ERα Phosphorylation at Y537 by Src Triggers E6-AP-ERα Binding, ERα Ubiquitylation, Promoter Occupancy, and Target Gene Expression

Abstract

Many transcription factors undergo transcription-coupled proteolysis. Although ligand binding activates ubiquitin proteolysis of estrogen receptor α (ERα), mechanisms governing this and its relationship to transcriptional activation were unclear. Data presented link cross talk between the Src kinase and liganded ERα with ERα activation and its ubiquitylation. Liganded ERα rapidly activates and recruits Src, which phosphorylates ERα at tyrosine 537 (Y537). This enhances ERα binding to the ubiquitin ligase/ERα coactivator, E6-associated protein (E6-AP), stimulating ERα ubiquitylation, target gene activation, and ultimately ERα loss. ERα phosphorylation by Src promotes ERα ubiquitylation by E6-AP and proteasomal degradation in vitro. Src inhibition impairs estrogen (E2)-activated ERα:E6-AP binding, reducing ERα degradation. ERα-Y537F shows little E2-stimulated degradation and activates native ERα target genes poorly. Src activation enhances ERα and E6-AP binding and their occupancy at ERα target gene promoters to enhance transcription. Thus, ERαY537 phosphorylation drives ERα:E6-AP binding to at least a subset of target promoters, linking transcriptional activation to ERα degradation and providing a novel mechanism to fine tune ERα action. The observation that ERα transcriptional activity can be briskly maintained in a context of reduced ERα levels raises the possibility that hormonally sensitive tissues may not always show robust ERα protein levels.

Estrogen (E2) exerts biological functions on target tissues through binding to its intracellular receptor, the estrogen receptor (ER). Together, they play important roles in cell growth and differentiation, as well as in the progression of breast cancer (1). Two subtypes of ER, ERα and ERβ, have been identified in humans (2–4). Although present findings may have relevance to both ERα and ERβ, our investigations were directed exclusively to ERα and herein, ER refers to ERα. When bound by E2, ER dimerizes and recruits various coregulators to the enhancer regions of its target genes to modulate gene transcription. ER coregulators comprise different families of molecules with diverse functions. These include acetylation, methylation, ubiquitylation, and phosphorylation (5). Here we investigated mechanisms regulating ER binding to E6-associated protein (E6-AP), a coactivator with known function as an ubiquitin ligase.

The cellular levels of ER protein are delicately regulated (6). It has been known for many years that E2 binding to ER not only activates ER transactivation, but also leads to rapid ubiquitin-dependent ER proteolysis (7, 8). The mechanisms that link ER activation to its proteolysis, however, have remained obscure. The stability of ligand-bound ER varies depending on the characteristics of the ligands (9). Moreover, certain ubiquitin proteasome components have been identified as steroid hormone receptor coactivators. These include not only the ubiquitin ligase, E6-AP (10), noted above, but also MDM2 (11), and BRCA1 (12, 13), in addition to ubiquitin-conjugating enzymes, UbcH7 (14) and Ubc9 (15, 16) and the 19S proteasomal subunit, yeast suppressor of Gal 1/thyroid receptor interacting protein 1 (17).

The function of these ubiquitin proteasome components as ER coactivators is intriguing. It has been proposed that, for certain ER targets, proteasome components may not only regulate ligand activated ER-coactivator complex turnover at the target gene promoter, but that this cyclic receptor proteolysis may also be intimately linked to its ability to activate transcription (18–20). Paradoxically, proteasome inhibition decreases ER transcriptional activity at some target promoters, despite an increase in ER protein levels. Moreover, ER mutations that abrogate coactivator binding also abolish ligand-activated receptor proteolysis (18, 20).

Ubiquitin-dependent protein degradation is a highly regulated process. Ubiquitin-activating enzyme (UBA) activates ubiquitin by forming a thioester bond between a cysteine residue in UBA and the C terminus of ubiquitin; ubiquitin is then transferred to an ubiquitin-conjugating enzyme (UBC); finally, ubiquitin is transferred to a lysine residue in the target protein through one of the ubiquitin-protein ligases. This permits formation of a polyubiquitin chain and leads to the degradation of the polyubiquitinated target protein by the 26S proteasome (21, 22). Phosphorylation is often the signal for proteasome-dependent protein degradation. Many ubiquitin ligases bind only to appropriately phosphorylated substrates (23). Substrate phosphorylation is usually tightly regulated to ensure the proper timing and extent of its recognition by the ubiquitin ligase that mediates its proteolysis. Specific phosphorylation events that trigger proteasomal degradation have been identified for certain cell cycle regulators (24), and for the progesterone receptor (25) and the androgen receptor (26).

c-Src is a tyrosine kinase involved in regulation of cell proliferation and survival and is frequently activated in human cancers (27). In breast cancer cells, c-Src is recruited to phosphorylated human epidermal growth factor receptor 2 (HER2) or epidermal growth factor receptor 1 (EGFR1), promoting synergistic activation of Ras/Raf/MAPK kinase/MAPK and phosphatidylinositol-3- kinase (PI3K)/protein kinase B signaling pathways (28). c-Src is also activated by E2. Liganded ER rapidly and transiently activates c-Src, leading to Ras and MAPK activation (29, 30). c-Src levels and/or activity are frequently increased in primary breast cancers (31–34). Our prior work demonstrated an inverse correlation between c-Src activity and ER protein levels in primary human breast cancers (35).

