was undetectable in the presence of OHT (Fig. 4A).
We next investigated whether increased concentrations of coactivators could boost the agonist activity of OHT with mutants at position 351. Expression vectors for coactivators of the p160 family (TIF2, SRC1, and AIB1) were transiently cotransfected in HeLa cells together with an expression vector for wild-type ER␣ and the reporter vector ERE3-TATA-CAT. Induction of estradiol-dependent transcription was maximal when cotransfecting the expression vector pSG5-TIF2.1 (Ref. 9; data not shown). Estradiol-induced transcriptional activity of HEG0 or of the Asp-351 mutants was increased by ϳ7-fold in the presence of TIF2.1 (Fig. 4B). Strikingly, basal activity of the wild type or of the HED351E mutant was increased 80-fold, while basal activity of all other mutants was still undetectable. No activity of wild-type ER␣ or any of the mutants was detected in the presence of the full antagonist ICI182,780 (Fig.  4B). Activity of HED351Y with OHT was stimulated 40-fold in the presence of TIF2.1, converting OHT into an almost full agonist (Fig. 4B). Activity of wild-type ER␣ or of HED351E in the presence of OHT was also stimulated, although levels of transcriptional activity reached were lower than with HED351Y. On the other hand, transcriptional activity of the HED351G, HED351A, or HED351V mutants in the presence of OHT remained very low even in the presence of excess TIF2.1 (Fig. 4B). Treatment of HeLa cells cotransfected with expression vectors for ER␣ mutants and pSG5-TIF2.1 with increasing concentrations of OHT demonstrates that, although transcrip-tional activity of the D351Y mutant reached a plateau at nanomolar concentrations of OHT, activity of the D351G, D351A, and D351V mutants remained very low in the nanomolar to micromolar range of OHT (Fig. 4C).
To investigate whether the cellular context could influence the degree of agonist activity observed in the presence of OHT with the different mutants at position 351, we repeated our experiments in breast carcinoma MDA-MB-231 cells. Levels of OHT-induced activity were undetectable in MDA-MB-231 cells transfected with expression vectors for HEG0, HED351E, HED351V, or HED351Y and the minimal ERE3-TATA promoter. When coactivator TIF2.1 was overexpressed in these cells, activity of HED351Y was clearly detectable in the presence of OHT. HEG0 and mutant HED351E were stimulated to lesser degrees, and mutant HED351V was even more weakly active (Fig. 5). Thus, similar to what was observed in HeLa cells, removing the negative charge at position 351 was not sufficient to increase the agonist activity of OHT in MDA-MB-231 cells.

Suppression of the Negative Charge at Position 351 Does Not Affect Relative Binding to Estrogen and Antiestrogens-
The absence of activity of the HED351G, HED351A, and HED351V mutants with OHT could be due to defective antiestrogen binding. HeLa cells were incubated with increasing concentrations of OHT premixed with 0.3 nM E2. The resulting competition curves were similar for all receptors tested including HED351Y, suggesting that the relative affinity for OHT versus E2 was comparable in all mutants (Fig. 6A). Similar results were obtained with ICI182,780 (data not shown). We also performed in vitro hormone binding assays with wild-type and mutant receptors to assess whether ER mutants have a reduced affinity for estradiol. Results of these experiments indicate that the D351A, D351V, or D351Y mutations did not significantly affect estradiol binding ( Fig. 6B and data not  shown). These results demonstrate that mutations at position 351 suppressing the negative charge do not grossly perturb the structure of the ligand binding domain, and that the lower activity of the resulting mutants with OHT cannot be attributed to a low affinity for this antiestrogen.
