Modulation of Estrogen Receptor α Function and Stability by Tamoxifen and a Critical Amino Acid (Asp-538) in Helix 12*

Estrogen receptor α (ER) is a ligand-activated transcription factor implicated in breast cancer growth. Selective estrogen receptor modulators (SERMs), such as tamoxifen (4-OHT), bind to the ER and affect the position of helix 12, thereby influencing coregulator binding and ER transcriptional activation. Previous studies have shown that a triple mutation in helix 12 (3m; D538A/E542A/D545A) caused a change in ER stability and obliterated 4-OHT action (Liu, H., Lee, E. S., de los Reyes, A., Zapf, J. W., and Jordan, V. C. (2001) Cancer Res. 61, 3632–3639). Two approaches were taken to determine the role of individual mutants (D538A, L540Q, E542A, and D545A) on the activity and stability of the 4-OHT·ER complex. First, mutants were evaluated using transient transfection into ER-negative T47D:C4:2 cells with an ERE3-luciferase reporter, and second, transforming growth factor α (TGFα) mRNA was used as a gene target in situ for stable transfectants of MDA-MB-231 cells. Transcriptional activity occurred in the presence of estrogen in all of the mutants, although a decreased response was observed in the L540Q, 3m, and D538A cells. The 3m and D538A mutants lacked any estrogenic responsiveness to 4-OHT, whereas the other mutations retained estrogen-like activity with 4-OHT. Unlike the other mutants, the ER was degraded in the D538A mutant with 4-OHT treatment. However, increasing the protein levels of the mutant with the proteasome inhibitor MG132 did not restore the ability of 4-OHT to induce TGFα mRNA. We suggest that Asp-538 is a critical amino acid in helix 12 that not only reduces the estrogen-like actions of 4-OHT but also facilitates the degradation of the 4-OHT·D538A complex. These data further illustrate the complex role of specific surface amino acids in the modulation of the concentration and the estrogenicity of the 4-OHT·ER complex.

tory transcription factors. Similar structural domains, designated A-F, are shared between the nuclear receptors (for review, see Refs. 1 and 2). Two transcriptional activation functions, activation function 1 (AF1) and activation function 2 (AF2), are present in the ER (see Fig. 1). AF1 is a constitutive activation function located in the A/B region, and AF2 is a ligand-dependent activation function in the E region or ligand binding domain (LBD). The activity of AF1 and AF2 is largely mediated by the cell and promoter context (3,4) and can be independent or synergistic (5).
The ER is an important therapeutic target for the treatment and prevention of breast cancer. Selective estrogen receptor modulators (SERMs) are compounds that bind to the ER and exert tissue-specific effects. Tamoxifen was the first SERM approved clinically for the treatment and prevention of breast cancer. Tamoxifen acts as an antiestrogen in the breast but has estrogenic properties in that it maintains bone density (6), lowers circulating cholesterol (7) and causes an increased risk of endometrial cancer in women over 50 (8). Raloxifene is a chemically related SERM that is used for the prevention of osteoporosis but also lowers cholesterol and reduces the risk of both breast cancer and endometrial cancer (9). ICI 182,780 is considered to be a pure antiestrogen in that it displays no agonist activity at the ER (10). This occurs because ICI 182,780 interferes with receptor dimerization (11) and increases ER protein turnover (12).
Analysis of the crystal structure of the ligand⅐ER complex has been instrumental in understanding ER conformation at the molecular level and highlights the importance of helix 12 in modulating estrogenic and antiestrogenic actions. Helix 12 is located in the LBD of the ER, but the composition and orientation of helix 12 differs depending on the ligand bound to the ER (13). When the ER LBD is complexed with the ER agonists estrogen (E 2 ) or diethylstilbestrol (DES), helix 12 is positioned over the ligand binding pocket (see Fig. 2A) (13,14). This proper positioning generates AF2 and forms a surface for the recruitment of coactivators. However, when 4-hydroxytamoxifen (4-OHT, the active metabolite of tamoxifen) or raloxifene is bound to the ER LBD, the antiestrogenic side chain displaces helix 12 from its normal position, thereby preventing the formation of a functional AF2 (Fig. 2B) (13,14). Having excluded AF2, the reported partial agonist activity of 4-OHT can only be mediated by AF1 (15). In a previous study, a binding site responsible for the estrogen-like action of 4-OHT was defined that is referred to as AF2b (16). This site contains two critical components: Asp-351 and a portion of helix 12 (Asp-538, Glu-542, and Asp-545). AF2b is proposed to be a docking site for coactivators or corepressors that modulate the estrogenicity of the 4-OHT or raloxifene ER complex (17)(18)(19). Therefore, different ligands induce different receptor conformations, and the positioning of helix 12 is the key event that permits discrimination between ER agonists and antagonists by influencing the interaction of the ER with coregulators.
