Report

A Transmembrane Intracellular Estrogen Receptor Mediates Rapid Cell Signaling

See allHide authors and affiliations

Science  11 Mar 2005:
Vol. 307, Issue 5715, pp. 1625-1630
DOI: 10.1126/science.1106943

Abstract

The steroid hormone estrogen regulates many functionally unrelated processes in numerous tissues. Although it is traditionally thought to control transcriptional activation through the classical nuclear estrogen receptors, it also initiates many rapid nongenomic signaling events. We found that of all G protein–coupled receptors characterized to date, GPR30 is uniquely localized to the endoplasmic reticulum, where it specifically binds estrogen and fluorescent estrogen derivatives. Activating GPR30 by estrogen resulted in intracellular calcium mobilization and synthesis of phosphatidylinositol 3,4,5-trisphosphate in the nucleus. Thus, GPR30 represents an intracellular transmembrane estrogen receptor that may contribute to normal estrogen physiology as well as pathophysiology.

Estrogen (17β-estradiol, E2) represents one of a family of steroid hormones that act through soluble intracellular receptors. Once activated, these receptors translocate to the nucleus, where they function as ligand-dependent transcription factors (1, 2). This mode of action of two such estrogen-binding receptors, ERα and ERβ, is reasonably well understood (3, 4). However, the existence of functional ERs associated with the plasma membrane has been debated (5). It has been suggested that such membrane receptors mediate the rapid nongenomic signaling events widely observed following stimulation of cells and tissues with estrogen, including the generation of the second messengers Ca2+ and nitric oxide as well as the activation of receptor tyrosine kinase and protein-lipid kinase pathways (1, 69). The cellular consequences can include adhesion, migration, survival, proliferation, and cancer. Novel receptors and novel forms of ER have been postulated to mediate many of these signal transduction events (8).

G protein–coupled receptors (GPCRs) represent the largest class of signaling molecules in the human genome (10). They are heptahelical transmembrane proteins (11) expressed on the cell surface, where the binding of agonists (12) initiates activation of heterotrimeric guanine nucleotide-binding proteins (G proteins) and downstream regulatory proteins (13) and effectors (14). It has been suggested that rapid nongenomic estrogen-mediated signaling may involve the GPCR, GPR30 (15) based on correlations of receptor expression with estrogen-mediated Erk activation (16), and cellular estrogen binding (17). Despite these assertions that a 7-transmembrane GPCR may be involved in estrogen-dependent signaling, no mechanisms have been demonstrated to explain the function of GPR30 in these processes.

To define the role of GPR30 in estrogen-mediated cell activation, we expressed GPR30 as a fusion protein with green fluorescent protein (GFP) in COS7 (monkey kidney fibroblast) cells (18). To examine the cellular localization of the receptor, GPR30-GFP was expressed with the β2-adrenergic receptor (β2-AR), a GPCR (Fig. 1A). Confocal fluorescence microscopy revealed that the β2-AR was localized to the plasma membrane, whereas GPR30-GFP was observed in an intracellular tubuloreticular network, including the nuclear envelope in some cases. To identify the specific subcellular localization of GPR30-GFP, we performed localization studies with markers for mitochondria, the endoplasmic reticulum, the Golgi apparatus, and the actin cytoskeleton. Staining of GPR30-GFP–expressing cells with either an antibody to KDEL (an endoplasmic reticulum retrieval sequence) or er-YFP (a subcellular localization marker targeting YFP to the endoplasmic reticulum using calreticulin and KDEL sequences) yielded nearly complete colocalization with GPR30-GFP. Coexpression of TGN38-GFP (a Golgi-resident protein) with GPR30-monomeric red fluorescent protein1 (mRFP1) also demonstrated colocalization. No colocalization was observed between GPR30-GFP and mitochondria or the actin cytoskeleton.

Fig. 1.