ER is phosphorylated at a number of sites in a ligand-activated manner through interaction with different signaling transduction pathways (36–38). ER-Y537 is close to the helix-12 of ER, which is indispensable for interaction with coactivators. Y537 can be phosphorylated by c-Src in vitro (39–41), but whether this phosphorylation event occurs in vivo and its potential role have been controversial. The present study was undertaken to investigate whether ligand-activated ER phosphorylation may, for at least a subset of ER target genes, stimulate both its action at transcriptional targets and its subsequent degradation.

We report herein that ligand binding to ER stimulates ER phosphorylation at Y537 by c-Src kinase both in vitro and in cells. Y537 phosphorylation not only increases ER binding to the E3 ubiquitin ligase E6-AP, but also triggers ER ubiquitylation, and ER/E6-AP complex recruitment to certain target gene promoters to enhance ER transcriptional activity. These data link ligand-activated ERY537 phosphorylation to mechanisms leading to ER degradation, supporting the notion that certain promoters are regulated by transcription-coupled ER proteolysis.

Materials and Methods

Antibodies

Antibodies to ER (both polyclonal antibody HC-20 and monoclonal antibody F10), ERpY537, and E6-AP, were obtained from Santa Cruz Biotechnology (Santa Cruz, CA); to Src from Cell Signaling Technology (Danvers, MA); to phosphotyrosine (4G10) from Millipore Corp. (Bedford, MA); to β-actin from Sigma (St. Louis, MO). For detection of E6-AP-bound ER, E6-AP was precipitated and ER immunoblotted with ER monoclonal antibody F10. Ubiquitylation of cellular ER was detected by immunoprecipitation with antiubiquitin antibody P4D1 (Santa Cruz Biotechnology) followed by immunoblotting with anti-ER antibody HC-20 or by ER-precipitation with HC-20 followed by blotting with P4D1. ERpY537 was detected by immunoprecipitation with anti-ERpY537 and blotting with anti-ER.

Small interfering RNA (siRNA)-mediated knockdown of c-Src or E6-AP

siRNA oligonucleotides to either c-Src or E6-AP were obtained as ON-TARGETplus SMARTpool from Thermo Fisher Scientific (Lafayette, CO). Each was supplied as four different siRNA oligonucleotide sequences. siRNA or scrambled oligonucleotide controls were transfected per manufacturer's instructions after which cells were E2 deprived by transfer to 5% charcoal stripped FBS (cFBS) for 72 h before addition of E2 for either 4 or 9 h.

BT-20 Src knockdown cells

BT-20 cells were infected with lentivirus containing either shRNA to Src or scrambled shRNA (RHS4430-101028217 and RHS4346; Open Biosystems, Huntsville, AL). Infected cells were selected with puromycin 0.5 μg/ml for 3 d and maintained as a pool thereafter.

Cell culture

Cells were cultured and E2 deprived in medium containing 5% cFBS for 72 h before E2 treatment as described elsewhere (35). The generation of MCF-7-pIND-Src was previously described (35). The MCF/Src line is a variant of MCF-pIND-Src with constitutive expression of activated Src^Y527F. MDA-MB-231 cells were transfected using lipofectamine with either pcDNA3-ER or pcDNA3-ERY537F and clones selected in G418. ER or ER-Y537F expression was verified by Western blotting.

Cycloheximide (CHX) chase

MDA-MB-231/ER and MDA-MB-231/ER-Y537F lines were cultured in DMEM containing 5% cFBS for 2 d. Cells were treated with 10 nm E2 for 2.5 h before addition of 10 μm CHX. Lysates were collected at the indicated times for ER immunoblotting. The decay of ER protein over time was quantitated by densitometry.

In vitro Src kinase assay

Recombinant ER was either incubated with 10 ng recombinant Src kinase (both from Millipore), or mock treated for 15 min at 30 C in Src kinase buffer [25 mm Tris (pH 7.5), 50 mm NaCl, 10 mm MgCl₂, 5 mm ATP, and 0.1 mm dithiothreitol]. As an additional control, recombinant Src was pretreated with the Src inhibitor PP1 10 μm for 30 min (dead Src), before the kinase reaction.