The estrogenic or antiestrogenic action of ligands at the ER depends on the subtle changes in ER shape that programs the ER to form an active or inactive transcription complex or to be degraded by the proteasome. The amount of available ER in the cell is controlled by a balance between synthesis and degradation. ER stability is influenced by the nature of the bound ligand such that ligand-induced conformational changes modulate the ability of the ER to interact with proteins involved in the degradation process (20). The transcriptional activity of the resulting ER pool is also influenced by the ligands present. The ER is activated if the ligand is estrogenic, and the established estrogens can be classified as class I or class II (21). Class I estrogens, such as DES or E 2 , are planar compounds that use the AF2 site for optimal action. Class II estrogens, represented by angular triphenylethylene compounds such as 4-OHT and fixed ring 4-hydroxy triphenyl pentene, utilize AF2b for activity. However, ligands such as SERMs or pure antiestrogens can block the activity of the ER by creating a ligand⅐ER complex that is inactive. Overall, the complex decision-making network depends upon the protein recognition sequences exposed on the external surface of the relevant SERM⅐ER complex in response to ligand binding.
Analysis of the helix 12 region of AF2b using the 3m mutation (D538A/E542A/D545A) yielded important insight into mechanism of 4-OHT agonism. The transforming growth factor ␣ (TGF␣) gene is recognized as a target of estrogen action and is involved in cell growth stimulation by estrogen (22,23), so the biological activity of the 4-OHT⅐ER complex can be assessed using Northern blotting for TGF␣ mRNA. Expression of TGF␣ mRNA is normally induced by E 2 and 4-OHT treatment in MDA-MB-231 human breast cancer cells stably transfected with the wild type ER (S30 cells) (24). The 3m mutation resulted in a decreased induction of TGF␣ in response to E 2 and no response to 4-OHT (18). Therefore, the 3m mutation abolished the agonist activity of 4-OHT and decreased the agonist activity of E 2 . In addition, a slight degradation of the ER was observed when the 3m mutant stable cell line was treated with E 2 , 4-OHT, and ICI (18). This is in contrast to stable cell lines containing the wild type ER, which displayed a large downregulation of the ER in the presence of E 2 and ICI, but an increase in ER protein with 4-OHT treatment. Although the effect of E 2 on the 3m mutation and the three individual amino acids comprising the 3m mutation has been studied using EREluciferase assays (4,(25)(26)(27), the majority of the studies were not performed in breast cancer cell lines and in a comprehensive manner. In addition, the precise interaction between 4-OHT and the individual mutations is not known.
Amino acid L540 is a nearby amino acid of interest on the underside of helix 12 when it is sealing estrogen in the hydrophilic pocket of the LBD. The L540Q mutation was initially generated by random chemical mutagenesis and is a dominant negative ER mutant (28 -31). Previous studies in MDA-MB-231 breast cancer cells have shown that an ERE-CAT reporter is activated by 4-OHT and ICI 164,384, but not by E 2 , in the presence of the L540Q mutant (27). Therefore, the L540Q mutation reverses the pharmacology of E 2 and ICI 182,780 that is normally observed at the wild type ER in MDA-MB-231 cells.
We have stably transfected individual mutant ER cDNAs into MDA-MB-231 human breast cancer cells to create an in vitro model to address the contribution of specific amino acids in helix 12 (D538A, E542A, D545A, and L540Q) to the agonist activity of 4-OHT at the AF2b site. We have found that Asp-538 is the critical amino acid in helix 12 that not only reduces the estrogen-like actions of 4-OHT but also enhances the degradation of the ER upon 4-OHT treatment.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagents-Stable cell lines were maintained in phenol red-free minimum essential media supplemented with 5% calf serum treated 3ϫ with dextran-coated charcoal, 0.5 mg/ml G418 (Geneticin, Invitrogen, Carlsbad, CA), 2 mM L-glutamine, 0.1 mM nonessential amino acids, 100 units/ml penicillin, 100 g/ml streptomycin, and 6 g/ml insulin. This media is referred to as stripped media, indicating that it is free of E 2 .
S30 cells are MDA-MB-231 human breast cancer cells stably transfected with wild type ER␣ (24) and are referred to as wild type cells. These cells are grown in stripped media. T47D:C4:2 cells are ER␣negative human breast cancer cells (32) that were propagated in phenol red-free RPMI media containing 10% fetal serum calf serum treated 3ϫ with dextran-coated charcoal, as well as the concentrations of amino acids, penicillin, streptomycin, and insulin described above. ER Ϫ represents a G418-resistant clone that is ER-negative. The 3m stable cell line (18) containing the triple mutation (D538A/E542A/D545A) was also grown in stripped media.
4-OHT and E 2 were purchased from Sigma (St. Louis, MO). ICI 182,780 was obtained from AstraZeneca (Macclesfield, England). Raloxifene was a generous gift from Eli Lilly and Co. (Indianapolis, IN). All drugs were dissolved in ethanol and stored at Ϫ20°C. MG132 was dissolved in Me 2 SO and obtained from Calbiochem (San Diego, CA).