Localization of GPR30 in the endoplasmic reticulum. (A) Confocal images of GPR30-GFP (or mRFP1) coexpressed with the β2-AR-mRFP1, an endoplasmic reticulum–targeted yellow fluorescent protein (er-YFP), or the Golgi marker TGN38-GFP in COS7 cells. The endoplasmic reticulum was also stained with an antibody to KDEL. Mitochondria and the actin cytoskeleton were visualized with Mito-Tracker Red CMXRos (Molecular Probes, Eugene, OR) and rhodamine-phalloidin, respectively. (B and C) Confocal images of full-length GPR30-GFP (B) or N-terminally FLAG-tagged, C-terminally truncated GPR30 (amino acids 1 to 349) expressed in COS7 cells (C). Antibodies to N-terminal GPR30, C-terminal GPR30, and M2 FLAG were used to stain both of the transfected cells. (D) N-terminally FLAG-tagged, C-terminally truncated GPR30 was expressed in Hec50 cells, which express endogenous GPR30, and stained with antibodies to C-terminal GPR30 and M2 FLAG. (E) Confocal images of cell lines stained with antibody to C-terminal GPR30 to determine endogenous GPR30 expression levels. All images were acquired with identical instrument settings at one sitting. (F) HTR8 (green) and JEG (red) cells were stained with antibody to C-terminal GPR30 (solid line) or secondary antibody alone (dotted line) and analyzed by flow cytometry for fluorescence intensity. HTR8 cells yield a 1.5-fold increase in fluorescence intensity above background, whereas JEG cells demonstrate a 15-fold increase. Data are representative of three independent experiments. In images where the location of the nucleus is not clear, the merged image includes the TO-PRO-3 staining of the nucleus (blue).

Because heterologous expression of GPCRs can result in retention of the receptor in the endoplasmic reticulum and Golgi, we examine the expression of endogenous GPR30. Antibodies were generated against the predicted N- and C-terminal peptides of GPR30, both of which recognized GPR30-GFP expressed in COS7 cells (Fig. 1B). The C-terminal antibody did not detect an amino-terminally epitope-tagged truncated form of GPR30 that lacked the C-terminal 24 amino acids (Fig. 1C), although it was still expressed in the endoplasmic reticulum. Both the C-terminally truncated form of GPR30 and endogenous GPR30 were expressed with the same localization pattern (Fig. 1D). Furthermore, this C-terminal antibody identified endogenous GPR30 in a number of cancer cell lines [breast cancers MCF7, SKBr3, and MDA-MB231; juxtaglomerular epithelioid granular (JEG) choriocarcinoma; and Hec50 uterine carcinoma] in a localization pattern identical to that observed for GPR30-GFP in transfected COS7 cells (Fig. 1E). MDA-MB-231 breast cancer cells exhibited somewhat lower expression of GPR30, whereas HTR8 normal extravillous trophoblasts, Ishikawa H cells (differentiated hormone-responsive uterine cells), and COS7 cells exhibited little to no expression. Whereas the aggressive cancer cell lines (JEG and Hec50) expressed high levels of GPR30, their associated normal cell lines (HTR8 and H, respectively) showed little to no expression of GPR30. JEG cells expressed a level of GPR30 at least 10 times as high as that of HTR8 cells (Fig. 1F), quantitatively confirming the results observed by confocal microscopy (Fig. 1E). Taken together, these results demonstrate that GPR30 is a 7-transmembrane receptor localized predominantly in the endoplasmic reticulum.

Intracellular localization has important implications for the function of a GPCR. Any natural ligand for the receptor must be membrane permeable and, furthermore, signal transduction molecules (e.g., G proteins and kinases) accessible to the receptor on internal membranes may be very different from those available to a receptor on the plasma membrane. To date, a cognate stoichiometric ligand for GPR30 has not been firmly established. Because GPR30-positive cells were responsive to estrogens, we determined whether GPR30 could directly bind to estrogen. We developed a family of fluorescent estrogens based on 17α-[4-aminomethyl-phenylethynyl]-estra-1, 3,5(10)-triene-3, 17β-diol, which contains a single reactive amine at the distal end of a linker that is attached to the 17α position of 17β-estradiol (19). This estrogen derivative was conjugated to the N-hydroxysuccinimidyl ester forms of various Alexa dyes (Fig. 2A). The resulting products, E2-Alexa 546 and E2-Alexa 633 were purified by reverse-phase high-performance liquid chromatography. To test the binding ability of these fluorescent estrogens, we expressed ERα-GFP in COS7 cells and stained the transfected cells with E2-Alexa 633. Unpermeabilized cells did not stain with E2-Alexa 633 (or 546) because of the impermeability of the charged Alexa dyes (Fig. 2B). However, transfected cells permeabilized with saponin demonstrated binding that coincided with the nuclear localization of ERα-GFP. Specificity of the binding to ERα was demonstrated by the ability of an excess of 17β-estradiol to compete for the binding of the E2-Alexa derivatives. Confocal fluorescence microscopy revealed that E2-Alexa 546 staining also colocalized nearly completely with GPR30-GFP expression in the endoplasmic reticulum and Golgi and that the binding could be competed by excess 17β-estradiol (Fig. 2C). No binding of the E2-Alexa 546 to the plasma membrane was observed, confirming the absence of GPR30 at the plasma membrane. We next stained the same panel of cell lines used in Fig. 1E with E2-Alexa 546 (Fig. 2D). In each cell type, the intensity and pattern of staining of the fluorescent estrogen was consistent with the level and pattern of GPR30 protein expression (Fig. 1E). To characterize the binding properties of the novel fluorescent estrogens further, we compared the level of E2-Alexa 633 binding to the expression level of ERα-GFP, ERβ-GFP, and GPR30-GFP in transfected COS7 cells, using β2-AR-GFP–expressing cells as a control. A direct linear correlation was observed, which supports direct binding of estrogen by GPR30 (Fig. 2E). Competition binding assays of E2-Alexa 633 binding with 17β-estradiol demonstrated a Ki for 17β-estradiol of approximately 6.6 nM for GPR30 (Fig. 2F). These results demonstrate that a classic GPCR superfamily member directly binds a sex steroid hormone and that GPR30 is an estrogen-binding receptor.