Suppression of the Negative Charge at Position Asp-351 Destabilizes the Active Conformation of the Ligand Binding Domain in the Absence of Hormone-Although mutations D351G, D351A, and D351V did not affect ER␣ function in the presence of estrogen, all inhibited its basal activity, even in the presence of overexpressed TIF2.1 (Figs. 2, 4, and 5). These results are similar to the phenotype associated with a previously described mutation in the ER␣ ligand binding domain, G400V, which results in a receptor (HE0) with very low basal levels of activity   FIG. 2. Mutations that suppress the negative charge at position 351 reduce basal activity of the ER without affecting estradiol-induced transcription. Expression vectors for wildtype or mutant ER␣s (0.5 g) were transiently transfected into HeLa cells together with the reporter vector ERE3-TATA-CAT (2 g) and the internal control vector CMV-␤Gal (2 g). CAT activity was measured in extracts from cells treated with varying concentrations of estradiol. The inset presents the basal activities of wild-type or mutant receptors.

FIG. 3. Mutations that suppress the negative charge at position 351 do not affect in vitro recruitment of TIF2 in the presence of estradiol.
GST beads bound to a fusion protein between GST and TIF2.1 or to GST alone were incubated with in vitro translated wild-type or mutant ERs in the presence of vehicle, estradiol, or antiestrogens (100 nM) overnight at 4°C. After washing, protein samples were released from the beads by boiling in Laemmli buffer and separated by electrophoresis on a 10% SDS-polyacrylamide gel. 10% of input protein is shown for comparison. due to diminished dimerization and DNA binding in the absence of hormone (31). We performed gel shift assays to compare the effect of the G400V and the D351A, D351Y, or D351E mutations on binding to the consensus ERE. Wild-type ER␣ and all Asp 351 mutants bound to an estrogen response element with similar efficiency in the absence and in the presence of hormone (Fig. 7, compare lanes 5-12 to lanes 1 and 2; note that concentrations of the Asp 351 mutants were higher than those of HEGO in this assay). By comparison, ER mutant HEO did not bind DNA in the absence of hormone (Fig. 7, compare  lanes 3 and 4 to lanes 1 and 2). These results suggest that, unlike mutation G400V, mutations at position 351 do not affect DNA binding and dimerization.
The absence of effect of the Asp-351 mutations on ER ligand binding and DNA binding properties suggest that the removal of the negative charge at this position specifically affects transcriptional activity of ER␣ in the absence of added estrogen. To confirm that mutations at position Asp-351 can repress transcriptional activity in the absence of hormone, we introduced mutation D351A in the HEY537A (34,35) and HEL536P (32) receptors, which are constitutively active ER␣ mutants that can recruit coactivators in the absence of hormone in vitro (33,34,35). While mutants HEY537A and HEL536P activated the ERE3-TATA promoter both in the absence or the presence of estradiol, but not antiestrogens (Fig. 8), mutation of D351A abolished the high levels of basal activity observed with both constitutive receptors. In contrast, activity of the double mutants HED351A,L536P and HED351A,Y537A in the presence of estradiol was comparable with that of their singly mutated counterparts (Fig. 8). These results confirm that Asp-351 plays a role in the active conformation of ER␣ that is more important in the absence of hormone than in its presence. DISCUSSION The presence and precise length of the antiestrogen side chain has been correlated with their antagonist activity (39,40). In human ER␣, aspartate 351 interacts with the tertiary amine present at the end of the side chains of Tam and Ral (16,23). We have investigated the precise role of this interaction in the pharmacology of partial antiestrogens. Hydrogen bonding between Asp-351 and the side chain of partial antiestrogens could potentially contribute to the affinity of the ER␣-antiestrogen complex. However, previous characterization of mutant D351Y indicated that the relative affinity for estradiol or for the antiestrogen tamoxifen was not affected by this mutation (41). Our experiments are consistent with this report and indicate that other mutations leading to replacement of Asp-351 by Ala or Val did not noticeably affect the relative affinity of the corresponding receptors for estradiol and OHT in transiently transfected cells. It is likely that the hydrogen bonds formed by the hydroxyl group of OHT and the numerous hydrophobic interactions with the steroid-like skeleton and the side chain contribute most of the affinity of interaction between ER␣ and partial antiestrogens.