Mutagenesis-Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit according to the manufacturer's instructions (Stratagene, La Jolla, CA). The ER␣PSG5 plasmid (HEGO, kindly provided by P. Chambon) was used as a template for PCR. Primers used were as follows: Generation of Stable Transfectants-MDA-MB-231 (clone 10A) cells (24) were grown in stripped media for 3-4 days prior to transfection. The cells were transfected with 10 g of the ER mutant in the pIRES-neo2 plasmid. 5 ϫ 10 6 cells were electroporated in a 0.4-cm cuvette (Bio-Rad, Hercules, CA) at a voltage of 0.250 kV and a high capacitance of 0.95 microfarad in phenol red-free minimal essential media with no additives. The cells were transferred to a 10-cm plate and incubated overnight in 10 ml of stripped media without G418, and the media was changed the next day. The following day, media containing 0.5 mg/ml G418 was added, and the cells were subsequently maintained in this media. Individual colonies appeared ϳ1 month after transfection, and these were isolated and screened for stable expression of the ER by Western blotting.
Transient Transfections and Luciferase Assays-The cells were transfected with 1 g of the ERE3-luciferase plasmid (33) and 1 g of the mutant or wild type ER␣PSG5 plasmid. To normalize for transfection efficiency, 0.2 g of the PCMV␤ plasmid (Clontech, Palo Alto, CA) were also transfected. 5 ϫ 10 6 cells were electroporated in a 0.4-cm cuvette (Bio-Rad, Hercules, CA) at a voltage of 0.320 kV and a high capacitance of 0.95 microfarad in serum-free media. The cells were transferred to 12-well plates and incubated overnight. The next day, the cells were treated with the appropriate compound for 24 h.
The cells were washed once with cold PBS, and 100 l of extraction buffer (0.1 M potassium phosphate (pH 7.5), 1% Triton X-100, 100 g/ml BSA, 2.5 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol) was added to each well. The cells were incubated on ice for 2 min, dislodged from the plates, and transferred to an Eppendorf tube. The lysate was centrifuged for 2 min at top speed in a microcentrifuge, and the supernatant was used for the assay. 50 l of the lysate was mixed with 350 l of reaction buffer (160 mM MgCl 2 , 75 mM glycylglycine (pH 7.8), 0.5 mg/ml BSA, 19 mg/ml ATP, and 15 mM Tris-HCl (pH 7.5)) and 100 l of luciferin (0.4 mg/ml). Luminescence was measured in a Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego, CA) for 10 s. ␤-Galactosidase activity was measured using 10 l of each sample and the Galacto-Light Plus detection system (Applied Biosystems, Bedford, MA). Data are reported as relative light units, which is the luciferase reading divided by the ␤-galactosidase reading.
Northern Blots-Stable transfectants were treated with compounds for 24 h. Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. 20 g of RNA was loaded per lane in a 1% agarose/0.66 M formaldehyde gel. The RNA was transferred to a MagnaGraph nylon transfer membrane (Osmonics, Minnetonka, MN) overnight in 10ϫ SSPE buffer (20ϫ SSPE is 3.6 M NaCl/0.2 M NaH 2 PO 4 /0.02 M EDTA (pH 7.4)). The RNA was fixed to the membrane by UV-cross-linking. The membrane was prehybridized in hybridization solution (0.5 M sodium phosphate, 10 mM EDTA, 1% BSA, 7% SDS (pH 7.2)) for a minimum of 2 h at 60°C. The TGF␣ probe (a gift from Dr. R. Derynck, Genentech, South San Francisco, CA) or the ER␣ probe (the EcoRI fragment from the ER␣PSG5 plasmid) was labeled with [ 32 P]dCTP using the Megaprime DNA labeling system (Amersham Biosciences, Piscataway, NJ), and the labeled probe was separated from free 32 P using Microspin columns (Amersham Biosciences) according to the manufacturer's instructions. The probe was heated at 95°C for 5 min, added to the hybridization buffer, and incubated at 60°C overnight. The next day, the membrane was washed for 30 min at 60°C with 1ϫ SSPE/0.1% SDS, 30 min with 0.5ϫ SSPE/ 0.05% SDS, and 2 ϫ 15 min with 0.1ϫ SSPE/0.1% SDS. To visualize TGF␣, the membrane was exposed to film overnight. Equal loading of samples was verified by stripping the membrane and reprobing with ␤-actin.
Protein Isolation and Western Blots-Cells were treated for 24 h with compound. To harvest protein, cells were washed once with PBS, scraped using a cell scraper into 10 ml of PBS, and transferred to a 15-ml conical tube. The cells were pelleted, and the supernatant was aspirated. The cell pellet was resuspended in 100 l of extraction buffer (50 mM HEPES, 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 10% glycerol, 0.5% Nonidet P-40, 10 mM ␤-glycerophosphate (pH 8), containing a 1:100 dilution of a freshly added protease inhibitor mixture (Sigma P8340, St. Louis, MO)), passed through a 22-gauge needle, and incubated on ice for 30 min. The cell lysate was centrifuged at 10,000 ϫ g for 10 min at 4°C, and the supernatant was transferred to a new tube. Samples were quantitated using the Bio-Rad protein assay kit. 20 g of cell lysate were separated on a 7.5% SDS-PAGE gel and transferred to nitrocellulose. The blot was blocked in blotto (2.5% dry milk/0.05% Tween/0.5ϫ PBS) for 1 h to overnight. Blots were probed with polyclonal ER␣ antibodies at 1:200 (G20, Santa Cruz Biotechnology, Santa Cruz, CA) and monoclonal ␤-actin antibodies at 1:20,000 (Sigma A5441, St. Louis, MO) for 1 h at room temperature. The mem-brane was then washed 3 ϫ 5 min with wash buffer (0.5ϫ PBS/0.05% Tween). The blot was incubated in a 1:3000 dilution of horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology) for 1 h. The membrane was washed 3 ϫ 5 min with wash buffer, and bands were visualized using chemiluminescence (ECL, Amersham Biosciences, Piscataway, NJ).