ER ubiquitylation and degradation assay

Recombinant human his-tagged E6-AP was purified by nickel column chromatography as described elsewhere (42). Ubiquitin (Ub), E1 (UBA), UbcH7, and 26S proteasome fraction were from Boston Biochem (Boston, MA). For in vitro ER ubiquitylation, recombinant ER was either reacted with Src kinase, dead Src (see above), or mock treated as above after which Src was inactivated by addition of 10 μm PP1 for 15 min (Sigma). The ER (1 μm) was then reacted for 30 min at 37 C with 200 ng ubiquitin, 100 ng UBA, 200 ng UbcH7, 100 ng E6-AP in buffer containing 25 mm Tris (pH 7.5), 50 mm NaCl, 10 mm MgCl₂, 5 mm ATP, and 0.1 mm dithiothreitol. ER was then immunoprecipitated and complexes were resolved by SDS-PAGE and transferred to nitrocellulose. The blot was first probed with antiubiquitin antibody and then stripped and reprobed with anti-ER antibody. Tyrosine phosphorylation of the ER was detected by probing ER precipitates with antiphosphotyrosine antibody, and ER-bound E6-AP was detected by blotting. In vitro degradation assays were carried out as for the in vitro ubiquitylation assays except that 50 nm 26S proteasome fraction was added to the mixture, and reactions were incubated at 37 C for the times specified.

RNA extraction and real-time RT-PCR

The total RNA was isolated using RNeasy kit (QIAGEN, Valencia, CA) according to the manufacturer's protocol. Total RNA (1 μg) was used as template for the cDNA synthesis using iScript cDNA synthesis kit (Bio-Rad Laboratories, Hercules, CA). Real-time PCR was performed for growth regulation by estrogen in breast cancer 1 (GREB1) and 36B4 as described elsewhere (43).

Chromatin immunoprecipitation (ChIP) assay

MCF-7 cells and MDA-MB-231-ER-wild type (WT) or MDA-MB-231-ER-Y537F cells were estradiol deprived by culture in phenol red-free Improved MEM and DMEM, respectively, with 5% cFBS for 3 d and then treated with 10 nm E2 for specified times. ChIP assays were performed as described elsewhere (43).

Results

E2 stimulates c-Src activation and ERY537 phosphorylation with similar kinetics

E2 binding to the ER is known to rapidly and transiently activate c-Src (29). To investigate whether E2-mediated c-Src activation may lead to ER-Y537 phosphorylation in cells, we first assayed the kinetics of this phosphorylation event after hormone-deprived MCF-7 cells were stimulated with E2. Within 5–10 min of hormone addition, E2 rapidly increased c-Src activity (detected with anti-Src-pY416), which peaked within 30–60 min, while total c-Src remained constant (Fig. 1A, top panel). ER-pY537, detected by immunoprecipitation with anti-ERpY537 antibody and blotting with anti-ER, was not present in E2-deprived cells but appeared within 5 min after E2 addition and peaked within 60 min (Fig. 1A).

Fig. 1.

ER phosphorylation at Y537 by E2-activated c-Src mediates ER degradation. A, MCF-7 cells were E2 deprived by growth in 5% cFBS for 72 h and then treated with 10 nm E2. Cell lysates were collected at the times indicated. Total ER, Src pY416, total Src, and β-actin were detected by immunoblot (top panel). ERpY537 was detected by immunoprecipitation (IP) with ERpY537 antibody followed by blotting with anti-ER (bottom panel). B, MDA-MB-231/ER and MDA-MB-231/ER-Y537F cells were E2 deprived in 5% cFBS for 48 h and then treated with 10 nm E2 or 10 μm PP1 plus 10 nm E2 for 30 min. ER and β-actin were assayed by immunoblotting (top panel). ERpY537 was precipitated, resolved, and detected by blotting for ER as in panel A. ERpY537 was barely detected in E2-deprived cells and increased notably 30 min after E2 addition, and this was attenuated by PP1. Data are representative of more than three experiments (bottom panel). C, MDA-MB-231/ER and MDA-MB-231/ER-Y537F cells were E2 deprived as in panel A and then treated with 10 nm E2 for 2.5 h before the addition of 10 mm CHX. ER was blotted at indicated times after CHX addition. D, Total Src was detected by Western from BT-20 control cells (C) and BT20shSrc cells (shSrc). Cells were E2 deprived and then treated with 10 nm E2 for indicated times. Total lysates (500 μg) were used to immunoprecipitate ER with HC20 antibody, followed by blotting with anti-ER (F10).

E2-stimulated ER phosphorylation at Y537 is c-Src dependent

To further assay the relationship between ERY537 phosphorylation and ER proteolysis, effects of E2 were compared in MDA-MB-231 human breast cancer lines stably expressing ER-WT or the mutant ER-Y537F. Cells were E2 deprived for 48 h and then treated with E2 for 30 min. ER and β-actin loading controls were assayed by Western blot (Fig. 1B, top). ERpY537, detected by anti-ERpY537 precipitation followed by immunoblotting for total ER is in Fig. 1B (bottom panel). ER-Y537 phosphorylation was minimal in E2-deprived MDA-MB-231-ER-WT and increased markedly within 30 min of E2 treatment, before detectable ER loss in MDA-MB-231-ER-WT, but was not detected in cells expressing ERY537F (Fig. 1B). Pretreatment of MDA-MB-231-ER-WT with the Src kinase inhibitor PP1 decreased E2-stimulated ERpY537 accumulation to 60% based on densitometry (Fig. 1B, bottom panel). The incomplete inhibition of ERpY537 by PP1, which is consistent in our assays, may reflect that the surge of Src activity stimulated by E2 cannot be completely blocked by PP1 and/or that other kinases which are not blocked by PP1 also contribute to phosphorylation at this Y537 site.

