Estrogen Receptor-α Hinge-Region Lysines 302 and 303 Regulate Receptor Degradation by the Proteasome

Abstract

Cellular levels of estrogen receptor-α (ERα) protein are regulated primarily by the ubiquitin-proteasome pathway. Dynamic interactions between ERα and the protein degradation machinery facilitate the down-regulation process by targeting receptor lysine residues for polyubiquitination. To date, the lysines that control receptor degradation have not been identified. Two receptor lysines, K302 and K303, located in the hinge-region of ERα, serve multiple regulatory functions, and we examined whether these might also regulate receptor polyubiquitination, turnover, and receptor-protein interactions. We used ERα-negative breast cancer C4-12 cells to generate cells stably expressing wild-type (wt)ERα or ERα with lysine-to-alanine substitutions at K302 and K303 (ERα-AA). In the unliganded state, ERα-AA displayed rapid polyubiquitination and enhanced basal turnover, as compared with wtERα, due to its elevated association with the ubiquitin ligase carboxy terminus of Hsc70-interacting protein (CHIP) and the proteasome-associated cochaperone Bag1. Treatment of C4-12 cells with either 17β-estradiol (E2) or the pure antiestrogen ICI 182,780 (ICI) induced rapid degradation of wtERα via the ubiquitin-proteasome pathway; however, in the presence of these ligands, ERα-AA was less efficiently degraded. Furthermore, ERα-AA was resistant to ICI-induced polyubiquitination, suggesting that these lysines are polyubiquitinated in response to the antiestrogen and demonstrate a novel role for these two lysines in the mechanism of action of ICI-induced receptor down-regulation. The reduced stability of ERα-AA in the unliganded state and the increased stability of ERα-AA in the liganded state were concordant with reporter gene assays demonstrating that ERα-AA has lower basal activity but higher E2 inducibility than wtERα. These data provide the first evidence that K302/303 protect ERα from basal degradation and are necessary for efficient E2- and ICI-induced turnover in breast cancer cells.

THE FEMALE SEX-STEROID hormone 17β-estradiol (E2) is essential for the normal growth and development of the reproductive system and the breast, and its action is mediated primarily by the estrogen receptor-α (ERα) (1). Aberrant E2 signaling through ERα has been strongly associated with disease development and the progression of breast cancer (2,3). Thus, appropriate ERα levels and subsequent E2 responses are precisely controlled, in part, by receptor turnover (4,5,6,7).

Cellular levels of ERα are maintained by distinct receptor degradation pathways that ultimately converge on the ubiquitin-26S proteasome system (7,8,9,10,11,12,13,14). Although both basal and ligand-induced ERα degradation are mediated by these proteolysis pathways (12,13,15,16,17), regulation of receptor degradation at the molecular level is highly dependent upon the physiological state and nature of the cellular stimulus. For example, in the unliganded state (i.e. basal receptor turnover), ERα is targeted for degradation by dynamic interactions with heat-shock proteins (Hsps), cochaperones, and the ubiquitin ligase carboxy terminus of Hsc70-interacting protein (CHIP) (18,19). In the presence of E2, hormone-bound receptors are targeted for degradation through a transcription-coupled pathway requiring new protein synthesis (thus blocked by the protein synthesis inhibitor cycloheximide) (10). However, neither transcriptional activity nor new protein synthesis are needed for degradation of ERα when bound by a class of drugs called selective estrogen receptor down-regulators (SERDs) [ICI 182,780 (ICI) or fulvestrant (Faslodex)] (12,16,20). Furthermore, drugs that inhibit Hsp90 function [e.g. geldanamycin (GA)] induce ERα down-regulation by altering receptor-Hsp90 interactions (19,21,22), in a ubiquitin ligase (CHIP)-dependent manner (18,19). In contrast, by dissociating receptor-chaperone complexes, selective estrogen receptor modulators [e.g. 4-hydroxytamoxifen (OHT)] stabilize and protect ERα from both basal turnover and GA-induced degradation pathways (13,19,23).

ERα protein turnover results from the formation of polyubiquitin chains on receptor lysines and its subsequent proteasomal degradation through the distinct degradation pathways described above. However, of the 29 lysines found within ERα, none have been identified as residues targeted for polyubiquitination and thus mediating receptor turnover. Two possible candidates for receptor polyubiquitination are lysines K302 and K303, found within the hinge-region of ERα. Lysines 302 and 303 have multiple regulatory functions, including receptor coactivator binding (24,25,26,27), and in the presence of E2, K302 is monoubiquitinated by BRCA1/BARD1, a ubiquitin ligase (28) The impact of K302 monoubiquitination on ERα stability is unknown, although these data reveal the availability of these hinge-region lysines for posttranslational modification, and we hypothesize that they may be suitable targets for polyubiquitination.

ERα matures into a form capable of ligand binding and transactivation via progression through dynamic receptor-cochaperone interactions (29). Several molecular chaperones, including Hsp70 and Hsp90, mediate ERα progression through this foldosome (30) by facilitating ERα interaction with cochaperones, including CHIP, Bag1 and p23 (Ptges3) (18,19,31,32). ERα hinge-region lysines 302 and 303 reside between the DNA-binding and ligand-binding domains and are within the receptor surface area that interacts with Hsp90 (31,33). Therefore, mutation of these residues may alter ERα-Hsp90-cochaperone interactions, resulting in altered receptor progression through the foldosome. We and others have shown that the cochaperone CHIP is an E3-ubiquitin ligase required for basal ERα ubiquitination and proteasomal degradation (18,19). We have also reported that GA increases ERα-Hsp90 association with CHIP, enhancing receptor degradation in the absence of ligand (19). The cochaperones Bag1 and p23 have also been found in Hsp90-ERα complexes (18); however, their precise role in receptor turnover remains unknown. Bag1 is found in early receptor-chaperone complexes and has been shown to link Hsp70 client proteins to the proteasome through its N-terminal ubiquitin-like domain (34,35), suggesting that Bag1 may promote receptor degradation. Conversely, p23 is found in mature receptor-Hsp complexes and has been found to enhance basal and ligand-induced receptor transactivation (36). In addition, p23 competes with CHIP for receptor association (37), suggesting that p23 may exert a stabilizing effect on ERα. These previously defined actions suggest that Bag1, and/or p23, may play a functional role in mediating receptor turnover.

In the present study, we used the ERα-negative breast cancer cell line C4-12 to stably express either wild-type (wt)ERα or ERα with lysine-to-alanine substitutions at K302 and K303 (ERα-AA) to investigate the role of these lysines in ERα polyubiquitination, turnover, and receptor-cochaperone interactions. We demonstrate that lysines K302 and K303, by impeding receptor-CHIP and Bag1 interactions, and promoting p23 interactions, protect unliganded ERα from basal turnover. Additionally, in the presence of ligand, these two lysine residues control proteasome-mediated turnover and promote ICI-induced receptor polyubiquitination/degradation, thus revealing a novel role for these lysines in regulating receptor turnover.