Quantitation and Statistics-Western and Northern blots were quantitated using the gel plot feature in Scion Image version 4.0.2. The results were statistically analyzed using SPSS 9.0.

RESULTS
Previous results in our laboratory showed that a triple mutation in helix 12 of the ER (3m, D538A/E542A/D545A) caused a change in ER stability and eliminated 4-OHT agonist activity (18). To analyze the role of individual amino acids in helix 12, multiple mutations were generated. These mutations include single point mutations of the 3m mutation (D538A/E542A/ D545A) and the L540Q mutation ( Figs. 1 and 2).
The transcriptional activity of the ER mutants was first tested by transient transfection of ER-negative T47D:C4:2 cells with the mutant ER cDNA and an ERE3-luciferase reporter  (14) and the 4-OHT⅐ER (B) (3ERT) (13) complex are depicted. Helix 12 (amino acids 536 -547 for E 2 and 536 -551 for 4-OHT) is shaded in yellow, the ligand is blue, Asp-351 is green, and Asp-538, Leu-540, Glu-542, and Asp-545 are red. Leu-540 is not visible in the 4-OHT⅐ER structure, because it is in the interior of the complex. In the diethylstilbestrol⅐ER structure, helix 12 comprises residues 538 -546, whereas in the 4-OHT structure, it comprises residues 536 -544 (13). For clarity, modified amino acids in helix 12 are shaded. The published crystal structures were obtained from the RCSB Protein Data Bank and are colored using Insight II. (33). Transfection with the empty vector PSG5 showed no induction of luciferase activity with any of the treatments used (Fig. 3). In addition, none of the mutants exhibited any response to treatment with the vehicle control ethanol (EtOH). E 2 treatment resulted in a statistically significant induction of luciferase activity with the wild type ER and all of the mutants when compared with the EtOH control. The greatest induction was observed with the E542A mutant. The wild type, D538A, D545A, and 3m mutants displayed an intermediate level, and the L540Q mutant displayed the smallest induction. The L540Q mutant was the only mutant that showed a slight induction of luciferase activity during ICI 182,780 treatment, but this was not statistically significant. In addition, the wild type, E542A, and D545A ERs displayed an induction of luciferase activity upon 4-OHT treatment, whereas the D538A, L540Q, and 3m mutants did not. When wild type, 3m, and D538A cells were treated with E 2 plus 4-OHT, the response of the cells was the same as that observed with 4-OHT alone, indicating that 4-OHT acts as a complete antiestrogen in these cells (data not shown). Therefore, of the three mutations present in the 3m mutant, the D538A mutation is responsible for the elimination of the agonist activity of 4-OHT at an ERE in T47D:C4:2 cells.
To further evaluate these mutants in a reproducible manner, stable transfectants were generated in ER-negative MDA-MB-231 cells. At least five clones were obtained representing each mutation, and the clones were screened for the presence of the ER using Western blotting. Two clones harboring each mutation were initially screened using Northern blotting for TGF␣ mRNA levels. Both of the clones studied showed similar TGF␣ levels in response to various treatments, so a single representative clone was chosen for further analysis. Each of the stable clones was screened to ensure that the proper mutation was present using reverse transcription-PCR and sequencing.
ER protein levels were compared between each of the stable cell lines (Fig. 4). All of the cell lines contained similar levels of ER protein, so the characteristics observed in each cell line were not a result of varying ER levels. A clone that was stably transfected but ER-negative by Western blot analysis was used as a control and designated ER Ϫ .
The transcriptional activity of the ER mutants was also analyzed using Northern blot analysis of TGF␣ mRNA. The advantage of this assay is that the TGF␣ gene is an endogenous gene in MDA-MB-231 cells, and induction of TGF␣ mRNA levels reflects a process that is inherent to these cells. Cells from the ER Ϫ clone were treated with EtOH, E 2 , 4-OHT, and E 2 plus 4-OHT, and no induction of TGF␣ mRNA was observed (data not shown). Wild type cells showed an induction of TGF␣ mRNA in response to E 2 and 4-OHT, but 4-OHT did not act as an antiestrogen in these cells, because it was not able to significantly block the E 2 response (Fig. 5). A similar pattern of mRNA expression was observed in the E542A and D545A mutants. Although the 3m and D538A mutants showed an increase in TGF␣ mRNA in response to E 2 treatment, the level of induction was less than that observed for the other mutants. In addition, no induction occurred with the 4-OHT treatment. This is in agreement with ERE-luciferase assay results and suggests that Asp-538 is the single amino acid within the 3m mutation required for the agonist activity of 4-OHT at the ER.