Loss of Y537 phosphorylation abolishes E2-stimulated ER proteolysis

To further investigate the requirement for ER-Y537 phosphorylation in E2-driven ER proteolysis, effects of E2 treatment after E2 deprivation on ER stability were compared in MDA-MB-231 cells bearing either ER-WT or ER-Y537F. After the addition of E2 for 2.5 h, CHX was added to the culture of cells to start the chase for ER degradation. With proteins equal at time zero, the E2-stimulated decay of ER-WT is appreciably faster than that of the nonphosphorylatable ER-Y537F. ER-WT was rapidly degraded with a calculated half-life of 3 h, whereas under the same conditions, ER-Y537F was stable. Its level was essentially unchanged 4 h after E2 stimulation (Fig. 1C). Thus, the Y537F mutation abrogated E2-stimulated ER loss. Notably, both proteins were very stable in the absence of E2 (data not shown).

Loss of Src stabilizes ER

BT-20 cells have low ER protein levels, high c-Src activity, and rapid ER turnover (35). We established BT-20shSrc using lentivirus vector to knock down Src expression (Fig. 1D). BT-20shControl cells (C) and BT-20shSrc (shSrc) cells were treated with E2 after E2 deprivation. Cell lysates were collected and ER was assayed by IP-blot. ER levels rapidly decreased 2 h after E2 stimulation in BT-20shControl cells but remained stable even after 4 h of E2 stimulation in BT-20shSrc cells (Fig. 1D). Thus loss of Src dramatically attenuated E2-stimulated ER loss.

Src kinase action promotes ER ubiquitylation

We next tested the effects of c-Src inhibition on E2-stimulated ER ubiquitylation in an MCF-7 line that overexpresses constitutively activated Src, MCF-7/Src. E2-deprived MCF-7/Src cells were stimulated with E2 with or without prior addition of either the Src inhibitor, PP1, or MG132. E2 stimulated the rapid appearance of an ubiquitylated ER with mobility compatible with monoubiquitylation. This monoubiquitylated ER was detected within 1 h, before the reduction in steady-state ER levels. Src inhibition reduced this E2-stimulated ER ubiquitylation, whereas proteasome inhibition by MG132 increased ubiquitylated ER levels (Fig. 2A). Within 4 h of E2 treatment, total ER levels were appreciably reduced and ER polyubiquitylation was readily detected. Src inhibition impaired both E2-stimulated ER polyubiquitylation and ER loss. MG132 blocked ER degradation and polyubiquitylated ER accumulated (Fig. 2B). Although polyubiquitylation of the ER was detected by 1 h after E2 addition (data not shown), it was more prominent at 4 h thereafter. Thus, detection of ER monoubiquitylation increases rapidly after E2 addition and precedes peak ER-polyubiquitylation. Both are activated by Src and both accumulate after proteasome inhibition.

Fig. 2.

Src promotes ER ubiquitylation and degradation. A, MCF-7/Src cells were E2 deprived by growth in 5% cFBS for 72 h before addition of either 10 nm E2 alone, 10 μm PP1 plus 10 nm E2, or 10 μm MG132 plus 10 nm E2 for 1 h. Total ER was detected by immunoblotting. Monoubiquitinated ER was detected by precipitation with antiubiquitin antibody followed by blotting with anti-ER (F10). The relative mobilities of total and monoubiquitinated ER are indicated. ERpY537 was detected by immunoprecipited (IP) anti-ERpY537 and blotting ER as described. B, MCF-7/Src cells were treated as in panel A, but recovered at 4 h after addition of E2. Polyubiquitinated ER was detected by immunoprecipitation with anti-ER (HC20) followed by immunoblotting with antiubiquitin. Corresponding ER levels are shown. C and D, Knockdown of cSrc and E6-AP impair E2-stimulated ER loss. MCF-7 cells were transfected with siRNA oligonucleotides for either cSrc or E6-AP or with appropriate scrambled siRNA controls (Control). Cells were then E2 deprived for 72 h before addition of 10 nm E2 for the indicated times. C, Knockdown was verified by immunoblotting at time 0 and 9 h. D, ER levels were detected by immunoblot in E2-deprived cells and at 4 and 9 h after E2 repletion. GAPDH, Glyceraldehyde-3-phosphate dehydrogenase; Ub, ubiquitinated.

Knockdown of either c-Src or E6-AP reduces E2-stimulated ER loss

E6-AP is an ubiquitin ligase known to function as an ER coactivator (10). To further evaluate the role of Src and test the requirement for E6-AP in hormone-mediated ER loss, cells were transiently transfected with siRNA to either c-Src or E6-AP or scrambled oligonucleotide controls for 12 h and then E2 deprived for 72 h and restimulated with E2. Within 4 and 9 h after E2 addition, ER loss was apparent. Knockdown of cSrc and of E6-AP each attenuated E2-driven ER loss (see Fig. 2, C and D).