RESULTS

Expression of wtERα and ERα-AA in C4-12 Cells

Site-directed mutagenesis was performed on wtERα in pcDNA. Lysines at positions 302 and 303 were mutated to alanines to create ERα-K302A, K303A (ERα-AA; Fig. 1A). Stable cell lines harboring wtERα or ERα-AA were established using C4-12 cells, an ERα-negative subline of MCF7 breast cancer cells (38). Two wtERα and three ERα-AA clones were chosen that had similar ERα expression levels. ERα expression level among the clones was 2-fold higher than MCF7 cells (Fig. 1B). ERα mRNA levels in wtERα and ERα-AA clones were 2- and 4-fold higher than MCF7 cells, respectively (Fig. 1C). The discrepancy between ERα-AA mRNA expression and protein levels was due to elevated basal ERα-AA degradation (described later).

Figure 1.

Expression of wtERα and ERα-AA in C4-12 Cells

Panel A, Schematic showing the mutation of hinge-region lysines to generate ERα-AA. Lysines 302 and 303 reside in the hinge region between the DNA-binding domain (letter D) and the ligand-binding domain (letter E). Panel B, Representative ERα protein levels from two wtERα clones and three ERα-AA clones. GAPDH was used as SDS-PAGE loading control. ERα-negative breast cancer C4-12 cells were transfected with 1 μg pcDNA-ERα (wtERα) or ERα-K302A, K303A (ERα-AA) using Lipofectamine/PLUS and then treated with 800 μg/ml G418 daily until single colonies were visible. Multiple clones from each transfection were selected and expanded. Panel C, RT-qPCR analysis of ERα mRNA levels. EF1α mRNA level was used as internal control.

Lysines 302/303 Reduce Basal Turnover of Unliganded ERα

Because apo-(unliganded) ERα is a short-lived protein and undergoes constant degradation (8,9,39), we investigated the effect of hinge-region lysine mutations on stability of the unliganded receptor. C4-12 cells stably expressing wtERα or ERα-AA were treated with the protein synthesis inhibitor cycloheximide (CHX) and apo-receptor degradation was then monitored by immunoblot. Levels of wtERα decreased in a time-dependent manner, displaying a half-life of 3.85 ± 0.3 h (Fig. 2A, upper panel, and 2C), in agreement with previous reports using CHX and other methods (14,39,40,41,42). In comparison, increased (P < 0.01) basal turnover of the mutant receptor ERα-AA was observed, and the half-life of ERα-AA was 1.04 ± 0.3 h (Fig. 2A, lower panel, and 2C). These results demonstrate that lysines 302/303 regulate ERα stability in the unliganded state by limiting basal receptor turnover.

Figure 2.

Lysines 302/303 Reduce Basal Turnover of Unliganded ERα

C4-12 cells stably expressing wtERα or ERα-AA were treated with the following: A, CHX (25 μg/ml) for 30 min before cell harvest at the indicated time points; B, CHX (25 μg/ml) for 30 min followed by GA (1 μm) for the indicated times. Experiments were performed in duplicate and repeated twice using two wtERα-expressing clones and three ERα-AA-expressing clones. GAPDH was used as SDS-PAGE loading control. C, Turnover rates of ERα and ERα-AA in the absence or presence of GA. The band density of the exposed film was evaluated with ImageJ software. Relative ERα levels (vs. untreated cells) are shown as means ± se. **, P < 0.01.

In the absence of ligand, apo-ERα is degraded by associating with Hsps, including Hsp90, and the E3-ubiquitin ligase CHIP (18,19). The Hsp90 inhibitor GA, by blocking ATP binding to Hsp90 (43,44,45,46), enhances ERα association with CHIP and increases receptor turnover (19,21,22). Because lysines 302 and 303 reside in the Hsp90-interacting domain of ERα (31,33), we investigated the effect of hinge-region lysine mutation on GA-induced ERα turnover. Treatment of C4-12 cells stably expressing wtERα with GA displayed an increased (P < 0.01) receptor turnover rate; the half-life of wtERα decreased from 3.85 ± 0.3 to 1.40 ± 0.3 h (Fig. 2B, upper panel, and 2C). GA treatment did not further increase ERα-AA basal turnover rate; receptor half-life was unchanged in the absence or presence of GA (1.04 ± 0.3 vs. 0.92 ± 0.3 h, respectively; Fig. 2B, bottom panel, and 2C). To summarize these findings, CHX and CHX/GA data are plotted together in Fig. 2C. As shown, GA-induced wtERα turnover occurred at the same rate as ERα-AA basal and GA-induced turnover, suggesting that loss of lysines 302/303 and Hsp90 inhibition by GA share a common molecular mechanism to promote ERα degradation.

Lysines 302/303 Protect Unliganded ERα from Polyubiquitination

Rapid ERα-AA protein turnover was observed in the absence of ligand (Fig. 2) and ERα-AA mRNA levels were increased in ERα-AA clones that had equal levels of protein as wtERα (Fig. 1C), suggesting that loss of lysines 302/303 resulted in a destabilized receptor in the absence of ligand. Because ERα turnover is mediated by the ubiquitin-proteasome pathway (8,9,10,11), we investigated the role of K302 and K303 in ubiquitination of ERα. Polyubiquitination assays in ERα-negative HeLa cells were performed as we have described previously (19). Briefly, cells were transfected with equal amounts of wtERα or ERα-AA in addition to a hemagglutinin (HA)-tagged ubiquitin or vector control plasmid. Transfected cells were then treated with the proteasome inhibitor MG132 and allowed to accumulate polyubiquitinated proteins. After MG132 treatment, immunoprecipitation was carried out using an ERα-specific antibody. Proteins were then resolved by SDS-PAGE, and the presence of ubiquitinated receptor was detected by immunoblotting with an HA antibody (polyubiquitinated species were detected as a high-molecular-weight ladder on the membrane). In the absence of MG132, wtERα polyubiquitination levels remained low (Fig. 3, lane 1) but subsequently increased after MG132 treatment (lane 3). In contrast, in the absence of MG132, total immunoprecipitated (lower panel) and polyubiquitinated forms of ERα-AA species were notably more abundant than untreated wtERα (lane 5). In addition, MG132 treatment resulted in greater accumulation of polyubiquitinated forms of ERα-AA and total ERα-AA protein compared with wtERα (Fig. 3, lane 6). Although it appeared that the mutant receptors may be more polyubiquitinated than wtERα, it was not possible with these methods to quantify the degree of polyubiquitination per receptor. The mutant receptor displayed enhanced basal turnover rate and possibly enhanced basal polyubiquitination, indicating that lysine residues 302 and 303 may protect ERα from basal degradation by limiting apo-receptor ubiquitination.

Figure 3.

Lysines 302/303 Protect Unliganded ERα from Polyubiquitination

HeLa cells were transfected with equal amounts (250 ng) of wtERα or ERα-AA along with 1 μg HA-ubiquitin, using Lipofectamine/PLUS, and 24 h later, the cells were treated with DMSO or MG132 for 5 h and then ERα immunoprecipitated with anti-ERα antibody. Precipitated proteins were then resolved by SDS-PAGE and Western blot performed using an HA antibody. Levels of immunoprecipitated ERα were also determined by probing with anti-ERα antibody (bottom panel).