Fig. 2.

Binding of a fluorescent estrogen analog to GPR30. (A) Structure of Alexa 546-modified estrogen derivative. (B and C) Confocal images of (B) ERα-GFP–expressing or (C) GPR30-GFP–expressing cells stained with E2 Alexa 546 or 633 under nonpermeabilized or saponin-permeabilized conditions in the absence or presence of excess competing 17β-estradiol (1 μM). In (B), the cell membrane in the merged image is shown by virtue of coexpression of β2-AR-mRFP1. (D) Confocal images of the same eight cell lines shown in Fig. 1E but now stained with E2 Alexa 546 following permeabilization with saponin. All images were acquired with identical instrument settings at one sitting. (E) ERα, ERβ, GPR30 and β2-AR GFP chimerae were expressed in COS7 cells and stained with E2 Alexa 633. Various receptor expression levels were defined by the use of markers in FL1 (GFP) and analyzed for the corresponding level of E2 Alexa 633 binding (specific binding was determined in the presence of 1 μM 17β-estradiol). (F) GPR30-GFP was expressed in COS7 cells and stained with 2 nM E2 Alexa 633 in the presence of the indicated concentration of 17β-estradiol. GPR30-GFP–positive cells were selected by a marker in FL1 (GFP) and analyzed for specific E2 Alexa 633 binding (as above). Data are representative of three independent experiments [(B) to (D)] or represent the mean ± SEM from three experiments [(E) and (F)].

To determine whether GPR30 mediates rapid nongenomic signaling, we expressed GPR30-GFP and ERα-GFP in COS7 cells and then loaded the cells with the Ca2+ indicator Indo1-AM. Whereas 17β-estradiol stimulation of mock-transfected cells yielded only a negligible mobilization of Ca2+ (Fig. 3B), stimulation of either ERα-GFP–expressing or GPR30-GFP–expressing cells resulted in a mobilization of intracellular calcium similar in magnitude to the response evoked by stimulation of purinergic receptors with adenosine triphosphate (Fig. 3, A and B). In contrast, an inactive isomer of estrogen, 17α-estradiol, did not induce a calcium response via either receptor when used at concentrations 1000 times as high as that of 17β-estradiol. Estrogen-mediated mobilization of intracellular Ca2+ was observed at 17β-estradiol concentrations below 0.1 nM, with an EC50 value of approximately 0.5 nM (Fig. 3C). These results suggest that the selectivity of GPR30 toward estrogen is similar to that of the ER. This is not unexpected, because the two hydroxyl groups of estrogen represent natural sites for ligand recognition and binding (20).

Fig. 3.