Src increases ligand-activated ER:E6-AP interaction

Although MCF-7/Src cells have a only modest increase in total Src protein compared with parental, this Src mutant is constitutively active. The strong Src activation in MCF-7/Src cells was shown by increased total protein tyrosine phosphorylation compared with MCF-7 cells (Fig. 3A). ERpY537 was not detected in E2-deprived MCF-7 or MCF-7/Src cells. E2 addition to E2-deprived cells rapidly increased ERpY537 in both cell lines, and ERpY537 was greater within 1 h in MCF-7/Src cells than in MCF-7 cells and modestly increased by proteasome inhibition (Fig. 3B). Because knockdown of E6-AP attenuated E2-driven ER loss (Fig. 2C), we tested the effects of E2 and Src on ER:E6-AP binding. In MCF-7 cells, E2 rapidly increased E6-AP-bound ER, and this was further enhanced by proteasome inhibitor MG132. This effect was even greater in MCF-7/Src cells, suggesting that Src stimulates E2-driven ER:E6-AP interaction (Fig. 3B).

Fig. 3.

Phosphorylation at Y537 increases ER interaction with E6-AP. A, MCF-7 and MCF-7/Src immunoblots show tyrosine-phosphorylated protein (reacted with antiphosphotyrosine antibody, pY), Src, and β-actin. B, MCF-7 and MCF-7/Src cells were E2 deprived before treatment with 10 nm E2, or 10 μm MG132 plus E2 for 1 h. Total ER and ERpY537 were detected as described. E6-AP was immunoprecipitated (IP), complexes were resolved by SDS-PAGE, and total E6-AP and E6-AP-bound ER were detected by immunoblotting. C, MDA-MB-231/ER and MDA-MB-231/ER-Y537F cells were E2 deprived before treatment with 10 nm E2, or 10 μm MG132 plus E2 for 30 min. Total E6-AP, ER, and β-actin levels are shown. E6-AP precipitates were resolved and blotted with ER antibody to show E6-AP-bound ER.

To test further whether ER-Y537 phosphorylation may prime the ER for E6-AP binding, the effect of E2 on ER:E6-AP coprecipitation was assayed in MDA-MB-231 cells that express either ER-WT or ER-Y537F. Immunoprecipitation of E6-AP 30 min after E2 stimulation of MDA-MB-231/ER cells and MDA-MB-231/ER-Y537F cells revealed detectable E6-AP-bound ER only in ER-WT-expressing cells (Fig. 3C). Even when E6-AP was immunoprecipitated from more MDA-MB-231/ER-Y537F lysate than from ER-WT-expressing cells, E6-AP-bound ER-Y537F was minimally detected compared with ER-WT (data not shown). Thus, ER-Y537F shows little detectable E2-stimulated E6-AP binding compared with ER-WT.

Src phosphorylates ER and accelerates E6-AP-mediated ER proteolysis in vitro

To test directly whether E6-AP-mediated ER proteolysis is stimulated by Src-dependent ER phosphorylation, the following in vitro reactions were performed. First, recombinant ER was reacted with recombinant Src kinase in vitro. Src kinase promoted ER tyrosine phosphorylation. Notably, this was further enhanced by prior addition of E2, suggesting a conformation dependence of Src action mediated by E2 binding to the ER (Fig. 4A). ER tyrosine phosphorylation was not detected when Src kinase was omitted from the reaction or when Src kinase was inactivated (dead Src) before its reaction with ER.

Fig. 4.

E6-AP-mediated ER ubiquitylation and degradation in vitro are enhanced by ER phosphorylation by Src. A, Recombinant ER was reacted with or without Src kinase, in the presence or absence of E2. In control reactions, Src was inhibited by PP1 for 30 min (dead Src) before addition to ER. ER was precipitated, resolved by SDS-PAGE, and blotted with antiphosphotyrosine antibody (ERpY) or for total ER. B, Recombinant ER protein was reacted with 1) UBA alone; 2) UBA and UbcH7; 3) UBA, UbcH7, and E6-AP; and 4 and 5) with UBA, UbcH7, and E6-AP after prior treatment of ER with 4) heat-inactivated Src or 5) active Src as described in Materials and Methods. ER was precipitated and complexes were resolved and immunoblotted to show total ER, ubiquitinated ER (Ub-ER), tyrosine phosphorylated ER (ERpY), and ER-bound Src and E6-AP. E6-AP and Src inputs are shown below. C, For in vitro ER degradation, reactions were carried out as in panel B with addition of 26S proteasome fraction and incubated for 45 min. ER immunoblotting shows that ER degradation is accelerated by pretreatment with Src in lane 5. IP, Immunoprecipitation.