FIG. 3. ER mutants display differential activation of an ERE-luciferase
reporter. ER-negative T47D:C4:2 cells were transiently transfected with 1 g of an ERE3-luciferase plasmid and 1 g of the ER plasmid. The cells were also transfected with 0.2 g of the PCMV␤ plasmid to normalize for transfection efficiency. The next day, the cells were treated with EtOH, E 2 (10 Ϫ9 M), ICI 182,780 (10 Ϫ6 M), or 4-OHT (10 Ϫ6 M) for 24 h before performing the luciferase and ␤-galactosidase assays. The data is presented as relative light units, which is the luciferase reading divided by the ␤-galactosidase reading, and the resulting value was multiplied by 1000. The graph represents values from three to four independent experiments, and the standard deviation is shown (*, p Ͻ 0.05).

FIG. 4. Protein expression levels are similar in wild type and mutant
ER stable transfectants. 20 g of protein lysate from each stable clone was run on a 7.5% SDS-PAGE gel, and a Western blot was performed using ER␣ and ␤-actin antibodies. ER Ϫ is a clone that was stably transfected but is ER-negative and G418-resistant.
The L540Q mutant produced an induction of TGF␣ mRNA with ICI 182,780 and 4-OHT treatment but not with E 2 treatment. In addition, raloxifene (Ral) treatment had no effect on TGF␣ mRNA levels with any of the transfected stable cell lines (Fig. 5).
Because the expression of E 2 -induced TGF␣ mRNA was lower in the D538A and 3m stable transfectants compared with all other transfectants, E 2 concentration response curves were completed to ensure that an E 2 response was present (Fig. 6). Wild type cells displayed an induction of TGF␣ mRNA at the lowest concentration of E 2 (10 Ϫ10 M), and the amount of TGF␣ mRNA continued to increase up to the highest concentration of E 2 (10 Ϫ6 M). An EC 50 value of 7 ϫ 10 Ϫ10 M was calculated for the wild type cells. A similar profile was observed in the D545A mutant, in that a 10 Ϫ10 M concentration of E 2 induced TGF␣ mRNA, and the increase continued up to 10 Ϫ6 M E 2 . This is consistent with an EC 50 of 2 ϫ 10 Ϫ10 M. The 3m and D538A mutants behaved similarly in that they first displayed an in-crease in TGF␣ mRNA at a 10 Ϫ9 M concentration of E 2 , and the induction continued with increasing concentrations of E 2 . Therefore, the 3m and D538A mutants required higher concentrations of E 2 before TGF␣ mRNA was transcribed, when compared with the wild type and D545A cells. The EC 50 of the 3m and D538A cells were 2 ϫ 10 Ϫ9 M and 4 ϫ 10 Ϫ9 M, respectively, which are ϳ10 times higher than the wild type and D545A cells. In addition, TGF␣ mRNA levels in the D538A mutant were lower at each concentration of E 2 tested in comparison to the other mutants.
In general, a SERM can negatively affect ER activity using two different mechanisms. First, the SERM could cause the degradation of the ER by creating a SERM⅐ER complex that is targeted for destruction. Second, a SERM⅐ER complex may not be degraded but be present and have no intrinsic activity. To distinguish between these possibilities, wild type and mutant ER protein levels were analyzed in the presence of ICI 182,780, E 2 , and 4-OHT (Fig. 7A). ICI 182,780 degraded the wild type ER at all concentrations tested, and the D545A and D538A mutants were also degraded by ICI 182,780 (Fig. 7B). In contrast, the L540Q mutant exhibited no changes in ER protein levels at any ICI 182,780 concentration. The 3m and E542A mutants displayed intermediate ER levels after ICI 182,780 treatment.
When wild type cells were treated with E 2 , less ER protein was observed by Western blot (Fig. 7C). E 2 treatment also resulted in a down-regulation of ER levels in the D538A and D545A mutants, whereas ER levels were stable in the 3m, L540Q, and E542A mutants. In the presence of 4-OHT, ER levels increased in the wild type cells and remained stable in the L540Q, E542A, and D545A mutants. Importantly, 4-OHT treatment reduced ER protein levels in the D538A and 3m stables. The stable cell lines that showed an induction of ER transcriptional activity upon 4-OHT treatment (wild type, E542A, and D545A) all contained ER protein levels that were maintained or increased when treated with 4-OHT. In contrast, FIG. 5. Northern blot analysis of TGF␣ in wild type and mutant ER stable cell lines. Stable cell lines were treated with EtOH, E 2 (10 Ϫ9 M), ICI 182,780 (10 Ϫ6 M), raloxifene (Ral, 10 Ϫ6 M), 4-OHT (10 Ϫ7 M), or combinations as indicated. RNA was harvested 24 h after treatment, and 20 g of each sample was run on an agarose gel. The membrane was probed with TGF␣ and ␤-actin. The Northern blots were repeated at least three times, and a representative blot is shown. The quantitation is a combination of at least three independent experiments, with the standard error shown.