Estrogen-mediated calcium mobilization and PI3K activation by GPR30 and ERα. (A and B) Intracellular calcium mobilization of Indo1-AM–loaded ERα-GFP–expressing or GPR30-GFP–expressing COS7 cells. The scales, though relative, are identical to each other in (A) and (B). In (A) and (B), 1 nM 17β-estradiol was used to stimulate the cells treated with the indicated inhibitors. In (B), COS7 refers to mock-transfected cells stimulated with 1 nM 17β-estradiol. (C) GPR30-GFP–expressing COS7 cells loaded with Indo1-AM were stimulated with the indicated concentration of 17β-estradiol. Mock GFP-expressing COS7 cells were used as a control (COS7). (D and E) Activation of PI3K was assessed by translocation of a PH-mRFP reporter in ERα-GFP–expressing or GPR30-GFP–expressing COS7 cells. In some merged images, the location of the nucleus is shown by TO-PRO-3 staining (blue). In all panels, the ligand concentrations used were 17β-estradiol (E2, 1 nM), 17α-estradiol (1 μM), 4-hydrotamoxifen (1 μM), EGF (100 ng/mL), and 1 nM 17β-estradiol for cells pretreated with inhibitors. Data are representative of three independent experiments [(A), (B), (D), and (E)] or the mean ± SEM from three experiments (C).

As numerous estrogen-mediated rapid signaling events are sensitive to pertussis toxin, which implies the involvement of Gi/o proteins, we determined whether ERα and GPR30 stimulated calcium mobilization in this pathway. Whereas pertussis toxin completely blocked ERα-mediated calcium mobilization (Fig. 3A), inhibition of GPR30-mediated calcium mobilization was only partial (∼50%) (Fig. 3B). Furthermore, because activation of mitogen-activated protein kinase signaling by GPR30 was suggested to occur through epidermal growth factor receptor (EGFR) transactivation (16) and GPCR-mediated calcium mobilization is mediated by phospholipase C (PLC)–dependent inositol 1,4,5-trisphosphate production, we assessed whether either of these pathways is involved in estrogen-stimulated calcium mobilization by ERα and GPR30. GPR30-mediated calcium mobilization was completely blocked by the EGFR inhibitor AG1478 but not by the PLC inhibitor U73122, whereas the ERα response was completely blocked by the PLC inhibitor but not substantially by the EGFR inhibitor. These results suggest that ER- and GPR30-initiated calcium mobilization are mediated by divergent signaling pathways.

Estrogen stimulates numerous intracellular kinase signaling pathways (8), including the phosphoinositide 3-kinase (PI3K)–Akt pathway, which leads to proliferative signaling. Such estrogen-dependent signaling has been observed even in apparently ER-negative cell lines (21). Activation of PI3K leads to the membrane-localized accumulation of phosphatidylinositol 3,4,5-trisphosphate (PIP3), which in turn acts to recruit Akt to the membrane through its pleckstrin homology (PH) domain. Molecular fusion of the isolated PH domain of Akt to fluorescent proteins has been widely used to assess PIP3 accumulation and localization (22). To investigate estrogen-mediated activation of PI3K, we transfected COS7 cells with a PH domain fused to monomeric red fluorescent protein1 (PH-mRFP1) and either ERα-GFP or GPR30-GFP. In unstimulated cells, the PH-mRFP1 reporter was localized primarily throughout the cytoplasm, with slightly higher concentrations found in the nucleus of some cells (Fig. 3, D and E). Upon stimulation with 17β-estradiol, ER-GFP–expressing cells displayed an almost complete translocation of PH-mRFP1 to the nucleus (Fig. 3D). A similar response was seen with GPR30-GFP–expressing cells (Fig. 3E), although the translocation on average was not as complete. Inhibition of PI3K by LY294002 prevented the translocation of the PH-mRFP1 to the nucleus by both receptors. Although tamoxifen [a selective estrogen receptor modulator (SERM)] induced calcium mobilization in both ERα-GFP–transfected and GPR30-GFP–transfected cells (23), it induced relocalization of the PH-mRFP1 only in GPR30-GFP–expressing cells. Furthermore, although PH-mRFP1 translocated to the nucleus, it also translocated to membranes of the endoplasmic reticulum containing GPR30-GFP. This result suggests that tamoxifen acts differentially on GPR30- and ER-mediated nuclear signal transduction and that tamoxifen activates GPR30 in a spatially different manner from estrogen. Accumulation of PIP3 by either receptor was not inhibited by pertussis toxin, indicating that ER-mediated calcium mobilization and PI3K activation occur by distinct mechanisms. PIP3 generation by GPR30, but not by ER, requires EGFR activation, indicating that EGFR transactivation is similarly involved for both ER and GPR30 with respect to calcium mobilization and PI3K activation.