ER ubiquitylation was next assayed in vitro. ER ubiquitylation was not detected in control reactions containing recombinant ER alone or ER mixed with ubiquitin-activating enzyme (UBA) (Fig 4B, lane 1) or UBA plus ubiquitin-conjugating enzyme (UbcH7) (Fig 4B, lane 2). When E6-AP was added to the reaction, ER polyubiquitylation was observed (Fig. 4B, lane 3). Pretreatment of ER with inactive Src kinase had no additional effect (Fig. 4B, lane 4), whereas pretreatment with Src kinase further increased ER polyubiquitylation (Fig. 4B, lane 5). ER binding to E6-AP in vitro was enhanced by prior tyrosine phosphorylation of the ER by Src (see ERpY in Fig. 4B, lane 5). Active Src was detected in ER precipitates. In contrast, inactive Src did not stably bind ER, failed to mediate tyrosine phosphorylation of the ER, and did not increase ER:E6-AP binding.

Finally, 26S proteasome fraction was added to the in vitro reactions described in Fig. 4B to permit ER degradation. ER degradation was detected within 45 min only in the reaction in which E6-AP was added and ER was pretreated with active Src (Fig. 4C, lane 5). This enhancement of ER proteolysis is consistent with the increased ER ubiquitylation observed under these conditions in Fig. 4B, lane 5.

Src kinase regulates ER transcriptional activity

E2-deprived MCF-7 and MCF-7/Src cells were treated with E2, and the expression of cellular ER-target genes was compared. E2 treatment for 3 h and 6 h induced pS2 and GREB1 expression in both cell lines, with greater induction in MCF-7/Src cells than in MCF-7 cells (Fig. 5, A and B). Src activation did not appreciably increase either pS2 or GREB1 expression in the absence of E2 in MCF-7/Src. Thus, Src activation enhances E2-mediated ER action on both pS2 and GREB1 expression.

Fig. 5.

Src enhances ER target gene transcription. A and B, MCF-7 and MCF-7/Src cells were E2 deprived in 5% cFBS for 72 h and then treated with 10 nm E2 for 3 or 6 h. pS2 (panel A) and GREB1 (panel B) mRNA levels were quantified by real-time PCR. Data from at least three experiments are presented as fold increase over time zero values for MCF-7 (mean ± sem). C and D, MDA-MB-231/ER and MDA-MB-231/ER-Y537F cells were E2 deprived and then treated with 10 nm E2 for 6, 12, and 24 h. pS2 (panel C) and GREB1 (panel D) mRNA expression was by real-time PCR. E and F, MCF-7 and MCF-7/Src cells were E2 deprived and then treated with 10 nm E2 for 75 and 135 min. E, ER binding to the GREB1 promoter ERE site was assayed by ChIP. Data from at least three experiments are presented as percent of input bound (mean ± sem). F, E6-AP binding to the GREB1 promoter ERE site was assayed by ChIP (mean relative binding over MCF-7, time zero control ± sem). G, MDA-MB-231/ER and MDA-MB-231/ER-Y537F cells were E2 deprived and then treated with E2 for 45, 75, and 105 min. Mean percent ER binding (±sem) to the GREB1 promoter was assayed as in panel E.

Next we investigated how loss of the potential to phosphorylate ER at Y537 would affect ER-dependent cellular gene expression. In MDA-MB-231-ER-WT, expression of both GREB1 and pS2 was induced by E2 (Fig. 5, panels C and D, respectively), as observed in MCF-7 cells. E2-induced expression of both of these genes was markedly attenuated in MDA-MB-231/ER-Y537F cells, suggesting that ER-Y537F is defective in E2-activated transactivation of both of these cellular ER target genes (Fig. 5, B and C).

The influence of Src on ER and E6-AP binding to the GREB1 estrogen response element (ERE) was assayed. E2-activated ER binding to the GREB1 promoter ERE motif was enhanced in MCF-7/Src compared with MCF-7 cells. At 75 and 135 min after E2 addition, ER levels were as yet unchanged, but ER binding to the GREB1 promoter was greater in MCF-7/Src than in MCF-7 cells (Fig. 5E). E6-AP binding to the GREB1 promoter was also assayed. In MCF-7 cells, very little E2 induced E6-AP binding was detectable at the GREB1 promoter; in contrast, constitutive Src expression enhanced detection of E6-AP binding to the GREB1 promoter ERE in MCF-7/Src cells (Fig. 5F).

The binding of ER-WT and ER-Y537F to the GREB1 promoter ERE was compared using ChIP. In MDA-MB-231 cells, E2 induced strong ER-WT binding to the GREB1 ERE, whereas ER-Y537F showed very little binding to the GREB1 promoter within the time frame of this assay (Fig. 5G).

