Proteasome-mediated proteolysis modulates the cellular concentration of estrogen receptor-α (ERα) and is induced by treatment of cells with 17β-estradiol. Herein, we show that multiple receptor agonists, including 17α-estradiol and estriol as well as the antagonist ICI-182780, stimulate proteasome-dependent proteolysis of ERα in a process that requires ligand binding to the receptor. Proteolysis of receptor depends on ligand concentration, and there exists a direct correlation between ligand-binding affinity and the half-maximal dose of ligand required to stimulate receptor degradation. Furthermore, introduction of a point mutation into the receptor ligand-binding pocket yields a stable receptor resistant to proteolysis. Interestingly, although all ligands stimulate receptor degradation, the extent to which overall ER levels are affected varies with each ligand and is not related to ligand-binding affinity or activation of transcription. These results demonstrate ligand-specific regulation of ERα proteolysis, and they introduce the concept that cellular receptor concentration is governed not only at the level of induction of proteolysis but also by the efficiency with which the receptor is degraded.
- nuclear receptor
cellular estrogen receptor levels are dynamic and are particularly sensitive to changes in circulating levels of 17β-estradiol. It has been demonstrated through a number of studies that the decline in estrogen receptor-α (ERα) upon exposure to 17β-estradiol results from a combination of mechanisms that control both receptor synthesis and degradation through transcriptional, posttranscriptional, and posttranslational mechanisms (20, 27, 32, 33, 35, 36). The most rapid of these regulatory mechanisms is the direct loss of ERα protein brought about by the induction of proteasome-mediated proteolysis (1, 21, 28).
Regulated proteolysis by proteasomes accounts for the turnover of most short-lived proteins, including many nuclear receptors (9, 16,23, 29, 41, 43, 45). Through a series of three enzymatic reactions, ubiquitin moieties are attached to a protein substrate, which targets it to the 26S proteasome. The molecular events that direct ERα into this pathway have not been clearly established. However, earlier studies that examine changes in receptor-binding capacity have shown that receptor levels can be controlled by both receptor agonists and antagonists, suggesting the possibility that receptor occupation by ligand may provide specificity (3, 4, 15,18, 22, 33).
In our original report of ERα protein regulation by proteolysis, we utilized a pituitary lactotrope model system, the PR1 cell line. The lactotrope cell population of the anterior pituitary is a major target of estrogen action. Animals that lack ERα show a decrease in lactotrope cell density and prolactin expression (37). In contrast, animals treated with estrogen for prolonged periods develop hyperplastic pituitaries comprised primarily of lactotrope cells. The F344 rat model is a classic system for the study of estrogen regulation of lactotrope growth and function (11, 44). The PR1 cell line is an in vitro correlate of this model that was derived from an estrogen-induced pituitary hyperplasia in F344 rats (30). Using this lactotrope cell line, we showed that estrogen shortened the half-life of ERα protein in the pituitary from >3 h to 1 h, which translated into a decrease in steady-state levels of ERα protein within 2 h of ligand exposure (1). During this time frame, there was no measurable change in ERα mRNA levels. We have capitalized on the identification of a time window in which proteolysis alone is responsible for changes in ER protein to further examine the regulation of proteasome-dependent degradation of endogenous ERα.
Herein, we demonstrate that proteasome-dependent proteolysis of ERα can be induced by multiple estrogens, including short-acting estrogens and the pure antiestrogen ICI-182780 (ICI). This response requires the direct binding of the various ligands to receptor, since introduction of a point mutation in a critical residue of the ERα ligand-binding pocket yields a receptor resistant to degradation. In addition, we find that binding affinity of ligand for receptor correlates with the potency with which ligand can induce proteolysis, further supporting a critical role for ligand binding in regulating receptor proteolysis. Interestingly, through quantitative assessment of changes in receptor level at varying concentrations of ligand, we observed that the overall decline in receptor levels achieved by proteolysis varies depending on the ligand that occupies the receptor. These results indicate that ligands can differentially regulate receptor proteolysis, and they suggest the possibility that cellular ER concentration is controlled by not only the induction of proteolysis but also its efficiency.