K302 and K303 Reduce ERα Association CHIP and Bag1 Complexes

We and others have previously shown that GA increases the association of ERα-Hsp90 complexes with the E3-ubiquitin ligase CHIP, increasing receptor degradation (18,19). Because apo-ERα-AA had a rapid basal turnover rate that was not further increased by GA (Fig. 2B), we investigated the possibility that lysine mutations mimic the effects of GA by enhancing ERα-AA association with CHIP. In addition, we wished to examine ERα interactions with the Bag1 cochaperone. Bag1 links Hsp70 to the proteasome (34) and has also been detected in ERα complexes (18), but its precise role in ERα turnover has not been previously explored. Finally, because GA inhibits p23 interaction with Hsp90 (47) and also blocks p23-mediated enhancement of receptor transactivation (47,48), we examined receptor-p23 interactions to determine whether p23 plays a stabilizing role on ERα. To investigate the role of K302/303 in Hsp90, CHIP, Bag1, and p23 interactions with the receptor protein, and to determine whether alterations in receptor-cochaperone interactions contributed to basal turnover of ERα-AA, we performed coimmunoprecipitation assays and analyzed Hsp90-cochaperone-receptor complexes in the presence or absence of GA. Complexes were immunoprecipitated with an ERα-specific antibody, and complexed proteins were then identified by immunoblot, as shown in Fig. 4A. Relative receptor-cochaperone levels for CHIP, Bag1, and p23 are shown in Fig. 4B. In untreated C4-12 cells stably expressing either wtERα or ERα-AA, comparable levels of ERα, Hsp90, CHIP, Bag1, and p23 proteins were observed (Fig. 4A, lanes 1 and 6). Before treatment, wtERα coimmunoprecipitated with Hsp90, CHIP, the cytosolic form of Bag1 (36 kDa), and p23 (Fig. 4A, upper panel, lane 2). ERα-AA also coimmunoprecipitated with these cochaperones, but association of ERα-AA with Bag1 and CHIP appeared to be enhanced, and only a weak association of ERα-AA with p23 was observed (Fig. 4A, lower panel, lane 7). Overall levels of immunoprecipitated Hsp90 did not change throughout the duration of the experiment for both wtERα and ERα-AA (Fig. 4A), suggesting that changes in receptor turnover were not simply due to changes in ERα-Hsp90 interaction. GA treatment resulted in an increase in association of CHIP and Bag1 with wtERα, with a concomitant decrease in p23 association (Fig. 4A, lane 3, and 4B). Although GA treatment further enhanced the interaction of ERα-AA with CHIP and Bag1, the p23 association was completely abolished (Fig. 4A, lanes 8 and 9, and 4B). These data suggest that K302/303 stabilize ERα by facilitating receptor progression through the foldosome, decreasing interaction with CHIP and Bag1, and increasing interaction with p23.

Figure 4.

K302 and K303 Reduce ERα Association with CHIP and Bag1 Complexes

A, C4-12 ERα stable cells were pretreated with CHX (25 μg/ml), followed by GA (1 μm) for 0–3 h. Proteins were harvested and receptor-cochaperone complexes immunoprecipitated with anti-ER or control IgG antibodies. Proteins were resolved by SDS-PAGE and Western blot performed using specific antibodies against ER, p23, Bag1, and CHIP. B, Relative cochaperone/ER levels for CHIP, Bag1, and p23. Cochaperone/ER levels were normalized to cochaperone/wtERα levels for untreated cells.

Depletion of CHIP and Bag1 Reduces ERα Turnover, Whereas p23 Knockdown Increases Receptor Turnover

Previously, we have demonstrated that knockdown of CHIP via small interference RNA (siRNA) abolished basal and GA-induced ERα down-regulation in both HeLa and MCF7 cells (19). To further establish a role for CHIP, Bag1, and p23 in regulating ERα turnover, we used siRNA to investigate turnover of wtERα and ERα-AA in the absence of these cochaperones. CHIP associated more strongly with rapidly degraded ERα-AA (Fig. 4), and ERα-AA was also resistant to GA-induced degradation (Fig. 2), suggesting that ERα-AA degradation is CHIP dependent. The siRNA against CHIP was performed in HeLa cells as we have described previously (19). HeLa cells were transfected with equal amounts of wtERα or ERα-AA plasmid, with or without the CHIP-siRNA expression construct (CHIPi) vector, and cells were then treated with CHX followed by vehicle or GA. Empty vector and mock transfection had no effect on CHIP levels (Fig. 5A, top panel). In addition, basal and GA-induced down-regulation in the vector control cells were not different from that of cells treated in Fig. 2 and were used as controls for CHIPi assays. Expression of CHIPi decreased the level of CHIP by over 60% (Fig. 5A, top panel), and this was sufficient to completely block basal turnover of both wtERα and ERα-AA (Fig. 5A, middle panel). CHIPi also blocked GA-induced turnover of wtERα and ERα-AA (Fig. 5A, bottom panel), thus confirming that CHIP mediates basal and GA-induced turnover of wtERα and ERα-AA.

Figure 5.

Depletion of CHIP and Bag1 Reduces ERα Turnover, Whereas p23 Knockdown Increases Receptor Turnover

A, For siRNA CHIP experiments, HeLa cells were transfected with equal amounts of wtERα or ERα-AA along with CHIPi, pcDNA empty vector, or transfection reagent only (Mock), and 24 h later, cells were pretreated with CHX followed by vehicle or GA (1 μm) for the time periods indicated. Western blots were then performed against CHIP and ERα, using specific antibodies. B and C, siRNA against Bag1 (Bag1i, 50 nm), p23 (p23i, 100 nm), and control scrambled oligo (Sc, 100 nm) were transfected into C4-12 stable cells using the siRNA transfection reagent Dharmafect1 for 3 d, and 72 h after transfection, cells were pretreated with CHX, followed by vehicle or GA for the indicated time periods. Protein levels were examined by Western blot using specific antibodies. The band density of exposed film was evaluated with ImageJ software. Scrambled siRNA and mock (M; Dharmafect1 only) were specificity and transfection controls. GAPDH was used as SDS-PAGE loading control. Experiments were performed in duplicate and repeated twice using two wtERα and three ERα-AA clones. **, P < 0.01.

(Figure continues next page.)

Similar to CHIP, association of Bag1 was stronger with ERα-AA than wtERα. Because this protein has been shown to link Hsp70 to the proteasome (34,35), C4-12 cells were transfected with siRNA against Bag1 or scrambled siRNA to investigate whether Bag1 is involved in ERα turnover. Scrambled siRNA or mock transfection had no effect on Bag1 levels (Fig. 5B, top panel), and basal and GA-induced ERα down-regulation was not different from untransfected cells (see Fig. 2C). Scrambled siRNA was therefore used as a control for Bag1 siRNA treatment. Knockdown of Bag1 in C4-12 stable clones delayed basal turnover of wtERα, increasing (P < 0.01) its half-life from 3.14 ± 0.3 h to more than 6 h (Fig. 5B, middle panel). Bag1 knockdown also delayed ERα-AA basal degradation, increasing (P < 0.01) the mutant receptor half-life from 1.2 ± 0.3 h (see Fig. 4A) to 3.4 ± 0.2 h (Fig. 5B, middle panel). GA-induced down-regulation of both wtERα and ERα-AA was delayed by Bag1 knockdown; wtERα half-life was increased (P < 0.01) from 1.40 ± 0.3 h to more than 6 h, and ERα-AA half-life increased (P < 0.01) from 0.92 ± 0.3 h to 3.3 ± 0.2 h (Fig. 5B, lower panel; and see Fig. 2C), thus confirming that Bag1 promotes both basal and GA-induced receptor turnover.