Discussion

Activation of many transcription factors, including c-Myc and Fos, is linked both to factor ubiquitylation and proteolysis (44, 45). Ubiquitylation may be required for factor activation and may serve to localize the receptor to promoter sites in the nucleus and nucleate formation of coactivator complexes (44, 45). Coactivators can enhance ubiquitylation of certain transcription factors (46). For many steroid hormone receptors (SR), including the ER (8), thyroid hormone receptor (47), progesterone receptor (48, 49), and most recently the androgen receptor (26), ligand binding not only activates a chain of events leading to transcriptional activity, but also predicates receptor ubiquitylation and proteolysis. Increasing data link the ubiquitin proteasome and ligand-activated SR activation. SR coactivators include ubiquitin-conjugating enzymes, ubiquitin ligases, and proteasome components (49). It is noteworthy in this regard that proteasome inhibition impairs ER transcriptional activity despite an increase in receptor levels (20). Despite this, and the observation that ubiquitylated ER and ubiquitin ligase coactivators cycle on and off ERE sites in a manner regulated by ligand binding (19), little was known of mechanisms linking ER activation with its proteolytic degradation. Indeed the coupling of these events has been controversial (50).

Rapid, ligand activated ER proteolysis, although linked to ER transcriptional activation, has been considered a nongenomic ER action. The present data indicate that this process is intimately linked with receptor transcriptional activation. Ligand binding mediates rapid, transient ER cross talk with several mitogenic pathways, leading to ER phosphorylation at multiple sites by kinases including Src, MAPK, and PI3K/AKT, which modulate ER transcriptional activity (30, 36–38, 51–55). Whether and how kinase-mediated ER phosphorylation could modulate coactivator and ERE binding and ER proteolysis is not fully understood. The ubiquitin-dependent degradation of many protein substrates is governed by their timed phosphorylation, which promotes their binding to an ubiquitin ligase (24). The present data indicate that E2-bound ER recruits Src, which phosphorylates ER at Y537. This in turn, promotes ER binding by the ubiquitin ligase E6-AP, promotes recruitment of this complex to ERE-bearing genes to increase ER transcriptional activity, and also triggers a mechanism that ultimately drives ER loss. Although both MAPK and Akt are known to phosphorylate ER, in our MCF-7 model neither MAPK kinase nor PI3K inhibition impaired E2-driven ER degradation (35).

There are 23 tyrosine residues in the full-length human ER. Whereas multiple tyrosine sites in ER could potentially be phosphorylated by c-Src, in vitro Src kinase reactions generate phosphorylation of, on average, only two tyrosine sites per ER molecule, and Y537 is one of these sites (40). Two N-terminal tyrosines, Y52 and Y219, can also be phosphorylated in vitro (56). Using a phosphorylation site prediction program (57), we found Y537 to be the single site most likely to be phosphorylated by Src among five tyrosine residues in the ER ligand-binding domain, based on the ligand-binding domain structure of estradiol-bound ER. Phosphorylation of this tyrosine, positioned at the hinge of helix-12, could potentially affect coactivator binding. Other tyrosine phosphorylation sites in ER may work together with ERpY537 to affect ER degradation and transactivation activity, and warrant further investigation. We observed that ERpY537 is not detected in E2-starved cells, and Src activation and the appearance of ERY537 phosphorylation share similar kinetics after E2 stimulation. That PP1 pretreatment did not abolish detection of ERpY537 may reflect incomplete drug-mediated Src inhibition or the existence of other kinases acting at this site. In vitro, both ER phosphorylation by Src and ER-Src coprecipitation were enhanced by addition of E2. Ligand binding may induce a conformational change favoring receptor interaction with this kinase.

ER phosphorylation at Y537 proved critical for E2-stimulated ER degradation. The ER-Y537F mutant failed to undergo ligand-activated proteolysis. In MCF-7 cells, both Src inhibitors and siRNA-mediated Src knockdown reduced, but did not fully abrogate, ligand-activated ubiquitylation and loss of cellular ER. This may reflect incomplete Src inhibition or knockdown, or the involvement of other Src kinase family members in Y537 phosphorylation. It is noteworthy that E2-stimulated ER monoubiquitylation was detected before peak ER polyubiquitylation, and both were impaired by prior Src inactivation. ER monoubiquitylation may serve to nucleate the receptor-coactivator complex, driving subsequent events including activation, polyubiquitylation, and receptor degradation.

A number of different ubiquitin ligases have been shown to act as ER coactivators, including E6-AP, BRCA1, and MDM2 (58–60). Transgenic E6-AP overexpression in the murine mammary gland decreases ER protein levels. Moreover, E6-AP knockout animals show increased mammary gland ER levels and mammary hyperplasia, raising the possibility that E6-AP acts not only as an ER coactivator, but also regulates ER levels in vivo (42). The present work provides both cellular and molecular evidence linking ligand activation of the ER to E6-AP recruitment, transcriptional activation, and receptor proteolysis. ER phosphorylation by Src enhanced subsequent E6-AP-mediated ER ubiquitylation and degradation in vitro. Moreover, ligand-activated coprecipitation of endogenous cellular ER:E6-AP was enhanced by Src-mediated phosphorylation of ER at Y537, which increased ER ubiquitylation and its proteasomal degradation.