PR1 lactotrope and HEK 293 cells were maintained under standard conditions of humidity and temperature in high-glucose DMEM medium (Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum (HyClone Laboratories, Logan, UT), 100 U/ml penicillin, and 100 μg/ml streptomycin (Life Technologies, Gaithersburg, MD).
For experiments involving steroid treatment, cells were grown in either Optimem medium (Life Technologies) or phenol red-free DMEM with 10% charcoal-stripped serum (34) and antibiotics. Cells were maintained in medium lacking steroid for a minimum of 3 days before treatment. Steroid stimulation proceeded for 2 h as previously described (1) unless otherwise indicated. The final concentration of ethanol (EtOH) in control and treated samples was 0.1%. 17β-Estradiol and 4-hydroxytamoxifen (4-OHTam) were purchased from Sigma Chemical (St. Louis, MO). 17α-Estradiol and estriol were purchased from Steraloids (Newport, RI). The antiestrogen ICI was a gift from Dr. Jack Gorski. In experiments utilizing the proteasome inhibitor ALLnL (Calbiochem, La Jolla, CA), samples were pretreated for 30 min with inhibitor before the addition of steroid.
Western blot analysis.
Western blot analysis was performed on whole cell extracts that were obtained by either direct dissolution of cell pellets in 2× sample buffer (125 mM Tris-base, 20% glycerol, 4% SDS, 10% β-mercaptoethanol, and bromphenol blue, pH 6.8) or by extraction in Totex buffer (19) followed by Bradford assay (5). Proteins were separated on a 7.5% acrylamide gel. Immunoblotting was performed as previously described (1), utilizing anti-rat ER (13) or anti-human ER (Santa Cruz Biotechnology, Santa Cruz, CA) antibodies as appropriate. Saturating concentrations of antibody were used in all cases to allow quantitative analysis of relative protein levels. Equivalent loading of lanes was verified by reprobing blots with antibody against α-tubulin (Calbiochem) or β-actin (Santa Cruz Biotechnology). Secondary antibodies were conjugated either to horseradish peroxidase or125I for visualization and quantification, respectively. Receptor levels were determined by phosphoimager analysis by use of Imagequant software (Molecular Dynamics, Sunnyvale, CA) or laser densitometry. Where densitometry was employed, an internal standard curve was generated for each blot. The EtOH-treated control was arbitrarily set at 100, and a series of dilutions of this control were run in parallel to generate the standard curve for the determination of receptor levels relative to the EtOH-treated control. Each experiment was repeated a minimum of three times to ensure reproducibility. The EC50 for the downregulation response for each ligand was calculated on the basis of nonlinear regression analysis (Prism 3.0; GraphPad Software, San Diego, CA). Statistical significance of differences between treatment groups was determined by one-way ANOVA followed by t-test analysis, with a 95% confidence interval (Microcal Origin, Microcal Software, Northampton, MA).
Transfections were performed utilizing Fugene reagent (Roche Molecular Biochemicals, Indianapolis, IN) or calcium phosphate precipitation in pituitary and 293 cells, respectively. To assess transcriptional activity of ERα, cells were transfected with either an ERE-tk-Luc reporter, which encodes a multimerized vitellogenin estrogen-response-element (ERE) enhancer fused to a thymidine kinase (tk) promoter (42), or a Prl-luc construct bearing 2.5 kb of the Prl regulatory region. The Prl-Luc construct was obtained from Dr. Richard Maurer. In addition, cells were cotransfected with a β-galactosidase reporter (CMV-βgal) to control for transfection efficiency. Cells were treated 24 h after transfection with EtOH or the indicated estrogen and were harvested after an additional 24 h. Assays for β-galactosidase (Tropix, Bedford, MA) and luciferase activity (Promega, Madison, WI) were performed as directed by the manufacturers. Luciferase activity was normalized to β-galactosidase activity, and degree of activation was determined relative to the ethanol control.
Generation of cell lines expressing wild-type and mutant estrogen receptors.