The cochaperone p23 is associated with mature Hsp90 complexes and enhances ERα transactivation (33,36). ERα-AA bound less strongly to p23 than wtERα; consequently, we transfected C4-12 cells with siRNA against p23 or scrambled siRNA to assess whether loss of p23 would destabilize ERα and enhance its turnover. Scrambled siRNA or mock transfection had no effect on p23 levels (Fig. 5C, top panel), and basal and GA-induced ERα down-regulation in cells transfected with scrambled siRNA or mock transfection were not different from untransfected cells (see Fig. 2C). Scrambled siRNA was therefore used as a control for p23 siRNA treatment. Knockdown of p23 enhanced wtERα turnover in CHX-treated cells; receptor half-life decreased from 3.1 ± 0.3 h to 0.85 ± 0.1 h (Fig. 5C, middle panel). Moreover, GA-induced turnover of wtERα was also increased after p23 knockdown (P < 0.01), with its half-life decreased from 1.5 ± 0.2 h to 0.75 ± 0.2 h (Fig. 5C, lower panel). However, p23 knockdown had no effect on basal or GA-induced turnover of ERα-AA, because the half-life of ERα-AA was similar in the presence or absence of p23 siRNA (1.04 ± 0.3 h vs. 0.94 ± 0.1 h; Fig. 5C, middle panel). This was not unexpected, because low levels of p23 were detected in ERα-AA-immunoprecipitated complexes (Fig. 4A). These results demonstrate that p23 exerts a stabilizing effect on ERα. Together, these data suggest that CHIP and Bag1 promote, whereas p23 inhibits, basal and GA-induced ERα degradation. Furthermore, K302 and K303 appear to be important for the association of ERα with these cochaperones during basal and GA-induced receptor turnover, by decreasing receptor association with the degradation-promoting cochaperones CHIP and Bag1 while simultaneously increasing association with the stabilizing cochaperone p23.

Hinge-Region Lysines Promote Ligand-Induced Receptor Turnover

Ligand binding dissociates ERα from Hsp90 complex and directs ERα toward alternative degradation pathways (12,13,15). To investigate the effect of hinge-region lysine mutations on ligand-mediated receptor turnover, C4-12 cells were treated with various ligands, and changes in ERα stability were monitored by immunoblot. As shown, E2 induced ERα down-regulation after transcriptional activation, decreasing the wtERα protein level in a time-dependent manner (Fig. 6A, upper panel) while impairing degradation of ERα-AA under the same experimental conditions (Fig. 6A, lower panel). ICI, which directly targets ERα for degradation (12,16,20,49), similarly reduced wtERα levels to less than 50% by 1 h (Fig. 6B, upper panel); this same level of reduction in ERα-AA levels was not seen until 3 h after ICI treatment (Fig. 6B, lower panel). CHX pretreatment did not significantly affect ICI-induced down-regulation of either receptor (supplemental Fig. S1, published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org), confirming that the slower decline of ERα-AA protein in the presence of ICI was due to impaired receptor degradation rather than elevated synthesis of the mutant ER. These data suggest two possible roles for ligand action on ERα-AA: 1) ligand binding may rescue ERα-AA from interaction with CHIP and Bag1, protecting it from the rapid basal turnover observed in Fig. 2B, or 2) ERα-AA may be less sensitive to ligand-induced degradation.

Figure 6.

Hinge-Region Lysines Promote Ligand-Induced Receptor Turnover

C4-12 ERα stable cells treated with E2 (10 nm) (A), ICI (100 nm) (B), or OHT (100 nm) (C) for the indicated times. Experiments were performed in duplicate and repeated twice using two ERα and three ERα-AA clones. GAPDH was used as SDS-PAGE loading control. The band density of exposed film was evaluated with ImageJ software. Relative ERα levels (vs. untreated cells) are shown in the corresponding graph as the mean ± se. **, P < 0.01.

By dissociating receptor-chaperone complexes, OHT stabilizes ERα and protects receptors from basal turnover (13,19,23). As expected, wtERα was stabilized by OHT (Fig. 6C, upper panel), and OHT caused ERα-AA levels to accumulate (Fig. 6C, lower panel), suggesting that OHT was able to antagonize rapid ERα-AA basal turnover, further implicating lysines 302/303 in protecting ERα from basal turnover.

Hinge-Region Lysines Promote Ligand-Induced Receptor Polyubiquitination

It is well known that both E2 and ICI stimulate receptor ubiquitination (9,49,50,51) and its subsequent degradation by the 26S proteasome (9). Consequently, our observation that these ligands were unable to efficiently degrade ERα-AA indicated that ERα-AA may be resistant to polyubiquitination. To investigate this possibility, we measured the ubiquitination of ERα and ERα-AA after inhibiting the proteasome with MG132 and stimulating receptor ubiquitination with E2 or ICI. ERα-negative HeLa cells were transiently transfected with equal amounts of wtERα or ERα-AA expression constructs, along with the HA-ubiquitin construct. Cells were then pretreated with dimethylsulfoxide (DMSO) or MG132 before treatment with DMSO, E2, or ICI. Subsequently, ERα was immunoprecipitated with an ERα-specific antibody, and HA-polyubiquitinated species of ERα were detected as a high-molecular-weight ladder on the membrane. As shown, MG132 treatment of cells containing wtERα resulted in accumulation of polyubiquitinated receptor forms (Fig. 7, left panel, lane 1 vs. 2). After E2 or ICI treatment, similar levels of ubiquitinated wtERα were observed, presumably due to proteasomal degradation of ubiquitinated receptors (Fig. 7, left panel, lanes 3 and 5). As expected, proteasome inhibition with MG132, before E2 or ICI treatment, resulted in the accumulation of polyubiquitinated wtERα (Fig. 7, left panel, lanes 4 and 6).

Figure 7.

Hinge-Region Lysines Promote Ligand-Induced Receptor Polyubiquitination

HeLa cells were transfected with equal amounts (250 ng) of wtERα or ERα-AA, along with 1 μg HA-ubiquitin using Lipofectamine/PLUS. Transfected cells were pretreated with vehicle (DMSO) or MG132 (25 μm) for 1 h, followed by DMSO, E2 (10 nm), or ICI (100 nm) for 4 h. ERα was then immunoprecipitated with anti-ERα antibody. Precipitated proteins were resolved by SDS-PAGE and Western blot performed with an HA antibody. Levels of immunoprecipitated ERα were also determined by probing with an anti-ERα antibody (lower panel).