ER phosphorylation of at Y537 regulates both receptor stability and transcriptional activity. Although early studies showed the transactivation activity of transfected ER-Y537F was similar to that of WT-ER in ERE reporter gene assays, the greater stability of ER-Y537F was not appreciated (61, 62). In contrast, we found that ER-Y537F was defective in activating cellular ER target genes including GREB1 and pS2. Our finding that ER-Y537F is more stable than WT-ER may explain the apparent discrepancy between the present data and earlier findings. Additionally, in transient transfection assays, the high levels of overexpressed WT-ER and ER-Y537F could mask the importance of phosphorylation to ER-coactivator interaction. It is noteworthy that a progesterone receptor phosphomutant also shows divergent transactivation activities in reporter-based transfection assays compared with action at native target genes in stably transfected cells (63). Despite its stability, little ER-Y537F was recruited to the GREB1 promoter compared with WT-ER after E2 stimulation, suggesting that ERY537 phosphorylation promotes not only E6-AP binding, but also complex recruitment to this chromosomal ERE promoter site. Src activation also led to higher ER occupancy at the ER target gene promoter, resulting in higher ER target gene transcription despite more rapid ER degradation after E2 stimulation.

Although ligand binding has been known to mediate ER ubiquitylation and receptor proteolysis, the role of the ubiquitin proteasome in receptor-mediated transcriptional activity has been unclear. For certain promoters, proteasomal inhibition causes loss of coactivator binding, loss of transcriptional activity (20), and dissociation of the receptor from ERE sites (19). In contrast, others found that proteasome inhibitors increased ER action on some promoters but impair it at others (50, 64). The ubiquitin proteasome may modulate receptor-mediated transcriptional activity in a number of ways (49). It may serve to limit receptor action, fine tuning, or titrating the receptor in the presence of ligand. In addition, receptor turnover may be needed for ER-coactivator complex remodeling to allow recharging the promoter once fired. Finally, for certain promoter sites, a continuous cycle of phosphorylation and ubiquitin-dependent receptor degradation may be needed for ongoing transcriptional activity. The proteasomal remodeling of the ER-coactivator complex may be required to disrupt the preinitiation complex to permit exchange of coactivators and activate elongation and ultimately RNA processing.

Although ER ubiquitylation was first observed more than 20 yr ago (7, 65), how E2 regulates this and how it might be driven by receptor cross talk with mitogenic kinases and ER phosphorylation were not known. The present data link E2 and mitogen-mediated Src activation to ER transactivation and raise the possibility that these may be coupled to receptor proteolysis for at least a subset of ER target genes. How Y537 phosphorylation may modulate ER interaction with other ubiquitin ligase ER-coactivators, global ER-target gene promoter selection, and transcriptional activation or repression warrants further investigation. Some ER coactivators are also degraded through the ubiquitin-proteasome pathway (66). Down-regulation of the ER coactivator, amplified in breast cancer (AIB1), also known as steroid receptor coactivator-3 (SRC-3), has been shown to reduce ER transcriptional activity and increase ER stability (18, 67). It will be of interest to determine whether ERY537 phosphorylation influences SRC-3 stability. It was recently demonstrated that ligand activation of the androgen receptor is also followed by phosphorylation at S515, which promotes recruitment of an E3 ligase, MDM2, and receptor ubiquitylation. This is required for efficient AR-mediated transcriptional activation (26).

Here we show that ER transcriptional activity depends not only on ER protein levels, but also on its phosphorylation status at Y537. E2- and Src-stimulated ER Y537 phosphorylation mediates not only efficient interaction with the ubiquitin ligase E6-AP, but also binding to target chromosome sites to activate gene transcription, providing a molecular basis for the link between ER transactivation and degradation (modeled in Fig. 6). Our observation that ER transcriptional activity can be briskly maintained in a context of declining ER levels raises the possibility that hormonally sensitive tissues may not always show robust ER protein levels. Similar mechanisms might also apply to other nuclear receptor family members.

Fig. 6.

ER action model. This schematic provides a model whereby ligand activation of the ER leads to rapid, transient recruitment of c-Src, which phosphorylates the ER at Y537. This phosphorylation event promotes E6-AP binding to ERpY537 to enhance receptor/coactivator recruitment to certain target gene promoters, transcriptional activation, and subsequent receptor proteolysis. Activated receptor tyrosine kinases (Her2, EFGR, IGF-1R) also activate c-Src and may potentiate the process. EGFR, Epidermal growth factor receptor; IGF-1R, IGF-1 receptor.

Constitutive Src activation is frequently observed in human cancers and promotes tumor cell survival, proliferation, invasion, and metastasis (32, 35). The present data support a novel role for Src in promoting the oncogenic effects of E2 in human breast cancers. The therapeutic utility of Src inhibitors, currently in clinical trials for breast and other hormonally regulated cancers, may be due, in significant part, to attenuation of ligand-activated ER action.

NURSA Molecule Pages^†:

• Nuclear Receptors: ER-α;

• Coregulators: E6AP;

• Ligands: 17β-estradiol.