Stable 293 cell lines were generated using calcium phosphate transfection of retroviral expression vector LHL-CA (26) containing wild-type (WT) and mutant ERs. The G521R-ERα was generated by PCR mutagenesis, in which primers encoding the mutation (5′-AGTAACAAACGCATG-GAGCATCTGTACAGC-3′) were used in two reactions, generating ER fragments between the XbaI site and the mutation and between the mutation and a region downstream of the ER stop codon that contained a BstEII site. The two reactions were used as a template in a third reaction to create a final fragment that was then digested with XbaI and BstEII and subcloned back into an ER-encoding vector. The resultant mutant was verified by sequencing and then subcloned into the LHL-CA expression vector. This vector contains a Moloney murine leukemia viral LTR driving the expression of a gene-encoding hygromycin resistance. Expression of WT and G521R ER is controlled by a CMV enhancer and actin promoter. The LHL-CA vector contains no viral coding regions. Selection of cell colonies with hygromycin B (200 μg/ml) began 24 h after transfection. Colonies were isolated and screened for ER expression by Western blot analysis by use of anti-human ER. All results obtained after experimental treatment were confirmed in multiple clones.
Whole cell binding assays were performed as previously described (1) by use of 10 pM [3H]estradiol (New England Nuclear/Du Pont, Boston, MA) on the basis of the reported disassociation constant of ERα in PR1 cells (7). All assays were performed in the presence of ALLnL to prevent the proteolysis of receptor that occurs during establishment of equilibrium conditions. For competition, cells were incubated with increasing concentrations of 17β-estradiol, 17α-estradiol, estriol, or ICI that ranged from 1 pM to 1 μM in log increments. Steroid binding proceeded for 2 h. Assays were performed with duplicate samples and were repeated twice. (Interassay variation was neglible and is included in Fig. 4 but is hidden by the data symbols.) EC50values were determined by nonlinear regression analysis based on single-site competition (Prism 3.0; GraphPad Software).
Initial experiments were performed to assess the specificity of inducible proteolysis of ERα. PR1 cells were exposed for a brief period of 2 h to various steroids, including testosterone, progesterone, cortisol, and estrogens (17β-estradiol, 17α-estradiol, and estriol). In addition, cells were treated with the receptor antagonists 4-OHTam and ICI. After treatment, changes in ERα levels were evaluated by Western blot analysis of whole cell lysates. Shown in Fig. 1, treatment of cells with 17β-estradiol resulted in a decline in steady-state levels of ERα protein (10). The acute loss of receptor protein could also be induced by treatment with other receptor ligands, including 17α-estradiol and estriol, but not by steroids such as testosterone, progesterone, and cortisol that do not bind specifically to ERα (Fig. 1 A and Table1). Unlike agonists, ERα antagonists showed differential activity in regulating receptor protein levels. ICI was similar to estrogens in its ability to decrease receptor levels (Fig. 1 B and Table 1). However, the partial antagonist 4-OHTam was without effect. These results indicate that downregulation of ERα may be restricted to specific ligands of the receptor (Fig.1 A).
The involvement of proteasomes in the degradation of ERα induced by 17α-estradiol, estriol, and ICI was tested using an inhibitor of the proteasome pathway, ALLnL. Proteasome activity was blocked by a 30-min preincubation of cells with proteasome inhibitor. In parallel, a control group was pretreated with DMSO, the inhibitor solvent. After inactivation of proteasomes, cells were treated with estrogens and ICI as before. As shown previously, treatment of cells with estrogen and ICI resulted in a significant decline in receptor level. The loss of receptor was effectively inhibited by ALLnL (Fig.2 A) and a second inhibitor, lactacystin (data not shown). Examination of the relative receptor levels illustrated in Fig. 2 C demonstrates that ALLnL stabilized the receptor level to that of control cells and completely prevented the rapid loss of receptor due to steroid treatment. ALLnL was also effective at blocking the actions of 17β-estradiol, 17α-estradiol, and estriol (Fig. 2 B). In Fig.2 B, it can be observed that inclusion of the proteasome inhibitor alone can slightly increase basal levels of the receptor. However, upon quantification of relative ER levels by phosphoimager analysis, this apparent increase was not statistically significant.