In ERα-AA-transfected cells, MG132 treatment also resulted in accumulation of polyubiquitinated receptor forms but to a greater extent than cells transfected with wtERα (Fig. 7, compare lanes 1 and 2 vs. 7 and 8). In the absence of MG132, E2 treatment did not further increase ERα-AA ubiquitination (Fig. 7, right panel, lane 9); however, MG132 pretreatment increased polyubiquitinated ERα-AA in the presence of E2 (Fig. 7, right panel, lane 10). Ubiquitination levels of ERα-AA and wtERα in the presence of E2 were similar (Fig. 7, lane 3 vs. 9). ICI induced modest polyubiquitination of ERα-AA in the absence of MG132 (Fig. 7, right panel, lane 11), although these levels were not different from ICI-induced polyubiquitination in the presence of MG132 (Fig. 7, right panel, lane 11 vs. 12). Importantly, ICI-treated ERα-AA protein levels were slightly higher than ICI-treated wtERα levels, but less polyubiquitination occurred in ERα-AA cells (Fig. 7, lane 12). An ERα mutant with lysine-to-arginine substitutions, ERα-K302R/K303R (ERα-RR), shared a similar ubiquitination profile to that of ERα-AA. In contrast to wtERα, both mutant receptors were heavily ubiquitinated in the absence of ligand, and no further ubiquitination was observed in response to ICI treatment (Fig. 7 and supplemental Fig. S2). These results indicate that K302 and K303 may be direct targets for polyubiquitination in the presence of ICI. We therefore report a previously undescribed role for these hinge-region lysines in mediating receptor polyubiquitination induced by the pure antiestrogen.

K302 and K303 Contribute to ERα Target Gene Transactivation

Although E2 binding increases ERα transactivation, apo-ERα is also capable of eliciting basal transcriptional activity (52). Mutating K302 and K303 resulted in rapid ERα turnover in the absence of ligand (Fig. 2) but increased receptor stability in the presence of E2 (Fig. 6). It was therefore of interest to examine whether these two hinge-region lysines play a functional role in ERα transactivation in the presence and absence of E2. To examine transcriptional competency of ERα-AA, basal and E2-induced receptor activity was examined using an E2-responsive chloramphenicol acetyltransferase (CAT) construct [estrogen response element (ERE)-CAT] (5). C4-12 stable cell lines were transiently transfected with an ERE-CAT reporter and treated with E2. The absolute CAT levels in untreated (DMSO) ERα-AA-expressing cells exhibited lower (P < 0.01) transcriptional output than cells expressing wtERα (0.15 ± 0.02 vs. 1.62 ± 0.09 pg/mg lysate; Fig. 8A). E2 treatment elicited a response in both cell lines, but CAT expression remained lower (P < 0.01) in ERα-AA-expressing cells vs. wtERα-expressing cells (0.98 ± 0.06 vs. 2.67 ± 0.06 pg/mg lysate), suggesting an overall reduction in ERα-AA-mediated transcriptional activity. Normalized CAT values (untreated CAT levels set to 1; Fig. 8B) revealed that E2-induced fold changes in CAT levels were higher (P < 0.01) for ERα-AA compared with wtERα (11.12 ± 2.58-fold vs. 2.11 ± 0.19-fold), suggesting that mutation of K302/303 results in overall lower transcriptional activity, with enhanced E2 inducibility.

Figure 8.

K302 and K303 Contribute to ERα Target Gene Transactivation

A, Estrogen-responsive CAT assays were performed in C4-12 ERα stable cells. Cells were transfected with 250 ng ERE-Vit-CAT and then treated with DMSO or E2 (10 nm) for 48 h. Cells were then lysed and total protein (100 μg) from each treatment group used to determine CAT levels. B, Relative levels of CAT expressed as fold change of E2-induced gene expression by setting untreated levels to 1. The E2-induced transactivation for ERα-AA was significantly higher than for wtERα. Results are expressed as mean ± se from three independent experiments. **, P < 0.01. C, Transactivation of cathepsin D expression by ERα-AA and wtERα. Induction of the E2-responsive gene cathepsin D was determined by RT-qPCR analysis after E2 treatment (10 nm) of C4-12 cells for the indicated time periods. Cathepsin D levels were normalized with EF1α. D, Relative levels of cathepsin D mRNA were expressed as fold change of E2-induced gene expression by setting untreated levels to 1. Results are the mean ± se from three independent experiments. **, P < 0.01. E–G, Cochaperone knockdown alters ERα-mediated transcriptional output. MCF7 cells (F), wtERα C4-12 cells (G), or ERα-AA C4-12 cells (H) were transfected with siRNA against CHIP, Bag1, and p23 for 2 d. E2 induction of cathepsin D was determined by RT-qPCR analysis after 4 h E2 treatment (10 nm). Cathepsin D levels were normalized with EF1α and expressed as mean ± se of results from two MCF7 replicates, two wtERα clones, or three ERα-AA clones. Scrambled oligo (Sc) was included as control. H, Lysines 302/303 do not modulate E2 sensitivity. ERE-Luciferase assays were performed in C4-12 ERα stable cells. Cells were transfected with 2x-ERE-ps2-luc and then treated with increasing dose of E2 (0–10 nm) for 12 h. Luciferase activity was measured and normalized to cotransfected CMV-β-gal, and compared with vehicle (DMSO)-treated cells. Values are expressed as mean ± se of three independent experiments, each performed in triplicate. The EC₅₀ values were calculated using the 95% confidence function of Prism software. The EC₅₀ values are shown as dashed lines and solid lines for wtERα and ERα-AA, respectively.

In a more physiologically relevant context, we investigated the expression of the endogenous E2-responsive gene cathepsin D in C4-12 cells. Induction of cathepsin D levels by E2 was observed in both wtERα and ERα-AA C4-12 cells (Fig. 8C). Similar to CAT assays, at all time points examined, absolute levels of cathepsin D mRNA were lower (P < 0.01) in cells expressing ERα-AA (Fig. 8C). However, when normalized (untreated mRNA level set to 1), the fold change of E2-induced cathepsin D mRNA expression was greater (P < 0.01) in cells expressing ERα-AA (Fig. 8D). The basal and E2-induced expression levels of cathepsin D mRNA in wtERα C4-12 cells were comparable to that in ERα-positive MCF7 cells (Fig. 8, E–G), whereas cathepsin D mRNA level in the parental ERα-negative C4-12 cells was not affected by E2 treatment (data not shown).

The ERα hinge region contains the receptor nuclear localization sequences (53). To determine whether the low transcription activity of ERα-AA is caused by altered cellular localization, we examined nuclear translocation of wtERα and ERα-AA. In the absence of ligand, ERα-AA was found equally distributed between cytosolic and nuclear fractions, whereas the majority of wtERα protein was found in the nuclear fraction (supplemental Fig. S3). This result is in agreement with coimmunoprecipitation data that revealed elevated association of ERα-AA with cytosolic cochaperones (Fig. 4). Both receptors efficiently translocated to the nucleus in response to E2 treatment, suggesting that the decreased ERα-AA transcription activity in response to E2 is not due to impaired nuclear localization.