We examined whether activation of PI3K by intracellular GPR30 also occurs in a breast cancer cell line (SKBr3) expressing endogenous GPR30 but lacking ERα and ERβ. PH-mRFP1-transfected SKBr3 cells responded to 17β-estradiol and tamoxifen stimulation with nuclear accumulation of PIP3, as demonstrated by translocation of the PH-mRFP1 (Fig. 4A). Unstimulated cells displayed PH-mRFP1 in the plasma membrane and cytoplasm. SKBr3 cells exhibit constitutive activation of ErbB2 signaling (24), which presumably induces PI3K-dependent plasma membrane localization of PH-mRFP1 in unstimulated cells. This is supported by the cytosolic localization of the PH-mRFP1 reporter upon treatment of unstimulated cells with PI3K and EGFR inhibitors (23). These same inhibitors also prevented the nuclear accumulation of PH-mRFP1 in estrogen-stimulated cells, which demonstrates that the estrogen-mediated signaling in SKBr3 cells occurs through the EGFR and PI3K, as demonstrated for GPR30 in COS7 cells. To demonstrate that GPR30 alone was responsible for the PI3K signaling observed in SKBr3 cells, we employed antisense-mediated depletion of endogenous GPR30. Expression of GPR30 antisense resulted in a marked reduction in GPR30 expression, whereas expression of β2-AR antisense did not affect GPR30 expression (Fig. 4B). Expression of GPR30 antisense also reduced binding of E2-Alexa 546 to transfected SKBr3 cells (Fig. 4, C and D). To determine whether GPR30 depletion would abrogate estrogen-mediated PI3K activation, we coexpressed antisense vectors (either GPR30 or β2-AR), PH-mRFP1 (to assess PIP3 synthesis and localization), and GFP (to mark transfected cells). 17β-estradiol stimulation of GPR30 antisense-expressing cells did not cause PH-mRFP1 translocation (Fig. 4E), which demonstrates that the GPR30 represents the sole estrogen-binding and functionally responsive estrogen receptor in SKBr3 cells.

Fig. 4.

Estrogen-mediated activation of PI3K in SKBr3 breast cancer cells. (A) SKBr3 cells expressing PH-mRFP1 were stimulated with buffer alone (unstim), 1 nM 17β-estradiol (17β E2), 1 μM tamoxifen, 1 μM 17α-estradiol (17α E2), or 1 nM 17β-estradiol for cells pretreated with LY294002 (E2 + LY) or AG1478 (E2 + AG). (B and C) SKBr3 cells were cotransfected with either a GPR30 or β2-AR antisense construct [receptor open reading frame cloned in the reverse orientation in pcDNA3.1/Hygro(–)] and actin-GFP at a molar ratio of 5:1. Following 3 days growth, cells were permeabilized and stained with either antibody to C-terminal GPR30 (B) or E2-Alexa 546 (C). (D) GPR30 protein expression levels and ligand-binding capacity of GPR30 antisense-expressing cells (GPR30 AS) and control β2-AR antisense-expressing cells (β2AR AS), prepared as in (B) and (C), were quantitated by flow cytometry. In each sample, transfected cells were gated on FL1 by virtue of actin-GFP expression and compared with untransfected (nongreen) cells, providing direct internal controls. (E) SKBr3 cells were cotransfected with GPR30 or β2-AR antisense constructs as in (B) and (C), along with PH-mRFP1 and GFP (to mark transfected cells), to assess PI3K activation and PIP3 accumulation. Only cells depleted of GPR30 fail to respond to estrogen stimulation with translocation of the PH-mRFP1 to the nucleus (GPR30 antisense + E2 versus β2-AR + E2). Data are representative of three independent experiments [(A) to (C) and (E)] or the mean ± SEM from three experiments (D).

Mechanisms of estrogen-mediated cellular activation have become increasingly complex. Rapid nongenomic ER-mediated signaling has been proposed to occur through distinct cellular localization of the classical ER, through classical heterotrimeric G proteins, and through many of the effectors traditionally associated with growth factors and GPCRs. Estrogen is one of the small number of membrane-permeable physiological ligands, consistent with the lack of requirement for GPR30 to be expressed on the cell surface. In some cell types, GPR30 represents the sole estrogen-responsive receptor. In direct comparison with ER-mediated signaling, GPR30-mediated signaling occurs via a distinct signal transduction pathway, and the effects of estrogen through this receptor may likely be a consequence of its intracellular localization.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1106943/DC1

Materials and Methods

References and Notes

References and Notes

View Abstract

Navigate This Article