The limitation of this proteolytic response to ligands of ERα suggested that ligand binding to receptor is an essential first step in directing the receptor to the proteasome pathway. To directly test this possibility, stable cell lines were generated that express either a WT-ERα or an ERα ligand-binding point mutant (G521R-ERα). HEK 293 cells were transfected with a retroviral expression vector that encodes hygromycin resistance and either WT-ERα or G521R-ERα. Stable colonies were selected after maintenance in hygromycin-containing medium and screened for the expression of ERα by Western blot. Multiple clones were identified for each line, and a representative line was selected for further analysis. The receptor transcriptional activity in the WT and G521R cell lines was confirmed by transient transfection of an estrogen-responsive reporter construct (ERE-tk-LUC). Shown in Fig. 3 A, cells that express WT-ERα exhibit increases in reporter gene activity in response to 17β-estradiol, 17α-estradiol, and estriol. In line with its antagonistic activity, ICI inhibited transcription, confirming the specificity of the transcriptional response. In contrast, cells expressing G521R-ERα did not exhibit ligand-dependent changes in transcription activity (Fig. 3 B). These results are consistent with the original characterizations of this mutation in mouse and human ERα (12). To examine the effect of introduction of this mutation on estrogen-induced proteolysis, cells were treated with 17β-estradiol, 17α-estradiol, estriol, or ICI for 2 h and analyzed for changes in ERα protein levels. Figure3 C shows that cells expressing WT-ERα respond to estrogens like PR1 cells by decreasing steady-state levels of ERα protein. Thus estrogen-induced proteolysis is faithfully reconstituted in this heterologous system. These cells, however, show a heightened sensitivity to ICI such that, within 2 h, WT-ERα is barely detectable. G521R-ERα levels, however, were unchanged by stimulation with either agonists or ICI (Fig. 3). These results demonstrate that ligand binding is essential in the induction of ERα proteolysis.
17β-Estradiol, 17α-estradiol, estriol, and ICI bind to ERα with varying affinities. To further examine the relationship between ligand binding and induction of proteolysis, PR1 cells were treated with different doses of estrogens ranging from 10−12 to 10−6 M (Fig. 4,top). Changes in ER levels were determined by quantification of receptor levels relative to an internal standard curve generated by dilution of extract from ethanol-treated control cells. For direct comparison, competitive binding assays were performed within the identical dose range (Fig. 4, bottom). Nonlinear regression analysis was used to calculate an EC50 for both the degradation response and receptor binding in PR1 cells. The rank order of potency indicates that 17β-estradiol is the most potent agonist and possesses approximately one order of magnitude greater activity in both binding and ability to induce degradation than estriol, 17α-estradiol, and ICI. Comparison of the EC50 values for binding and degradation for individual ligands (Table2) clearly shows a direct correlation between ligand affinity for receptor and the ability of individual ligands to induce receptor proteolysis. Thus ligands stimulate proteolysis at concentrations consistent with their binding affinity, providing further evidence that ligand binding triggers the events that lead to receptor proteolysis by proteasomes. It also points out that proteolysis can be modulated by hormone concentration.
Interestingly, although all ligands induce proteolysis, the maximum change in receptor level affected by these ligands varies. Examination of the dose curves illustrated in Fig. 4 A shows that the plateau level of receptor achieved by stimulation with estriol and 17α-estradiol is higher than that reached by 17β-estradiol and ICI. This is also supported by data in Table 1, which shows that at a saturable dose for all ligands (10 nM), significant differences exist between the relative receptor levels. 17β-Estradiol and ICI are similar in their ability to degrade the receptor. Both induce a loss of ∼50% of receptor. 17α-Estradiol and estriol, however, affect smaller declines in the receptor that are significantly different from 17β-estradiol and ICI. This ligand-specific variation cannot be accounted for by differences in binding, since all of the ligands tested are capable of occupying 100% of receptor at the 10 nM dose. Furthermore, 17α-estradiol, estriol, and ICI share similar binding affinities for the receptor, yet show different capacities to alter ER levels.