E2-induced cathepsin D expression was also examined after siRNA knockdown of cochaperones to determine the relative contribution of each cochaperone to receptor transcriptional activity. CHIP knockdown increased basal and E2-induced cathepsin D mRNA levels in both wtERα- and ERα-AA-expressing cells (Fig. 8, E–G). Notably, CHIP knockdown had a greater effect on ERα-AA-mediated gene expression. CHIPi increased basal and E2-induced wtERα activity by 2-fold, whereas it increased basal and E2-induced ERα-AA activity by 3-fold. The enhanced interaction between ERα-AA and CHIP is clearly involved in decreasing the transcriptional capacity of the mutant receptor. Bag1 knockdown did not significantly alter cathepsin D expression mediated by either ERα-AA or wt-ERα. Knockdown of p23 significantly decreased both basal and E2-induced cathepsin D levels in wtERα-expressing cells but not in ERα-AA cells, in agreement with a previous report that p23 enhances receptor activity (54). It is not surprising that knockdown of p23 had no effect on ERα-AA-mediated cathepsin D expression, because ERα-AA does not significantly interact with p23 (Fig. 4).

In both CAT and cathepsin D assays, mutation of K302 and K303 resulted in lower transcriptional output with or without ligand, suggesting that these residues are critical for full ERα transcriptional competence. The effect of hinge-region mutation on ERα sensitivity to E2 has recently been investigated, with disparate findings (27,55). The discordant reports on this issue appear be due to experiments using different cellular environments. To shed light on this issue, we examined the sensitivity of ERα-AA to E2 in the previously unexplored C4-12 cellular background. C4-12 ERα clones were transfected with the estrogen-responsive luciferase reporter 2x-ERE-pS2-luc (5) and then treated with E2 (range 10⁻¹⁶ to 10⁻⁹ m). A dose-responsive increase in luciferase activity was observed for both wtERα- and ERα-AA-transfected cells after E2 treatment (Fig. 8H). Sigmoidal curve-fit analysis was then used to determine the concentration of E2 inducing 50% maximal luciferase activity (E2 EC₅₀). There was no significant difference in sensitivity of wtERα and ERα-AA to E2; EC₅₀ was 10^−12.315 m vs. 10^−12.44 m for wtERα vs. ERα-AA, respectively (Fig. 8H; indicated by dashed and solid vertical lines on the x-axis). Taken together, these results demonstrate a role for K302/303 in promoting both basal and E2-induced transactivation, without altering hormone sensitivity.

DISCUSSION

Protein turnover and degradation pathways, which ultimately converge on the ubiquitin-26S proteasome system (7,8,9,10,11,12,13,14), are the predominant mechanisms for regulating cellular levels of ERα (51,56). Distinct mechanisms that down-regulate ERα and other steroid hormone nuclear receptors promote lysine polyubiquitination and subsequent proteasome-mediated receptor degradation (57). However, none of the 29 ERα lysine residues have been identified as direct polyubiquitination sites that stimulate ERα turnover. Although previous studies have suggested that ERα lysines K302 and K303, found within the hinge region, can serve multiple regulatory functions (25,27), the role of these two lysines in receptor turnover has not been established. In the present study, we focused on how K302 and K303 control ERα ubiquitination and turnover. By mutating these two lysines, we demonstrate that K302 and K303 promote ERα stability in the unliganded state, allow for efficient receptor turnover in response to E2, and finally promote polyubiquitination and turnover in response to the antiestrogen ICI. The potential roles of lysines 302/303 in ERα degradation pathways have been summarized in a model in Fig. 9.

Figure 9.

Lysines 302/303 Protect ERα from Basal Turnover and Promote E2 and SERD-Induced Degradation

ERα is degraded via three distinct down-regulatory pathways that converge on the 26S proteasome. 1) Basal turnover (left pathway). ERα protein folding and maturation begins in a multiprotein Hsp70/90 chaperone complex (shown as a simplified complex at top). In the absence of ligand, ERα is ubiquitinated by CHIP, and ubiquitinated receptors are recognized by Bag1 and delivered to the proteasome. Apo-ERα turnover (basal turnover) is enhanced by both GA and lysine 302/303 mutation, through increased complex association with CHIP and Bag1.2) Transcription-coupled ERα turnover (right pathway). ERα is dynamically maintained in a mature receptor complex that includes p23. Upon binding of E2, ERα disassociates from the Hsp90 complex and is thus protected from CHIP-mediated degradation. Activation of E2-responsive target genes results in receptor ubiquitination and degradation in a transcription-coupled manner. OHT stabilizes receptor-DNA complexes and blocks both basal and transcription-coupled turnover. ERα-AA blocks transcription and transcription-coupled turnover by limiting p23-mediated receptor maturation. 3) SERD-mediated ERα degradation (middle pathway). The antiestrogen ICI stimulates ERα release from Hsp complexes and blocks receptor transactivation, sequestering ERα in the nuclear matrix, and triggering rapid receptor ubiquitination and degradation. Lysines 302/303 are required for ICI-induced polyubiquitination and turnover. Because ERα degradation is dependent on the 26S proteasome, the proteasome inhibitor MG132 blocks all receptor turnover pathways.

It is possible that lysine mutations resulted in a misfolded, unstable receptor. In the present study, it is not possible to determine whether ERα-AA is misfolded, because the mutant receptor favors the CHIP/proteasome-dependent pathway, which degrades both mature and misfolded receptors (18,19). However, we do not believe that ERα-AA was misfolded, because crystal structures are not possible in this region due to the flexible nature of the hinge region (58), suggesting mutation does not disrupt secondary protein structures. Additionally, Fuqua et al. (27) reported that a similar mutant, ERα-K303R, bound E2 and OHT with the same affinity as wtERα.

ERα lysines 302/303, located within the Hsp90/ER interface (31,33), may influence receptor stability by altering receptor-Hsp90-cochaperone interactions. Although both receptors associated similarly with Hsp90, coimmunoprecipitation analysis revealed an increased interaction of ERα-AA with CHIP and decreased interaction with p23, compared with wtERα. We have previously shown that Hsp90 inhibitor GA induces receptor association with CHIP and dissociation from p23, resulting in receptor ubiquitination and degradation by the 26S proteasome (19). These observations suggest that loss of lysines 302/303 and GA may promote receptor degradation through the same CHIP-mediated protein degradation pathway. In support of this notion, both ERα-AA turnover and GA induced wtERα could be blocked by OHT (Fig. 6), which interrupts ERα interaction with Hsp90/cochaperones.

The cochaperone Bag1 has been also found to associate with ERα (18), but to date, an association between Bag1 and ERα turnover has not been established. Bag1 functions as a nucleotide exchange factor (59) that may destabilize protein-Hsp complexes and promote delivery of ubiquitinated client proteins to the proteasome. The glucocorticoid receptor (GR), another Hsp70/90 client, is also ubiquitinated by CHIP (37), and after CHIP-mediated ubiquitination, Bag1 delivers GR to the proteasome (60). Similar to the results of the present study with ERα and CHIP, mutations in the Hsp90-interacting residues of GR likewise resulted in altered GR-Hsp90-cochaperone dynamics and receptors that were immune to GA-induced turnover (48,61,62). Therefore, the mechanism by which ERα is delivered to the proteasome is likely similar to Bag1-mediated delivery of GR. Our results further suggest that Bag1 promotes basal and GA-induced receptor degradation, because ERα-AA-Bag1 association increased after GA treatment, whereas Bag1 siRNA delayed receptor turnover (Figs. 4 and 5). As with GR, Bag1 may again cooperate with CHIP, delivering polyubiquitinated ERα to the proteasome through its proteasome-recognition domain (34,60). CHIP and Bag1 cooperation may therefore represent a common basal turnover mechanism for nuclear receptor degradation.