FIG. 6. Concentration-dependent induction of TGF␣ mRNA expression by E 2 in wild type and mutant ER stable cell lines.
Stable cell lines were treated with EtOH or varying concentrations of E 2 (10 Ϫ10 -10 Ϫ6 M) for 24 h and processed as described in Fig. 5. The top blot represents TGF␣, and the bottom blot represents ␤-actin. The EC 50 for each dose-response curve was calculated by first subtracting the EtOH value and then setting the 10 Ϫ6 M concentration to 100. The EC 50 is the concentration at which the E 2 -induced TGF␣ mRNA increase is half-maximal. the D538A and 3m mutants exhibited no transcriptional activity in response to 4-OHT, and they were the only mutants that showed a decrease in ER levels with 4-OHT treatment. This observation prompted us to explore the possibility that the decreased transcriptional activity of the D538A mutant was a result of decreased ER protein levels in response to 4-OHT.
In the D538A mutant, the degradation of ER protein by 4-OHT could be a result of 4-OHT-induced transcriptional down-regulation of the ER message, or 4-OHT-induced posttranslational degradation. To distinguish between these possibilities, Northern blot analyses for ER␣ were performed in wild type and D538A cells that were treated with EtOH or 4-OHT (Fig. 7D). A comparison of ER mRNA and protein levels showed that 4-OHT up-regulated both ER mRNA and protein in the wild type cells. In contrast, 4-OHT treatment did not change ER mRNA levels and decreased ER protein levels in the D538A cells by 70%. This indicated that the 4-OHT-mediated decrease in ER protein levels in the D538A mutant was not a result of decreased ER mRNA stability.
ER protein levels were measured in all of the stably transfected cell lines to establish whether the proteasome is involved in the degradation of the ER. The cells were treated with the proteasome inhibitor MG132 before the addition of ligand, and a Western blot was performed to analyze ER levels (Fig. 8). Preincubation with MG132 elevated the amount of ER protein present, indicating that the ligand-induced degradation of the ER could be mediated by the proteasome. However, subtle differences were observed, depending on the mutation and the ligand evaluated, especially in the case of ICI 182,780. MG132 was able to prevent the ICI 182,780-mediated degradation in D538A and E542A cells, but smaller increases were detected in wild type and D545A cells. It is possible that MG132 was not able to restore ER levels in the ICI 182,780-treated cells to control levels, because a relatively high concentration of ICI 182,780 (10 Ϫ6 M) was used. Experiments using lower concentrations of ICI 182,780 (10 Ϫ6 -10 Ϫ8 M) in combination with MG132 were performed in the wild type cells (data not shown), but MG132 treatment did not restore ER levels in the ICI 182,780-treated cells to control levels. In addition, MCF-7 cells were treated under the same conditions as described in Fig. 8, and the results were essentially the same (data not shown) in that MG132 treatment could not fully abrogate ICI 182,780 treatment. Therefore, MG132 was unable to restore ER protein to control levels in wild type, D545A, and MCF-7 cells treated with ICI 182,780. This is in contrast to MG132 treatment in E 2 -treated cells, where the ER levels are greater than control cells. L540Q cells were not included in this experiment, because the ER levels in these cells remained the same after treatment with E 2 , ICI 182,780, and 4-OHT (Fig. 7). These data indicate that the degradation of the ER in the stably transfected cell lines occurs through the proteasome.
When D538A cells were treated with MG132 and 4-OHT, ER protein levels were increased 2.3-fold, compared with the decreased levels of the receptor normally observed with 4-OHT treatment alone (Fig. 8). The D538A stable cells were treated with a range of MG132 concentrations and 4-OHT or E 2 to FIG. 7. Western blot analysis of ER levels in wild type and mutant stable cell lines. A, stable cell lines were treated with EtOH, E 2 (10 Ϫ9 M), 4-OHT (10 Ϫ6 M), or ICI 182,780 (10 Ϫ8 -10 Ϫ6 M) for 24 h. 20 g of each sample was loaded onto a SDS-PAGE gel. The membrane was probed with ER␣ antibodies, and ␤-actin antibodies were used to ensure even loading. The Western blots were repeated at least three times, and a representative blot is shown. B and C, the quantitation is a combination of three independent experiments, with the standard error shown (*, p Ͻ 0.05). D, wild type and D538A stable cell lines were treated with EtOH or 4-OHT (10 Ϫ7 M) for 24 h, and RNA was processed for a Northern blot as described in Fig. 5. ER mRNA and protein levels (from C) were compared. The quantitation is a combination of three independent experiments, and the standard error is shown. determine whether preventing the degradation of the 4-OHT⅐D538A complex could restore 4-OHT agonist activity (Fig. 9). TGF␣ mRNA was induced with E 2 treatment, but no induction of TGF␣ mRNA levels was observed with MG132 treatment alone or MG132 in combination with 4-OHT. Therefore, increasing the amount of D538A ER that would normally be available in the presence of 4-OHT did not restore the agonist activity of 4-OHT. Therefore, adjusting the levels of the 4-OHT⅐D538A complex did not have an effect, because the complex has no intrinsic activity.