It has been suggested that amino acid Asp-351 is key to the antiestrogenic character of partial antiestrogens (28). This conclusion was based on the observation that mutant D351Y has increased transcriptional activity in the presence of tamoxifen and raloxifene in breast cells MDA-MB-231. A possible role of the hydrogen bond formed by Asp-351 may be to maintain the side chain of OHT or Ral in an antagonist position. Among the mutations at position 351 tested here, D351Y was unique in its capacity to yield a receptor that was stimulated by OHT and, to a lesser degree, by RU39,411 and Ral (data not shown), but not by full antiestrogen ICI182,780. We speculate that a tyrosine at position 351 may destabilize the inactive conformation and/or stabilize a partially active structure of the ER bound to partial antiestrogens. However, our results indicate that disruption of the hydrogen bond between Asp-351 and the side chain of partial antiestrogens is not sufficient for this effect. Indeed, replacement of Asp-351 by Gly, Ala, or Val did not increase the agonist activity of partial antiestrogens in HeLa and in MDA-MB-231 cells, in the presence or absence of overexpressed TIF2.1, on either a minimal promoter (ERE3-TATA-CAT) or a more complex promoter (ERE3-tk-CAT, data not shown). Moreover, our results are consistent with the observation that the steroid analog RU486, which has a tertiary amine-containing side chain, is an antagonist not only for the progesterone receptor, which contains a glutamic acid at the position corresponding to Asp-351 of ER␣, but also with the glucocorticoid receptor, which contains a glycine instead (19).
We observed a marked reduction in the basal activity of the HED351G, HED351A, HED351V, and HED351Y mutants, but not of HED351E. In contrast, estradiol-induced transactivation and TIF2.1 recruitment in vitro were not affected under the conditions of our assays. The DNA and ligand binding properties of these receptors were not affected either. Introduction of mutation D351A in the constitutively active receptors HEL536P and HEY537A confirmed that abolition of the negative charge at position Asp-351 inhibits folding of the unliganded receptor in an active conformation, while estradiol binding was sufficient to stabilize the active conformation despite the D351A mutation. Examination of the active structure of the ER␣ ligand binding domain reveals that Asp-351 interacts with the amide of the peptide bond between helix 12 amino acids Leu-539 and Leu-540 (16,23). Accurate positioning of helix 12 is crucial for recruitment of coactivators that interact with nuclear receptors through LXXLL motifs. We propose that positioning of helix 12 in the active conformation is unstable in the absence of hormone. In this model, disruption of the interaction between Asp-351 and helix 12 would be sufficient to fully destabilize the active configuration in the unliganded receptor. On the other hand, the active conformation of helix 12 in the presence of estradiol is brought about by a network of intraand intermolecular interactions, such that the contribution of Asp-351 may be less important in the liganded receptor. The larger stimulation of wild-type ER␣ activity by TIF2.1 overexpression in the absence of hormone than in the presence of estradiol is consistent with a role of this coactivator in stabilizing the unstable active conformation of unliganded ER␣ LBD. Hydrophobic mutations of Asp-351 fully blocked this positive effect of TIF2.1 on basal activity of ER␣, supporting our hypothesis that Asp-351 is important for the active structure of the unliganded ER LBD.
Previous studies with antiestrogen analogs have demonstrated the importance for the antagonist activity of antiestrogens of hydrogen bond acceptor groups at the position corresponding to the tertiary amine in OHT (42). In this light, our results that the hydrogen bond between Asp-351 and the antiestrogen side chain can be disrupted without affecting antagonist activity are unexpected. However, we cannot exclude possible tissue-specific effects or the possibility that interaction of the antiestrogen side chain with Asp-351 may prevent this amino acid from stabilizing an active conformation of the ER in a manner similar to our observations with the unliganded receptor. Mutagenizing Asp-351 to hydrophobic residues would prevent such a stabilizing role in addition to disrupting the hydrogen bond with the antiestrogen side chain.
In conclusion, our data demonstrate that ER␣ mutants lacking the negative charge at amino acid Asp-351 are efficiently repressed by partial antiestrogens in HeLa cells. These results are not consistent with a crucial role of amino acid Asp-351 in maintaining the side chain of OHT in an antagonist conformation.