Although CHIP and Bag1 are found in early receptor complexes, the ERα cochaperone p23 is found in late/mature receptor complexes (36). p23 has been shown to enhance both basal and ligand-induced ERα transactivation (54) and also to compete with CHIP for Hsp90 binding (37). We found that wtERα was rapidly degraded upon p23 knockdown (Fig. 5C), suggesting that p23 exerts a stabilizing effect on the receptor. In addition, ERα-AA preferentially associated with CHIP and Bag1 and also had less affinity for p23 (Fig. 4). Knockdown of p23 expression decreased wtERα-mediated cathepsin D gene expression but not ERα-AA. In contrast, CHIP knockdown had a greater effect on ERα-AA-mediated gene expression (Fig. 8, E–G). These results suggest that p23 positively regulates ERα activity by stabilizing receptors, whereas CHIP limits ERα function by promoting receptor degradation.

Taken together, these data indicate that lysines 302/303 may encourage receptor association with p23, facilitating progression of ERα through the foldosome and increasing receptor transactivation potential. Numerous mutations that stabilize ERα in the presence of ligand also block E2-mediated receptor transactivation (63). Indeed, ERα-AA was stabilized in the presence of E2, and the mutant receptor was less transcriptionally competent than wtERα (Fig. 8). Alterations in the hinge region may reduce basal ERα-AA-mediated transactivation due to disruption of an ERα prototypical nuclear localization sequence (pNLS) located between K299 and K303 (53). We observed increased cytosolic retention of unstimulated ERα-AA, which may contribute to the low basal transcription activity observed in ERα-AA cells and the elevated ERα-AA interaction with cytosolic CHIP. Elevated basal ERα-AA degradation may also explain the discrepancy between ERα-AA mRNA expression and protein levels. In untreated cells, the level of ERα-AA mRNA in the clones was twice that of wtERα (Fig. 1). Because the half-life of apo-ERα-AA was significantly less than wtERα (Fig. 2), and ERα-AA displayed elevated polyubiquitination in the absence of ligand (Fig. 3), it is likely that ERα-AA clones maintained similar protein levels as wtERα clones due to rapid ERα-AA protein degradation.

Upon ligand binding, nuclear receptors dissociate from Hsp90-CHIP complexes and are directed toward alternative down-regulatory pathways (13,15,23). Treatment with E2 moves ERα toward a transcription-coupled degradation pathway (57). Concordantly, we observed increased polyubiquitination and turnover of wtERα after E2. In contrast to the wild-type receptor, ERα-AA was stabilized by E2. Although ERα-AA was more stable than wtERα after E2 treatment, E2-induced polyubiquitination of ERα-AA did not appear to be different from wtERα. The stabilization of ERα-AA by E2, without decreased polyubiquitination, may be due to its protection from rapid basal turnover observed in the unliganded condition (Fig. 2). Alternatively, K302/303 may be required for efficient E2-induced turnover, either by interacting with degradation machinery directly or by serving as sites of posttranslational modification that recruit degradation machinery. In fact, K302 and K303 have been reported to be sites for acetylation (64) and sumoylation (65), in addition to K302 monoubiquitination (28), so it is possible that altered receptor stability was due to loss of a posttranslational modification site. However, a recent report has shown ERα to be acetylated at lysines 266/268 and specifically not at lysines 302/303 (66). Because lysines 266/268 are also sumoylation sites (65), E2-induced monoubiquitination of lysines 302/303 by BRCA1/BARD1 (28) remains the likely signal for initiating E2-induced polyubiquitination and receptor turnover.

In contrast to E2 treatment, K302 and K303 appeared to play a significant role in ICI-induced receptor polyubiquitination (Fig. 7). ERα-AA was more stable than wtERα upon ICI treatment, and the mutant receptor had markedly diminished polyubiquitination. Because additional ERα-AA polyubiquitination did not occur in the presence of the antiestrogen, this raises the strong possibility that lysines 302/303 are ICI-induced polyubiquitination targets. This is the first report to identify lysines whose mutation results in altered ICI-induced receptor ubiquitination, providing insight into the mechanism of ICI action. Both ERα-AA and ERα-RR were resistant to ICI-induced polyubiquitination. Lysine mutation to alanine removes positive charges, whereas lysine mutation to arginine preserves positive charges. Because both ERα-AA and ERα-RR have a similar ubiquitination profile, we suggest that the charge of the residues is not responsible for directing receptor ubiquitination. Rather, it may be the loss of posttranslational modifications of these lysines that is responsible for the decrease in ICI-induced polyubiquitination, raising the possibility that lysines 302/303 are preferential ubiquitination sites in response to ICI. Furthermore, it has been recently shown that K302 is monoubiquitinated in the presence of ligand (28); it is very likely that additional K302 polyubiquitin attachment could occur in the presence of ICI, thus facilitating receptor degradation.

In conclusion, we propose that lysines 302/303 regulate basal ERα turnover pathways by preventing interaction with the cochaperones CHIP and Bag1 in the absence of ligand. We also report that K302 and K303 appear to function as polyubiquitination sites in the presence of ICI. These results reveal that K302 and K303 play a multifaceted role in regulating receptor stability and also highlight a previously undescribed role for these hinge-region lysines in the mechanism of ICI action. Using mass spectrometry, we are investigating which of the 29 ERα lysines are ubiquitinated during receptor degradation and attempting to identify the specific ubiquitin ligase(s) involved in these processes. It is well established that deregulation of ERα stability occurs in breast cancer cells. Consequently, understanding the role of receptor lysines in ERα turnover will aid in understanding the mechanisms of antiestrogen therapies and may also facilitate the development of novel ERα down-regulators.

MATERIALS AND METHODS

Materials

The following antibodies and reagents were used in this study: anti-ERα (HC-20; Santa Cruz Biotechnology, Inc., Santa Cruz, CA); anti-glyceraldehyde phosphate dehydrogenase (GAPDH) (Chemicon International, Temecula, CA); anti-HA (Roche, Indianapolis, IN); anti-p23 (Abcam, Cambridge, MA); anti-CHIP and anti-Bag1 (Affinity Bioreagents, Golden, CO); anti-Hsp90 (Stressgen, San Diego, CA); SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL); protease inhibitor cocktail set III and protein G-agarose beads (Calbiochem-Novabiochem, La Jolla, CA); Lipofectamine/PLUS reagents, G418, and cell culture reagents (Invitrogen, Carlsbad, CA); TrueBlot antimouse IgG beads (eBioscience, San Diego, CA); FuGENE6 and CAT-ELISA kit (Roche Applied Science); ICI (Tocris Cookson Ltd., Ellisville, MO); CHX, E2, GA, MG132, and OHT (Sigma Chemical Co., St. Louis, MO); passive lysis buffer and luciferase assay system (Promega, Madison, WI); fetal bovine serum and dextran-coated charcoal-stripped fetal bovine serum (Hyclone Laboratories, Inc., Logan, UT); cell culture supplementary reagents (Life Technologies, Inc., Rockville, MD); and siRNA and DharmaFect1 transfection reagent (Dharmacon, Lafayette, CO).