DISCUSSION
The finding that Asp-538 is an essential amino acid that modulates estrogen action with 4-OHT supports and extends the idea that helix 12 plays a vital role in the mechanics of estrogen action (13,14). Asp-538 appears to be a central control mechanism for both the intrinsic activity of the 4-OHT⅐ER complex and the processing and degradation of the complex by the proteasome. These observations introduce a new dimension for consideration with SERMs as modulators of estrogen responsive genes.
Proteolysis is involved in the regulation of a variety of cellular functions such as cell cycle progression, oncogenesis, transcription, development, tissue growth, elimination of abnormal proteins, and antigen processing (34). Degradation of proteins can occur through three major pathways, which include mechanisms mediated through lysosomes, calpains (calciumdependent cysteine proteases), and the proteasome. It has been shown previously that the ER is a ubiquitinated protein and that ubiquitination targets the protein to the proteasome, which causes ER degradation (35)(36)(37). Our data indicate that the proteasome is responsible for the degradation of ER protein observed in all of the stably transfected cell lines (Fig. 8).
The stability of the ER complex has been shown to be influenced by the bound ligand. E 2 and ICI 182,780 decrease ER levels, and 4-OHT increases the accumulation of the ER in MCF-7 human breast cancer cells (38) and pituitary lactotrope PR1 cells (20). A similar situation occurs when MDA-MB-231 cells are stably transfected with the wild type ER (Fig. 7). However, the typical pattern of ER stability is changed by mutation of residues in helix 12. For example, treatment of cells with E 2 did not down-regulate the ER in the 3m, L540Q, and E542A mutants, whereas down-regulation is observed in the D538A and D545A mutants. ICI 182,780 degraded the ER in all of the stable cell lines except for the L540Q cells, and ICI 182,780 had a reduced affect in the 3m and E542A cells. 4-OHT either increased or did not affect ER levels in L540Q, E542A, and D545A cells but had a dramatic effect on the degradation of the ER in 3m and D538A cells. This suggests that the presence of an aspartic acid at 538 prevents degradation of the ER when liganded by 4-OHT.
Alterations in other amino acids in helix 12 are reported to result in changes in the stability of the ER. The protein levels of the L539A/L540A double mutant remained constant upon E 2 treatment, whereas 4-OHT induced degradation of the ER (38). The mouse ER mutants L543A/L544A (L539A/L540A in human) and M547A/L548A (M543A/L544A in human) (26) as well as the human ER mutants L540Q, E542A/D545A, and L540Q/ E542A/D545A (27) are not degraded by ICI 182,780. This indicates that helix 12 is important for maintaining the proper regulation of the ER protein in response to ligands, especially ICI 182,780.
The signal that ultimately targets the ER for ubiquitination and subsequent degradation has not been definitively established. Several possibilities have been proposed that modify the shape of a protein so that is it recognized by the E3 ubiquitin protein ligase, such as ER phosphorylation, binding of ancillary proteins to the ER, or binding of ligand. Modulation of the ubiquitination machinery or masking of a degradation signal are also possibilities (for review, see Refs. 39 -42). It is likely that a combination of these mechanisms could contribute to the ER degradation observed in the stable cell lines.
In addition to degradation, another consequence of changing the shape or charge distribution of the external surface of the ligand⅐ER complex is the modulation of ER transcriptional activity. Transcriptional activity was measured initially using transient transfection of the ER mutant and an ERE3-luciferase reporter into T47D:C4:2 cells (Fig. 3), but a further study of TGF␣ mRNA in the stably transfected MDA-MB-231 cells extended our observations (Figs. 5 and 6). The results of both of these assay approaches were consistent. In wild type cells, E 2 and 4-OHT induced transcriptional activity but ICI 182,780 did not. The 3m stable cells exhibited no response to ICI 182,780 and 4-OHT and showed a decreased response to E 2 compared with the wild type cells. The E542A and D545A cells showed essentially the same transcriptional activity as wild type cells. Little transcription occurred with E 2 treatment in the L540Q mutant, but activation was observed upon ICI 182,780 and 4-OHT treatment. Remarkably, ICI 182,780 can produce a biological effect when the receptor is stable. The crystal structure of a pure antiestrogen and ER␤, but not ER␣, is available. Although the ER␤:ICI 164,384 (a pure antiestrogen related to ICI 182,780) crystal structure has been solved, helix 12 is invisible in the experimental electron density maps (43), so structure/function speculations are not yet possible.
The D538A mutant exhibits decreased transcription in the presence of E 2 and no transcription in the presence of ICI 182,780 and 4-OHT. This indicates that the D538A mutation is the single mutation responsible for the decreased activity of the 3m triple mutation. These data are important, because it is now possible to redefine the components of AF2b as Asp-351 and Asp-538, which must interact with AF1, because only these regions are required for the estrogen-like activity of 4-OHT. By analogy with our approach to defining the precise amino acids on helix 12, we are currently addressing the question of the one or more critical amino acids in AF-1 that may be required for SERM activity.