Recent studies have suggested that the transcriptional activation of receptor correlates with receptor proteolysis (24). However, in WT-ERα stable cells, 17α-estradiol, estriol, and 17β-estradiol appear to activate transcription equivalently despite exhibiting differences in modulating receptor proteolysis (Fig.2 A). To directly compare the ability of agonists to modulate proteolysis and activate gene expression, ERα transcriptional activity was assessed in PR1 cells over a range of doses (Fig.5). Cells were transfected with a reporter construct representative of a natural estrogen-regulated gene in lactotropes (Prl-Luc) and treated with varying doses of 17β-estradiol, 17α-estradiol, estriol, and ICI. Consistent with its pure antiestrogenic activity, ICI inhibited the transcriptional activity of ERα in PR1 cells (data not shown). Examination of the transcriptional response at different doses shows that ligand ability to activate transcription parallels the ligand-binding affinity. The EC50 values for transcriptional activation are ∼0.09 nM, 1 nM, and 0.4 nM for 17β-estradiol, 17α-estradiol, and estriol, respectively. Interestingly, however, differences in transactivation capacity between these ligands are abolished at doses of 1 nM and higher. Thus it appears that, although ligand binding is required to induce both receptor transcriptional activity and proteolysis, ligand-specific differences in the proteolytic response are not reflected in ligand capacity to activate ERα transcriptional function.
Proteasome-mediated proteolysis has recently garnered increased attention as a conserved mechanism responsible for the control of steroid receptor protein stability. Because of its potential importance to the regulation of receptor function, much interest has been generated in determining the mechanism through which steroids target the receptor for proteolysis. Although it has been demonstrated that estradiol treatment of cells can stimulate proteolysis, the molecular signals that engage the receptor in the proteasome pathway have not been delineated. In this study, we demonstrate that proteolysis of ERα is a specific response that is induced by ligand binding to receptor. ERα ligands, however, show varying capacity to limit intracellular ERα concentration through this pathway. These results demonstrate the existence of ligand-specific regulation of receptor proteolysis by proteasomes.
Multiple ligands stimulate the destruction of ERα, including potent agonists such as 17β-estradiol as well as the weak or short-acting agonists estriol and 17α-estradiol. Ligand binding induces a “transformation” of ERα in which the receptor becomes tightly associated with components of the nucleus. Release of ERα from this complex(es) requires high salt (0.4 M) extraction. In studies with GFP-ERα fusion protein, estradiol stimulation results in a formation of nuclear foci, providing a visual representation of the strong interactions of ERα with nuclear factors (39). Weak agonists, like estriol and 17α-estradiol, are so called because of their ability to activate some but not all of ERα responses, depending on the length of time of exposure. For example, a single administration of estriol can induce short-term estrogenic responses, such as water imbibition of the uterus, but not long-term growth responses (17). This selective regulation of receptor activity by estriol is not correlated with its ability to activate transcription, because estriol and estradiol are equally efficient at stimulating transcription in vitro (25), and under chronic stimulation, estriol functionally mimics estradiol (Fig. 5 and Ref.2). Rather, it is more closely associated with the inability of estriol-bound receptor to sustain tight nuclear interactions (2, 8). This led to the concept that factors that form complexes with ERα in the nucleus may have a role in the selective regulation of the receptor by estriol and other weak agonists. Our studies show that estriol and 17α-estradiol are less efficient at lowering receptor levels than 17β-estradiol or ICI, suggesting that fewer receptors are targeted to the proteasome pathway by these ligands despite 100% receptor occupancy. These data suggest that, like receptor transformation, proteolysis may be sensitive to the weak or short-term interactions between receptor and other nuclear factors. This notion is supported by a recent study showing that coactivator interaction with ERα is necessary for 17β-estradiol-induced downregulation (24). Our data, however, do not agree with the further conclusion suggested in the latter report, that receptor turnover is linked to the efficiency of receptor transcriptional activity. We show instead that estriol and 17α-estradiol both activate transcription to comparable levels as 17β-estradiol (Fig. 5), yet they have diminished capacity to decrease receptor concentration (Table 1). Our data, thus, better support the hypothesis that interactions within the nucleus may have a role in providing specificity to the proteolytic response.