Plasmid Constructs

pcDNA-ERα and pcDNA-ERα-K302A, K303A constructs were kindly provided by Dr. H. Nakshatri (Indiana University School of Medicine). ERα lysines 302 and 303 within the pcDNA plasmid were changed to alanines by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit (Stratagene, Palo Alto, CA) to generate ERα-K302A, K303A. ERE-Vit-CAT, 2x-ERE-pS2-Luc, CMV-β-gal, HA-ubiquitin, and pBS/U6/CHIPi have been described previously (19).

Cell Lines

The human cervical carcinoma HeLa cell line and the breast cancer cell lines MCF7 and the ERα-negative MCF7-derived C4-12 cells (generously provided by Dr. W. Welshons, University of Missouri) are routinely maintained in our laboratory and have been described previously (23). Before all experiments involving transient transfection and/or hormone treatment, cells were cultured in hormone-free medium (phenol red-free MEM with 3% charcoal-stripped fetal bovine serum) for 2 d.

Stable Transfection of ERα

C4-12 cells were transfected with ERα constructs using Lipofectamine/PLUS reagent and exposed to antibiotic (G418; 0.8 mg/ml) for 3 wk. Multiple single G418-resistant clones were selected and expanded, and ERα levels were determined by immunoblot.

Protein Extraction, Coimmunoprecipitation, and Immunoblot

Soluble cell lysates were prepared in ER extraction lysis buffer [50 mm Tris (pH 7.4), 150 mm NaCl, 5 mm EDTA, 1% Triton X-100, ATPase inhibitors (1 mm Na₃O₄V, 25 mm NaF, 20 μm MoNa₂O₄)], and protease inhibitor cocktail set III. Receptor-chaperone complexes were immunoprecipitated with an ERα antibody (HC-20; Santa Cruz). Complexes were pelleted with antirabbit IgG agarose beads (TrueBlot; eBiosciences). Beads were washed in Tris-buffered saline with ATPase inhibitors and 1 mm phenylmethylsulfonyl fluoride (PMSF). Samples were boiled in 2× SDS loading buffer and proteins resolved by SDS-PAGE. Western blot was performed using antibodies specific for ERα, Hsp90, CHIP, Bag1, and p23. To prepare nuclear extracts, cells were resuspended in hypotonic buffer (20 mm HEPES, 0.5 mm MgCl₂, 0.5 mm dithiothreitol, 5 mm KCl, 2 mm CaCl₂, 8.55% sucrose, 1 mm PMSF) and cell membranes disrupted with in a Dounce homogenizer on ice (30 strokes using pestle B). Fractured cells were centrifuged at 2500 rpm for 10 min at 4 C. Nuclei pellets were washed twice with hypotonic buffer, and nuclear extracts were prepared with ER extraction lysis buffer.

Polyubiquitination Assays

HeLa cells were transiently transfected with ERα or ERα-AA and HA-tagged ubiquitin for 24 h using Lipofectamine/PLUS, according to manufacturer’s guidelines. Cells were pretreated with 25 μm MG132 for 1 h to block proteasome activity. Cells were then treated with DMSO, E2 (10 nm), or ICI (100 nm) for 4 h. After treatment, cells were lysed in ER extraction buffer, and 500 μg lysate was precleared with protein G-agarose for 30 min at 4 C and immunoprecipitated using anti-ERα antibody or IgG at 4 C overnight followed by addition of 30 μl protein G-agarose beads for 30 min. Beads were briefly centrifuged, washed three times with Tris-buffered saline with 0.1 m PMSF, and resuspended in 2× SDS loading buffer. Proteins were separated by electrophoresis and transferred to polyvinylidene difluoride membrane. Blots were probed for ubiquitinated ERα using anti-HA antibody.

RNA Interference (siRNA)

siRNA transfection reagent Dharmafect1 and SMARTpool siRNA targeting Bag1, p23, and scrambled siRNA were purchased from Dharmacon. Bag1 and p23 siRNA were transfected into C4-12 cells according to the manufacturer’s protocol. At 24 and 48 h, the medium was changed. Seventy-two hours after transfection, cells were pretreated with DMSO or CHX (25 μg/ml) and then treated with or without GA (1 μm) for the indicated times. Cells were lysed and Western blotting performed using specific antibodies. CHIP siRNA was generated by transfection of pBS/U6/CHIPi plasmid into HeLa cells using Lipofectamine/PLUS; pcDNA vector was used as nontargeting control, as described previously (19). Mock transfection was transfection reagent only.

Estrogen-Responsive Reporter Gene Assays

For luciferase assays, C4-12 cells were transfected with 250 ng 2xERE-ps2-Luc using Fugene. Twenty-four hours later, the medium was changed and cells treated with E2 (10⁻¹⁶ to 10⁻⁹ m) for 12 h. At the end of the experiment, cell lysates were prepared for reporter enzyme assays. Luciferase activity was determined using the Promega Luciferase Assay System. Luciferase activity was normalized to β-galactosidase activity as determined by the Galacto-Light Plus chemiluminescent reporter assay (Applied Biosystems, Foster City, CA). For estrogen-responsive CAT assays, C4-12 cells were transfected with 250 ng ERE-CAT for 24 h using Fugene. The medium was then changed and cells treated with vehicle (DMSO) or E2 (10 nm) for 48 h. Cell lysates were prepared and protein quantified using the Bio-Rad BCA Protein Assay Kit, and 100 μg total protein from each treatment group was used to determine CAT levels with the colorimetric Roche CAT ELISA kit. ERα expression, determined by immunoblot of vehicle-treated cells, was quantified and used to adjust CAT levels to account for any slight difference in stable clone ERα expression level and eliminate any possibility that elevated CAT levels were due to elevated ERα expression in a clone.

Quantitative Real-Time RT-PCR (RT-qPCR)

Total RNA was prepared by RNAeasy Mini Kit (QIAGEN, Valencia, CA), according to the protocol provided by the manufacturer. RNA (2 μg) was reverse transcribed in a total volume of 25 μl containing 400 U Moloney murine leukemia virus reverse transcriptase (New England Biolabs, Beverly, MA), 400 ng random hexamers (Promega), 80 U RNase inhibitor, and 1 mm deoxynucleotide triphosphates. The resulting cDNA was used in subsequent RT-qPCR performed in 20 μl Roche LightCycler Mix with 5 pmol forward and reverse primers for cathepsin D forward primer 5′-GTACATGATCCCCTGTGAGAAGGT-3′ and reverse primer, 5′-GGGACAGCTTGTAGCCTTTGC-3′ (5), TaqMan primers for EF1α forward primer 5′-CTGAACCATCCAGGCCAAAT-3′ and reverse primer 5′-GCCGTGTGGCAATCCAAT-3′, and EF1α TaqMan probe 5′-FAM-AGCGCCGGCTATGCCCCTG-TAMRA-3′. The relative concentration of mRNA was calculated using the ΔΔCt method according to Relative Quantitation of Gene Expression (Applied Biosystems) with EF1α mRNA as an internal control.

Quantification and Statistical Analysis

Films were quantified with ImageJ software (http://rsb.info.nih.gov/ij/). Statistical analyses were performed using Prism software. P values were determined by Student’s t test and ANOVA. EC₅₀ values were calculated using sigmoidal dose-response curve-fit analysis.