The external surface of the ER affects protein stability and transcriptional activity, but the amount of the ligand⅐ER complex does not necessarily correlate with activity. Because 4-OHT treatment results in rapid degradation of the D538A protein with an associated lack of transcriptional activity, we increased the level of the 4-OHT⅐D538A ER complex using an inhibitor of the proteasome. However, activity of the D538A mutant was not restored in the TGF␣ assay (Fig. 9), indicating that the 4-OHT⅐D538A complex has no intrinsic activity. An interesting point is that the amount of TGF␣ mRNA was reduced when the proteasome inhibitor was combined with E 2 in the D538A cells. This observation is in agreement with a study by Lonard et al. (44), which suggests that proteasomal degradation is required for E 2 -mediated ER transcription and that coactivator binding is required for ligand-mediated degradation of the ER.
Our studies and the known crystal structures of the ligand⅐ER complex emphasize the idea that the amino acids present in helix 12 are located in unique positions to influence the interaction of coregulators with the ER. In the E 2 ⅐ER complex, helix 12 is positioned over the ligand binding pocket and the charged residues Asp-538, Glu-542, and Asp-545 are on the outside of the complex ( Fig. 2A) (14). Leu-540 is positioned more toward the inside of the ligand binding pocket. Glu-542 is uniquely positioned in the DES⅐ER crystal structure as an N-terminal capping amino acid that stabilizes the conformation of the coactivator GRIP1 peptide when this peptide is bound to the ER (13,45). Contacts are also made to the GRIP1 peptide in the DES⅐ER structure by Asp-538 (13). In the 4-OHT⅐ER structure, the side chain of 4-OHT repositions helix 12 so that it binds to and blocks the GRIP1 coactivator binding surface (Fig. 2B) (13). In fact, the side chains of Leu-540, Met-543, and Leu-544 on the inner hydrophobic surface of helix 12 mimic the interactions made by the coactivator's nuclear receptor binding motif LXXLL (45). None of the helix 12 mutations utilized in our study are residues that contact the ligand or Asp-351, which also modulates the estrogenicity of 4-OHT (16,18,46).
An active transcription complex contains coactivators that enhance the transcriptional activity of the ER. An inactive complex contains corepressor or is in a conformation that is unable to bind coactivators (47). Much of the transcriptional activity of the 4-OHT⅐D538A complex could be explained by the differential binding of coregulators. For example, the D538A mutation results in decreased transcriptional activity in the presence of E 2 . Because Asp-538 contacts a coactivator, the decreased agonist activity of E 2 in the D538A mutant (Figs. 3, 5, and 6) could be a result of a decreased ability to recruit a coactivator. The agonist activity of 4-OHT in the wild type cells is mediated by constitutive AF1 activity, because the AF2 coactivator binding site has been disrupted by the side chain of 4-OHT. The D538A mutant eliminates the agonist activity of 4-OHT, suggesting that the D538A mutation in AF2 has allosterically affected AF1. The charge alterations produced by this mutation could favor the recruitment of a corepressor, because it has been demonstrated that 4-OHT can induce the formation of a ER⅐corepressor complex on the promoter (48,49).
The dominant negative activity of the L540Q mutant occurs as a result of competition for ERE binding, formation of inactive heterodimers with the wild type receptor, and transcriptional silencing (31). In addition, the L540Q mutant recruits a coregulator protein called repressor of estrogen receptor activity (REA) (50). These mechanisms contribute to the activity of the L540Q mutant in vitro, but activity is also observed in vivo. The L540Q mutant was introduced into T47D human breast cancer cells using adenoviral infection, and when these cells were injected into athymic mice, tumor formation was inhibited (51). Injection of adenoviruses encoding the L540Q mutation into pre-existing T47D tumors resulted in tumor regression.
In summary, ER action is a complex and tightly regulated system involving interactions between the ligand, the receptor, and effectors that are all coordinated to modulate the appropriate action (52). Helix 12, more specifically Asp-538, is central to these diverse interactions. The model we propose illustrates an unusually dramatic regulation of ER degradation and efficacy of the SERM⅐ER complex. The SERM 4-OHT normally causes an accumulation of the ER complex that is promiscuous and can induce estrogen-like action at the TGF␣ target gene. 4-OHT degrades the ER complex if a specific amino acid (Asp-538) is mutated, but if the 4-OHT⅐D538A complex is prevented from being destroyed by the proteasome, the complex is not estrogen-like. We believe that the modulation of the estrogenic and antiestrogenic properties of the SERM⅐ER complex occur through the multiple dimensions of ER destruction and the interaction of the ER with other coregulatory proteins. The finding that Asp-538 in helix 12 can control both ER stability and the intrinsic activity of the 4-OHT⅐ER complex may not only provide new opportunities in drug design but also provide a new insight into the regulation of ER protein concentrations within the cell.