We demonstrate that ICI induces proteasome-dependent degradation of ERα in lactotrope cells. Early studies by Gibson et al. (14) demonstrated that ICI-164384 could elicit a rapid decrease in uterine ERα, and it was suggested that the mechanism through which ICI operated involved proteolytic processing of the receptor (14). Dauvois et al. (10) also demonstrated that ICI-164384 shortened the half-life of mouse ERα, supporting the direct action of ICI on protein stability. We show that ICI targets the receptor to proteasomes in lactotrope cells, in agreement with studies by Stenoien et al. (40). In the PR1 cells, however, 17β-estradiol and ICI are equally effective, and dose-response studies indicate that higher doses of ICI are required to elicit an effect equivalent to that of 17β-estradiol. This differs from the response in MCF-7 cells (40), in which ICI appears to induce a complete loss of receptor and is more effective than estradiol when saturating concentrations of ligand are used. Differences in the effectiveness of ICI were also apparent in 293 stable cell lines expressing the wild-type receptor (Fig. 3), in which ICI stimulation resulted in a greater reduction in receptor levels, similar to what is seen in MCF-7 cells. These results suggest the possibility that regulation of proteolysis of ERα displays cell-type specificity in addition to ligand dependency.
Demonstration that ligand-induced proteolysis is abolished by a point mutant in the ligand-binding pocket provides direct evidence that ligand binding is required to target receptor to proteasomes. This is also supported by deletion of the entire ligand-binding domain (24), and by the correlation between ligand-binding affinity and potency of ligand to induce proteolysis. Ligand binding alone, however, is insufficient to induce proteolysis, since ligands of the triphenyltheylene class, such as tamoxifen, are incapable of downregulating ERα (see Fig. 1 and Ref. 4). Structural studies of the ERα ligand-binding domain demonstrate that tamoxifen binding disrupts the organization of critical helix 12 of ERα and prevents interaction of the receptor with coactivators (6,38). It is unlikely, however, that tamoxifen's activity in disrupting coactivator receptor interaction is responsible for its lack of ability to induce proteolysis of receptor, because ICI binding produces a similar structural occlusion to the coactivator interface yet is a potent stimulator of receptor degradation (31). The demonstration that not all receptor ligands can target receptor to proteasomes, however, highlights the observation that, although ligand binding is required to induce proteolysis, signals downstream of binding are also necessary to direct receptor into the proteasome pathway.
Varying doses of ligand were utilized in this study to allow for the quantitative analysis of changes in ERα levels due to proteolysis. Only a few studies have directly examined proteasome-dependent proteolysis of ERα, and commonly, saturating concentrations of ligand are utilized to induce proteolysis. In addition, changes in receptor level are assessed by qualitative determination of relative receptor protein levels. Although these other studies provide important insight relevant to the induction of proteolysis, they overlook ligand-specific differences in the regulation of ERα degradation that might provide insight into underlying regulatory mechanisms. Our results indicate that proteolysis can be modulated by ligand concentration. They also demonstrate that ligands that induce proteolysis can differentially control the numbers of receptors that are degraded. These findings illustrate an important concept, that regulation of receptor stability is governed not only at the level of proteolysis induction but also by the efficiency with which receptor is degraded. As such, small changes in receptor stability may be misinterpreted as an absence of proteasome-dependent proteolysis, when instead, they may simply reflect changes in efficiency of receptor processing. Our results highlight the importance of the quantitative assessment of changes in receptor levels and provide evidence for ligand-specific regulation of proteasome-dependent proteolysis of ERα.
We thank Drs. Jack Gorski, Jyoti Watters, Fern Murdoch, Mike Fritsch, and Shigeki Miyamoto for helpful discussion throughout the course of this investigation and for critical reading of the manuscript. In addition, we thank Dr. Richard Maurer for the gift of the Prl-Luc reporter gene construct.
This work was supported by National Cancer Institute Grant K01 CA-79090 to E. T. Alarid.
Address for reprint requests and other correspondence: E. T. Alarid, Dept. of Physiology, 120 Service Memorial Institute, 1300 University Ave., Madison, WI 53706 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2002 the American Physiological Society